From: Paul Eggert Date: Tue, 1 Jan 2019 01:57:29 +0000 (-0800) Subject: Merge from origin/emacs-26 X-Git-Tag: archive/raspbian/1%29.2+1-2+rpi1^2~5^2~19^2~4279 X-Git-Url: https://dgit.raspbian.org/%22http://www.example.com/cgi/%22/%22http:/www.example.com/cgi/%22?a=commitdiff_plain;h=ba809612c06bcabd05dcbaae79da39dcb0925992;p=emacs.git Merge from origin/emacs-26 2fcf2df Fix copyright years by hand 26bed8b Update copyright year to 2019 2814292 Fix value of default frame height. (Bug#33921) --- ba809612c06bcabd05dcbaae79da39dcb0925992 diff --cc build-aux/config.guess index 28bec27b307,5dacc291236..3ea87fbf0b7 --- a/build-aux/config.guess +++ b/build-aux/config.guess @@@ -1,8 -1,8 +1,8 @@@ #! /bin/sh # Attempt to guess a canonical system name. - # Copyright 1992-2018 Free Software Foundation, Inc. + # Copyright 1992-2019 Free Software Foundation, Inc. -timestamp='2018-01-01' +timestamp='2018-12-07' # This file is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by diff --cc build-aux/config.sub index a5081b8e0d1,cb90ee49099..872516579f2 --- a/build-aux/config.sub +++ b/build-aux/config.sub @@@ -1,8 -1,8 +1,8 @@@ #! /bin/sh # Configuration validation subroutine script. - # Copyright 1992-2018 Free Software Foundation, Inc. + # Copyright 1992-2019 Free Software Foundation, Inc. -timestamp='2018-01-01' +timestamp='2018-12-16' # This file is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by diff --cc build-aux/update-copyright index f2fc97e368f,19d881316af..d80549ea8dd --- a/build-aux/update-copyright +++ b/build-aux/update-copyright @@@ -3,9 -3,9 +3,9 @@@ eval '(exit $?0)' && eval 'exec perl -w if 0; # Update an FSF copyright year list to include the current year. -my $VERSION = '2018-01-04.14:48'; # UTC +my $VERSION = '2018-03-07.03:47'; # UTC - # Copyright (C) 2009-2018 Free Software Foundation, Inc. + # Copyright (C) 2009-2019 Free Software Foundation, Inc. # # This program is free software: you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by diff --cc doc/misc/tramp.texi index 49c684f5bc9,bfb6d748f24..7587059f393 --- a/doc/misc/tramp.texi +++ b/doc/misc/tramp.texi @@@ -12,8 -12,18 +12,8 @@@ @c This is *so* much nicer :) @footnotestyle end -@c Macro for formatting a file name according to the respective -@c syntax. Macro arguments should not have any leading or trailing -@c whitespace. Not very elegant, but I don't know it better. - -@macro trampfn {method, userhost, localname} -@value{prefix}@c -\method\@value{postfixhop}@c -\userhost\@value{postfix}\localname\ -@end macro - @copying - Copyright @copyright{} 1999--2018 Free Software Foundation, Inc. + Copyright @copyright{} 1999--2019 Free Software Foundation, Inc. @quotation Permission is granted to copy, distribute and/or modify this document diff --cc doc/misc/trampver.texi index af7b6786a6a,aecbbe261c8..f5cc2bae179 --- a/doc/misc/trampver.texi +++ b/doc/misc/trampver.texi @@@ -2,15 -2,15 +2,15 @@@ @c texi/trampver.texi. Generated from trampver.texi.in by configure. @c This is part of the Emacs manual. - @c Copyright (C) 2003-2018 Free Software Foundation, Inc. + @c Copyright (C) 2003-2019 Free Software Foundation, Inc. @c See file doclicense.texi for copying conditions. -@c In the Tramp GIT, the version number is auto-frobbed from -@c configure.ac, so you should edit that file and run -@c "autoconf && ./configure" to change the version number. -@set trampver 2.3.4.26.2 +@c In the Tramp GIT, the version number is auto-frobbed from tramp.el, +@c and the bug report address is auto-frobbed from configure.ac. +@set trampver 2.4.1 +@set tramp-bug-report-address tramp-devel@@gnu.org -@c Other flags from configuration +@c Other flags from configuration. @set instprefix /usr/local @set lispdir /usr/local/share/emacs/site-lisp @set infodir /usr/local/share/info diff --cc etc/HELLO index 5b14bbbd754,ae52e94b065..a56a73bc1d2 --- a/etc/HELLO +++ b/etc/HELLO @@@ -22,88 -17,68 +22,88 @@@ Non-ASCII examples LANGUAGE (NATIVE NAME) HELLO ---------------------- ----- -Amharic ($,1O M[MmN{(B) $,1M`MKM](B -Arabic ($,1-g.$-y-q-h.*.1-i(B) $,1-g.$-s.1.$-g.%(B $,1-y.$.*.#.%(B -Armenian ($,1+p+a+u+e, +e+v(B) $,1+2+a, ,'(B $,1+q+e+f(B -Bengali ($,17,7>6b727>(B) $,17(7.787M6u7>70(B -Braille $,2(3(1('('(5(B -Burmese ($,1H9H\H4HZH9HL(B) $,1H9H$HZHYH"Hmule-unicode-0100-24ffAmharic (አማርኛ) ሠላም +Arabic (العربيّة) السّلام عليكم +Armenian (հայերեն) Բարև ձեզ +Bengali (বাংলা) নমস্কার +mule-unicode-2500-33ffBraille ⠓⠑⠇⠇⠕ +mule-unicode-0100-24ffBurmese (မြန်မာ) မင်္ဂလာပါ +C printf ("Hello, world!\n"); +unicodeCherokee (ᏣᎳᎩ ᎦᏬᏂᎯᏍᏗ) ᎣᏏᏲ / ᏏᏲ +Comanche /kəˈmæntʃiː/ Haa marʉ́awe + +Cree (ᓀᐦᐃᔭᐍᐏᐣ) ᑕᓂᓯ / ᐙᒋᔮ + +latin-iso8859-2Czech (čeština) Dobrý den +latin-iso8859-1Danish (dansk) Hej / Goddag / Halløj Dutch (Nederlands) Hallo / Dag +unicodeEfik /ˈɛfɪk/ Mɔkɔm + Emacs emacs --no-splash -f view-hello-file -English /$(O+S,0!,D?$(O*y(Bl,0!$(O*h(B/ Hello -Esperanto Saluton (E,C6(Bo,C~(Ban,Cx(Bo ,Cf(Biu,C<(Ba,C}(Bde) -Estonian (eesti keel) Tere p,Ad(Bevast / Tere ,Au(Bhtust -Finnish (suomi) Hei / Hyv,Add(B p,Ad(Biv,Add(B -French (fran,Ag(Bais) Bonjour / Salut -Georgian ($,1JEJ0J@J7J5J4J:J8(B) $,1J2J0J;J0J@JOJ=J1J0(B -German (Deutsch) Guten Tag / Gr,A|_(B Gott -Greek (,Fekkgmij\(B) ,FCei\(B ,Fsar(B -Greek, ancient ($,1p1,Fkkgmij^(B) ,FO$,1pv,Fk](B ,Fte(B ,Fja$,1q6(B ,Fl]ca(B ,Fwa$,1r6,Fqe(B -Gujarati ($,19W:!9\9p9~9d: (B) $,19h9n9x:-9d:'(B -Hebrew ($,1-",q-(,y-*(B) ,Hylem(B -Hungarian (magyar) Sz,Bi(Bp j,Bs(B napot! -Hindi ($,15y55B5f6 (B) $,15h5n5x6-5d6'(B / $,15h5n5x6-5U5~5p(B $,16D(B -Italian (italiano) Ciao / Buon giorno -Javanese (Jawa) System.out.println("Sugeng siang!"); -Kannada ($,1>u?(?M?(?!(B) $,1?(?.?8?M>u?>?0(B -Khmer ($,1\7\V\?\V\!\r\8\b\:(B) $,1\'\f\:\V\4\?\]\:(B -Lao ((1>RJRERG(B) (1JP:R-4U(B / (1"mcKib*!4U(B -Malayalam ($,1@N@R@O@^@S@"(B) $,1@H@N@X@m@5@^@P@"(B -Maltese (il-Malti) Bon,Cu(Bu / Sa,C11(Ba -Mathematics $,1x (B p $,1x((B world $,1s"(B hello p $,2!a(B -Mongolian (,L\^]S^[(B ,Lem[(B) ,LAPY](B ,LQPY]P(B ,Lcc(B? -Norwegian (norsk) Hei / God dag -Oriya ($,1:s;\;?:f(B) $,1;6;A;#;?;,;G(B -Polish (j,Bj(Bzyk polski) Dzie,Bq(B dobry! / Cze,B6f(B! -Russian (,L`caaZXY(B) ,L7T`P$(O+Z,LRabRcYbU(B! -Sinhala ($,1B#B2ABB$A}(B) $,1AFAzB4AvB=B AqB*(B -Slovak (sloven,Bh(Bina) Dobr,A}(B de,Br(B -Slovenian (sloven,B9h(Bina) Pozdravljeni! -Spanish (espa,Aq(Bol) ,A!(BHola! -Swedish (svenska) Hej / Goddag / Hall,Ae(B -Tamil ($,1&=r>!=W>!(B) $,1=h=n=x>-=U=~=p=B(B -Thai (,T@RIRd7B(B) ,TJGQJ4U$CQ:(B / ,TJGQJ4U$hP(B -Tibetan ($(7"7"]"2!;"G#!"2!;(B) $(7"7"!#C!;"E"S"G!;"7"2"[!;"D"["#"G!>(B -Tigrigna ($,1NUP-MmN{(B) $,1MpMKM[NU(B -Turkish (T,A|(Brk,Ag(Be) Merhaba -Ukrainian (,LcZ`Pw]alZP(B) ,L2vbPn(B -Vietnamese (ti,1*(Bng Vi,1.(Bt) Ch,A`(Bo b,1U(Bn - -Japanese ($BF|K\8l(B) $B$3$s$K$A$O(B / (I:]FAJ(B -Chinese ($AVPND(B,$AFUM(;0(B,$A::So(B) $ADc:C(B -Cantonese ($(0GnM$(B,$(0N]0*Hd(B) $(0*/=((B, $(0+$)p(B -Korean ($(CGQ1[(B) $(C>H3gGO<H3gGO=J4O1n(B - - + +Emoji 👋 +English ipa/ˈɪŋɡlɪʃ/ Hello +latin-iso8859-3Esperanto Saluton (Eĥoŝanĝo ĉiuĵaŭde) +latin-iso8859-15Estonian (eesti keel) Tere päevast / Tere õhtust +latin-iso8859-1Finnish (suomi) Hei / Hyvää päivää +French (français) Bonjour / Salut +mule-unicode-0100-24ffGeorgian (ქართველი) გამარჯობა +latin-iso8859-1German (Deutsch) Guten Tag / Grüß Gott +greek-iso8859-7Greek (ελληνικά) Γειά σας +mule-unicode-0100-24ffGreek, ancient (ἑλληνική) Οὖλέ τε καὶ μέγα χαῖρε +Gujarati (ગુજરાતી) નમસ્તે +hebrew-iso8859-8Hebrew (עברית) שלום +latin-iso8859-2Hungarian (magyar) Szép jó napot! +mule-unicode-0100-24ffHindi (हिंदी) नमस्ते / नमस्कार । +unicodeInuktitut (ᐃᓄᒃᑎᑐᑦ) ᐊᐃ + +latin-iso8859-1Italian (italiano) Ciao / Buon giorno +Javanese (Jawa) System.out.println("Sugeng siang!"); +mule-unicode-0100-24ffKannada (ಕನ್ನಡ) ನಮಸ್ಕಾರ +Khmer (ភាសាខ្មែរ) ជំរាបសួរ +laoLao (ພາສາລາວ) ສະບາຍດີ / ຂໍໃຫ້ໂຊກດີ +mule-unicode-0100-24ffMalayalam (മലയാളം) നമസ്കാരം +unicodeMaldivian (ދިވެހި) އައްސަލާމު ޢަލައިކުމް / ކިހިނެހް؟ + +latin-iso8859-3Maltese (il-Malti) Bonġu / Saħħa +unicodeMathematics ∀ p ∈ world • hello p □ +cyrillic-iso8859-5Mongolian (монгол хэл) Сайн байна уу? +latin-iso8859-1Norwegian (norsk) Hei / God dag +mule-unicode-0100-24ffOriya (ଓଡ଼ିଆ) ଶୁଣିବେ +latin-iso8859-2Polish (język polski) Dzień dobry! / Cześć! +cyrillic-iso8859-5Russian (русский) Здра́вствуйте! +mule-unicode-0100-24ffSinhala (සිංහල) ආයුබෝවන් +latin-iso8859-2Slovak (slovenčina) Dobrý deň +Slovenian (slovenščina) Pozdravljeni! +Spanish (espalatin-iso8859-1ñol) ¡Hola! +Swedish (svenska) Hej / Goddag / Hallå +mule-unicode-0100-24ffTamil (தமிழ்) வணக்கம் +Telugu (తెలుగు) నమస్కారం +thai-tis620Thai (ภาษาไทย) สวัสดีครับ / สวัสดีค่ะ +tibetanTibetan (བོད་སྐད་) བཀྲ་ཤིས་བདེ་ལེགས༎ +mule-unicode-0100-24ffTigrigna (ትግርኛ) ሰላማት +latin-iso8859-9Turkish (Türkçe) Merhaba +cyrillic-iso8859-5Ukrainian (українська) Вітаю +vietnamese-viscii-lowerVietnamese (tiếng vietnamese-viscii-upperVvietnamese-viscii-loweriệt) vietnamese-viscii-upperChào bạn + + + +japanese-jisx0208Japanese (日本語) こんにちは katakana-jisx0201/ ｺﾝﾆﾁﾊ +chinese-gb2312Chinese (中文,普通话,汉语) 你好 +chinese-big5-1Cantonese (粵語,廣東話) 早晨, 你好 +korean-ksc5601Korean (한글) 안녕하세요 / 안녕하십니까 + + + +unicode + - Copyright (C) 2001-2018 Free Software Foundation, Inc. + Copyright (C) 2001-2019 Free Software Foundation, Inc. + This file is part of GNU Emacs. + GNU Emacs is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or diff --cc etc/NEWS.26 index 55bdaf11172,00000000000..9ee1a4f284e mode 100644,000000..100644 --- a/etc/NEWS.26 +++ b/etc/NEWS.26 @@@ -1,1946 -1,0 +1,1946 @@@ +GNU Emacs NEWS -- history of user-visible changes. + - Copyright (C) 2016-2018 Free Software Foundation, Inc. ++Copyright (C) 2016-2019 Free Software Foundation, Inc. +See the end of the file for license conditions. + +Please send Emacs bug reports to 'bug-gnu-emacs@gnu.org'. +If possible, use 'M-x report-emacs-bug'. + +This file is about changes in Emacs version 26. + +See file HISTORY for a list of GNU Emacs versions and release dates. +See files NEWS.25, NEWS.24, ..., NEWS.18, and NEWS.1-17 for changes +in older Emacs versions. + +You can narrow news to a specific version by calling 'view-emacs-news' +with a prefix argument or by typing 'C-u C-h C-n'. + + +* Installation Changes in Emacs 26.2 + +--- +** Building Emacs with the '--with-xwidgets' option now requires WebKit2. +To build Emacs with xwidgets support, you will need to install the +webkit2gtk-4.0 package; version 2.12 or later is required. +(This change was actually made in Emacs 26.1, but was not called out +in its NEWS.) + ++++ +** Installing Emacs now installs the emacs-module.h file. +The emacs-module.h file is now installed in the system-wide include +directory as part of the Emacs installation. This allows to build +Emacs modules outside of the Emacs source tree. + + +* Startup Changes in Emacs 26.2 + + +* Changes in Emacs 26.2 + +--- +** Emacs is now compliant with the latest version 11.0 of the Unicode Standard. + +--- +** New variable 'xft-ignore-color-fonts'. +Default t means don't try to load color fonts when using Xft, as they +often cause crashes. Set it to nil if you really need those fonts. + + +* Editing Changes in Emacs 26.2 + + +* Changes in Specialized Modes and Packages in Emacs 26.2 + +** Dired + ++++ +*** The 'Z' command on a directory name compresses all of its files. +It produces a compressed '.tar.gz' archive with all the files in the +directory and all of its subdirectories. For symmetry, 'Z' on a +'.tar.gz' or a '.tgz' archive extracts all the archived files into the +current directory; thus, typing 'Z' on a '.tar.gz' archive created by +a previous 'Z' command will extract the archived files into a +directory whose name is the archive name sans the '.tar.gz' extension. +(This change was actually made in Emacs 25.1 but was only +partially called out in its NEWS; 'tgz' handling was added in 26.1.) + +** Ibuffer + +--- +*** New toggle 'ibuffer-do-toggle-lock', bound to 'L'. + +** Imenu + +--- +*** The value for 'imenu-auto-rescan-maxout' has been increased to 600000. + +** Gnus + +--- +*** Mailutils movemail will now be used if found at runtime. +The default value of 'mail-source-movemail-program' is now "movemail". +This ensures that the movemail program from GNU Mailutils will be used +if found in 'exec-path', even if it was not found at build time. To +use a different program, customize 'mail-source-movemail-program' to the +absolute file name of the desired executable. + +** Shadowfile + +--- +*** shadowfile.el has been rewritten to support Tramp file names. + +** Shell mode + +--- +*** Shell mode buffers now have 'scroll-conservatively' set to 101. +This is so as to better emulate the scrolling behavior of a text +terminal when new output is added to the screen buffer. To get back +the previous behavior, reset 'scroll-conservatively' to zero (or any +other value you like) in a function and add it to 'shell-mode-hook'. +(This change was actually made in Emacs 26.1, but was not called out +in its NEWS.) + +** VC + +--- +*** VC support for Mercurial was improved. +Emacs now avoids invoking 'hg' as much as possible, for faster operation. +(This and the following changes were actually made in Emacs 26.1, but +were not called out in its NEWS.) + +--- +**** New vc-hg options. +The new option 'vc-hg-parse-hg-data-structures' controls whether vc-hg +will try parsing the Mercurial data structures directly instead of +running 'hg'; it defaults to t (set to nil if you want the pre-26.1 +behavior). +The new option 'vc-hg-symbolic-revision-styles' controls how versions +in a Mercurial repository are presented symbolically on the mode line. +The new option 'vc-hg-use-file-version-for-mode-line-version' controls +whether the version shown on the mode line is that of the visited file +or of the repository working copy. + +--- +**** Display of Mercurial revisions in the mode line has changed. +Previously, the mode line displayed the local number (1, 2, 3, ...) of +the revision. Starting with Emacs 26.1, the default has changed, and +it now shows the global revision number, in the form of its changeset +hash value. To get back the previous behavior, customize the new +option 'vc-hg-symbolic-revision-styles' to the value '("{rev}")'. + + +* New Modes and Packages in Emacs 26.2 + + +* Incompatible Lisp Changes in Emacs 26.2 + +--- +** shadowfile config files have changed their syntax. +Existing files "~/.emacs.d/shadows" and "~/.emacs.d/shadow_todo" must +be removed prior using the changed 'shadow-*' commands. + ++++ +** 'thread-alive-p' has been renamed to 'thread-live-p'. +The old name is an alias of the new name. Future Emacs version will +obsolete it. + +--- +** 'while-no-input' does not return due to input from subprocesses. +Input that arrived from subprocesses while some code executed inside +the 'while-no-input' form injected an internal buffer-switch event +that counted as input and would cause 'while-no-input' to return, +perhaps prematurely. These buffer-switch events are now by default +ignored by 'while-no-input'; if you need to get the old behavior, +remove 'buffer-switch' from the list of events in +'while-no-input-ignore-events'. + + +* Lisp Changes in Emacs 26.2 + ++++ +** The new function 'read-answer' accepts either long or short answers +depending on the new customizable variable 'read-answer-short'. + ++++ +** New function 'assoc-delete-all'. +Like 'assq-delete-all', but uses 'equal' for comparison. + +--- +** The function 'thing-at-point' behaves as before Emacs 26.1. +The behavior of 'thing-at-point' when called with argument 'list' has +changed in Emacs 26.1, in that it didn't consider text inside comments +and strings as a potential list. This change is now reverted, and +'thing-at-point' behaves like it did before Emacs 26.1. + +--- +** To cater to use cases where comments and strings are to be ignored +when looking for a list, the function 'list-at-point' now takes an +optional argument to do so. + + +* Changes in Emacs 26.2 on Non-Free Operating Systems + +** macOS features can now be detected at run-time as well as at +build-time. See nextstep/INSTALL for details. +(This change was actually made in Emacs 26.1, but was undocumented and +not called out in its NEWS.) + + +* Installation Changes in Emacs 26.1 + +** By default libgnutls is now required when building Emacs. +Use 'configure --with-gnutls=no' to build even when GnuTLS is missing. + +** GnuTLS version 2.12.2 or later is now required, instead of merely +version 2.6.6 or later. + +** The new option 'configure --with-mailutils' causes Emacs to rely on +GNU Mailutils to retrieve email. It is recommended, and is the +default if GNU Mailutils is installed. When '--with-mailutils' is not +in effect, the Emacs build procedure by default continues to build and +install a limited 'movemail' substitute that retrieves POP3 email only +via insecure channels. To avoid this problem, use either +'--with-mailutils' or '--without-pop' when configuring; '--without-pop' +is the default on platforms other than native MS-Windows. + +** The new option 'configure --enable-gcc-warnings=warn-only' causes +GCC to issue warnings without stopping the build. This behavior is +now the default in developer builds. As before, use +'--disable-gcc-warnings' to suppress GCC's warnings, and +'--enable-gcc-warnings' to stop the build if GCC issues warnings. + +** When GCC warnings are enabled, '--enable-check-lisp-object-type' is +now enabled by default when configuring. + +** The Emacs server now has socket-launching support. +This allows socket based activation, where an external process like +systemd can invoke the Emacs server process upon a socket connection +event and hand the socket over to Emacs. Emacs uses this socket to +service emacsclient commands. This new functionality can be disabled +with the configure option '--disable-libsystemd'. + +** A systemd user unit file is provided. +Use it in the standard way: 'systemctl --user enable emacs'. (If your +Emacs is installed in a non-standard location, you may need to copy +the emacs.service file to eg ~/.config/systemd/user/) + +** New configure option '--disable-build-details' attempts to build an +Emacs that is more likely to be reproducible; that is, if you build +and install Emacs twice, the second Emacs is a copy of the first. +Deterministic builds omit the build date from the output of the +'emacs-version' and 'erc-cmd-SV' functions, and the leave the +following variables nil: 'emacs-build-system', 'emacs-build-time', +'erc-emacs-build-time'. + +** Emacs can now be built with support for Little CMS. +If the lcms2 library is installed, Emacs will enable features built on +top of that library. The new configure option '--without-lcms2' can +be used to build without lcms2 support even if it is installed. Emacs +linked to Little CMS exposes color management functions in Lisp: the +color metrics 'lcms-cie-de2000' and 'lcms-cam02-ucs', as well as +functions for conversion to and from CIE CAM02 and CAM02-UCS. + +** The configure option '--with-gameuser' now defaults to 'no', +as this appears to be the most common configuration in practice. +When it is 'no', the shared game directory and the auxiliary program +update-game-score are no longer needed and are not installed. + +** Emacs no longer works on IRIX. We expect that Emacs users are not +affected by this, as SGI stopped supporting IRIX in December 2013. + + +* Startup Changes in Emacs 26.1 + +** New option '--fg-daemon'. This is the same as '--daemon', except +it runs in the foreground and does not fork. This is intended for +modern init systems such as systemd, which manage many of the traditional +aspects of daemon behavior themselves. '--bg-daemon' is now an alias +for '--daemon'. + +** New option '--module-assertions'. +When given this option, Emacs will perform expensive correctness +checks when dealing with dynamic modules. This is intended for module +authors that wish to verify that their module conforms to the module +requirements. The option makes Emacs abort if a module-related +assertion triggers. + +** Emacs now supports 24-bit colors on capable text terminals. +Terminal is automatically initialized to use 24-bit colors if the +required capabilities are found in terminfo. See the FAQ node +"(efaq) Colors on a TTY" for more information. + +** Emacs now obeys the X resource "scrollBar" at startup. +The effect is similar to that of "toolBar" resource on the tool bar. + + +* Changes in Emacs 26.1 + +** Option 'buffer-offer-save' can be set to new value, 'always'. +When set to 'always', the command 'save-some-buffers' will always +offer this buffer for saving. + +** Security vulnerability related to Enriched Text mode is removed. + +*** Enriched Text mode does not evaluate Lisp in 'display' properties. +This feature allows saving 'display' properties as part of text. +Emacs 'display' properties support evaluation of arbitrary Lisp forms +as part of processing the property for display, so displaying Enriched +Text could be vulnerable to executing arbitrary malicious Lisp code +included in the text (e.g., sent as part of an email message). +Therefore, execution of arbitrary Lisp forms in 'display' properties +decoded by Enriched Text mode is now disabled by default. Customize +the new option 'enriched-allow-eval-in-display-props' to a non-nil +value to allow Lisp evaluation in decoded 'display' properties. + +This vulnerability was introduced in Emacs 21.1. To work around that +in Emacs versions before 25.3, append the following to your ~/.emacs +init file: + + (eval-after-load "enriched" + '(defun enriched-decode-display-prop (start end &optional param) + (list start end))) + +** Functions in 'write-contents-functions' can fully short-circuit the +'save-buffer' process. Previously, saving a buffer that was not +visiting a file would always prompt for a file name. Now it only does +so if 'write-contents-functions' is nil (or all its functions return +nil). + +** New variable 'executable-prefix-env' for inserting magic signatures. +This variable affects the format of the interpreter magic number +inserted by 'executable-set-magic'. If non-nil, the magic number now +takes the form "#!/usr/bin/env interpreter", otherwise the value +determined by 'executable-prefix', which is by default +"#!/path/to/interpreter". By default, 'executable-prefix-env' is nil, +so the default behavior is not changed. + +** The variable 'emacs-version' no longer includes the build number. +This is now stored separately in a new variable, 'emacs-build-number'. + +** Emacs now provides a limited form of concurrency with Lisp threads. +Concurrency in Emacs Lisp is "mostly cooperative", meaning that +Emacs will only switch execution between threads at well-defined +times: when Emacs waits for input, during blocking operations related +to threads (such as mutex locking), or when the current thread +explicitly yields. Global variables are shared among all threads, but +a 'let' binding is thread-local. Each thread also has its own current +buffer and its own match data. + +See the chapter "(elisp) Threads" in the ELisp manual for full +documentation of these facilities. + +** The new user variable 'electric-quote-chars' provides a list +of curved quotes for 'electric-quote-mode', allowing user to choose +the types of quotes to be used. + +** The new user option 'electric-quote-context-sensitive' makes +'electric-quote-mode' context sensitive. If it is non-nil, you can +type an ASCII apostrophe to insert an opening or closing quote, +depending on context. Emacs will replace the apostrophe by an opening +quote character at the beginning of the buffer, the beginning of a +line, after a whitespace character, and after an opening parenthesis; +and it will replace the apostrophe by a closing quote character in all +other cases. + +** The new variable 'electric-quote-inhibit-functions' controls when +to disable electric quoting based on context. Major modes can add +functions to this list; Emacs will temporarily disable +'electric-quote-mode' whenever any of the functions returns non-nil. +This can be used by major modes that derive from 'text-mode' but allow +inline code segments, such as 'markdown-mode'. + +** The new user variable 'dired-omit-case-fold' allows the user to +customize the case-sensitivity of dired-omit-mode. It defaults to +the same sensitivity as that of the filesystem for the corresponding +dired buffer. + +** Emacs now uses double buffering to reduce flicker when editing and +resizing graphical Emacs frames on the X Window System. This support +requires the DOUBLE-BUFFER extension, which major X servers have +supported for many years. If your system has this extension, but an +Emacs built with double buffering misbehaves on some displays you use, +you can disable the feature by adding + + '(inhibit-double-buffering . t) + +to default-frame-alist. Or inject this parameter into the selected +frame by evaluating this form: + + (modify-frame-parameters nil '((inhibit-double-buffering . t))) + +** The customization group 'wp', whose label was "text", is now +deprecated. Use the new group 'text', which inherits from 'wp', +instead. + +** The new function 'call-shell-region' executes a command in an +inferior shell with the buffer region as input. + +** The new user option 'shell-command-dont-erase-buffer' controls +if the output buffer is erased between shell commands; if non-nil, +the output buffer is not erased; this variable also controls where +to set the point in the output buffer: beginning of the output, +end of the buffer or save the point. +When 'shell-command-dont-erase-buffer' is nil, the default value, +the behavior of 'shell-command', 'shell-command-on-region' and +'async-shell-command' is as usual. + +** The new user option 'async-shell-command-display-buffer' controls +whether the output buffer of an asynchronous command is shown +immediately, or only when there is output. + +** New user option 'mouse-select-region-move-to-beginning'. +This option controls the position of point when double-clicking +mouse-1 on the end of a parenthetical grouping or string-delimiter: +the default value nil keeps point at the end of the region, setting it +to non-nil moves point to the beginning of the region. + +** New user option 'mouse-drag-and-drop-region'. +This option allows you to drag the entire region of text to another +place or another buffer. Its behavior is customizable via the new +options 'mouse-drag-and-drop-region-cut-when-buffers-differ', +'mouse-drag-and-drop-region-show-tooltip', and +'mouse-drag-and-drop-region-show-cursor'. + +** The new user option 'confirm-kill-processes' allows the user to +skip a confirmation prompt for killing subprocesses when exiting +Emacs. When set to t (the default), Emacs will prompt for +confirmation before killing subprocesses on exit, which is the same +behavior as before. + +** 'find-library-name' will now fall back on looking at 'load-history' +to try to locate libraries that have been loaded with an explicit path +outside 'load-path'. + +** Faces in 'minibuffer-prompt-properties' no longer overwrite properties +in the text in functions like 'read-from-minibuffer', but instead are +added to the end of the face list. This allows users to say things +like '(read-from-minibuffer (propertize "Enter something: " 'face 'bold))'. + +** The new variable 'extended-command-suggest-shorter' has been added +to control whether to suggest shorter 'M-x' commands or not. + +** icomplete now respects 'completion-ignored-extensions'. + +** Non-breaking hyphens are now displayed with the 'nobreak-hyphen' +face instead of the 'escape-glyph' face. + +** Approximations to quotes are now displayed with the new 'homoglyph' +face instead of the 'escape-glyph' face. + +** New face 'header-line-highlight'. +This face is the header-line analogue of 'mode-line-highlight'; it +should be the preferred mouse-face for mouse-sensitive elements in the +header line. + +** 'C-x h' ('mark-whole-buffer') will now avoid marking the prompt +part of minibuffers. + +** 'fill-paragraph' no longer marks the buffer as changed unless it +actually changed something. + +** The locale language name 'ca' is now mapped to the language +environment 'Catalan', which has been added. + +** 'align-regexp' has a separate history for its interactive argument. +'align-regexp' no longer shares its history with all other +history-less functions that use 'read-string'. + +** The networking code has been reworked so that it's more +asynchronous than it was (when specifying :nowait t in +'make-network-process'). How asynchronous it is varies based on the +capabilities of the system, but on a typical GNU/Linux system the DNS +resolution, the connection, and (for TLS streams) the TLS negotiation +are all done without blocking the main Emacs thread. To get +asynchronous TLS, the TLS boot parameters have to be passed in (see +the manual for details). + +Certain process oriented functions (like 'process-datagram-address') +will block until socket setup has been performed. The recommended way +to deal with asynchronous sockets is to avoid interacting with them +until they have changed status to "run". This is most easily done +from a process sentinel. + +** 'make-network-process' and 'open-network-stream' sometimes allowed +:service to be an integer string (e.g., :service "993") and sometimes +required an integer (e.g., :service 993). This difference has been +eliminated, and integer strings work everywhere. + +** It is possible to disable attempted recovery on fatal signals. +Two new variables support disabling attempts to recover from stack +overflow and to avoid automatic auto-save when Emacs is delivered a +fatal signal. 'attempt-stack-overflow-recovery', if set to nil, +will disable attempts to recover from C stack overflows; Emacs will +then crash as with any other fatal signal. +'attempt-orderly-shutdown-on-fatal-signal', if set to nil, will +disable attempts to auto-save the session and shut down in an orderly +fashion when Emacs receives a fatal signal; instead, Emacs will +terminate immediately. Both variables are non-nil by default. +These variables are for users who would like to avoid the small +probability of data corruption due to techniques Emacs uses to recover +in these situations. + +** File local and directory local variables are now initialized each +time the major mode is set, not just when the file is first visited. +These local variables will thus not vanish on setting a major mode. + +** A second dir-local file (.dir-locals-2.el) is now accepted. +See the doc string of 'dir-locals-file' for more information. + +** Connection-local variables can be used to specify local variables +with a value depending on the connected remote server. For details, +see the node "(elisp) Connection Local Variables" in the ELisp manual. + +** International domain names (IDNA) are now encoded via the new +puny.el library, so that one can visit Web sites with non-ASCII URLs. + +** The new 'list-timers' command lists all active timers in a buffer, +where you can cancel them with the 'c' command. + +** 'switch-to-buffer-preserve-window-point' now defaults to t. +Applications that call 'switch-to-buffer' and want to show the buffer at +the position of its point should use 'pop-to-buffer-same-window' in lieu +of 'switch-to-buffer'. + +** The new variable 'debugger-stack-frame-as-list' allows displaying +all call stack frames in a Lisp backtrace buffer as lists. Both +debug.el and edebug.el have been updated to heed to this variable. + +** Values in call stack frames are now displayed using 'cl-prin1'. +The old behavior of using 'prin1' can be restored by customizing the +new option 'debugger-print-function'. + +** NUL bytes in text copied to the system clipboard are now replaced with "\0". + +** The new variable 'x-ctrl-keysym' has been added to the existing +roster of X keysyms. It can be used in combination with another +variable of this kind to swap modifiers in Emacs. + +** New input methods: 'cyrillic-tuvan', 'polish-prefix', 'uzbek-cyrillic'. + +** The 'dutch' input method no longer attempts to support Turkish too. +Also, it no longer converts 'IJ' and 'ij' to the compatibility +characters U+0132 LATIN CAPITAL LIGATURE IJ and U+0133 LATIN SMALL +LIGATURE IJ. + +** File name quoting by adding the prefix "/:" is now possible for the +local part of a remote file name. Thus, if you have a directory named +"/~" on the remote host "foo", you can prevent it from being +substituted by a home directory by writing it as "/foo:/:/~/file". + +** The new variable 'maximum-scroll-margin' allows having effective +settings of 'scroll-margin' up to half the window size, instead of +always restricting the margin to a quarter of the window. + +** Emacs can scroll horizontally using mouse, touchpad, and trackbar. +You can enable this by customizing 'mouse-wheel-tilt-scroll'. If you +want to reverse the direction of the scroll, customize +'mouse-wheel-flip-direction'. + +** The default GnuTLS priority string now includes %DUMBFW. +This is to avoid bad behavior in some firewalls, which causes the +connection to be closed by the remote host. + +** Emacsclient changes + +*** Emacsclient has a new option '-u' / '--suppress-output'. +This option suppresses display of return values from the server +process. + +*** Emacsclient has a new option '-T' / '--tramp'. +This helps with using a local Emacs session as the server for a remote +emacsclient. With appropriate setup, one can now set the EDITOR +environment variable on a remote machine to emacsclient, and +use the local Emacs to edit remote files via Tramp. See the node +"(emacs) emacsclient Options" in the user manual for the details. + +*** Emacsclient now accepts command-line options in ALTERNATE_EDITOR +and '--alternate-editor'. For example, ALTERNATE_EDITOR="emacs -Q -nw". +Arguments may be quoted "like this", so that for example an absolute +path containing a space may be specified; quote escaping is not +supported. + +** New user option 'dig-program-options' and extended functionality +for DNS-querying functions 'nslookup-host', 'dns-lookup-host', +and 'run-dig'. Each function now accepts an optional name server +argument interactively (with a prefix argument) and non-interactively. + +** 'describe-key-briefly' now ignores mouse movement events. + +** The new variable 'eval-expression-print-maximum-character' prevents +large integers from being displayed as characters by 'M-:' and similar +commands. + +** Two new commands for finding the source code of Emacs Lisp +libraries: 'find-library-other-window' and 'find-library-other-frame'. + +** The new variable 'display-raw-bytes-as-hex' allows you to change +the display of raw bytes from octal to hex. + +** You can now provide explicit field numbers in format specifiers. +For example, '(format "%2$s %1$s %2$s" "X" "Y")' produces "Y X Y". + +** Emacs now supports optional display of line numbers in the buffer. +This is similar to what 'linum-mode' provides, but much faster and +doesn't usurp the display margin for the line numbers. Customize the +buffer-local variable 'display-line-numbers' to activate this optional +display. Alternatively, you can use the 'display-line-numbers-mode' +minor mode or the global 'global-display-line-numbers-mode'. When +using these modes, customize 'display-line-numbers-type' with the same +value as you would use with 'display-line-numbers'. + +Line numbers are not displayed at all in minibuffer windows and in +tooltips, as they are not useful there. + +Lisp programs can disable line-number display for a particular screen +line by putting the 'display-line-numbers-disable' text property or +overlay property on the first character of that screen line. This is +intended for add-on packages that need a finer control of the display. + +Lisp programs that need to know how much screen estate is used up for +line-number display in a window can use the new function +'line-number-display-width'. + +'linum-mode' and all similar packages are henceforth becoming obsolete. +Users and developers are encouraged to switch to this new feature +instead. + +** The new user option 'arabic-shaper-ZWNJ-handling' controls how to +handle ZWNJ in Arabic text rendering. + + +* Editing Changes in Emacs 26.1 + +** New variable 'column-number-indicator-zero-based'. +Traditionally, in Column Number mode, the displayed column number +counts from zero starting at the left margin of the window. This +behavior is now controlled by 'column-number-indicator-zero-based'. +If you would prefer for the displayed column number to count from one, +you may set this variable to nil. (Behind the scenes, there is now a +new mode line construct, '%C', which operates exactly as '%c' does +except that it counts from one.) + +** New single-line horizontal scrolling mode. +The 'auto-hscroll-mode' variable can now have a new special value, +'current-line', which causes only the line where the cursor is +displayed to be horizontally scrolled when lines are truncated on +display and point moves outside the left or right window margin. + +** New mode line constructs '%o' and '%q', and user option +'mode-line-percent-position'. '%o' displays the "degree of travel" of +the window through the buffer. Unlike the default '%p', this +percentage approaches 100% as the window approaches the end of the +buffer. '%q' displays the percentage offsets of both the start and +the end of the window, e.g. "5-17%". The new option +'mode-line-percent-position' makes it easier to switch between '%p', +'%P', and these new constructs. + +** Two new user options 'list-matching-lines-jump-to-current-line' and +'list-matching-lines-current-line-face' to show the current line +highlighted in *Occur* buffer. + +** The 'occur' command can now operate on the region. + +** New bindings for 'query-replace-map'. +'undo', undo the last replacement; bound to 'u'. +'undo-all', undo all replacements; bound to 'U'. + +** 'delete-trailing-whitespace' deletes whitespace after form feed. +In modes where form feed was treated as a whitespace character, +'delete-trailing-whitespace' would keep lines containing it unchanged. +It now deletes whitespace after the last form feed thus behaving the +same as in modes where the character is not whitespace. + +** Emacs no longer prompts about editing a changed file when the file's +content is unchanged. Instead of only checking the modification time, +Emacs now also checks the file's actual content before prompting the user. + +** Various casing improvements. + +*** 'upcase', 'upcase-region' et al. convert title case characters +(such as Dz) into their upper case form (such as DZ). + +*** 'capitalize', 'upcase-initials' et al. make use of title-case forms +of initial characters (correctly producing for example Džungla instead +of incorrect DŽungla). + +*** Characters which turn into multiple ones when cased are correctly handled. +For example, fi ligature is converted to FI when upper cased. + +*** Greek small sigma is correctly handled when at the end of the word. +Strings such as ΌΣΟΣ are now correctly converted to Όσος when +capitalized instead of incorrect Όσοσ (compare lowercase sigma at the +end of the word). + +** Emacs can now auto-save buffers to visited files in a more robust +manner via the new mode 'auto-save-visited-mode'. Unlike +'auto-save-visited-file-name', this mode uses the normal saving +procedure and therefore obeys saving hooks. +'auto-save-visited-file-name' is now obsolete. + +** New behavior of 'mark-defun'. +Prefix argument selects that many (or that many more) defuns. +Negative prefix arg flips the direction of selection. Also, +'mark-defun' between defuns correctly selects N following defuns (or +-N previous for negative arguments). Finally, comments preceding the +defun are selected unless they are separated from the defun by a blank +line. + +** New command 'replace-buffer-contents'. +This command replaces the contents of the accessible portion of the +current buffer with the contents of the accessible portion of a +different buffer while keeping point, mark, markers, and text +properties as intact as possible. + +** New commands 'apropos-local-variable' and 'apropos-local-value'. +These are buffer-local versions of 'apropos-variable' and +'apropos-value', respectively. They show buffer-local variables whose +names and values, respectively, match a given pattern. + +** More user control of reordering bidirectional text for display. +The two new variables, 'bidi-paragraph-start-re' and +'bidi-paragraph-separate-re', allow customization of what exactly are +paragraphs, for the purposes of bidirectional display. + +** New variable 'x-wait-for-event-timeout'. +This controls how long Emacs will wait for updates to the graphical +state to take effect (making a frame visible, for example). + + +* Changes in Specialized Modes and Packages in Emacs 26.1 + +** Emacs 26.1 comes with Org v9.1.6. +See the file ORG-NEWS for user-visible changes in Org. + +** New function 'cl-generic-p'. + +** Dired + +*** You can answer 'all' in 'dired-do-delete' to delete recursively all +remaining directories without more prompts. + +*** Dired supports wildcards in the directory part of the file names. + +*** You can now use '`?`' in 'dired-do-shell-command'. +It gets replaced by the current file name, like ' ? '. + +*** A new option 'dired-always-read-filesystem' defaulting to nil. +If non-nil, buffers visiting files are reverted before they are +searched; for instance, in 'dired-mark-files-containing-regexp' a +non-nil value of this option means the file is revisited in a +temporary buffer; this temporary buffer is the actual buffer searched: +the original buffer visiting the file is not modified. + +*** Users can now customize mouse clicks in Dired in a more flexible way. +The new command 'dired-mouse-find-file' can be bound to a mouse click +and used to visit files/directories in Dired in the selected window. +The new command 'dired-mouse-find-file-other-frame' similarly visits +files/directories in another frame. You can write your own commands +that invoke 'dired-mouse-find-file' with non-default optional +arguments, to tailor the effects of mouse clicks on file names in +Dired buffers. + +*** In wdired, when editing files to contain slash characters, +the resulting directories are automatically created. Whether to do +this is controlled by the 'wdired-create-parent-directories' variable. + +*** 'W' is now bound to 'browse-url-of-dired-file', and is useful for +viewing HTML files and the like. + +*** New variable 'dired-clean-confirm-killing-deleted-buffers' +controls whether Dired asks to kill buffers visiting deleted files and +directories. The default is t, so Dired asks for confirmation, to +keep previous behavior. + +** html2text is now marked obsolete. + +** smerge-refine-regions can refine regions in separate buffers. + +** Info menu and index completion uses substring completion by default. +This can be customized via the 'info-menu' category in +'completion-category-overrides'. + +** The ancestor buffer is shown by default in 3-way merges. +A new option 'ediff-show-ancestor' and a new toggle +'ediff-toggle-show-ancestor'. + +** TeX: Add luatex and xetex as alternatives to pdftex. + +** Electric-Buffer-menu + +*** Key 'U' is bound to 'Buffer-menu-unmark-all' and key 'M-DEL' is +bound to 'Buffer-menu-unmark-all-buffers'. + +** hideshow mode got four key bindings that are analogous to outline +mode bindings: 'C-c @ C-a', 'C-c @ C-t', 'C-c @ C-d', and 'C-c @ C-e'. + +** bs + +*** Two new commands 'bs-unmark-all', bound to 'U', and +'bs-unmark-previous', bound to . + +** Buffer-menu + +*** Two new commands 'Buffer-menu-unmark-all', bound to 'U' and +'Buffer-menu-unmark-all-buffers', bound to 'M-DEL'. + +** Checkdoc + +*** 'checkdoc-arguments-in-order-flag' now defaults to nil. + +** Gnus + +*** The ~/.newsrc file will now only be saved if the native select +method is an NNTP select method. + +*** A new command for sorting articles by readedness marks has been +added: 'C-c C-s C-m C-m'. + +*** In 'message-citation-line-format' the '%Z' format is now the time +zone name instead of the numeric form. The '%z' format continues to +be the numeric form. The new behavior is compatible with +'format-time-string'. + +** Ibuffer + +*** New command 'ibuffer-jump'. + +*** New filter commands 'ibuffer-filter-by-basename', +'ibuffer-filter-by-file-extension', 'ibuffer-filter-by-directory', +'ibuffer-filter-by-starred-name', 'ibuffer-filter-by-modified' +and 'ibuffer-filter-by-visiting-file'; bound respectively +to '/b', '/.', '//', '/*', '/i' and '/v'. + +*** Two new commands 'ibuffer-filter-chosen-by-completion' +and 'ibuffer-and-filter', the second bound to '/&'. + +*** The commands 'ibuffer-pop-filter', 'ibuffer-pop-filter-group', +'ibuffer-or-filter' and 'ibuffer-filter-disable' have the alternative +bindings '/', '/S-', '/|' and '/DEL', respectively. + +*** The data format specifying filters has been extended to allow +explicit logical 'and', and a more flexible form for logical 'not'. +See 'ibuffer-filtering-qualifiers' doc string for full details. + +*** A new command 'ibuffer-copy-buffername-as-kill'; bound +to 'B'. + +*** New command 'ibuffer-change-marks'; bound to '* c'. + +*** A new command 'ibuffer-mark-by-locked' to mark +all locked buffers; bound to '% L'. + +*** A new option 'ibuffer-locked-char' to indicate +locked buffers; Ibuffer shows a new column displaying +'ibuffer-locked-char' for locked buffers. + +*** A new command 'ibuffer-unmark-all-marks' to unmark +all buffers without asking confirmation; bound to +'U'; 'ibuffer-do-replace-regexp' bound to 'r'. + +*** A new command 'ibuffer-mark-by-content-regexp' to mark buffers +whose content matches a regexp; bound to '% g'. + +*** Two new options 'ibuffer-never-search-content-name' and +'ibuffer-never-search-content-mode' used by +'ibuffer-mark-by-content-regexp'. + +** Browse-URL + +*** Support for opening links to man pages in Man or WoMan mode. + +** Comint + +*** New user option 'comint-move-point-for-matching-input' to control +where to place point after 'C-c M-r' and 'C-c M-s'. + +*** New user option 'comint-terminfo-terminal'. +This option allows control of the value of the TERM environment +variable Emacs puts into the environment of the Comint mode and its +derivatives, such as Shell mode and Compilation Shell minor-mode. The +default is "dumb", for compatibility with previous behavior. + +** Compilation mode + +*** Messages from CMake are now recognized. + +*** The number of errors, warnings, and informational messages is now +displayed in the mode line. These are updated as compilation +proceeds. + +** Grep + +*** Grep commands will now use GNU grep's '--null' option if +available, which allows distinguishing the filename from contents if +they contain colons. This can be controlled by the new custom option +'grep-use-null-filename-separator'. + +*** The grep/rgrep/lgrep functions will now ask about saving files +before running. This is controlled by the 'grep-save-buffers' +variable. + +** Edebug + +*** Edebug can be prevented from pausing 1 second after reaching a +breakpoint (e.g. with "f" and "o") by customizing the new option +'edebug-sit-on-break'. + +*** New customizable option 'edebug-max-depth'. +This allows you to enlarge the maximum recursion depth when +instrumenting code. + +*** 'edebug-prin1-to-string' now aliases 'cl-prin1-to-string'. +This means edebug output is affected by variables 'cl-print-readably' +and 'cl-print-compiled'. To completely restore the previous printing +behavior, use + + (fset 'edebug-prin1-to-string #'prin1-to-string) + +** Eshell + +*** 'eshell-input-filter's value is now a named function +'eshell-input-filter-default', and has a new custom option +'eshell-input-filter-initial-space' to ignore adding commands prefixed +with blank space to eshell history. + +** EUDC + +*** Backward compatibility support for BBDB versions less than 3 +(i.e., BBDB 2.x) is deprecated and will likely be removed in the next +major release of Emacs. Users of BBDB 2.x should plan to upgrade to +BBDB 3.x. + +** eww + +*** New 'M-RET' command for opening a link at point in a new eww buffer. + +*** A new 's' command for switching to another eww buffer via the minibuffer. + +*** The 'o' command ('shr-save-contents') has moved to 'O' to avoid collision +with the 'o' command from 'image-map'. + +*** A new command 'C' ('eww-toggle-colors') can be used to toggle +whether to use the HTML-specified colors or not. The user can also +customize the 'shr-use-colors' variable. + +*** Images that are being loaded are now marked with gray +"placeholder" images of the size specified by the HTML. They are then +replaced by the real images asynchronously, which will also now +respect width/height HTML specs (unless they specify widths/heights +bigger than the current window). + +*** The 'w' command on links is now 'shr-maybe-probe-and-copy-url'. +'shr-copy-url' now only copies the url at point; users who wish to +avoid accidentally accessing remote links may rebind 'w' and 'u' in +'eww-link-keymap' to it. + +** Ido + +*** The commands 'find-alternate-file-other-window', +'dired-other-window', 'dired-other-frame', and +'display-buffer-other-window' are now remapped to Ido equivalents if +Ido mode is active. + +** Images + +*** Images are automatically scaled before displaying based on the +'image-scaling-factor' variable (if Emacs supports scaling the images +in question). + +*** It's now possible to specify aspect-ratio preserving combinations +of :width/:max-height and :height/:max-width keywords. In either +case, the "max" keywords win. (Previously some combinations would, +depending on the aspect ratio of the image, just be ignored and in +other instances this would lead to the aspect ratio not being +preserved.) + +*** Images inserted with 'insert-image' and related functions get a +keymap put into the text properties (or overlays) that span the +image. This keymap binds keystrokes for manipulating size and +rotation, as well as saving the image to a file. These commands are +also available in 'image-mode'. + +*** A new library for creating and manipulating SVG images has been +added. See the "(elisp) SVG Images" section in the ELisp reference +manual for details. + +*** New setf-able function to access and set image parameters is +provided: 'image-property'. + +*** New commands 'image-scroll-left' and 'image-scroll-right' +for 'image-mode' that complement 'image-scroll-up' and +'image-scroll-down': they have the same prefix arg behavior and stop +at image boundaries. + +** Image-Dired + +*** Now provides a minor mode 'image-dired-minor-mode' which replaces +the function 'image-dired-setup-dired-keybindings'. + +*** Thumbnail generation is now asynchronous. +The number of concurrent processes is limited by the variable +'image-dired-queue-active-limit'. + +*** 'image-dired-thumbnail-storage' has a new option 'standard-large' +for generating 256x256 thumbnails according to the Thumbnail Managing +Standard. + +*** Inherits movement keys from 'image-mode' for viewing full images. +This includes the usual char, line, and page movement commands. + +*** All the -options types have been changed to argument lists +instead of shell command strings. This change affects +'image-dired-cmd-create-thumbnail-options', +'image-dired-cmd-create-temp-image-options', +'image-dired-cmd-rotate-thumbnail-options', +'image-dired-cmd-rotate-original-options', +'image-dired-cmd-write-exif-data-options', +'image-dired-cmd-read-exif-data-options', and introduces +'image-dired-cmd-pngnq-options', 'image-dired-cmd-pngcrush-options', +'image-dired-cmd-create-standard-thumbnail-options'. + +*** Recognizes more tools by default, including pngnq-s9 and OptiPNG. + +*** 'find-file' and related commands now work on thumbnails and +displayed images, providing a default argument of the original file name +via an addition to 'file-name-at-point-functions'. + +** The default 'Info-default-directory-list' no longer checks some obsolete +directory suffixes (gnu, gnu/lib, gnu/lib/emacs, emacs, lib, lib/emacs) +when searching for info directories. + +** The commands that add ChangeLog entries now prefer a VCS root directory +for the ChangeLog file, if none already exists. Customize +'change-log-directory-files' to nil for the old behavior. + +** Support for non-string values of 'time-stamp-format' has been removed. + +** Message + +*** 'message-use-idna' now defaults to t (because Emacs comes with +built-in IDNA support now). + +*** When sending HTML messages with embedded images, and you have +exiftool installed, and you rotate images with EXIF data (i.e., +JPEGs), the rotational information will be inserted into the outgoing +image in the message. (The original image will not have its +orientation affected.) + +*** The 'message-valid-fqdn-regexp' variable has been removed, since +there are now top-level domains added all the time. Message will no +longer warn about sending emails to top-level domains it hasn't heard +about. + +*** 'message-beginning-of-line' (bound to 'C-a') understands folded headers. +In 'visual-line-mode' it will look for the true beginning of a header +while in non-'visual-line-mode' it will move the point to the indented +header's value. + +** Package + +*** The new variable 'package-gnupghome-dir' has been added to control +where the GnuPG home directory (used for signature verification) is +located and whether GnuPG's option '--homedir' is used or not. + +*** Deleting a package no longer respects 'delete-by-moving-to-trash'. + +** Python + +*** The new variable 'python-indent-def-block-scale' has been added. +It controls the depth of indentation of arguments inside multi-line +function signatures. + +** Tramp + +*** The method part of remote file names is mandatory now. +A valid remote file name starts with "/method:host:" or +"/method:user@host:". + +*** The new pseudo method "-" is a marker for the default method. +"/-::" is the shortest remote file name then. + +*** The command 'tramp-change-syntax' allows you to choose an +alternative remote file name syntax. + +*** New connection method "sg", which supports editing files under a +different group ID. + +*** New connection method "doas" for OpenBSD hosts. + +*** New connection method "gdrive", which allows access to Google +Drive onsite repositories. + +*** Gateway methods in Tramp have been removed. +Instead, the Tramp manual documents how to configure ssh and PuTTY +accordingly. + +*** Setting the "ENV" environment variable in +'tramp-remote-process-environment' enables reading of shell +initialization files. + +*** Tramp is able now to send SIGINT to remote asynchronous processes. + +*** Variable 'tramp-completion-mode' is obsoleted. + +** 'auto-revert-use-notify' is set back to t in 'global-auto-revert-mode'. + +** JS mode + +*** JS mode now sets 'comment-multi-line' to t. + +*** New variable 'js-indent-align-list-continuation', when set to nil, +will not align continuations of bracketed lists, but will indent them +by the fixed width 'js-indent-level'. + +** CSS mode + +*** Support for completing attribute values, at-rules, bang-rules, +HTML tags, classes and IDs using the 'completion-at-point' command. +Completion candidates for HTML classes and IDs are retrieved from open +HTML mode buffers. + +*** CSS mode now binds 'C-h S' to a function that will show +information about a CSS construct (an at-rule, property, pseudo-class, +pseudo-element, with the default being guessed from context). By +default the information is looked up on the Mozilla Developer Network, +but this can be customized using 'css-lookup-url-format'. + +*** CSS colors are fontified using the color they represent as the +background. For instance, #ff0000 would be fontified with a red +background. + +** Emacs now supports character name escape sequences in character and +string literals. The syntax variants '\N{character name}' and +'\N{U+code}' are supported. + +** Prog mode has some support for multi-mode indentation. +This allows better indentation support in modes that support multiple +programming languages in the same buffer, like literate programming +environments or ANTLR programs with embedded Python code. + +A major mode can provide indentation context for a sub-mode. To +support this, modes should use 'prog-first-column' instead of a +literal zero and avoid calling 'widen' in their indentation functions. +See the node "(elisp) Mode-Specific Indent" in the ELisp manual for +more details. + +** ERC + +*** New variable 'erc-default-port-tls' used to connect to TLS IRC +servers. + +** URL + +*** The new function 'url-cookie-delete-cookie' can be used to +programmatically delete all cookies, or cookies from a specific +domain. + +*** 'url-retrieve-synchronously' now takes an optional timeout parameter. + +*** The URL package now supports HTTPS over proxies supporting CONNECT. + +*** 'url-user-agent' now defaults to 'default', and the User-Agent +string is computed dynamically based on 'url-privacy-level'. + +** VC and related modes + +*** 'vc-dir-mode' now binds 'vc-log-outgoing' to 'O'; and has various +branch-related commands on a keymap bound to 'B'. + +*** 'vc-region-history' is now bound to 'C-x v h', replacing the older +'vc-insert-headers' binding. + +*** New user option 'vc-git-print-log-follow' to follow renames in Git logs +for a single file. + +** CC mode + +*** Opening a .h file will turn C or C++ mode depending on language used. +This is done with the help of the 'c-or-c++-mode' function, which +analyzes buffer contents to infer whether it's a C or C++ source file. + +** New option 'cpp-message-min-time-interval' to allow user control +of progress messages in cpp.el. + +** New DNS mode command 'dns-mode-ipv6-to-nibbles' to convert IPv6 addresses +to a format suitable for reverse lookup zone files. + +** Ispell + +*** Enchant is now supported as a spell-checker. +Enchant is a meta-spell-checker that uses providers such as Hunspell +to do the actual checking. With it, users can use spell-checkers not +directly supported by Emacs, such as Voikko, Hspell and AppleSpell, +more easily share personal word-lists with other programs, and +configure different spelling-checkers for different languages. +(Version 2.1.0 or later of Enchant is required.) + +** Flymake + +*** Flymake has been completely redesigned. +Flymake now annotates arbitrary buffer regions, not just lines. It +supports arbitrary diagnostic types, not just errors and warnings (see +variable 'flymake-diagnostic-types-alist'). + +It also supports multiple simultaneous backends, meaning that you can +check your buffer from different perspectives (see variable +'flymake-diagnostic-functions'). Backends for Emacs Lisp mode are +provided. + +The old Flymake behavior is preserved in the so-called "legacy +backend", which has been updated to benefit from the new UI features. + +** Term + +*** 'term-char-mode' now makes its buffer read-only. +The buffer is made read-only to prevent changes from being made by +anything other than the process filter; and movements of point away +from the process mark are counter-acted so that the cursor is in the +correct position after each command. This is needed to avoid states +which are inconsistent with the state of the terminal understood by +the inferior process. + +New user options 'term-char-mode-buffer-read-only' and +'term-char-mode-point-at-process-mark' control these behaviors, and +are non-nil by default. Customize these options to nil if you want +the previous behavior. + +** Xref + +*** When an *xref* buffer is needed, 'TAB' quits and jumps to an xref. +A new command 'xref-quit-and-goto-xref', bound to 'TAB' in *xref* +buffers, quits the window before jumping to the destination. In many +situations, the intended window configuration is restored, just as if +the *xref* buffer hadn't been necessary in the first place. + + +* New Modes and Packages in Emacs 26.1 + +** New Elisp data-structure library 'radix-tree'. + +** New library 'xdg' with utilities for some XDG standards and specs. + +** HTML + +*** A new submode of 'html-mode', 'mhtml-mode', is now the default +mode for *.html files. This mode handles indentation, +fontification, and commenting for embedded JavaScript and CSS. + +** New mode 'conf-toml-mode' is a sub-mode of 'conf-mode', specialized +for editing TOML files. + +** New mode 'conf-desktop-mode' is a sub-mode of 'conf-unix-mode', +specialized for editing freedesktop.org desktop entries. + +** New minor mode 'pixel-scroll-mode' provides smooth pixel-level scrolling. + +** New major mode 'less-css-mode' (a minor variant of 'css-mode') for +editing Less files. + +** New package 'auth-source-pass' integrates 'auth-source' with the +password manager password-store (http://passwordstore.org). + + +* Incompatible Lisp Changes in Emacs 26.1 + +** 'password-data' is now a hash-table so that 'password-read' can use +any object for the 'key' argument. + +** Command 'dired-mark-extension' now automatically prepends a '.' to the +extension when not present. The new command 'dired-mark-suffix' behaves +similarly but it doesn't prepend a '.'. + +** Certain cond/pcase/cl-case forms are now compiled using a faster jump +table implementation. This uses a new bytecode op 'switch', which +isn't compatible with previous Emacs versions. This functionality can +be disabled by setting 'byte-compile-cond-use-jump-table' to nil. + +** If 'comment-auto-fill-only-comments' is non-nil, 'auto-fill-function' +is now called only if either no comment syntax is defined for the +current buffer or the self-insertion takes place within a comment. + +** The alist 'ucs-names' is now a hash table. + +** 'if-let' and 'when-let' now support binding lists as defined by the +SRFI-2 (Scheme Request for Implementation 2). + +** 'C-up', 'C-down', 'C-left' and 'C-right' are now defined in term +mode to send the same escape sequences that xterm does. This makes +things like 'forward-word' in readline work. + +** Customizable variable 'query-replace-from-to-separator' +now doesn't propertize the string value of the separator. +Instead, text properties are added by 'query-replace-read-from'. +Additionally, the new nil value restores pre-24.5 behavior +of not providing replacement pairs via the history. + +** Some obsolete functions, variables, and faces have been removed: + +*** 'make-variable-frame-local'. Variables cannot be frame-local any more. + +*** From subr.el: 'window-dot', 'set-window-dot', 'read-input', +'show-buffer', 'eval-current-buffer', 'string-to-int'. + +*** 'icomplete-prospects-length'. + +*** All the default-FOO variables that hold the default value of the +FOO variable. Use 'default-value' and 'setq-default' to access and +change FOO, respectively. The exhaustive list of removed variables is: +'default-mode-line-format', 'default-header-line-format', +'default-line-spacing', 'default-abbrev-mode', 'default-ctl-arrow', +'default-truncate-lines', 'default-left-margin', 'default-tab-width', +'default-case-fold-search', 'default-left-margin-width', +'default-right-margin-width', 'default-left-fringe-width', +'default-right-fringe-width', 'default-fringes-outside-margins', +'default-scroll-bar-width', 'default-vertical-scroll-bar', +'default-indicate-empty-lines', 'default-indicate-buffer-boundaries', +'default-fringe-indicator-alist', 'default-fringe-cursor-alist', +'default-scroll-up-aggressively', 'default-scroll-down-aggressively', +'default-fill-column', 'default-cursor-type', +'default-cursor-in-non-selected-windows', +'default-buffer-file-coding-system', 'default-major-mode', and +'default-enable-multibyte-characters'. + +*** Many variables obsoleted in 22.1 referring to face symbols. + +** The variable 'text-quoting-style' is now a customizable option. +It controls whether to and how to translate ASCII quotes in messages +and help output. Its possible values and their semantics remain +unchanged from Emacs 25. In particular, when this variable's value is +'grave', all quotes in formats are output as-is. + +** Functions like 'check-declare-file' and 'check-declare-directory' +now generate less chatter and more-compact diagnostics. The auxiliary +function 'check-declare-errmsg' has been removed. + +** The regular expression character class '[:blank:]' now matches +Unicode horizontal whitespace as defined in the Unicode Technical +Standard #18. If you only want to match space and tab, use '[ \t]' +instead. + +** 'min' and 'max' no longer round their results. +Formerly, they returned a floating-point value if any argument was +floating-point, which was sometimes numerically incorrect. For +example, on a 64-bit host (max 1e16 10000000000000001) now returns its +second argument instead of its first. + +** The variable 'old-style-backquotes' has been made internal and +renamed to 'lread--old-style-backquotes'. No user code should use +this variable. + +** 'default-file-name-coding-system' now defaults to a coding system +that does not process CRLF. For example, it defaults to 'utf-8-unix' +instead of to 'utf-8'. Before this change, Emacs would sometimes +mishandle file names containing these control characters. + +** 'file-attributes', 'file-symlink-p' and 'make-symbolic-link' no +longer quietly mutate the target of a local symbolic link, so that +Emacs can access and copy them reliably regardless of their contents. +The following changes are involved. + +*** 'file-attributes' and 'file-symlink-p' no longer prepend "/:" to +symbolic links whose targets begin with "/" and contain ":". For +example, if a symbolic link "x" has a target "/y:z:", '(file-symlink-p +"x")' now returns "/y:z:" rather than "/:/y:z:". + +*** 'make-symbolic-link' no longer looks for file name handlers of +target when creating a symbolic link. For example, +'(make-symbolic-link "/y:z:" "x")' now creates a symbolic link to +"/y:z:" instead of failing. + +*** 'make-symbolic-link' removes the remote part of a link target if +target and newname have the same remote part. For example, +'(make-symbolic-link "/x:y:a" "/x:y:b")' creates a link with the +literal string "a"; and '(make-symbolic-link "/x:y:a" "/x:z:b")' +creates a link with the literal string "/x:y:a" instead of failing. + +*** 'make-symbolic-link' now expands a link target with leading "~" +only when the optional third arg is an integer, as when invoked +interactively. For example, '(make-symbolic-link "~y" "x")' now +creates a link with target the literal string "~y"; to get the old +behavior, use '(make-symbolic-link (expand-file-name "~y") "x")'. To +avoid this expansion in interactive use, you can now prefix the link +target with "/:". For example, '(make-symbolic-link "/:~y" "x" 1)' +now creates a link to literal "~y". + +** 'file-truename' returns a quoted file name if the target of a +symbolic link has remote file name syntax. + +** Module functions are now implemented slightly differently; in +particular, the function 'internal--module-call' has been removed. +Code that depends on undocumented internals of the module system might +break. + +** The argument LOCKNAME of 'write-region' is propagated to file name +handlers now. + +** When built against recent versions of GTK+, Emacs always uses +gtk_window_move for moving frames and ignores the value of the +variable 'x-gtk-use-window-move'. The variable is now obsolete. + +** Several functions that create or rename files now treat their +destination argument specially only when it is a directory name, i.e., +when it ends in '/' on GNU and other POSIX-like systems. When the +destination argument D of one of these functions is an existing +directory and the intent is to act on an entry in that directory, D +should now be a directory name. For example, (rename-file "e" "f/") +renames to 'f/e'. Although this formerly happened sometimes even when +D was not a directory name, as in (rename-file "e" "f") where 'f' +happened to be a directory, the old behavior often contradicted the +documentation and had inherent races that led to security holes. A +call like (rename-file C D) that used the old, undocumented behavior +can be written as (rename-file C (file-name-as-directory D)), a +formulation portable to both older and newer versions of Emacs. +Affected functions include 'add-name-to-file', 'copy-directory', +'copy-file', 'format-write-file', 'gnus-copy-file', +'make-symbolic-link', 'rename-file', 'thumbs-rename-images', and +'write-file'. + +** The list returned by 'overlays-at' is now in decreasing priority order. +The documentation of this function always said the order should be +that of decreasing priority, if the 2nd argument of the function is +non-nil, but the code returned the list in the increasing order of +priority instead. Now the code does what the documentation says it +should do. + +** 'format' now avoids allocating a new string in more cases. +'format' was previously documented to return a newly-allocated string, +but this documentation was not correct, as (eq x (format x)) returned +t when x was the empty string. 'format' is no longer documented to +return a newly-allocated string, and the implementation now takes +advantage of the doc change to avoid making copies of strings in +common cases like (format "foo") and (format "%s" "foo"). + +** The function 'eldoc-message' now accepts a single argument. +Programs that called it with multiple arguments before should pass +them through 'format' first. Even that is discouraged: for ElDoc +support, you should set 'eldoc-documentation-function' instead of +calling 'eldoc-message' directly. + +** Using '&rest' or '&optional' incorrectly is now an error. +For example giving '&optional' without a following variable, or +passing '&optional' multiple times: + + (defun foo (&optional &rest x)) + (defun bar (&optional &optional x)) + +Previously, Emacs would just ignore the extra keyword, or give +incorrect results in certain cases. + +** The pinentry.el library has been removed. +That package (and the corresponding change in GnuPG and pinentry) +was intended to provide a way to input passphrase through Emacs with +GnuPG 2.0. However, the change to support that was only implemented +in GnuPG >= 2.1 and didn't get backported to GnuPG 2.0. And with +GnuPG 2.1 and later, pinentry.el is not needed at all. So the +library was useless, and we removed it. GnuPG 2.0 is no longer +supported by the upstream project. + +To adapt to the change, you may need to set 'epa-pinentry-mode' to the +symbol 'loopback'. Alternatively, leave 'epa-pinentry-mode' at its +default value of nil, and remove the 'allow-emacs-pinentry' setting +from your 'gpg-agent.conf' configuration file, usually found in the +'~/.gnupg' directory. + +Note that previously, it was said that passphrase input through +minibuffer would be much less secure than other graphical pinentry +programs. However, these days the difference is insignificant: the +'read-password' function sufficiently protects input from leakage to +message logs. Emacs still doesn't use secure memory to protect +passphrases, but it was also removed from other pinentry programs as +the attack is unrealistic on modern computer systems which don't +utilize swap memory usually. + +** The function 'display-buffer-in-major-side-window' no longer exists. +It has been renamed as internal function 'window--make-major-side-window', +however applications should instead call 'display-buffer-in-side-window' +(passing the SIDE and SLOT parameters as elements of ALIST). This approach +is backwards-compatible with versions of Emacs in which the old function +exists. See the node "Displaying Buffers in Side Windows" in the ELisp +manual for more details. + +* Lisp Changes in Emacs 26.1 + +** The function 'assoc' now takes an optional third argument TESTFN. +This argument, when non-nil, is used for comparison instead of +'equal'. + +** New optional argument TESTFN in 'alist-get', 'map-elt' and 'map-put'. +If non-nil, the argument specifies a function to use for comparison, +instead of, respectively, 'assq' and 'eql'. + +** New function 'seq-set-equal-p' to check if SEQUENCE1 and SEQUENCE2 +contain the same elements, regardless of the order. + +** The new function 'mapbacktrace' applies a function to all frames of +the current stack trace. + +** The new function 'file-name-case-insensitive-p' tests whether a +given file is on a case-insensitive filesystem. + +** Several accessors for the value returned by 'file-attributes' +have been added. They are: 'file-attribute-type', +'file-attribute-link-number', 'file-attribute-user-id', +'file-attribute-group-id', 'file-attribute-access-time', +'file-attribute-modification-time', +'file-attribute-status-change-time', 'file-attribute-size', +'file-attribute-modes', 'file-attribute-inode-number', +'file-attribute-device-number' and 'file-attribute-collect'. + +** The new function 'buffer-hash' computes a fast, non-consing hash of +a buffer's contents. + +** 'interrupt-process' now consults the list 'interrupt-process-functions', +to determine which function has to be called in order to deliver the +SIGINT signal. This allows Tramp to send the SIGINT signal to remote +asynchronous processes. The hitherto existing implementation has been +moved to 'internal-default-interrupt-process'. + +** The new function 'read-multiple-choice' prompts for multiple-choice +questions, with a handy way to display help texts. + +** 'comment-indent-function' values may now return a cons to specify a +range of indentation. + +** New optional argument TEXT in 'make-temp-file'. + +** New function 'define-symbol-prop'. + +** New function 'secure-hash-algorithms' to list the algorithms that +'secure-hash' supports. +See the node "(elisp) Checksum/Hash" in the ELisp manual for details. + +** Emacs now exposes the GnuTLS cryptographic API with the functions +'gnutls-macs' and 'gnutls-hash-mac'; 'gnutls-digests' and +'gnutls-hash-digest'; 'gnutls-ciphers' and 'gnutls-symmetric-encrypt' +and 'gnutls-symmetric-decrypt'. +See the node "(elisp) GnuTLS Cryptography" in the ELisp manual for details. + +** The function 'gnutls-available-p' now returns a list of capabilities +supported by the GnuTLS library used by Emacs. + +** Emacs now supports records for user-defined types, via the new +functions 'make-record', 'record', and 'recordp'. Records are now +used internally to represent cl-defstruct and defclass instances, for +example. + +If your program defines new record types, you should use +package-naming conventions for naming those types. This is so any +potential conflicts with other types are avoided. + +** 'save-some-buffers' now uses 'save-some-buffers-default-predicate' +to decide which buffers to ask about, if the PRED argument is nil. +The default value of 'save-some-buffers-default-predicate' is nil, +which means ask about all file-visiting buffers. + +** string-(to|as|make)-(uni|multi)byte are now declared obsolete. + +** New variable 'while-no-input-ignore-events' which allow +setting which special events 'while-no-input' should ignore. +It is a list of symbols. + +** New function 'undo-amalgamate-change-group' to get rid of +undo-boundaries between two states. + +** New var 'definition-prefixes' is a hash table mapping prefixes to +the files where corresponding definitions can be found. This can be +used to fetch definitions that are not yet loaded, for example for +'C-h f'. + +** New var 'syntax-ppss-table' to control the syntax-table used in +'syntax-ppss'. + +** 'define-derived-mode' can now specify an :after-hook form, which +gets evaluated after the new mode's hook has run. This can be used to +incorporate configuration changes made in the mode hook into the +mode's setup. + +** Autoload files are now generated without timestamps. +Set 'autoload-timestamps' to a non-nil value to get timestamps in +autoload files. + +** 'gnutls-boot' now takes a parameter ':complete-negotiation' that +says that negotiation should complete even on non-blocking sockets. + +** There is now a new variable 'flyspell-sort-corrections-function' +that allows changing the way corrections are sorted. + +** The new command 'fortune-message' has been added, which displays +fortunes in the echo area. + +** New function 'func-arity' returns information about the argument list +of an arbitrary function. This generalizes 'subr-arity' for functions +that are not built-in primitives. We recommend using this new +function instead of 'subr-arity'. + +** New function 'region-bounds' can be used in the interactive spec +to provide region boundaries (for rectangular regions more than one) +to an interactively callable function as a single argument instead of +two separate arguments 'region-beginning' and 'region-end'. + +** 'parse-partial-sexp' state has a new element. +Element 10 is non-nil when the last character scanned might be the +first character of a two character construct, i.e., a comment +delimiter or escaped character. Its value is the syntax of that last +character. + +** 'parse-partial-sexp's state, element 9, has now been confirmed as +permanent and documented, and may be used by Lisp programs. Its value +is a list of currently open parenthesis positions, starting with the +outermost parenthesis. + +** 'read-color' will now display the color names using the color itself +as the background color. + +** The function 'redirect-debugging-output' now works on platforms +other than GNU/Linux. + +** The new function 'string-version-lessp' compares strings by +interpreting consecutive runs of numerical characters as numbers, and +compares their numerical values. According to this predicate, +"foo2.png" is smaller than "foo12.png". + +** Numeric comparisons and 'logb' no longer return incorrect answers +due to internal rounding errors. For example, '(< most-positive-fixnum +(+ 1.0 most-positive-fixnum))' now correctly returns t on 64-bit hosts. + +** The functions 'ffloor', 'fceiling', 'ftruncate' and 'fround' now +accept only floating-point arguments, as per their documentation. +Formerly, they quietly accepted integer arguments and sometimes +returned nonsensical answers, e.g., '(< N (ffloor N))' could return t. + +** On hosts like GNU/Linux x86-64 where a 'long double' fraction +contains at least EMACS_INT_WIDTH - 3 bits, 'format' no longer returns +incorrect answers due to internal rounding errors when formatting +Emacs integers with '%e', '%f', or '%g' conversions. For example, on +these hosts '(eql N (string-to-number (format "%.0f" N)))' now returns +t for all Emacs integers N. + +** Calls that accept floating-point integers (for use on hosts with +limited integer range) now signal an error if arguments are not +integral. For example '(decode-char 'ascii 0.5)' now signals an +error. + +** Functions 'string-trim-left', 'string-trim-right' and 'string-trim' +now accept optional arguments which specify the regexp of a substring +to trim. + +** The new function 'char-from-name' converts a Unicode name string +to the corresponding character code. + +** New functions 'sxhash-eq' and 'sxhash-eql' return hash codes of a +Lisp object suitable for use with 'eq' and 'eql' correspondingly. If +two objects are 'eq' ('eql'), then the result of 'sxhash-eq' +('sxhash-eql') on them will be the same. + +** Function 'sxhash' has been renamed to 'sxhash-equal' for +consistency with the new functions. For compatibility, 'sxhash' +remains as an alias to 'sxhash-equal'. + +** 'make-hash-table' now defaults to a rehash threshold of 0.8125 +instead of 0.8, to avoid rounding glitches. + +** New function 'add-variable-watcher' can be used to call a function +when a symbol's value is changed. This is used to implement the new +debugger command 'debug-on-variable-change'. + +** New variable 'print-escape-control-characters' causes 'prin1' and +'print' to output control characters as backslash sequences. + +** Time conversion functions that accept a time zone rule argument now +allow it to be OFFSET or a list (OFFSET ABBR), where the integer +OFFSET is a count of seconds east of Universal Time, and the string +ABBR is a time zone abbreviation. The affected functions are +'current-time-string', 'current-time-zone', 'decode-time', +'format-time-string', and 'set-time-zone-rule'. + +** 'format-time-string' now formats '%q' to the calendar quarter. + +** New built-in function 'mapcan'. +It avoids unnecessary consing (and garbage collection). + +** 'car' and 'cdr' compositions 'cXXXr' and 'cXXXXr' are now part of Elisp. + +** 'gensym' is now part of Elisp. + +** Low-level list functions like 'length' and 'member' now do a better +job of signaling list cycles instead of looping indefinitely. + +** The new functions 'make-nearby-temp-file' and 'temporary-file-directory' +can be used for creation of temporary files on remote or mounted directories. + +** On GNU platforms when operating on a local file, 'file-attributes' +no longer suffers from a race when called while another process is +altering the filesystem. On non-GNU platforms 'file-attributes' +attempts to detect the race, and returns nil if it does so. + +** The new function 'file-local-name' can be used to specify arguments +of remote processes. + +** The new functions 'file-name-quote', 'file-name-unquote' and +'file-name-quoted-p' can be used to quote / unquote file names with +the prefix "/:". + +** The new error 'file-missing', a subcategory of 'file-error', is now +signaled instead of 'file-error' if a file operation acts on a file +that does not exist. + +** The function 'delete-directory' no longer signals an error when +operating recursively and when some other process deletes the directory +or its files before 'delete-directory' gets to them. + +** New error type 'user-search-failed' like 'search-failed' but +avoids debugger like 'user-error'. + +** The function 'line-number-at-pos' now takes a second optional +argument 'absolute'. If this parameter is nil, the default, this +function keeps on returning the line number taking potential narrowing +into account. If this parameter is non-nil, the function ignores +narrowing and returns the absolute line number. + +** The function 'color-distance' now takes a second optional argument +'metric'. When non-nil, it should be a function of two arguments that +accepts two colors and returns a number. + +** Changes in Frame and Window Handling + +*** Resizing a frame no longer runs 'window-configuration-change-hook'. +'window-size-change-functions' should be used instead. + +*** The new function 'frame-size-changed-p' can tell whether a frame has +been resized since the last time 'window-size-change-functions' has been +run. + +*** The function 'frame-geometry' now also returns the width of a +frame's outer border. + +*** New frame parameters and changed semantics for older ones: + +**** 'z-group' positions a frame above or below all others. + +**** 'min-width' and 'min-height' specify the absolute minimum size of a +frame. + +**** 'parent-frame' makes a frame the child frame of another Emacs +frame. The section "(elisp) Child Frames" in the ELisp manual +describes the intrinsics of that relationship. + +**** 'delete-before' triggers deletion of one frame before that of +another. + +**** 'mouse-wheel-frame' specifies another frame whose windows shall be +scrolled instead. + +**** 'no-other-frame' has 'next-frame' and 'previous-frame' skip this +frame. + +**** 'skip-taskbar' removes a frame's icon from the taskbar and has +'Alt-' skip this frame. + +**** 'no-focus-on-map' avoids that a frame gets input focus when mapped. + +**** 'no-accept-focus' means that a frame does not want to get input +focus via the mouse. + +**** 'undecorated' removes the window manager decorations from a frame. + +**** 'override-redirect' tells the window manager to disregard this +frame. + +**** 'width' and 'height' now allow the specification of pixel values +and ratios. + +**** 'left' and 'top' now allow the specification of ratios. + +**** 'keep-ratio' preserves size and position of child frames when their +parent frame is resized. + +**** 'no-special-glyphs' suppresses display of truncation and +continuation glyphs in a frame. + +**** 'auto-hide-function' and 'minibuffer-exit' handle auto hiding of +frames and exiting from minibuffer individually. + +**** 'fit-frame-to-buffer-margins' and 'fit-frame-to-buffer-sizes' +handle fitting a frame to its buffer individually. + +**** 'drag-internal-border', 'drag-with-header-line', +'drag-with-mode-line', 'snap-width', 'top-visible' and 'bottom-visible' +allow dragging and resizing frames with the mouse. + +**** 'minibuffer' is now set to the default minibuffer window when +initially specified as nil and is not reset to nil when initially +specifying a minibuffer window. + +*** The new function 'frame-list-z-order' returns a list of all frames +in Z (stacking) order. + +*** The function 'x-focus-frame' optionally tries to not activate its +frame. + +*** The variable 'focus-follows-mouse' has a third meaningful value +'auto-raise' to indicate that the window manager automatically raises a +frame when the mouse pointer enters it. + +*** The new function 'frame-restack' puts a frame above or below +another on the display. + +*** The new face 'internal-border' specifies the background of a frame's +internal border. + +*** The NORECORD argument of 'select-window' now has a meaningful value +'mark-for-redisplay' which is like any other non-nil value but marks +WINDOW for redisplay. + +*** Support for side windows is now official. The display action +function 'display-buffer-in-side-window' will display its buffer in a +side window. Functions for toggling all side windows on a frame, +changing and reversing the layout of side windows and returning the +main (major non-side) window of a frame are provided. For details +consult the section "(elisp) Side Windows" in the ELisp manual. + +*** Support for atomic windows - rectangular compositions of windows +treated by 'split-window', 'delete-window' and 'delete-other-windows' +like a single live window - is now official. For details consult the +section "(elisp) Atomic Windows" in the ELisp manual. + +*** New 'display-buffer' alist entry 'window-parameters' allows the +assignment of window parameters to the window used for displaying the +buffer. + +*** New function 'display-buffer-reuse-mode-window' is an action function +suitable for use in 'display-buffer-alist'. For example, to avoid +creating a new window when opening man pages when there's already one, +use + +(add-to-list 'display-buffer-alist + '("\\`\\*Man .*\\*\\'" . + (display-buffer-reuse-mode-window + (inhibit-same-window . nil) + (mode . Man-mode)))) + +*** New window parameter 'no-delete-other-windows' prevents that +its window gets deleted by 'delete-other-windows'. + +*** New window parameters 'mode-line-format' and 'header-line-format' +allow the buffer-local formats for this window to be overridden. + +*** New command 'window-swap-states' swaps the states of two live +windows. + +*** New functions 'window-pixel-width-before-size-change' and +'window-pixel-height-before-size-change' support detecting which +window changed size when 'window-size-change-functions' are run. + +*** The new function 'window-lines-pixel-dimensions' returns the pixel +dimensions of a window's text lines. + +*** The new function 'window-largest-empty-rectangle' returns the +dimensions of the largest rectangular area not occupying any text in a +window's body. + +*** The semantics of 'mouse-autoselect-window' has changed slightly. +For details see the section "(elisp) Mouse Window Auto-selection" in +the ELisp manual. + +*** 'select-frame-by-name' now may return a frame on another display +if it does not find a suitable one on the current display. + +** 'tcl-auto-fill-mode' is now declared obsolete. +Its functionality can be replicated simply by setting +'comment-auto-fill-only-comments'. + +** New pcase pattern 'rx' to match against an rx-style regular expression. +For details, see the doc string of 'rx--pcase-macroexpander'. + +** New functions to set region from secondary selection and vice versa. +The new functions 'secondary-selection-to-region' and +'secondary-selection-from-region' let you set the beginning and the +end of the region from those of the secondary selection and vice +versa. + +** New function 'lgstring-remove-glyph' can be used to modify a +gstring returned by the underlying layout engine (e.g. m17n-flt, +uniscribe). + + +* Changes in Emacs 26.1 on Non-Free Operating Systems + +** Intercepting hotkeys on Windows 7 and later now works better. +The new keyboard hooking code properly grabs system hotkeys such as +'Win-*' and 'Alt-TAB', in a way that Emacs can get at them before the +system. This makes the 'w32-register-hot-key' functionality work +again on all versions of MS-Windows starting with Windows 7. On +Windows NT and later you can now register any hotkey combination. (On +Windows 9X, the previous limitations, spelled out in the Emacs manual, +still apply.) + +** 'convert-standard-filename' no longer mirrors slashes on MS-Windows. +Previously, on MS-Windows this function converted slash characters in +file names into backslashes. It no longer does that. If your Lisp +program used 'convert-standard-filename' to prepare file names to be +passed to subprocesses (which is not the recommended usage of that +function), you will now have to mirror slashes in your application +code. One possible way is this: + + (let ((start 0)) + (while (string-match "/" file-name start) + (aset file-name (match-beginning 0) ?\\) + (setq start (match-end 0)))) + +** GUI sessions on MS-Windows now treat SIGINT like Posix platforms do. +The effect of delivering a Ctrl-C (SIGINT) signal to a GUI Emacs on +MS-Windows is now the same as on Posix platforms -- Emacs saves the +session and exits. In particular, this will happen if you start +emacs.exe from the Windows shell, then type Ctrl-C into that shell's +window. + +** 'signal-process' supports SIGTRAP on Windows XP and later. +The 'kill' emulation on Windows now maps SIGTRAP to a call to the +'DebugBreakProcess' API. This causes the receiving process to break +execution and return control to the debugger. If no debugger is +attached to the receiving process, the call is typically ignored. +This is in contrast to the default action on POSIX Systems, where it +causes the receiving process to terminate with a core dump if no +debugger has been attached to it. + +** 'set-mouse-position' and 'set-mouse-absolute-pixel-position' work +on macOS. + +** Emacs can now be run as a GUI application from the command line on +macOS. + +** 'ns-appearance' and 'ns-transparent-titlebar' change the appearance +of frame decorations on macOS 10.9+. + +** 'ns-use-thin-smoothing' enables thin font smoothing on macOS 10.8+. + +** 'process-attributes' on Darwin systems now returns more information. + +** Mousewheel and trackpad scrolling on macOS 10.7+ now behaves more +like the macOS default. The new variables 'ns-mwheel-line-height', +'ns-use-mwheel-acceleration' and 'ns-use-mwheel-momentum' can be used +to customize the behavior. + + +---------------------------------------------------------------------- +This file is part of GNU Emacs. + +GNU Emacs is free software: you can redistribute it and/or modify +it under the terms of the GNU General Public License as published by +the Free Software Foundation, either version 3 of the License, or +(at your option) any later version. + +GNU Emacs is distributed in the hope that it will be useful, +but WITHOUT ANY WARRANTY; without even the implied warranty of +MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +GNU General Public License for more details. + +You should have received a copy of the GNU General Public License +along with GNU Emacs. If not, see . + + +Local variables: +coding: utf-8 +mode: outline +paragraph-separate: "[ ]*$" +end: diff --cc etc/refcards/ru-refcard.tex index 0c4cfbe88fd,d44a104f6ba..43c18d58e29 --- a/etc/refcards/ru-refcard.tex +++ b/etc/refcards/ru-refcard.tex @@@ -40,8 -40,8 +40,8 @@@ \newlength{\ColThreeWidth} \setlength{\ColThreeWidth}{25mm} -\newcommand{\versionemacs}[0]{26} % version of Emacs this is for +\newcommand{\versionemacs}[0]{27} % version of Emacs this is for - \newcommand{\cyear}[0]{2018} % copyright year + \newcommand{\cyear}[0]{2019} % copyright year \newcommand\shortcopyrightnotice[0]{\vskip 1ex plus 2 fill \centerline{\footnotesize \copyright\ \cyear\ Free Software Foundation, Inc. diff --cc lisp/calc/calc-alg.el index 2f23399841e,8e3476d191e..41ffc83d86f --- a/lisp/calc/calc-alg.el +++ b/lisp/calc/calc-alg.el @@@ -1,6 -1,6 +1,6 @@@ -;;; calc-alg.el --- algebraic functions for Calc +;;; calc-alg.el --- algebraic functions for Calc -*- lexical-binding:t -*- - ;; Copyright (C) 1990-1993, 2001-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1990-1993, 2001-2019 Free Software Foundation, Inc. ;; Author: David Gillespie diff --cc lisp/calc/calc-ext.el index 761eb97a816,f2e70906e94..1456fb28570 --- a/lisp/calc/calc-ext.el +++ b/lisp/calc/calc-ext.el @@@ -1,6 -1,6 +1,6 @@@ -;;; calc-ext.el --- various extension functions for Calc +;;; calc-ext.el --- various extension functions for Calc -*- lexical-binding:t -*- - ;; Copyright (C) 1990-1993, 2001-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1990-1993, 2001-2019 Free Software Foundation, Inc. ;; Author: David Gillespie diff --cc lisp/calc/calc-poly.el index 41083b77480,7e3e423868c..5fba85e059d --- a/lisp/calc/calc-poly.el +++ b/lisp/calc/calc-poly.el @@@ -1,6 -1,6 +1,6 @@@ -;;; calc-poly.el --- polynomial functions for Calc +;;; calc-poly.el --- polynomial functions for Calc -*- lexical-binding:t -*- - ;; Copyright (C) 1990-1993, 2001-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1990-1993, 2001-2019 Free Software Foundation, Inc. ;; Author: David Gillespie diff --cc lisp/calc/calc-units.el index 6e58eaf225f,8794d1f3c67..86bebe6a9ed --- a/lisp/calc/calc-units.el +++ b/lisp/calc/calc-units.el @@@ -1,6 -1,6 +1,6 @@@ -;;; calc-units.el --- unit conversion functions for Calc +;;; calc-units.el --- unit conversion functions for Calc -*- lexical-binding:t -*- - ;; Copyright (C) 1990-1993, 2001-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1990-1993, 2001-2019 Free Software Foundation, Inc. ;; Author: David Gillespie diff --cc lisp/calc/calc.el index f155b8283b7,ef850a44fe6..c1f23cc94e4 --- a/lisp/calc/calc.el +++ b/lisp/calc/calc.el @@@ -1,6 -1,6 +1,6 @@@ -;;; calc.el --- the GNU Emacs calculator +;;; calc.el --- the GNU Emacs calculator -*- lexical-binding:t -*- - ;; Copyright (C) 1990-1993, 2001-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1990-1993, 2001-2019 Free Software Foundation, Inc. ;; Author: David Gillespie ;; Keywords: convenience, extensions diff --cc lisp/calc/calccomp.el index 75c7adc59ec,d81cc04fe50..91eadfbb4e8 --- a/lisp/calc/calccomp.el +++ b/lisp/calc/calccomp.el @@@ -1,6 -1,6 +1,6 @@@ -;;; calccomp.el --- composition functions for Calc +;;; calccomp.el --- composition functions for Calc -*- lexical-binding:t -*- - ;; Copyright (C) 1990-1993, 2001-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1990-1993, 2001-2019 Free Software Foundation, Inc. ;; Author: David Gillespie diff --cc lisp/calendar/appt.el index 0259dd1e1e5,da041f024f8..40cb9f7cbdb --- a/lisp/calendar/appt.el +++ b/lisp/calendar/appt.el @@@ -1,6 -1,6 +1,6 @@@ -;;; appt.el --- appointment notification functions +;;; appt.el --- appointment notification functions -*- lexical-binding:t -*- - ;; Copyright (C) 1989-1990, 1994, 1998, 2001-2018 Free Software + ;; Copyright (C) 1989-1990, 1994, 1998, 2001-2019 Free Software ;; Foundation, Inc. ;; Author: Neil Mager diff --cc lisp/calendar/cal-dst.el index 8392e81b16f,e78f19f803f..567ba9c8a12 --- a/lisp/calendar/cal-dst.el +++ b/lisp/calendar/cal-dst.el @@@ -1,6 -1,6 +1,6 @@@ -;;; cal-dst.el --- calendar functions for daylight saving rules +;;; cal-dst.el --- calendar functions for daylight saving rules -*- lexical-binding:t -*- - ;; Copyright (C) 1993-1996, 2001-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1993-1996, 2001-2019 Free Software Foundation, Inc. ;; Author: Paul Eggert ;; Edward M. Reingold diff --cc lisp/calendar/calendar.el index 71fb76ce213,a15f15cf307..de8f758fae8 --- a/lisp/calendar/calendar.el +++ b/lisp/calendar/calendar.el @@@ -1,6 -1,6 +1,6 @@@ -;;; calendar.el --- calendar functions +;;; calendar.el --- calendar functions -*- lexical-binding:t -*- - ;; Copyright (C) 1988-1995, 1997, 2000-2018 Free Software Foundation, + ;; Copyright (C) 1988-1995, 1997, 2000-2019 Free Software Foundation, ;; Inc. ;; Author: Edward M. Reingold diff --cc lisp/calendar/diary-lib.el index 1cc59784c86,164363c2b70..1be2a05eee3 --- a/lisp/calendar/diary-lib.el +++ b/lisp/calendar/diary-lib.el @@@ -1,6 -1,6 +1,6 @@@ -;;; diary-lib.el --- diary functions +;;; diary-lib.el --- diary functions -*- lexical-binding:t -*- - ;; Copyright (C) 1989-1990, 1992-1995, 2001-2018 Free Software + ;; Copyright (C) 1989-1990, 1992-1995, 2001-2019 Free Software ;; Foundation, Inc. ;; Author: Edward M. Reingold diff --cc lisp/calendar/holidays.el index 62b9d778e2e,40bc066c9ec..2b080c30738 --- a/lisp/calendar/holidays.el +++ b/lisp/calendar/holidays.el @@@ -1,6 -1,6 +1,6 @@@ -;;; holidays.el --- holiday functions for the calendar package +;;; holidays.el --- holiday functions for the calendar package -*- lexical-binding:t -*- - ;; Copyright (C) 1989-1990, 1992-1994, 1997, 2001-2018 Free Software + ;; Copyright (C) 1989-1990, 1992-1994, 1997, 2001-2019 Free Software ;; Foundation, Inc. ;; Author: Edward M. Reingold diff --cc lisp/calendar/solar.el index ddaf7451bd9,aa5ab91f16e..23bc7611e5f --- a/lisp/calendar/solar.el +++ b/lisp/calendar/solar.el @@@ -1,6 -1,6 +1,6 @@@ -;;; solar.el --- calendar functions for solar events +;;; solar.el --- calendar functions for solar events -*- lexical-binding:t -*- - ;; Copyright (C) 1992-1993, 1995, 1997, 2001-2018 Free Software + ;; Copyright (C) 1992-1993, 1995, 1997, 2001-2019 Free Software ;; Foundation, Inc. ;; Author: Edward M. Reingold diff --cc lisp/calendar/timeclock.el index 646f5298fe4,769beddc3c4..0562f4a998d --- a/lisp/calendar/timeclock.el +++ b/lisp/calendar/timeclock.el @@@ -1,6 -1,6 +1,6 @@@ -;;; timeclock.el --- mode for keeping track of how much you work +;;; timeclock.el --- mode for keeping track of how much you work -*- lexical-binding:t -*- - ;; Copyright (C) 1999-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1999-2019 Free Software Foundation, Inc. ;; Author: John Wiegley ;; Created: 25 Mar 1999 diff --cc lisp/cus-theme.el index 995c55b2b20,c195f4afe26..bc9d1d4f7d6 --- a/lisp/cus-theme.el +++ b/lisp/cus-theme.el @@@ -1,6 -1,6 +1,6 @@@ -;;; cus-theme.el -- custom theme creation user interface +;;; cus-theme.el -- custom theme creation user interface -*- lexical-binding: t -*- ;; - ;; Copyright (C) 2001-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2001-2019 Free Software Foundation, Inc. ;; ;; Author: Alex Schroeder ;; Maintainer: emacs-devel@gnu.org diff --cc lisp/custom.el index a08f7fda705,f0125742d1f..53b8045f058 --- a/lisp/custom.el +++ b/lisp/custom.el @@@ -1,6 -1,6 +1,6 @@@ -;;; custom.el --- tools for declaring and initializing options +;;; custom.el --- tools for declaring and initializing options -*- lexical-binding: t -*- ;; - ;; Copyright (C) 1996-1997, 1999, 2001-2018 Free Software Foundation, + ;; Copyright (C) 1996-1997, 1999, 2001-2019 Free Software Foundation, ;; Inc. ;; ;; Author: Per Abrahamsen diff --cc lisp/delim-col.el index 076d4dc5c3d,d8116f3544b..a968b32052b --- a/lisp/delim-col.el +++ b/lisp/delim-col.el @@@ -1,9 -1,9 +1,9 @@@ ;;; delim-col.el --- prettify all columns in a region or rectangle - ;; Copyright (C) 1999-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1999-2019 Free Software Foundation, Inc. -;; Author: Vinicius Jose Latorre -;; Maintainer: Vinicius Jose Latorre +;; Author: Vinicius Jose Latorre +;; Maintainer: Vinicius Jose Latorre ;; Version: 2.1 ;; Keywords: internal ;; X-URL: http://www.emacswiki.org/cgi-bin/wiki/ViniciusJoseLatorre diff --cc lisp/ecomplete.el index 3bfab4743cb,c54c110867b..9fc3f57833c --- a/lisp/ecomplete.el +++ b/lisp/ecomplete.el @@@ -1,6 -1,6 +1,6 @@@ -;;; ecomplete.el --- electric completion of addresses and the like +;;; ecomplete.el --- electric completion of addresses and the like -*- lexical-binding:t -*- - ;; Copyright (C) 2006-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2006-2019 Free Software Foundation, Inc. ;; Author: Lars Magne Ingebrigtsen ;; Keywords: mail diff --cc lisp/emacs-lisp/lisp-mnt.el index 5c623a3ab8c,6cad17a4a1b..91c76158a31 --- a/lisp/emacs-lisp/lisp-mnt.el +++ b/lisp/emacs-lisp/lisp-mnt.el @@@ -1,6 -1,6 +1,6 @@@ -;;; lisp-mnt.el --- utility functions for Emacs Lisp maintainers +;;; lisp-mnt.el --- utility functions for Emacs Lisp maintainers -*- lexical-binding:t -*- - ;; Copyright (C) 1992, 1994, 1997, 2000-2018 Free Software Foundation, + ;; Copyright (C) 1992, 1994, 1997, 2000-2019 Free Software Foundation, ;; Inc. ;; Author: Eric S. Raymond diff --cc lisp/emulation/viper-keym.el index cc0b7ebc379,c0b7a5b5c9c..8bb75d65afa --- a/lisp/emulation/viper-keym.el +++ b/lisp/emulation/viper-keym.el @@@ -1,6 -1,6 +1,6 @@@ -;;; viper-keym.el --- Viper keymaps +;;; viper-keym.el --- Viper keymaps -*- lexical-binding:t -*- - ;; Copyright (C) 1994-1997, 2000-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1994-1997, 2000-2019 Free Software Foundation, Inc. ;; Author: Michael Kifer ;; Package: viper diff --cc lisp/emulation/viper-macs.el index cfb46cc19a8,90d33d5fa7b..37ab81d78ef --- a/lisp/emulation/viper-macs.el +++ b/lisp/emulation/viper-macs.el @@@ -1,6 -1,6 +1,6 @@@ -;;; viper-macs.el --- functions implementing keyboard macros for Viper +;;; viper-macs.el --- functions implementing keyboard macros for Viper -*- lexical-binding:t -*- - ;; Copyright (C) 1994-1997, 2000-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1994-1997, 2000-2019 Free Software Foundation, Inc. ;; Author: Michael Kifer ;; Package: viper diff --cc lisp/emulation/viper-util.el index aa456551a68,04edc90c88a..a7e7af3bf85 --- a/lisp/emulation/viper-util.el +++ b/lisp/emulation/viper-util.el @@@ -1,6 -1,6 +1,6 @@@ -;;; viper-util.el --- Utilities used by viper.el +;;; viper-util.el --- Utilities used by viper.el -*- lexical-binding:t -*- - ;; Copyright (C) 1994-1997, 1999-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1994-1997, 1999-2019 Free Software Foundation, Inc. ;; Author: Michael Kifer ;; Package: viper diff --cc lisp/erc/erc-services.el index ac49a3e12ef,b6bceff205c..886ba60eb47 --- a/lisp/erc/erc-services.el +++ b/lisp/erc/erc-services.el @@@ -1,6 -1,6 +1,6 @@@ -;;; erc-services.el --- Identify to NickServ +;;; erc-services.el --- Identify to NickServ -*- lexical-binding:t -*- - ;; Copyright (C) 2002-2004, 2006-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2002-2004, 2006-2019 Free Software Foundation, Inc. ;; Maintainer: emacs-devel@gnu.org diff --cc lisp/filecache.el index 9dd631001da,eb4a64b768f..fb099c1802a --- a/lisp/filecache.el +++ b/lisp/filecache.el @@@ -1,6 -1,6 +1,6 @@@ -;;; filecache.el --- find files using a pre-loaded cache +;;; filecache.el --- find files using a pre-loaded cache -*- lexical-binding:t -*- - ;; Copyright (C) 1996, 2000-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1996, 2000-2019 Free Software Foundation, Inc. ;; Author: Peter Breton ;; Created: Sun Nov 10 1996 diff --cc lisp/font-lock.el index ee4b78b590e,d3828cf6b47..1475911195a --- a/lisp/font-lock.el +++ b/lisp/font-lock.el @@@ -1,6 -1,6 +1,6 @@@ -;;; font-lock.el --- Electric font lock mode +;;; font-lock.el --- Electric font lock mode -*- lexical-binding:t -*- - ;; Copyright (C) 1992-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1992-2019 Free Software Foundation, Inc. ;; Author: Jamie Zawinski ;; Richard Stallman diff --cc lisp/fringe.el index 583a0e2c20b,31d80a8a77d..92387a21571 --- a/lisp/fringe.el +++ b/lisp/fringe.el @@@ -1,6 -1,6 +1,6 @@@ -;;; fringe.el --- fringe setup and control +;;; fringe.el --- fringe setup and control -*- lexical-binding:t -*- - ;; Copyright (C) 2002-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2002-2019 Free Software Foundation, Inc. ;; Author: Simon Josefsson ;; Maintainer: emacs-devel@gnu.org diff --cc lisp/gnus/gnus.el index 1ac02b4531c,c3a57800e39..e5574d4f4f0 --- a/lisp/gnus/gnus.el +++ b/lisp/gnus/gnus.el @@@ -1,6 -1,6 +1,6 @@@ -;;; gnus.el --- a newsreader for GNU Emacs +;;; gnus.el --- a newsreader for GNU Emacs -*- lexical-binding:t -*- - ;; Copyright (C) 1987-1990, 1993-1998, 2000-2018 Free Software + ;; Copyright (C) 1987-1990, 1993-1998, 2000-2019 Free Software ;; Foundation, Inc. ;; Author: Masanobu UMEDA diff --cc lisp/gnus/mm-decode.el index 3e6883b2a4b,5dad2acbdd0..80a2e50c498 --- a/lisp/gnus/mm-decode.el +++ b/lisp/gnus/mm-decode.el @@@ -1,6 -1,6 +1,6 @@@ -;;; mm-decode.el --- Functions for decoding MIME things +;;; mm-decode.el --- Functions for decoding MIME things -*- lexical-binding:t -*- - ;; Copyright (C) 1998-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1998-2019 Free Software Foundation, Inc. ;; Author: Lars Magne Ingebrigtsen ;; MORIOKA Tomohiko diff --cc lisp/gnus/mm-extern.el index fbae669ce94,32ba831a0da..c3054432d51 --- a/lisp/gnus/mm-extern.el +++ b/lisp/gnus/mm-extern.el @@@ -1,6 -1,6 +1,6 @@@ -;;; mm-extern.el --- showing message/external-body +;;; mm-extern.el --- showing message/external-body -*- lexical-binding:t -*- - ;; Copyright (C) 2000-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2000-2019 Free Software Foundation, Inc. ;; Author: Shenghuo Zhu ;; Keywords: message external-body diff --cc lisp/gnus/mm-util.el index ba54b4e7074,bbaab536f1a..00a8a532d27 --- a/lisp/gnus/mm-util.el +++ b/lisp/gnus/mm-util.el @@@ -1,6 -1,6 +1,6 @@@ -;;; mm-util.el --- Utility functions for Mule and low level things +;;; mm-util.el --- Utility functions for Mule and low level things -*- lexical-binding:t -*- - ;; Copyright (C) 1998-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1998-2019 Free Software Foundation, Inc. ;; Author: Lars Magne Ingebrigtsen ;; MORIOKA Tomohiko diff --cc lisp/help.el index ad782f74cac,c23d4d8fe54..5fa1876b38e --- a/lisp/help.el +++ b/lisp/help.el @@@ -1,6 -1,6 +1,6 @@@ -;;; help.el --- help commands for Emacs +;;; help.el --- help commands for Emacs -*- lexical-binding:t -*- - ;; Copyright (C) 1985-1986, 1993-1994, 1998-2018 Free Software + ;; Copyright (C) 1985-1986, 1993-1994, 1998-2019 Free Software ;; Foundation, Inc. ;; Maintainer: emacs-devel@gnu.org diff --cc lisp/hfy-cmap.el index ee6e18edb0a,310805652f2..a34efa60b4e --- a/lisp/hfy-cmap.el +++ b/lisp/hfy-cmap.el @@@ -1,6 -1,6 +1,6 @@@ -;;; hfy-cmap.el --- Fallback colour name -> rgb mapping for `htmlfontify' +;;; hfy-cmap.el --- Fallback color name -> rgb mapping for `htmlfontify' - ;; Copyright (C) 2002-2003, 2009-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2002-2003, 2009-2019 Free Software Foundation, Inc. ;; Emacs Lisp Archive Entry ;; Package: htmlfontify diff --cc lisp/macros.el index 4078b983ec6,27a14694ee6..ba6a840d60c --- a/lisp/macros.el +++ b/lisp/macros.el @@@ -1,6 -1,6 +1,6 @@@ -;;; macros.el --- non-primitive commands for keyboard macros +;;; macros.el --- non-primitive commands for keyboard macros -*- lexical-binding:t -*- - ;; Copyright (C) 1985-1987, 1992, 1994-1995, 2001-2018 Free Software + ;; Copyright (C) 1985-1987, 1992, 1994-1995, 2001-2019 Free Software ;; Foundation, Inc. ;; Maintainer: emacs-devel@gnu.org diff --cc lisp/mail/binhex.el index fa2ea3d8471,eb00b87f4c1..19fcbaaf475 --- a/lisp/mail/binhex.el +++ b/lisp/mail/binhex.el @@@ -1,6 -1,6 +1,6 @@@ -;;; binhex.el --- decode BinHex-encoded text +;;; binhex.el --- decode BinHex-encoded text -*- lexical-binding:t -*- - ;; Copyright (C) 1998-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1998-2019 Free Software Foundation, Inc. ;; Author: Shenghuo Zhu ;; Keywords: binhex news diff --cc lisp/mail/flow-fill.el index db2a30ad15e,8451315a12e..06279d4041c --- a/lisp/mail/flow-fill.el +++ b/lisp/mail/flow-fill.el @@@ -1,6 -1,6 +1,6 @@@ -;;; flow-fill.el --- interpret RFC2646 "flowed" text +;;; flow-fill.el --- interpret RFC2646 "flowed" text -*- lexical-binding:t -*- - ;; Copyright (C) 2000-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2000-2019 Free Software Foundation, Inc. ;; Author: Simon Josefsson ;; Keywords: mail diff --cc lisp/mail/footnote.el index f5d280ae1ea,be3a878f832..a7802929dca --- a/lisp/mail/footnote.el +++ b/lisp/mail/footnote.el @@@ -1,9 -1,8 +1,9 @@@ -;;; footnote.el --- footnote support for message mode +;;; footnote.el --- footnote support for message mode -*- lexical-binding:t -*- - ;; Copyright (C) 1997, 2000-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1997, 2000-2019 Free Software Foundation, Inc. -;; Author: Steven L Baur +;; Author: Steven L Baur (1997-2011) +;; Boruch Baum (2017-) ;; Keywords: mail, news ;; Version: 0.19 diff --cc lisp/mail/hashcash.el index 37b2d94e5f5,9fdc7ea756c..59200eaab88 --- a/lisp/mail/hashcash.el +++ b/lisp/mail/hashcash.el @@@ -1,6 -1,6 +1,6 @@@ -;;; hashcash.el --- Add hashcash payments to email +;;; hashcash.el --- Add hashcash payments to email -*- lexical-binding:t -*- - ;; Copyright (C) 2003-2005, 2007-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2003-2005, 2007-2019 Free Software Foundation, Inc. ;; Written by: Paul Foley (1997-2002) ;; Maintainer: emacs-devel@gnu.org diff --cc lisp/mail/ietf-drums.el index 0af3221fc33,6501a99f7bc..66006573c14 --- a/lisp/mail/ietf-drums.el +++ b/lisp/mail/ietf-drums.el @@@ -1,6 -1,6 +1,6 @@@ -;;; ietf-drums.el --- Functions for parsing RFC822bis headers +;;; ietf-drums.el --- Functions for parsing RFC822bis headers -*- lexical-binding:t -*- - ;; Copyright (C) 1998-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1998-2019 Free Software Foundation, Inc. ;; Author: Lars Magne Ingebrigtsen ;; This file is part of GNU Emacs. diff --cc lisp/mail/rfc2231.el index 103af55248a,31112a7505a..0a0983c25f8 --- a/lisp/mail/rfc2231.el +++ b/lisp/mail/rfc2231.el @@@ -1,6 -1,6 +1,6 @@@ -;;; rfc2231.el --- Functions for decoding rfc2231 headers +;;; rfc2231.el --- Functions for decoding rfc2231 headers -*- lexical-binding:t -*- - ;; Copyright (C) 1998-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1998-2019 Free Software Foundation, Inc. ;; Author: Lars Magne Ingebrigtsen ;; This file is part of GNU Emacs. diff --cc lisp/mail/sendmail.el index 6fc91a3acd9,58a2cd16011..9315d6ac627 --- a/lisp/mail/sendmail.el +++ b/lisp/mail/sendmail.el @@@ -1,6 -1,6 +1,6 @@@ -;;; sendmail.el --- mail sending commands for Emacs +;;; sendmail.el --- mail sending commands for Emacs -*- lexical-binding:t -*- - ;; Copyright (C) 1985-1986, 1992-1996, 1998, 2000-2018 Free Software + ;; Copyright (C) 1985-1986, 1992-1996, 1998, 2000-2019 Free Software ;; Foundation, Inc. ;; Maintainer: emacs-devel@gnu.org diff --cc lisp/mail/smtpmail.el index 8a1e86b7750,5b28c271038..49645aba7a1 --- a/lisp/mail/smtpmail.el +++ b/lisp/mail/smtpmail.el @@@ -1,6 -1,6 +1,6 @@@ -;;; smtpmail.el --- simple SMTP protocol (RFC 821) for sending mail +;;; smtpmail.el --- simple SMTP protocol (RFC 821) for sending mail -*- lexical-binding:t -*- - ;; Copyright (C) 1995-1996, 2001-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1995-1996, 2001-2019 Free Software Foundation, Inc. ;; Author: Tomoji Kagatani ;; Maintainer: Simon Josefsson diff --cc lisp/mail/uudecode.el index b8f74e3a839,a78202938e4..32e668baab3 --- a/lisp/mail/uudecode.el +++ b/lisp/mail/uudecode.el @@@ -1,6 -1,6 +1,6 @@@ -;;; uudecode.el -- elisp native uudecode +;;; uudecode.el -- elisp native uudecode -*- lexical-binding:t -*- - ;; Copyright (C) 1998-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1998-2019 Free Software Foundation, Inc. ;; Author: Shenghuo Zhu ;; Keywords: uudecode news diff --cc lisp/mail/yenc.el index 25b4ebb9bda,f7a12d6c9e2..75b62c14117 --- a/lisp/mail/yenc.el +++ b/lisp/mail/yenc.el @@@ -1,6 -1,6 +1,6 @@@ -;;; yenc.el --- elisp native yenc decoder +;;; yenc.el --- elisp native yenc decoder -*- lexical-binding:t -*- - ;; Copyright (C) 2002-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2002-2019 Free Software Foundation, Inc. ;; Author: Jesper Harder ;; Keywords: yenc news diff --cc lisp/net/ange-ftp.el index 8828010ab41,f400c562939..f2ee8950494 --- a/lisp/net/ange-ftp.el +++ b/lisp/net/ange-ftp.el @@@ -1,6 -1,6 +1,6 @@@ -;;; ange-ftp.el --- transparent FTP support for GNU Emacs +;;; ange-ftp.el --- transparent FTP support for GNU Emacs -*- lexical-binding:t -*- - ;; Copyright (C) 1989-1996, 1998, 2000-2018 Free Software Foundation, + ;; Copyright (C) 1989-1996, 1998, 2000-2019 Free Software Foundation, ;; Inc. ;; Author: Andy Norman (ange@hplb.hpl.hp.com) diff --cc lisp/net/eudc.el index a28fa6aa17a,bc550fbc113..3c9c01d0f96 --- a/lisp/net/eudc.el +++ b/lisp/net/eudc.el @@@ -1,6 -1,6 +1,6 @@@ -;;; eudc.el --- Emacs Unified Directory Client +;;; eudc.el --- Emacs Unified Directory Client -*- lexical-binding:t -*- - ;; Copyright (C) 1998-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1998-2019 Free Software Foundation, Inc. ;; Author: Oscar Figueiredo ;; Pavel Janík diff --cc lisp/net/imap.el index 042b0f9a2c9,f4ae7b94495..504900eabe9 --- a/lisp/net/imap.el +++ b/lisp/net/imap.el @@@ -1,6 -1,6 +1,6 @@@ -;;; imap.el --- imap library +;;; imap.el --- imap library -*- lexical-binding:t -*- - ;; Copyright (C) 1998-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1998-2019 Free Software Foundation, Inc. ;; Author: Simon Josefsson ;; Keywords: mail diff --cc lisp/net/newst-backend.el index 40096ca4c1e,94101fb0b25..bb7b7bcc1b0 --- a/lisp/net/newst-backend.el +++ b/lisp/net/newst-backend.el @@@ -1,6 -1,6 +1,6 @@@ -;;; newst-backend.el --- Retrieval backend for newsticker. +;;; newst-backend.el --- Retrieval backend for newsticker -*- lexical-binding:t -*- - ;; Copyright (C) 2003-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2003-2019 Free Software Foundation, Inc. ;; Author: Ulf Jasper ;; Filename: newst-backend.el diff --cc lisp/net/ntlm.el index 142c37510ec,d3899e45eae..88c561910cb --- a/lisp/net/ntlm.el +++ b/lisp/net/ntlm.el @@@ -1,6 -1,6 +1,6 @@@ -;;; ntlm.el --- NTLM (NT LanManager) authentication support +;;; ntlm.el --- NTLM (NT LanManager) authentication support -*- lexical-binding:t -*- - ;; Copyright (C) 2001, 2007-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2001, 2007-2019 Free Software Foundation, Inc. ;; Author: Taro Kawagishi ;; Maintainer: Thomas Fitzsimmons diff --cc lisp/net/pop3.el index 2a6807e1aca,f73607081c5..bcf56c9ab3f --- a/lisp/net/pop3.el +++ b/lisp/net/pop3.el @@@ -1,6 -1,6 +1,6 @@@ -;;; pop3.el --- Post Office Protocol (RFC 1460) interface +;;; pop3.el --- Post Office Protocol (RFC 1460) interface -*- lexical-binding:t -*- - ;; Copyright (C) 1996-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1996-2019 Free Software Foundation, Inc. ;; Author: Richard L. Pieri ;; Maintainer: emacs-devel@gnu.org diff --cc lisp/net/rfc2104.el index 57bca2e8788,3b000399b99..68c35aa3130 --- a/lisp/net/rfc2104.el +++ b/lisp/net/rfc2104.el @@@ -1,6 -1,6 +1,6 @@@ -;;; rfc2104.el --- RFC2104 Hashed Message Authentication Codes +;;; rfc2104.el --- RFC2104 Hashed Message Authentication Codes -*- lexical-binding:t -*- - ;; Copyright (C) 1998-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1998-2019 Free Software Foundation, Inc. ;; Author: Simon Josefsson ;; Keywords: mail diff --cc lisp/net/rlogin.el index 015e04f4075,f73638699d4..bf6f5359aad --- a/lisp/net/rlogin.el +++ b/lisp/net/rlogin.el @@@ -1,6 -1,6 +1,6 @@@ -;;; rlogin.el --- remote login interface +;;; rlogin.el --- remote login interface -*- lexical-binding:t -*- - ;; Copyright (C) 1992-1995, 1997-1998, 2001-2018 Free Software + ;; Copyright (C) 1992-1995, 1997-1998, 2001-2019 Free Software ;; Foundation, Inc. ;; Author: Noah Friedman diff --cc lisp/net/shr-color.el index 6303141c898,fcc307b929c..2d6cf68d8ff --- a/lisp/net/shr-color.el +++ b/lisp/net/shr-color.el @@@ -1,6 -1,6 +1,6 @@@ -;;; shr-color.el --- Simple HTML Renderer color management +;;; shr-color.el --- Simple HTML Renderer color management -*- lexical-binding:t -*- - ;; Copyright (C) 2010-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2010-2019 Free Software Foundation, Inc. ;; Author: Julien Danjou ;; Keywords: html diff --cc lisp/net/sieve-manage.el index 8c70ae037ab,18a3dcc4c09..72731b8fdae --- a/lisp/net/sieve-manage.el +++ b/lisp/net/sieve-manage.el @@@ -1,6 -1,6 +1,6 @@@ -;;; sieve-manage.el --- Implementation of the managesieve protocol in elisp +;;; sieve-manage.el --- Implementation of the managesieve protocol in elisp -*- lexical-binding:t -*- - ;; Copyright (C) 2001-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2001-2019 Free Software Foundation, Inc. ;; Author: Simon Josefsson ;; Albert Krewinkel diff --cc lisp/net/socks.el index 5ee6eea933f,c2a8b699cd5..6356707a1db --- a/lisp/net/socks.el +++ b/lisp/net/socks.el @@@ -1,6 -1,6 +1,6 @@@ -;;; socks.el --- A Socks v5 Client for Emacs +;;; socks.el --- A Socks v5 Client for Emacs -*- lexical-binding:t -*- - ;; Copyright (C) 1996-2000, 2002, 2007-2018 Free Software Foundation, + ;; Copyright (C) 1996-2000, 2002, 2007-2019 Free Software Foundation, ;; Inc. ;; Author: William M. Perry diff --cc lisp/obsolete/starttls.el index 0dc2663870a,00000000000..b89e612e4a3 mode 100644,000000..100644 --- a/lisp/obsolete/starttls.el +++ b/lisp/obsolete/starttls.el @@@ -1,305 -1,0 +1,305 @@@ +;;; starttls.el --- STARTTLS functions + - ;; Copyright (C) 1999-2018 Free Software Foundation, Inc. ++;; Copyright (C) 1999-2019 Free Software Foundation, Inc. + +;; Author: Daiki Ueno +;; Author: Simon Josefsson +;; Created: 1999/11/20 +;; Keywords: TLS, SSL, OpenSSL, GnuTLS, mail, news +;; Obsolete-since: 27.1 + +;; This file is part of GNU Emacs. + +;; GNU Emacs is free software: you can redistribute it and/or modify +;; it under the terms of the GNU General Public License as published by +;; the Free Software Foundation, either version 3 of the License, or +;; (at your option) any later version. + +;; GNU Emacs is distributed in the hope that it will be useful, +;; but WITHOUT ANY WARRANTY; without even the implied warranty of +;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +;; GNU General Public License for more details. + +;; You should have received a copy of the GNU General Public License +;; along with GNU Emacs. If not, see . + +;;; Commentary: + +;; This module defines some utility functions for STARTTLS profiles. + +;; [RFC 2595] "Using TLS with IMAP, POP3 and ACAP" +;; by Chris Newman (1999/06) + +;; This file now contains a combination of the two previous +;; implementations both called "starttls.el". The first one is Daiki +;; Ueno's starttls.el which uses his own "starttls" command line tool, +;; and the second one is Simon Josefsson's starttls.el which uses +;; "gnutls-cli" from GnuTLS. +;; +;; If "starttls" is available, it is preferred by the code over +;; "gnutls-cli", for backwards compatibility. Use +;; `starttls-use-gnutls' to toggle between implementations if you have +;; both tools installed. It is recommended to use GnuTLS, though, as +;; it performs more verification of the certificates. + +;; The GnuTLS support requires GnuTLS 0.9.90 (released 2003-10-08) or +;; later, from , or "starttls" +;; from . + +;; Usage is similar to `open-network-stream'. For example: +;; +;; (when (setq tmp (starttls-open-stream +;; "test" (current-buffer) "yxa.extundo.com" 25)) +;; (accept-process-output tmp 15) +;; (process-send-string tmp "STARTTLS\n") +;; (accept-process-output tmp 15) +;; (message "STARTTLS output:\n%s" (starttls-negotiate tmp)) +;; (process-send-string tmp "EHLO foo\n")) + +;; An example run yields the following output: +;; +;; 220 yxa.extundo.com ESMTP Sendmail 8.12.11/8.12.11/Debian-3; Wed, 26 May 2004 19:12:29 +0200; (No UCE/UBE) logging access from: c494102a.s-bi.bostream.se(OK)-c494102a.s-bi.bostream.se [217.215.27.65] +;; 220 2.0.0 Ready to start TLS +;; 250-yxa.extundo.com Hello c494102a.s-bi.bostream.se [217.215.27.65], pleased to meet you +;; 250-ENHANCEDSTATUSCODES +;; 250-PIPELINING +;; 250-EXPN +;; 250-VERB +;; 250-8BITMIME +;; 250-SIZE +;; 250-DSN +;; 250-ETRN +;; 250-AUTH DIGEST-MD5 CRAM-MD5 PLAIN LOGIN +;; 250-DELIVERBY +;; 250 HELP +;; nil +;; +;; With the message buffer containing: +;; +;; STARTTLS output: +;; *** Starting TLS handshake +;; - Server's trusted authorities: +;; [0]: C=SE,ST=Stockholm,L=Stockholm,O=YXA,OU=CA,CN=yxa.extundo.com,EMAIL=staff@yxa.extundo.com +;; - Certificate type: X.509 +;; - Got a certificate list of 2 certificates. +;; +;; - Certificate[0] info: +;; # The hostname in the certificate matches 'yxa.extundo.com'. +;; # valid since: Wed May 26 12:16:00 CEST 2004 +;; # expires at: Wed Jul 26 12:16:00 CEST 2023 +;; # serial number: 04 +;; # fingerprint: 7c 04 4b c1 fa 26 9b 5d 90 22 52 3c 65 3d 85 3a +;; # version: #1 +;; # public key algorithm: RSA +;; # Modulus: 1024 bits +;; # Subject's DN: C=SE,ST=Stockholm,L=Stockholm,O=YXA,OU=Mail server,CN=yxa.extundo.com,EMAIL=staff@yxa.extundo.com +;; # Issuer's DN: C=SE,ST=Stockholm,L=Stockholm,O=YXA,OU=CA,CN=yxa.extundo.com,EMAIL=staff@yxa.extundo.com +;; +;; - Certificate[1] info: +;; # valid since: Sun May 23 11:35:00 CEST 2004 +;; # expires at: Sun Jul 23 11:35:00 CEST 2023 +;; # serial number: 00 +;; # fingerprint: fc 76 d8 63 1a c9 0b 3b fa 40 fe ed 47 7a 58 ae +;; # version: #3 +;; # public key algorithm: RSA +;; # Modulus: 1024 bits +;; # Subject's DN: C=SE,ST=Stockholm,L=Stockholm,O=YXA,OU=CA,CN=yxa.extundo.com,EMAIL=staff@yxa.extundo.com +;; # Issuer's DN: C=SE,ST=Stockholm,L=Stockholm,O=YXA,OU=CA,CN=yxa.extundo.com,EMAIL=staff@yxa.extundo.com +;; +;; - Peer's certificate issuer is unknown +;; - Peer's certificate is NOT trusted +;; - Version: TLS 1.0 +;; - Key Exchange: RSA +;; - Cipher: ARCFOUR 128 +;; - MAC: SHA +;; - Compression: NULL + +;;; Code: + +(defgroup starttls nil + "Support for `Transport Layer Security' protocol." + :version "21.1" + :group 'mail) + +(defcustom starttls-gnutls-program "gnutls-cli" + "Name of GnuTLS command line tool. +This program is used when GnuTLS is used, i.e. when +`starttls-use-gnutls' is non-nil." + :version "22.1" + :type 'string + :group 'starttls) + +(defcustom starttls-program "starttls" + "The program to run in a subprocess to open an TLSv1 connection. +This program is used when the `starttls' command is used, +i.e. when `starttls-use-gnutls' is nil." + :type 'string + :group 'starttls) + +(defcustom starttls-use-gnutls (not (executable-find starttls-program)) + "Whether to use GnuTLS instead of the `starttls' command." + :version "22.1" + :type 'boolean + :group 'starttls) + +(defcustom starttls-extra-args nil + "Extra arguments to `starttls-program'. +These apply when the `starttls' command is used, i.e. when +`starttls-use-gnutls' is nil." + :type '(repeat string) + :group 'starttls) + +(defcustom starttls-extra-arguments nil + "Extra arguments to `starttls-gnutls-program'. +These apply when GnuTLS is used, i.e. when `starttls-use-gnutls' is non-nil. + +For example, non-TLS compliant servers may require +\(\"--protocols\" \"ssl3\"). Invoke \"gnutls-cli --help\" to +find out which parameters are available." + :version "22.1" + :type '(repeat string) + :group 'starttls) + +(defcustom starttls-process-connection-type nil + "Value for `process-connection-type' to use when starting STARTTLS process." + :version "22.1" + :type 'boolean + :group 'starttls) + +(defcustom starttls-connect "- Simple Client Mode:\n\n" + "Regular expression indicating successful connection. +The default is what GnuTLS's \"gnutls-cli\" outputs." + ;; GnuTLS cli.c:main() prints this string when it is starting to run + ;; in the application read/write phase. If the logic, or the string + ;; itself, is modified, this must be updated. + :version "22.1" + :type 'regexp + :group 'starttls) + +(defcustom starttls-failure "\\*\\*\\* Handshake has failed" + "Regular expression indicating failed TLS handshake. +The default is what GnuTLS's \"gnutls-cli\" outputs." + ;; GnuTLS cli.c:do_handshake() prints this string on failure. If the + ;; logic, or the string itself, is modified, this must be updated. + :version "22.1" + :type 'regexp + :group 'starttls) + +(defcustom starttls-success "- Compression: " + "Regular expression indicating completed TLS handshakes. +The default is what GnuTLS's \"gnutls-cli\" outputs." + ;; GnuTLS cli.c:do_handshake() calls, on success, + ;; common.c:print_info(), that unconditionally print this string + ;; last. If that logic, or the string itself, is modified, this + ;; must be updated. + :version "22.1" + :type 'regexp + :group 'starttls) + +(defun starttls-negotiate-gnutls (process) + "Negotiate TLS on PROCESS opened by `open-starttls-stream'. +This should typically only be done once. It typically returns a +multi-line informational message with information about the +handshake, or nil on failure." + (let (buffer info old-max done-ok done-bad) + (if (null (setq buffer (process-buffer process))) + ;; XXX How to remove/extract the TLS negotiation junk? + (signal-process (process-id process) 'SIGALRM) + (with-current-buffer buffer + (save-excursion + (setq old-max (goto-char (point-max))) + (signal-process (process-id process) 'SIGALRM) + (while (and (processp process) + (eq (process-status process) 'run) + (save-excursion + (goto-char old-max) + (not (or (setq done-ok (re-search-forward + starttls-success nil t)) + (setq done-bad (re-search-forward + starttls-failure nil t)))))) + (accept-process-output process 1 100) + (sit-for 0.1)) + (setq info (buffer-substring-no-properties old-max (point-max))) + (delete-region old-max (point-max)) + (if (or (and done-ok (not done-bad)) + ;; Prevent mitm that fake success msg after failure msg. + (and done-ok done-bad (< done-ok done-bad))) + info + (message "STARTTLS negotiation failed: %s" info) + nil)))))) + +(defun starttls-negotiate (process) + (if starttls-use-gnutls + (starttls-negotiate-gnutls process) + (signal-process (process-id process) 'SIGALRM))) + +(defun starttls-open-stream-gnutls (name buffer host port) + (message "Opening STARTTLS connection to `%s:%s'..." host port) + (let* (done + (old-max (with-current-buffer buffer (point-max))) + (process-connection-type starttls-process-connection-type) + (process (apply #'start-process name buffer + starttls-gnutls-program "-s" host + "-p" (if (integerp port) + (int-to-string port) + port) + starttls-extra-arguments))) + (set-process-query-on-exit-flag process nil) + (while (and (processp process) + (eq (process-status process) 'run) + (with-current-buffer buffer + (goto-char old-max) + (not (setq done (re-search-forward + starttls-connect nil t))))) + (accept-process-output process 0 100) + (sit-for 0.1)) + (if done + (with-current-buffer buffer + (delete-region old-max done)) + (delete-process process) + (setq process nil)) + (message "Opening STARTTLS connection to `%s:%s'...%s" + host port (if done "done" "failed")) + process)) + +;;;###autoload +(defun starttls-open-stream (name buffer host port) + "Open a TLS connection for a port to a host. +Returns a subprocess object to represent the connection. +Input and output work as for subprocesses; `delete-process' closes it. +Args are NAME BUFFER HOST PORT. +NAME is name for process. It is modified if necessary to make it unique. +BUFFER is the buffer (or `buffer-name') to associate with the process. + Process output goes at end of that buffer, unless you specify + a filter function to handle the output. + BUFFER may be also nil, meaning that this process is not associated + with any buffer +Third arg is name of the host to connect to, or its IP address. +Fourth arg PORT is an integer specifying a port to connect to. +If `starttls-use-gnutls' is nil, this may also be a service name, but +GnuTLS requires a port number." + (if starttls-use-gnutls + (starttls-open-stream-gnutls name buffer host port) + (message "Opening STARTTLS connection to `%s:%s'" host (format "%s" port)) + (let* ((process-connection-type starttls-process-connection-type) + (process (apply #'start-process + name buffer starttls-program + host (format "%s" port) + starttls-extra-args))) + (set-process-query-on-exit-flag process nil) + process))) + +(defun starttls-available-p () + "Say whether the STARTTLS programs are available." + (and (not (memq system-type '(windows-nt ms-dos))) + (executable-find (if starttls-use-gnutls + starttls-gnutls-program + starttls-program)))) + +(defalias 'starttls-any-program-available 'starttls-available-p) +(make-obsolete 'starttls-any-program-available 'starttls-available-p + "2011-08-02") + +(provide 'starttls) + +;;; starttls.el ends here diff --cc lisp/obsolete/tls.el index fb7c20c843a,00000000000..d17ddad7ee5 mode 100644,000000..100644 --- a/lisp/obsolete/tls.el +++ b/lisp/obsolete/tls.el @@@ -1,302 -1,0 +1,302 @@@ +;;; tls.el --- TLS/SSL support via wrapper around GnuTLS + - ;; Copyright (C) 1996-1999, 2002-2018 Free Software Foundation, Inc. ++;; Copyright (C) 1996-1999, 2002-2019 Free Software Foundation, Inc. + +;; Author: Simon Josefsson +;; Keywords: comm, tls, gnutls, ssl +;; Obsolete-since: 27.1 + +;; This file is part of GNU Emacs. + +;; GNU Emacs is free software: you can redistribute it and/or modify +;; it under the terms of the GNU General Public License as published by +;; the Free Software Foundation, either version 3 of the License, or +;; (at your option) any later version. + +;; GNU Emacs is distributed in the hope that it will be useful, +;; but WITHOUT ANY WARRANTY; without even the implied warranty of +;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +;; GNU General Public License for more details. + +;; You should have received a copy of the GNU General Public License +;; along with GNU Emacs. If not, see . + +;;; Commentary: + +;; This package implements a simple wrapper around "gnutls-cli" to +;; make Emacs support TLS/SSL. +;; +;; Usage is the same as `open-network-stream', i.e.: +;; +;; (setq tmp (open-tls-stream "test" (current-buffer) "news.mozilla.org" 563)) +;; ... +;; # +;; (process-send-string tmp "mode reader\n") +;; 200 secnews.netscape.com Netscape-Collabra/3.52 03615 NNRP ready ... +;; nil +;; (process-send-string tmp "quit\n") +;; 205 +;; nil + +;; To use this package as a replacement for ssl.el by William M. Perry +;; , you need to evaluate the following: +;; +;; (defalias 'open-ssl-stream 'open-tls-stream) + +;;; Code: + +(require 'gnutls) + +(autoload 'format-spec "format-spec") +(autoload 'format-spec-make "format-spec") + +(defgroup tls nil + "Transport Layer Security (TLS) parameters." + :group 'comm) + +(defcustom tls-end-of-info + (concat + "\\(" + ;; `openssl s_client' regexp. See ssl/ssl_txt.c lines 219-220. + ;; According to apps/s_client.c line 1515 `---' is always the last + ;; line that is printed by s_client before the real data. + "^ Verify return code: .+\n---\n\\|" + ;; `gnutls' regexp. See src/cli.c lines 721-. + "^- Simple Client Mode:\n" + "\\(\n\\|" ; ignore blank lines + ;; According to GnuTLS v2.1.5 src/cli.c lines 640-650 and 705-715 + ;; in `main' the handshake will start after this message. If the + ;; handshake fails, the programs will abort. + "^\\*\\*\\* Starting TLS handshake\n\\)*" + "\\)") + "Regexp matching end of TLS client informational messages. +Client data stream begins after the last character this matches. +The default matches the output of \"gnutls-cli\" (version 2.0.1)." + :version "22.2" + :type 'regexp + :group 'tls) + +(defcustom tls-program + '("gnutls-cli --x509cafile %t -p %p %h" + "gnutls-cli --x509cafile %t -p %p %h --protocols ssl3") + "List of strings containing commands to start TLS stream to a host. +Each entry in the list is tried until a connection is successful. +%h is replaced with the server hostname, %p with the port to +connect to, and %t with a file name containing trusted certificates. +The program should read input on stdin and write output to stdout. + +See `tls-checktrust' on how to check trusted root certs. + +Also see `tls-success' for what the program should output after +successful negotiation." + :type + '(choice + (const :tag "Default list of commands" + ("gnutls-cli --x509cafile %t -p %p %h" + "gnutls-cli --x509cafile %t -p %p %h --protocols ssl3")) + (list :tag "Choose commands" + :value + ("gnutls-cli --x509cafile %t -p %p %h" + "gnutls-cli --x509cafile %t -p %p %h --protocols ssl3") + (set :inline t + ;; FIXME: add brief `:tag "..."' descriptions. + ;; (repeat :inline t :tag "Other" (string)) + ;; No trust check: + (const "gnutls-cli --insecure -p %p %h") + (const "gnutls-cli --insecure -p %p %h --protocols ssl3")) + (repeat :inline t :tag "Other" (string))) + (list :tag "List of commands" + (repeat :tag "Command" (string)))) + :version "26.1" ; remove s_client + :group 'tls) + +(defcustom tls-process-connection-type nil + "Value for `process-connection-type' to use when starting TLS process." + :version "22.1" + :type 'boolean + :group 'tls) + +(defcustom tls-success "- Handshake was completed\\|SSL handshake has read " + "Regular expression indicating completed TLS handshakes. +The default is what GnuTLS's \"gnutls-cli\" outputs." +;; or OpenSSL's \"openssl s_client\" + :version "22.1" + :type 'regexp + :group 'tls) + +(defcustom tls-checktrust nil + "Indicate if certificates should be checked against trusted root certs. +If this is `ask', the user can decide whether to accept an +untrusted certificate. You may have to adapt `tls-program' in +order to make this feature work properly, i.e., to ensure that +the external program knows about the root certificates you +consider trustworthy, e.g.: + +\(setq tls-program + \\='(\"gnutls-cli --x509cafile /etc/ssl/certs/ca-certificates.crt -p %p %h\" + \"gnutls-cli --x509cafile /etc/ssl/certs/ca-certificates.crt -p %p %h --protocols ssl3\"))" + :type '(choice (const :tag "Always" t) + (const :tag "Never" nil) + (const :tag "Ask" ask)) + :version "23.1" ;; No Gnus + :group 'tls) + +(defcustom tls-untrusted + "- Peer's certificate is NOT trusted\\|Verify return code: \\([^0] \\|.[^ ]\\)" + "Regular expression indicating failure of TLS certificate verification. +The default is what GnuTLS's \"gnutls-cli\" returns in the event of +unsuccessful verification." +;; or OpenSSL's \"openssl s_client\" + :type 'regexp + :version "23.1" ;; No Gnus + :group 'tls) + +(defcustom tls-hostmismatch + "# The hostname in the certificate does NOT match" + "Regular expression indicating a host name mismatch in certificate. +When the host name specified in the certificate doesn't match the +name of the host you are connecting to, gnutls-cli issues a +warning to this effect. There is no such feature in openssl. Set +this to nil if you want to ignore host name mismatches." + :type 'regexp + :version "23.1" ;; No Gnus + :group 'tls) + +(defcustom tls-certtool-program "certtool" + "Name of GnuTLS certtool. +Used by `tls-certificate-information'." + :version "22.1" + :type 'string + :group 'tls) + +(defalias 'tls-format-message + (if (fboundp 'format-message) 'format-message + ;; for Emacs < 25, and XEmacs, don't worry about quote translation. + 'format)) + +(defun tls-certificate-information (der) + "Parse X.509 certificate in DER format into an assoc list." + (let ((certificate (concat "-----BEGIN CERTIFICATE-----\n" + (base64-encode-string der) + "\n-----END CERTIFICATE-----\n")) + (exit-code 0)) + (with-current-buffer (get-buffer-create " *certtool*") + (erase-buffer) + (insert certificate) + (setq exit-code (condition-case () + (call-process-region (point-min) (point-max) + tls-certtool-program + t (list (current-buffer) nil) t + "--certificate-info") + (error -1))) + (if (/= exit-code 0) + nil + (let ((vals nil)) + (goto-char (point-min)) + (while (re-search-forward "^\\([^:]+\\): \\(.*\\)" nil t) + (push (cons (match-string 1) (match-string 2)) vals)) + (nreverse vals)))))) + +(defun open-tls-stream (name buffer host port) + "Open a TLS connection for a port to a host. +Returns a subprocess-object to represent the connection. +Input and output work as for subprocesses; `delete-process' closes it. +Args are NAME BUFFER HOST PORT. +NAME is name for process. It is modified if necessary to make it unique. +BUFFER is the buffer (or buffer name) to associate with the process. + Process output goes at end of that buffer, unless you specify + a filter function to handle the output. + BUFFER may be also nil, meaning that this process is not associated + with any buffer +Third arg is name of the host to connect to, or its IP address. +Fourth arg PORT is an integer specifying a port to connect to." + (let ((cmds tls-program) + (use-temp-buffer (null buffer)) + process cmd done) + (if use-temp-buffer + (setq buffer (generate-new-buffer " TLS")) + ;; BUFFER is a string but does not exist as a buffer object. + (unless (and (get-buffer buffer) + (buffer-name (get-buffer buffer))) + (generate-new-buffer buffer))) + (with-current-buffer buffer + (message "Opening TLS connection to `%s'..." host) + (while (and (not done) (setq cmd (pop cmds))) + (let ((process-connection-type tls-process-connection-type) + (formatted-cmd + (format-spec + cmd + (format-spec-make + ?t (car (gnutls-trustfiles)) + ?h host + ?p (if (integerp port) + (int-to-string port) + port))))) + (message "Opening TLS connection with `%s'..." formatted-cmd) + (setq process (start-process + name buffer shell-file-name shell-command-switch + formatted-cmd)) + (while (and process + (memq (process-status process) '(open run)) + (progn + (goto-char (point-min)) + (not (setq done (re-search-forward + tls-success nil t))))) + (unless (accept-process-output process 1) + (sit-for 1))) + (message "Opening TLS connection with `%s'...%s" formatted-cmd + (if done "done" "failed")) + (if (not done) + (delete-process process) + ;; advance point to after all informational messages that + ;; `openssl s_client' and `gnutls' print + (let ((start-of-data nil)) + (while + (not (setq start-of-data + ;; the string matching `tls-end-of-info' + ;; might come in separate chunks from + ;; `accept-process-output', so start the + ;; search where `tls-success' ended + (save-excursion + (if (re-search-forward tls-end-of-info nil t) + (match-end 0))))) + (accept-process-output process 1)) + (if start-of-data + ;; move point to start of client data + (goto-char start-of-data))) + (setq done process)))) + (when (and done + (or + (and tls-checktrust + (save-excursion + (goto-char (point-min)) + (re-search-forward tls-untrusted nil t)) + (or + (and (not (eq tls-checktrust 'ask)) + (message "The certificate presented by `%s' is \ +NOT trusted." host)) + (not (yes-or-no-p + (tls-format-message "\ +The certificate presented by `%s' is NOT trusted. Accept anyway? " host))))) + (and tls-hostmismatch + (save-excursion + (goto-char (point-min)) + (re-search-forward tls-hostmismatch nil t)) + (not (yes-or-no-p + (format "Host name in certificate doesn't \ +match `%s'. Connect anyway? " host)))))) + (setq done nil) + (delete-process process)) + ;; Delete all the informational messages that could confuse + ;; future uses of `buffer'. + (delete-region (point-min) (point))) + (message "Opening TLS connection to `%s'...%s" + host (if done "done" "failed")) + (when use-temp-buffer + (if done (set-process-buffer process nil)) + (kill-buffer buffer)) + done)) + +(provide 'tls) + +;;; tls.el ends here diff --cc lisp/play/bubbles.el index 6379c708be5,8d161775ffd..239fbe4e07c --- a/lisp/play/bubbles.el +++ b/lisp/play/bubbles.el @@@ -1,6 -1,6 +1,6 @@@ -;;; bubbles.el --- Puzzle game for Emacs +;;; bubbles.el --- Puzzle game for Emacs -*- lexical-binding:t -*- - ;; Copyright (C) 2007-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2007-2019 Free Software Foundation, Inc. ;; Author: Ulf Jasper ;; URL: http://ulf.epplejasper.de/ diff --cc lisp/play/gamegrid.el index 79825c4aaa5,7999194207b..430a2071bf8 --- a/lisp/play/gamegrid.el +++ b/lisp/play/gamegrid.el @@@ -1,6 -1,6 +1,6 @@@ -;;; gamegrid.el --- library for implementing grid-based games on Emacs +;;; gamegrid.el --- library for implementing grid-based games on Emacs -*- lexical-binding:t -*- - ;; Copyright (C) 1997-1998, 2001-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1997-1998, 2001-2019 Free Software Foundation, Inc. ;; Author: Glynn Clements ;; Version: 1.02 diff --cc lisp/printing.el index c1a73df14c1,d3240fe532c..27856eb09fc --- a/lisp/printing.el +++ b/lisp/printing.el @@@ -1,9 -1,9 +1,9 @@@ ;;; printing.el --- printing utilities - ;; Copyright (C) 2000-2001, 2003-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2000-2001, 2003-2019 Free Software Foundation, Inc. -;; Author: Vinicius Jose Latorre -;; Maintainer: Vinicius Jose Latorre +;; Author: Vinicius Jose Latorre +;; Maintainer: Vinicius Jose Latorre ;; Keywords: wp, print, PostScript ;; Version: 6.9.3 ;; X-URL: http://www.emacswiki.org/cgi-bin/wiki/ViniciusJoseLatorre diff --cc lisp/progmodes/cperl-mode.el index f9c390cd729,f25e24ba717..b83947f9034 --- a/lisp/progmodes/cperl-mode.el +++ b/lisp/progmodes/cperl-mode.el @@@ -1,6 -1,6 +1,6 @@@ -;;; cperl-mode.el --- Perl code editing commands for Emacs +;;; cperl-mode.el --- Perl code editing commands for Emacs -*- lexical-binding:t -*- - ;; Copyright (C) 1985-1987, 1991-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1985-1987, 1991-2019 Free Software Foundation, Inc. ;; Author: Ilya Zakharevich ;; Bob Olson diff --cc lisp/progmodes/ebnf-abn.el index 66f1d398df4,ff18c968bfa..b8441a79d98 --- a/lisp/progmodes/ebnf-abn.el +++ b/lisp/progmodes/ebnf-abn.el @@@ -1,9 -1,9 +1,9 @@@ ;;; ebnf-abn.el --- parser for ABNF (Augmented BNF) - ;; Copyright (C) 2001-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2001-2019 Free Software Foundation, Inc. -;; Author: Vinicius Jose Latorre -;; Maintainer: Vinicius Jose Latorre +;; Author: Vinicius Jose Latorre +;; Maintainer: Vinicius Jose Latorre ;; Keywords: wp, ebnf, PostScript ;; Old-Version: 1.2 ;; Package: ebnf2ps diff --cc lisp/progmodes/ebnf-bnf.el index 7defe9877b2,ed8419214cc..f28f49090d0 --- a/lisp/progmodes/ebnf-bnf.el +++ b/lisp/progmodes/ebnf-bnf.el @@@ -1,9 -1,9 +1,9 @@@ ;;; ebnf-bnf.el --- parser for EBNF - ;; Copyright (C) 1999-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1999-2019 Free Software Foundation, Inc. -;; Author: Vinicius Jose Latorre -;; Maintainer: Vinicius Jose Latorre +;; Author: Vinicius Jose Latorre +;; Maintainer: Vinicius Jose Latorre ;; Keywords: wp, ebnf, PostScript ;; Old-Version: 1.10 ;; Package: ebnf2ps diff --cc lisp/progmodes/ebnf-dtd.el index 2dec3f9159b,735ccbf593d..98284faf297 --- a/lisp/progmodes/ebnf-dtd.el +++ b/lisp/progmodes/ebnf-dtd.el @@@ -1,9 -1,9 +1,9 @@@ ;;; ebnf-dtd.el --- parser for DTD (Data Type Description for XML) - ;; Copyright (C) 2001-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2001-2019 Free Software Foundation, Inc. -;; Author: Vinicius Jose Latorre -;; Maintainer: Vinicius Jose Latorre +;; Author: Vinicius Jose Latorre +;; Maintainer: Vinicius Jose Latorre ;; Keywords: wp, ebnf, PostScript ;; Old-Version: 1.1 ;; Package: ebnf2ps diff --cc lisp/progmodes/ebnf-ebx.el index 0dc82fc3bff,9261280be16..5f9f9b2c197 --- a/lisp/progmodes/ebnf-ebx.el +++ b/lisp/progmodes/ebnf-ebx.el @@@ -1,9 -1,9 +1,9 @@@ ;;; ebnf-ebx.el --- parser for EBNF used to specify XML (EBNFX) - ;; Copyright (C) 2001-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2001-2019 Free Software Foundation, Inc. -;; Author: Vinicius Jose Latorre -;; Maintainer: Vinicius Jose Latorre +;; Author: Vinicius Jose Latorre +;; Maintainer: Vinicius Jose Latorre ;; Keywords: wp, ebnf, PostScript ;; Old-Version: 1.2 ;; Package: ebnf2ps diff --cc lisp/progmodes/ebnf-iso.el index 06aaf8a3f55,27457682e8b..88644bbe897 --- a/lisp/progmodes/ebnf-iso.el +++ b/lisp/progmodes/ebnf-iso.el @@@ -1,9 -1,9 +1,9 @@@ ;;; ebnf-iso.el --- parser for ISO EBNF - ;; Copyright (C) 1999-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1999-2019 Free Software Foundation, Inc. -;; Author: Vinicius Jose Latorre -;; Maintainer: Vinicius Jose Latorre +;; Author: Vinicius Jose Latorre +;; Maintainer: Vinicius Jose Latorre ;; Keywords: wp, ebnf, PostScript ;; Old-Version: 1.9 ;; Package: ebnf2ps diff --cc lisp/progmodes/ebnf-otz.el index 5857aa306ba,7b9bd33e4e1..cbc058e4767 --- a/lisp/progmodes/ebnf-otz.el +++ b/lisp/progmodes/ebnf-otz.el @@@ -1,9 -1,9 +1,9 @@@ ;;; ebnf-otz.el --- syntactic chart OpTimiZer - ;; Copyright (C) 1999-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1999-2019 Free Software Foundation, Inc. -;; Author: Vinicius Jose Latorre -;; Maintainer: Vinicius Jose Latorre +;; Author: Vinicius Jose Latorre +;; Maintainer: Vinicius Jose Latorre ;; Keywords: wp, ebnf, PostScript ;; Old-Version: 1.0 ;; Package: ebnf2ps diff --cc lisp/progmodes/ebnf-yac.el index eac0bfc878a,c667ed54d8e..3365f647c78 --- a/lisp/progmodes/ebnf-yac.el +++ b/lisp/progmodes/ebnf-yac.el @@@ -1,9 -1,9 +1,9 @@@ ;;; ebnf-yac.el --- parser for Yacc/Bison - ;; Copyright (C) 1999-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1999-2019 Free Software Foundation, Inc. -;; Author: Vinicius Jose Latorre -;; Maintainer: Vinicius Jose Latorre +;; Author: Vinicius Jose Latorre +;; Maintainer: Vinicius Jose Latorre ;; Keywords: wp, ebnf, PostScript ;; Old-Version: 1.4 ;; Package: ebnf2ps diff --cc lisp/progmodes/ebnf2ps.el index e29eb74a05b,770acc987f6..7e7efacf1e7 --- a/lisp/progmodes/ebnf2ps.el +++ b/lisp/progmodes/ebnf2ps.el @@@ -1,9 -1,9 +1,9 @@@ -;;; ebnf2ps.el --- translate an EBNF to a syntactic chart on PostScript +;;; ebnf2ps.el --- translate an EBNF to a syntactic chart on PostScript -*- lexical-binding:t -*- - ;; Copyright (C) 1999-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1999-2019 Free Software Foundation, Inc. -;; Author: Vinicius Jose Latorre -;; Maintainer: Vinicius Jose Latorre +;; Author: Vinicius Jose Latorre +;; Maintainer: Vinicius Jose Latorre ;; Keywords: wp, ebnf, PostScript ;; Version: 4.4 ;; X-URL: http://www.emacswiki.org/cgi-bin/wiki/ViniciusJoseLatorre diff --cc lisp/progmodes/flymake-proc.el index e969c5d9920,673f83e3396..7cdbb266fe1 --- a/lisp/progmodes/flymake-proc.el +++ b/lisp/progmodes/flymake-proc.el @@@ -1,10 -1,10 +1,10 @@@ ;;; flymake-proc.el --- Flymake backend for external tools -*- lexical-binding: t; -*- - ;; Copyright (C) 2003-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2003-2019 Free Software Foundation, Inc. ;; Author: Pavel Kobyakov -;; Maintainer: Leo Liu -;; Version: 0.3 +;; Maintainer: João Távora +;; Version: 1.0 ;; Keywords: c languages tools ;; This file is part of GNU Emacs. diff --cc lisp/progmodes/flymake.el index 7b100da42b8,16d97b6ccaf..50a9880a14e --- a/lisp/progmodes/flymake.el +++ b/lisp/progmodes/flymake.el @@@ -1,11 -1,10 +1,11 @@@ ;;; flymake.el --- A universal on-the-fly syntax checker -*- lexical-binding: t; -*- - ;; Copyright (C) 2003-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2003-2019 Free Software Foundation, Inc. ;; Author: Pavel Kobyakov -;; Maintainer: Leo Liu -;; Version: 0.3 +;; Maintainer: João Távora +;; Version: 1.0.3 +;; Package-Requires: ((emacs "26.1")) ;; Keywords: c languages tools ;; This file is part of GNU Emacs. diff --cc lisp/progmodes/idlw-shell.el index 46e2ecaa397,883616cd285..a3b079830b2 --- a/lisp/progmodes/idlw-shell.el +++ b/lisp/progmodes/idlw-shell.el @@@ -1,6 -1,6 +1,6 @@@ -;; idlw-shell.el --- run IDL as an inferior process of Emacs. +;; idlw-shell.el --- run IDL as an inferior process of Emacs. -*- lexical-binding:t -*- - ;; Copyright (C) 1999-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1999-2019 Free Software Foundation, Inc. ;; Authors: J.D. Smith ;; Carsten Dominik diff --cc lisp/ps-def.el index d0cd7625a41,e3c9504a01f..20631338902 --- a/lisp/ps-def.el +++ b/lisp/ps-def.el @@@ -1,11 -1,11 +1,11 @@@ ;;; ps-def.el --- XEmacs and Emacs definitions for ps-print -*- lexical-binding: t -*- - ;; Copyright (C) 2007-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2007-2019 Free Software Foundation, Inc. -;; Author: Vinicius Jose Latorre +;; Author: Vinicius Jose Latorre ;; Kenichi Handa (multi-byte characters) ;; Maintainer: Kenichi Handa (multi-byte characters) -;; Vinicius Jose Latorre +;; Vinicius Jose Latorre ;; Keywords: wp, print, PostScript ;; X-URL: http://www.emacswiki.org/cgi-bin/wiki/ViniciusJoseLatorre ;; Package: ps-print diff --cc lisp/ps-mule.el index 2658aec5e6d,9a32b05526e..c88d6d26797 --- a/lisp/ps-mule.el +++ b/lisp/ps-mule.el @@@ -1,11 -1,11 +1,11 @@@ ;;; ps-mule.el --- provide multi-byte character facility to ps-print - ;; Copyright (C) 1998-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1998-2019 Free Software Foundation, Inc. -;; Author: Vinicius Jose Latorre +;; Author: Vinicius Jose Latorre ;; Kenichi Handa (multi-byte characters) ;; Maintainer: Kenichi Handa (multi-byte characters) -;; Vinicius Jose Latorre +;; Vinicius Jose Latorre ;; Keywords: wp, print, PostScript, multibyte, mule ;; Package: ps-print diff --cc lisp/rtree.el index fe24cd18719,ff160f207b2..9a0a649abf1 --- a/lisp/rtree.el +++ b/lisp/rtree.el @@@ -1,6 -1,6 +1,6 @@@ -;;; rtree.el --- functions for manipulating range trees +;;; rtree.el --- functions for manipulating range trees -*- lexical-binding:t -*- - ;; Copyright (C) 2010-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2010-2019 Free Software Foundation, Inc. ;; Author: Lars Magne Ingebrigtsen diff --cc lisp/term.el index 9f8f1f703a6,cbef68dc0ac..f49777f94c2 --- a/lisp/term.el +++ b/lisp/term.el @@@ -1,6 -1,6 +1,6 @@@ -;;; term.el --- general command interpreter in a window stuff +;;; term.el --- general command interpreter in a window stuff -*- lexical-binding: t -*- - ;; Copyright (C) 1988, 1990, 1992, 1994-1995, 2001-2018 Free Software + ;; Copyright (C) 1988, 1990, 1992, 1994-1995, 2001-2019 Free Software ;; Foundation, Inc. ;; Author: Per Bothner diff --cc lisp/textmodes/page-ext.el index 92fce4d364b,a9199fffe0d..23e2d28f77e --- a/lisp/textmodes/page-ext.el +++ b/lisp/textmodes/page-ext.el @@@ -1,6 -1,6 +1,6 @@@ -;;; page-ext.el --- extended page handling commands +;;; page-ext.el --- extended page handling commands -*- lexical-binding:t -*- - ;; Copyright (C) 1990-1991, 1993-1994, 2001-2018 Free Software + ;; Copyright (C) 1990-1991, 1993-1994, 2001-2019 Free Software ;; Foundation, Inc. ;; Author: Robert J. Chassell diff --cc lisp/url/url-file.el index 02542ccbccc,23fc97828ff..b953ce76940 --- a/lisp/url/url-file.el +++ b/lisp/url/url-file.el @@@ -1,6 -1,6 +1,6 @@@ -;;; url-file.el --- File retrieval code +;;; url-file.el --- File retrieval code -*- lexical-binding:t -*- - ;; Copyright (C) 1996-1999, 2004-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1996-1999, 2004-2019 Free Software Foundation, Inc. ;; Keywords: comm, data, processes diff --cc lisp/whitespace.el index af06f7ccb14,79bc5c88348..2463e655b19 --- a/lisp/whitespace.el +++ b/lisp/whitespace.el @@@ -1,9 -1,9 +1,9 @@@ ;;; whitespace.el --- minor mode to visualize TAB, (HARD) SPACE, NEWLINE -*- lexical-binding: t -*- - ;; Copyright (C) 2000-2018 Free Software Foundation, Inc. + ;; Copyright (C) 2000-2019 Free Software Foundation, Inc. -;; Author: Vinicius Jose Latorre -;; Maintainer: Vinicius Jose Latorre +;; Author: Vinicius Jose Latorre +;; Maintainer: Vinicius Jose Latorre ;; Keywords: data, wp ;; Version: 13.2.2 ;; X-URL: http://www.emacswiki.org/cgi-bin/wiki/ViniciusJoseLatorre diff --cc lisp/windmove.el index 75b9e0cf599,cf7b65a6ed7..65270d9bbe4 --- a/lisp/windmove.el +++ b/lisp/windmove.el @@@ -1,6 -1,6 +1,6 @@@ -;;; windmove.el --- directional window-selection routines +;;; windmove.el --- directional window-selection routines -*- lexical-binding:t -*- ;; - ;; Copyright (C) 1998-2018 Free Software Foundation, Inc. + ;; Copyright (C) 1998-2019 Free Software Foundation, Inc. ;; ;; Author: Hovav Shacham (hovav@cs.stanford.edu) ;; Created: 17 October 1998 diff --cc m4/acl.m4 index b64aa849c88,9ee7bbbde47..5234a80f93f --- a/m4/acl.m4 +++ b/m4/acl.m4 @@@ -1,7 -1,7 +1,7 @@@ # acl.m4 - check for access control list (ACL) primitives -# serial 22 +# serial 23 - # Copyright (C) 2002, 2004-2018 Free Software Foundation, Inc. + # Copyright (C) 2002, 2004-2019 Free Software Foundation, Inc. # This file is free software; the Free Software Foundation # gives unlimited permission to copy and/or distribute it, # with or without modifications, as long as this notice is preserved. diff --cc m4/c-strtod.m4 index 9282cee58e9,5b6e8297310..1694a2396c5 --- a/m4/c-strtod.m4 +++ b/m4/c-strtod.m4 @@@ -1,6 -1,6 +1,6 @@@ -# c-strtod.m4 serial 15 +# c-strtod.m4 serial 16 - # Copyright (C) 2004-2006, 2009-2018 Free Software Foundation, Inc. + # Copyright (C) 2004-2006, 2009-2019 Free Software Foundation, Inc. # This file is free software; the Free Software Foundation # gives unlimited permission to copy and/or distribute it, # with or without modifications, as long as this notice is preserved. diff --cc m4/extensions.m4 index 71a854f8bfa,1a8e899d369..fd1ce817f06 --- a/m4/extensions.m4 +++ b/m4/extensions.m4 @@@ -1,7 -1,7 +1,7 @@@ -# serial 17 -*- Autoconf -*- +# serial 18 -*- Autoconf -*- # Enable extensions on systems that normally disable them. - # Copyright (C) 2003, 2006-2018 Free Software Foundation, Inc. + # Copyright (C) 2003, 2006-2019 Free Software Foundation, Inc. # This file is free software; the Free Software Foundation # gives unlimited permission to copy and/or distribute it, # with or without modifications, as long as this notice is preserved. diff --cc m4/lstat.m4 index 3694e4ceaf7,ef6b457a94d..ace163867e2 --- a/m4/lstat.m4 +++ b/m4/lstat.m4 @@@ -1,6 -1,6 +1,6 @@@ -# serial 31 +# serial 32 - # Copyright (C) 1997-2001, 2003-2018 Free Software Foundation, Inc. + # Copyright (C) 1997-2001, 2003-2019 Free Software Foundation, Inc. # # This file is free software; the Free Software Foundation # gives unlimited permission to copy and/or distribute it, diff --cc m4/symlink.m4 index afaa941744e,277614e0443..508e94dd410 --- a/m4/symlink.m4 +++ b/m4/symlink.m4 @@@ -1,7 -1,7 +1,7 @@@ -# serial 7 +# serial 8 # See if we need to provide symlink replacement. - dnl Copyright (C) 2009-2018 Free Software Foundation, Inc. + dnl Copyright (C) 2009-2019 Free Software Foundation, Inc. dnl This file is free software; the Free Software Foundation dnl gives unlimited permission to copy and/or distribute it, dnl with or without modifications, as long as this notice is preserved. diff --cc nt/README.W32 index f0147b4c68f,d9f7e23a184..0fac1c32988 --- a/nt/README.W32 +++ b/nt/README.W32 @@@ -1,7 -1,7 +1,7 @@@ - Copyright (C) 2001-2018 Free Software Foundation, Inc. + Copyright (C) 2001-2019 Free Software Foundation, Inc. See the end of the file for license conditions. - Emacs version 26.1.90 for MS-Windows + Emacs version 27.0.50 for MS-Windows This README file describes how to set up and run a precompiled distribution of the latest version of GNU Emacs for MS-Windows. You diff --cc src/bidi.c index a53a2295c09,216279cbc3a..a62b888432e --- a/src/bidi.c +++ b/src/bidi.c @@@ -1,8 -1,6 +1,8 @@@ /* Low-level bidirectional buffer/string-scanning functions for GNU Emacs. - Copyright (C) 2000-2001, 2004-2005, 2009-2019 Free Software - Foundation, Inc. + - Copyright (C) 2000-2001, 2004-2005, 2009-2018 Free Software Foundation, Inc. ++Copyright (C) 2000-2001, 2004-2005, 2009-2019 Free Software Foundation, Inc. + +Author: Eli Zaretskii This file is part of GNU Emacs. diff --cc src/regex-emacs.c index 5cb7bba158e,00000000000..b667a43a37f mode 100644,000000..100644 --- a/src/regex-emacs.c +++ b/src/regex-emacs.c @@@ -1,5094 -1,0 +1,5094 @@@ +/* Emacs regular expression matching and search + - Copyright (C) 1993-2018 Free Software Foundation, Inc. ++ Copyright (C) 1993-2019 Free Software Foundation, Inc. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 3, or (at your option) + any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see . */ + +/* TODO: + - structure the opcode space into opcode+flag. + - replace (succeed_n + jump_n + set_number_at) with something that doesn't + need to modify the compiled regexp so that re_search can be reentrant. + - get rid of on_failure_jump_smart by doing the optimization in re_comp + rather than at run-time, so that re_search can be reentrant. +*/ + +#include + +#include "regex-emacs.h" + +#include + +#include "character.h" +#include "buffer.h" +#include "syntax.h" +#include "category.h" + +/* Maximum number of duplicates an interval can allow. Some systems + define this in other header files, but we want our value, so remove + any previous define. Repeat counts are stored in opcodes as 2-byte + unsigned integers. */ +#ifdef RE_DUP_MAX +# undef RE_DUP_MAX +#endif +#define RE_DUP_MAX (0xffff) + +/* Make syntax table lookup grant data in gl_state. */ +#define SYNTAX(c) syntax_property (c, 1) + +/* Convert the pointer to the char to BEG-based offset from the start. */ +#define PTR_TO_OFFSET(d) POS_AS_IN_BUFFER (POINTER_TO_OFFSET (d)) +/* Strings are 0-indexed, buffers are 1-indexed; pun on the boolean + result to get the right base index. */ +#define POS_AS_IN_BUFFER(p) \ + ((p) + (NILP (gl_state.object) || BUFFERP (gl_state.object))) + +#define RE_MULTIBYTE_P(bufp) ((bufp)->multibyte) +#define RE_TARGET_MULTIBYTE_P(bufp) ((bufp)->target_multibyte) +#define RE_STRING_CHAR(p, multibyte) \ + (multibyte ? STRING_CHAR (p) : *(p)) +#define RE_STRING_CHAR_AND_LENGTH(p, len, multibyte) \ + (multibyte ? STRING_CHAR_AND_LENGTH (p, len) : ((len) = 1, *(p))) + +#define RE_CHAR_TO_MULTIBYTE(c) UNIBYTE_TO_CHAR (c) + +#define RE_CHAR_TO_UNIBYTE(c) CHAR_TO_BYTE_SAFE (c) + +/* Set C a (possibly converted to multibyte) character before P. P + points into a string which is the virtual concatenation of STR1 + (which ends at END1) or STR2 (which ends at END2). */ +#define GET_CHAR_BEFORE_2(c, p, str1, end1, str2, end2) \ + do { \ + if (target_multibyte) \ + { \ + re_char *dtemp = (p) == (str2) ? (end1) : (p); \ + re_char *dlimit = (p) > (str2) && (p) <= (end2) ? (str2) : (str1); \ + while (dtemp-- > dlimit && !CHAR_HEAD_P (*dtemp)) \ + continue; \ + c = STRING_CHAR (dtemp); \ + } \ + else \ + { \ + (c = ((p) == (str2) ? (end1) : (p))[-1]); \ + (c) = RE_CHAR_TO_MULTIBYTE (c); \ + } \ + } while (false) + +/* Set C a (possibly converted to multibyte) character at P, and set + LEN to the byte length of that character. */ +#define GET_CHAR_AFTER(c, p, len) \ + do { \ + if (target_multibyte) \ + (c) = STRING_CHAR_AND_LENGTH (p, len); \ + else \ + { \ + (c) = *p; \ + len = 1; \ + (c) = RE_CHAR_TO_MULTIBYTE (c); \ + } \ + } while (false) + +/* 1 if C is an ASCII character. */ +#define IS_REAL_ASCII(c) ((c) < 0200) + +/* 1 if C is a unibyte character. */ +#define ISUNIBYTE(c) (SINGLE_BYTE_CHAR_P ((c))) + +/* The Emacs definitions should not be directly affected by locales. */ + +/* In Emacs, these are only used for single-byte characters. */ +#define ISDIGIT(c) ((c) >= '0' && (c) <= '9') +#define ISCNTRL(c) ((c) < ' ') +#define ISXDIGIT(c) (0 <= char_hexdigit (c)) + +/* The rest must handle multibyte characters. */ + +#define ISBLANK(c) (IS_REAL_ASCII (c) \ + ? ((c) == ' ' || (c) == '\t') \ + : blankp (c)) + +#define ISGRAPH(c) (SINGLE_BYTE_CHAR_P (c) \ + ? (c) > ' ' && !((c) >= 0177 && (c) <= 0240) \ + : graphicp (c)) + +#define ISPRINT(c) (SINGLE_BYTE_CHAR_P (c) \ + ? (c) >= ' ' && !((c) >= 0177 && (c) <= 0237) \ + : printablep (c)) + +#define ISALNUM(c) (IS_REAL_ASCII (c) \ + ? (((c) >= 'a' && (c) <= 'z') \ + || ((c) >= 'A' && (c) <= 'Z') \ + || ((c) >= '0' && (c) <= '9')) \ + : alphanumericp (c)) + +#define ISALPHA(c) (IS_REAL_ASCII (c) \ + ? (((c) >= 'a' && (c) <= 'z') \ + || ((c) >= 'A' && (c) <= 'Z')) \ + : alphabeticp (c)) + +#define ISLOWER(c) lowercasep (c) + +#define ISPUNCT(c) (IS_REAL_ASCII (c) \ + ? ((c) > ' ' && (c) < 0177 \ + && !(((c) >= 'a' && (c) <= 'z') \ + || ((c) >= 'A' && (c) <= 'Z') \ + || ((c) >= '0' && (c) <= '9'))) \ + : SYNTAX (c) != Sword) + +#define ISSPACE(c) (SYNTAX (c) == Swhitespace) + +#define ISUPPER(c) uppercasep (c) + +#define ISWORD(c) (SYNTAX (c) == Sword) + +#define SIGN_EXTEND_CHAR(c) ((signed char) (c)) + +/* Use alloca instead of malloc. This is because using malloc in + re_search* or re_match* could cause memory leaks when C-g is used + in Emacs (note that SAFE_ALLOCA could also call malloc, but does so + via 'record_xmalloc' which uses 'unwind_protect' to ensure the + memory is freed even in case of non-local exits); also, malloc is + slower and causes storage fragmentation. On the other hand, malloc + is more portable, and easier to debug. + + Because we sometimes use alloca, some routines have to be macros, + not functions -- 'alloca'-allocated space disappears at the end of the + function it is called in. */ + +/* This may be adjusted in main(), if the stack is successfully grown. */ +ptrdiff_t emacs_re_safe_alloca = MAX_ALLOCA; +/* Like USE_SAFE_ALLOCA, but use emacs_re_safe_alloca. */ +#define REGEX_USE_SAFE_ALLOCA \ + USE_SAFE_ALLOCA; sa_avail = emacs_re_safe_alloca + +/* Assumes a 'char *destination' variable. */ +#define REGEX_REALLOCATE(source, osize, nsize) \ + (destination = SAFE_ALLOCA (nsize), \ + memcpy (destination, source, osize)) + +/* True if 'size1' is non-NULL and PTR is pointing anywhere inside + 'string1' or just past its end. This works if PTR is NULL, which is + a good thing. */ +#define FIRST_STRING_P(ptr) \ + (size1 && string1 <= (ptr) && (ptr) <= string1 + size1) + +/* (Re)Allocate N items of type T using malloc, or fail. */ +#define TALLOC(n, t) ((t *) xmalloc ((n) * sizeof (t))) +#define RETALLOC(addr, n, t) ((addr) = (t *) xrealloc (addr, (n) * sizeof (t))) + +#define BYTEWIDTH 8 /* In bits. */ + +/* Type of source-pattern and string chars. */ +typedef const unsigned char re_char; + +static void re_compile_fastmap (struct re_pattern_buffer *); +static ptrdiff_t re_match_2_internal (struct re_pattern_buffer *bufp, + re_char *string1, size_t size1, + re_char *string2, size_t size2, + ptrdiff_t pos, + struct re_registers *regs, + ptrdiff_t stop); + +/* These are the command codes that appear in compiled regular + expressions. Some opcodes are followed by argument bytes. A + command code can specify any interpretation whatsoever for its + arguments. Zero bytes may appear in the compiled regular expression. */ + +typedef enum +{ + no_op = 0, + + /* Succeed right away--no more backtracking. */ + succeed, + + /* Followed by one byte giving n, then by n literal bytes. */ + exactn, + + /* Matches any (more or less) character. */ + anychar, + + /* Matches any one char belonging to specified set. First + following byte is number of bitmap bytes. Then come bytes + for a bitmap saying which chars are in. Bits in each byte + are ordered low-bit-first. A character is in the set if its + bit is 1. A character too large to have a bit in the map is + automatically not in the set. + + If the length byte has the 0x80 bit set, then that stuff + is followed by a range table: + 2 bytes of flags for character sets (low 8 bits, high 8 bits) + See RANGE_TABLE_WORK_BITS below. + 2 bytes, the number of pairs that follow (upto 32767) + pairs, each 2 multibyte characters, + each multibyte character represented as 3 bytes. */ + charset, + + /* Same parameters as charset, but match any character that is + not one of those specified. */ + charset_not, + + /* Start remembering the text that is matched, for storing in a + register. Followed by one byte with the register number, in + the range 0 to one less than the pattern buffer's re_nsub + field. */ + start_memory, + + /* Stop remembering the text that is matched and store it in a + memory register. Followed by one byte with the register + number, in the range 0 to one less than 're_nsub' in the + pattern buffer. */ + stop_memory, + + /* Match a duplicate of something remembered. Followed by one + byte containing the register number. */ + duplicate, + + /* Fail unless at beginning of line. */ + begline, + + /* Fail unless at end of line. */ + endline, + + /* Succeeds if at beginning of buffer. */ + begbuf, + + /* Analogously, for end of buffer/string. */ + endbuf, + + /* Followed by two byte relative address to which to jump. */ + jump, + + /* Followed by two-byte relative address of place to resume at + in case of failure. */ + on_failure_jump, + + /* Like on_failure_jump, but pushes a placeholder instead of the + current string position when executed. */ + on_failure_keep_string_jump, + + /* Just like 'on_failure_jump', except that it checks that we + don't get stuck in an infinite loop (matching an empty string + indefinitely). */ + on_failure_jump_loop, + + /* Just like 'on_failure_jump_loop', except that it checks for + a different kind of loop (the kind that shows up with non-greedy + operators). This operation has to be immediately preceded + by a 'no_op'. */ + on_failure_jump_nastyloop, + + /* A smart 'on_failure_jump' used for greedy * and + operators. + It analyzes the loop before which it is put and if the + loop does not require backtracking, it changes itself to + 'on_failure_keep_string_jump' and short-circuits the loop, + else it just defaults to changing itself into 'on_failure_jump'. + It assumes that it is pointing to just past a 'jump'. */ + on_failure_jump_smart, + + /* Followed by two-byte relative address and two-byte number n. + After matching N times, jump to the address upon failure. + Does not work if N starts at 0: use on_failure_jump_loop + instead. */ + succeed_n, + + /* Followed by two-byte relative address, and two-byte number n. + Jump to the address N times, then fail. */ + jump_n, + + /* Set the following two-byte relative address to the + subsequent two-byte number. The address *includes* the two + bytes of number. */ + set_number_at, + + wordbeg, /* Succeeds if at word beginning. */ + wordend, /* Succeeds if at word end. */ + + wordbound, /* Succeeds if at a word boundary. */ + notwordbound, /* Succeeds if not at a word boundary. */ + + symbeg, /* Succeeds if at symbol beginning. */ + symend, /* Succeeds if at symbol end. */ + + /* Matches any character whose syntax is specified. Followed by + a byte which contains a syntax code, e.g., Sword. */ + syntaxspec, + + /* Matches any character whose syntax is not that specified. */ + notsyntaxspec, + + at_dot, /* Succeeds if at point. */ + + /* Matches any character whose category-set contains the specified + category. The operator is followed by a byte which contains a + category code (mnemonic ASCII character). */ + categoryspec, + + /* Matches any character whose category-set does not contain the + specified category. The operator is followed by a byte which + contains the category code (mnemonic ASCII character). */ + notcategoryspec +} re_opcode_t; + +/* Common operations on the compiled pattern. */ + +/* Store NUMBER in two contiguous bytes starting at DESTINATION. */ + +#define STORE_NUMBER(destination, number) \ + do { \ + (destination)[0] = (number) & 0377; \ + (destination)[1] = (number) >> 8; \ + } while (false) + +/* Same as STORE_NUMBER, except increment DESTINATION to + the byte after where the number is stored. Therefore, DESTINATION + must be an lvalue. */ + +#define STORE_NUMBER_AND_INCR(destination, number) \ + do { \ + STORE_NUMBER (destination, number); \ + (destination) += 2; \ + } while (false) + +/* Put into DESTINATION a number stored in two contiguous bytes starting + at SOURCE. */ + +#define EXTRACT_NUMBER(destination, source) \ + ((destination) = extract_number (source)) + +static int +extract_number (re_char *source) +{ + unsigned leading_byte = SIGN_EXTEND_CHAR (source[1]); + return (leading_byte << 8) + source[0]; +} + +/* Same as EXTRACT_NUMBER, except increment SOURCE to after the number. + SOURCE must be an lvalue. */ + +#define EXTRACT_NUMBER_AND_INCR(destination, source) \ + ((destination) = extract_number_and_incr (&source)) + +static int +extract_number_and_incr (re_char **source) +{ + int num = extract_number (*source); + *source += 2; + return num; +} + +/* Store a multibyte character in three contiguous bytes starting + DESTINATION, and increment DESTINATION to the byte after where the + character is stored. Therefore, DESTINATION must be an lvalue. */ + +#define STORE_CHARACTER_AND_INCR(destination, character) \ + do { \ + (destination)[0] = (character) & 0377; \ + (destination)[1] = ((character) >> 8) & 0377; \ + (destination)[2] = (character) >> 16; \ + (destination) += 3; \ + } while (false) + +/* Put into DESTINATION a character stored in three contiguous bytes + starting at SOURCE. */ + +#define EXTRACT_CHARACTER(destination, source) \ + do { \ + (destination) = ((source)[0] \ + | ((source)[1] << 8) \ + | ((source)[2] << 16)); \ + } while (false) + + +/* Macros for charset. */ + +/* Size of bitmap of charset P in bytes. P is a start of charset, + i.e. *P is (re_opcode_t) charset or (re_opcode_t) charset_not. */ +#define CHARSET_BITMAP_SIZE(p) ((p)[1] & 0x7F) + +/* Nonzero if charset P has range table. */ +#define CHARSET_RANGE_TABLE_EXISTS_P(p) ((p)[1] & 0x80) + +/* Return the address of range table of charset P. But not the start + of table itself, but the before where the number of ranges is + stored. '2 +' means to skip re_opcode_t and size of bitmap, + and the 2 bytes of flags at the start of the range table. */ +#define CHARSET_RANGE_TABLE(p) (&(p)[4 + CHARSET_BITMAP_SIZE (p)]) + +/* Extract the bit flags that start a range table. */ +#define CHARSET_RANGE_TABLE_BITS(p) \ + ((p)[2 + CHARSET_BITMAP_SIZE (p)] \ + + (p)[3 + CHARSET_BITMAP_SIZE (p)] * 0x100) + +/* Return the address of end of RANGE_TABLE. COUNT is number of + ranges (which is a pair of (start, end)) in the RANGE_TABLE. '* 2' + is start of range and end of range. '* 3' is size of each start + and end. */ +#define CHARSET_RANGE_TABLE_END(range_table, count) \ + ((range_table) + (count) * 2 * 3) + +/* If REGEX_EMACS_DEBUG is defined, print many voluminous messages + (if the variable regex_emacs_debug is positive). */ + +#ifdef REGEX_EMACS_DEBUG + +/* Use standard I/O for debugging. */ +# include + +static int regex_emacs_debug = -100000; + +# define DEBUG_STATEMENT(e) e +# define DEBUG_PRINT(...) if (regex_emacs_debug > 0) printf (__VA_ARGS__) +# define DEBUG_COMPILES_ARGUMENTS +# define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \ + if (regex_emacs_debug > 0) print_partial_compiled_pattern (s, e) +# define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \ + if (regex_emacs_debug > 0) print_double_string (w, s1, sz1, s2, sz2) + + +/* Print the fastmap in human-readable form. */ + +static void +print_fastmap (char *fastmap) +{ + unsigned was_a_range = 0; + unsigned i = 0; + + while (i < (1 << BYTEWIDTH)) + { + if (fastmap[i++]) + { + was_a_range = 0; + putchar (i - 1); + while (i < (1 << BYTEWIDTH) && fastmap[i]) + { + was_a_range = 1; + i++; + } + if (was_a_range) + { + printf ("-"); + putchar (i - 1); + } + } + } + putchar ('\n'); +} + + +/* Print a compiled pattern string in human-readable form, starting at + the START pointer into it and ending just before the pointer END. */ + +static void +print_partial_compiled_pattern (re_char *start, re_char *end) +{ + int mcnt, mcnt2; + re_char *p = start; + re_char *pend = end; + + if (start == NULL) + { + fprintf (stderr, "(null)\n"); + return; + } + + /* Loop over pattern commands. */ + while (p < pend) + { + fprintf (stderr, "%td:\t", p - start); + + switch ((re_opcode_t) *p++) + { + case no_op: + fprintf (stderr, "/no_op"); + break; + + case succeed: + fprintf (stderr, "/succeed"); + break; + + case exactn: + mcnt = *p++; + fprintf (stderr, "/exactn/%d", mcnt); + do + { + fprintf (stderr, "/%c", *p++); + } + while (--mcnt); + break; + + case start_memory: + fprintf (stderr, "/start_memory/%d", *p++); + break; + + case stop_memory: + fprintf (stderr, "/stop_memory/%d", *p++); + break; + + case duplicate: + fprintf (stderr, "/duplicate/%d", *p++); + break; + + case anychar: + fprintf (stderr, "/anychar"); + break; + + case charset: + case charset_not: + { + register int c, last = -100; + register int in_range = 0; + int length = CHARSET_BITMAP_SIZE (p - 1); + int has_range_table = CHARSET_RANGE_TABLE_EXISTS_P (p - 1); + + fprintf (stderr, "/charset [%s", + (re_opcode_t) *(p - 1) == charset_not ? "^" : ""); + + if (p + *p >= pend) + fprintf (stderr, " !extends past end of pattern! "); + + for (c = 0; c < 256; c++) + if (c / 8 < length + && (p[1 + (c/8)] & (1 << (c % 8)))) + { + /* Are we starting a range? */ + if (last + 1 == c && ! in_range) + { + fprintf (stderr, "-"); + in_range = 1; + } + /* Have we broken a range? */ + else if (last + 1 != c && in_range) + { + fprintf (stderr, "%c", last); + in_range = 0; + } + + if (! in_range) + fprintf (stderr, "%c", c); + + last = c; + } + + if (in_range) + fprintf (stderr, "%c", last); + + fprintf (stderr, "]"); + + p += 1 + length; + + if (has_range_table) + { + int count; + fprintf (stderr, "has-range-table"); + + /* ??? Should print the range table; for now, just skip it. */ + p += 2; /* skip range table bits */ + EXTRACT_NUMBER_AND_INCR (count, p); + p = CHARSET_RANGE_TABLE_END (p, count); + } + } + break; + + case begline: + fprintf (stderr, "/begline"); + break; + + case endline: + fprintf (stderr, "/endline"); + break; + + case on_failure_jump: + EXTRACT_NUMBER_AND_INCR (mcnt, p); + fprintf (stderr, "/on_failure_jump to %td", p + mcnt - start); + break; + + case on_failure_keep_string_jump: + EXTRACT_NUMBER_AND_INCR (mcnt, p); + fprintf (stderr, "/on_failure_keep_string_jump to %td", + p + mcnt - start); + break; + + case on_failure_jump_nastyloop: + EXTRACT_NUMBER_AND_INCR (mcnt, p); + fprintf (stderr, "/on_failure_jump_nastyloop to %td", + p + mcnt - start); + break; + + case on_failure_jump_loop: + EXTRACT_NUMBER_AND_INCR (mcnt, p); + fprintf (stderr, "/on_failure_jump_loop to %td", + p + mcnt - start); + break; + + case on_failure_jump_smart: + EXTRACT_NUMBER_AND_INCR (mcnt, p); + fprintf (stderr, "/on_failure_jump_smart to %td", + p + mcnt - start); + break; + + case jump: + EXTRACT_NUMBER_AND_INCR (mcnt, p); + fprintf (stderr, "/jump to %td", p + mcnt - start); + break; + + case succeed_n: + EXTRACT_NUMBER_AND_INCR (mcnt, p); + EXTRACT_NUMBER_AND_INCR (mcnt2, p); + fprintf (stderr, "/succeed_n to %td, %d times", + p - 2 + mcnt - start, mcnt2); + break; + + case jump_n: + EXTRACT_NUMBER_AND_INCR (mcnt, p); + EXTRACT_NUMBER_AND_INCR (mcnt2, p); + fprintf (stderr, "/jump_n to %td, %d times", + p - 2 + mcnt - start, mcnt2); + break; + + case set_number_at: + EXTRACT_NUMBER_AND_INCR (mcnt, p); + EXTRACT_NUMBER_AND_INCR (mcnt2, p); + fprintf (stderr, "/set_number_at location %td to %d", + p - 2 + mcnt - start, mcnt2); + break; + + case wordbound: + fprintf (stderr, "/wordbound"); + break; + + case notwordbound: + fprintf (stderr, "/notwordbound"); + break; + + case wordbeg: + fprintf (stderr, "/wordbeg"); + break; + + case wordend: + fprintf (stderr, "/wordend"); + break; + + case symbeg: + fprintf (stderr, "/symbeg"); + break; + + case symend: + fprintf (stderr, "/symend"); + break; + + case syntaxspec: + fprintf (stderr, "/syntaxspec"); + mcnt = *p++; + fprintf (stderr, "/%d", mcnt); + break; + + case notsyntaxspec: + fprintf (stderr, "/notsyntaxspec"); + mcnt = *p++; + fprintf (stderr, "/%d", mcnt); + break; + + case at_dot: + fprintf (stderr, "/at_dot"); + break; + + case categoryspec: + fprintf (stderr, "/categoryspec"); + mcnt = *p++; + fprintf (stderr, "/%d", mcnt); + break; + + case notcategoryspec: + fprintf (stderr, "/notcategoryspec"); + mcnt = *p++; + fprintf (stderr, "/%d", mcnt); + break; + + case begbuf: + fprintf (stderr, "/begbuf"); + break; + + case endbuf: + fprintf (stderr, "/endbuf"); + break; + + default: + fprintf (stderr, "?%d", *(p-1)); + } + + fprintf (stderr, "\n"); + } + + fprintf (stderr, "%td:\tend of pattern.\n", p - start); +} + + +static void +print_compiled_pattern (struct re_pattern_buffer *bufp) +{ + re_char *buffer = bufp->buffer; + + print_partial_compiled_pattern (buffer, buffer + bufp->used); + printf ("%zu bytes used/%zu bytes allocated.\n", + bufp->used, bufp->allocated); + + if (bufp->fastmap_accurate && bufp->fastmap) + { + printf ("fastmap: "); + print_fastmap (bufp->fastmap); + } + + printf ("re_nsub: %zu\t", bufp->re_nsub); + printf ("regs_alloc: %d\t", bufp->regs_allocated); + printf ("can_be_null: %d\t", bufp->can_be_null); + fflush (stdout); + /* Perhaps we should print the translate table? */ +} + + +static void +print_double_string (re_char *where, re_char *string1, ssize_t size1, + re_char *string2, ssize_t size2) +{ + ssize_t this_char; + + if (where == NULL) + printf ("(null)"); + else + { + if (FIRST_STRING_P (where)) + { + for (this_char = where - string1; this_char < size1; this_char++) + putchar (string1[this_char]); + + where = string2; + } + + for (this_char = where - string2; this_char < size2; this_char++) + putchar (string2[this_char]); + } +} + +#else /* not REGEX_EMACS_DEBUG */ + +# define DEBUG_STATEMENT(e) +# define DEBUG_PRINT(...) +# define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) +# define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) + +#endif /* not REGEX_EMACS_DEBUG */ + +typedef enum +{ + REG_NOERROR = 0, /* Success. */ + REG_NOMATCH, /* Didn't find a match (for regexec). */ + + /* POSIX regcomp return error codes. (In the order listed in the + standard.) An older version of this code supported the POSIX + API; this version continues to use these names internally. */ + REG_BADPAT, /* Invalid pattern. */ + REG_ECOLLATE, /* Not implemented. */ + REG_ECTYPE, /* Invalid character class name. */ + REG_EESCAPE, /* Trailing backslash. */ + REG_ESUBREG, /* Invalid back reference. */ + REG_EBRACK, /* Unmatched left bracket. */ + REG_EPAREN, /* Parenthesis imbalance. */ + REG_EBRACE, /* Unmatched \{. */ + REG_BADBR, /* Invalid contents of \{\}. */ + REG_ERANGE, /* Invalid range end. */ + REG_ESPACE, /* Ran out of memory. */ + REG_BADRPT, /* No preceding re for repetition op. */ + + /* Error codes we've added. */ + REG_EEND, /* Premature end. */ + REG_ESIZE, /* Compiled pattern bigger than 2^16 bytes. */ + REG_ERPAREN, /* Unmatched ) or \); not returned from regcomp. */ + REG_ERANGEX, /* Range striding over charsets. */ + REG_ESIZEBR /* n or m too big in \{n,m\} */ +} reg_errcode_t; + +static const char *re_error_msgid[] = + { + [REG_NOERROR] = "Success", + [REG_NOMATCH] = "No match", + [REG_BADPAT] = "Invalid regular expression", + [REG_ECOLLATE] = "Invalid collation character", + [REG_ECTYPE] = "Invalid character class name", + [REG_EESCAPE] = "Trailing backslash", + [REG_ESUBREG] = "Invalid back reference", + [REG_EBRACK] = "Unmatched [ or [^", + [REG_EPAREN] = "Unmatched ( or \\(", + [REG_EBRACE] = "Unmatched \\{", + [REG_BADBR] = "Invalid content of \\{\\}", + [REG_ERANGE] = "Invalid range end", + [REG_ESPACE] = "Memory exhausted", + [REG_BADRPT] = "Invalid preceding regular expression", + [REG_EEND] = "Premature end of regular expression", + [REG_ESIZE] = "Regular expression too big", + [REG_ERPAREN] = "Unmatched ) or \\)", + [REG_ERANGEX ] = "Range striding over charsets", + [REG_ESIZEBR ] = "Invalid content of \\{\\}", + }; + +/* For 'regs_allocated'. */ +enum { REGS_UNALLOCATED, REGS_REALLOCATE, REGS_FIXED }; + +/* If 'regs_allocated' is REGS_UNALLOCATED in the pattern buffer, + 're_match_2' returns information about at least this many registers + the first time a 'regs' structure is passed. */ +enum { RE_NREGS = 30 }; + +/* The searching and matching functions allocate memory for the + failure stack and registers. Otherwise searching and matching + routines would have much smaller memory resources at their + disposal, and therefore might fail to handle complex regexps. */ + +/* Failure stack declarations and macros; both re_compile_fastmap and + re_match_2 use a failure stack. These have to be macros because of + SAFE_ALLOCA. */ + + +/* Approximate number of failure points for which to initially allocate space + when matching. If this number is exceeded, we allocate more + space, so it is not a hard limit. */ +#define INIT_FAILURE_ALLOC 20 + +/* Roughly the maximum number of failure points on the stack. Would be + exactly that if failure always used TYPICAL_FAILURE_SIZE items. + This is a variable only so users of regex can assign to it; we never + change it ourselves. We always multiply it by TYPICAL_FAILURE_SIZE + before using it, so it should probably be a byte-count instead. */ +/* Note that 4400 was enough to cause a crash on Alpha OSF/1, + whose default stack limit is 2mb. In order for a larger + value to work reliably, you have to try to make it accord + with the process stack limit. */ +size_t emacs_re_max_failures = 40000; + +union fail_stack_elt +{ + re_char *pointer; + /* This should be the biggest 'int' that's no bigger than a pointer. */ + long integer; +}; + +typedef union fail_stack_elt fail_stack_elt_t; + +typedef struct +{ + fail_stack_elt_t *stack; + size_t size; + size_t avail; /* Offset of next open position. */ + size_t frame; /* Offset of the cur constructed frame. */ +} fail_stack_type; + +#define FAIL_STACK_EMPTY() (fail_stack.frame == 0) + + +/* Define macros to initialize and free the failure stack. */ + +#define INIT_FAIL_STACK() \ + do { \ + fail_stack.stack = \ + SAFE_ALLOCA (INIT_FAILURE_ALLOC * TYPICAL_FAILURE_SIZE \ + * sizeof (fail_stack_elt_t)); \ + fail_stack.size = INIT_FAILURE_ALLOC; \ + fail_stack.avail = 0; \ + fail_stack.frame = 0; \ + } while (false) + + +/* Double the size of FAIL_STACK, up to a limit + which allows approximately 'emacs_re_max_failures' items. + + Return 1 if succeeds, and 0 if either ran out of memory + allocating space for it or it was already too large. + + REGEX_REALLOCATE requires 'destination' be declared. */ + +/* Factor to increase the failure stack size by. + This used to be 2, but 2 was too wasteful + because the old discarded stacks added up to as much space + were as ultimate, maximum-size stack. */ +#define FAIL_STACK_GROWTH_FACTOR 4 + +#define GROW_FAIL_STACK(fail_stack) \ + (((fail_stack).size >= emacs_re_max_failures * TYPICAL_FAILURE_SIZE) \ + ? 0 \ + : ((fail_stack).stack \ + = REGEX_REALLOCATE ((fail_stack).stack, \ + (fail_stack).size * sizeof (fail_stack_elt_t), \ + min (emacs_re_max_failures * TYPICAL_FAILURE_SIZE, \ + ((fail_stack).size * FAIL_STACK_GROWTH_FACTOR)) \ + * sizeof (fail_stack_elt_t)), \ + ((fail_stack).size \ + = (min (emacs_re_max_failures * TYPICAL_FAILURE_SIZE, \ + ((fail_stack).size * FAIL_STACK_GROWTH_FACTOR)))), \ + 1)) + + +/* Push a pointer value onto the failure stack. + Assumes the variable 'fail_stack'. Probably should only + be called from within 'PUSH_FAILURE_POINT'. */ +#define PUSH_FAILURE_POINTER(item) \ + fail_stack.stack[fail_stack.avail++].pointer = (item) + +/* This pushes an integer-valued item onto the failure stack. + Assumes the variable 'fail_stack'. Probably should only + be called from within 'PUSH_FAILURE_POINT'. */ +#define PUSH_FAILURE_INT(item) \ + fail_stack.stack[fail_stack.avail++].integer = (item) + +/* These POP... operations complement the PUSH... operations. + All assume that 'fail_stack' is nonempty. */ +#define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer +#define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer + +/* Individual items aside from the registers. */ +#define NUM_NONREG_ITEMS 3 + +/* Used to examine the stack (to detect infinite loops). */ +#define FAILURE_PAT(h) fail_stack.stack[(h) - 1].pointer +#define FAILURE_STR(h) (fail_stack.stack[(h) - 2].pointer) +#define NEXT_FAILURE_HANDLE(h) fail_stack.stack[(h) - 3].integer +#define TOP_FAILURE_HANDLE() fail_stack.frame + + +#define ENSURE_FAIL_STACK(space) \ +while (REMAINING_AVAIL_SLOTS <= space) { \ + if (!GROW_FAIL_STACK (fail_stack)) \ + return -2; \ + DEBUG_PRINT ("\n Doubled stack; size now: %zu\n", (fail_stack).size);\ + DEBUG_PRINT (" slots available: %zu\n", REMAINING_AVAIL_SLOTS);\ +} + +/* Push register NUM onto the stack. */ +#define PUSH_FAILURE_REG(num) \ +do { \ + char *destination; \ + long n = num; \ + ENSURE_FAIL_STACK(3); \ + DEBUG_PRINT (" Push reg %ld (spanning %p -> %p)\n", \ + n, regstart[n], regend[n]); \ + PUSH_FAILURE_POINTER (regstart[n]); \ + PUSH_FAILURE_POINTER (regend[n]); \ + PUSH_FAILURE_INT (n); \ +} while (false) + +/* Change the counter's value to VAL, but make sure that it will + be reset when backtracking. */ +#define PUSH_NUMBER(ptr,val) \ +do { \ + char *destination; \ + int c; \ + ENSURE_FAIL_STACK(3); \ + EXTRACT_NUMBER (c, ptr); \ + DEBUG_PRINT (" Push number %p = %d -> %d\n", ptr, c, val); \ + PUSH_FAILURE_INT (c); \ + PUSH_FAILURE_POINTER (ptr); \ + PUSH_FAILURE_INT (-1); \ + STORE_NUMBER (ptr, val); \ +} while (false) + +/* Pop a saved register off the stack. */ +#define POP_FAILURE_REG_OR_COUNT() \ +do { \ + long pfreg = POP_FAILURE_INT (); \ + if (pfreg == -1) \ + { \ + /* It's a counter. */ \ + /* Discard 'const', making re_search non-reentrant. */ \ + unsigned char *ptr = (unsigned char *) POP_FAILURE_POINTER (); \ + pfreg = POP_FAILURE_INT (); \ + STORE_NUMBER (ptr, pfreg); \ + DEBUG_PRINT (" Pop counter %p = %ld\n", ptr, pfreg); \ + } \ + else \ + { \ + regend[pfreg] = POP_FAILURE_POINTER (); \ + regstart[pfreg] = POP_FAILURE_POINTER (); \ + DEBUG_PRINT (" Pop reg %ld (spanning %p -> %p)\n", \ + pfreg, regstart[pfreg], regend[pfreg]); \ + } \ +} while (false) + +/* Check that we are not stuck in an infinite loop. */ +#define CHECK_INFINITE_LOOP(pat_cur, string_place) \ +do { \ + ptrdiff_t failure = TOP_FAILURE_HANDLE (); \ + /* Check for infinite matching loops */ \ + while (failure > 0 \ + && (FAILURE_STR (failure) == string_place \ + || FAILURE_STR (failure) == NULL)) \ + { \ + eassert (FAILURE_PAT (failure) >= bufp->buffer \ + && FAILURE_PAT (failure) <= bufp->buffer + bufp->used); \ + if (FAILURE_PAT (failure) == pat_cur) \ + { \ + cycle = 1; \ + break; \ + } \ + DEBUG_PRINT (" Other pattern: %p\n", FAILURE_PAT (failure)); \ + failure = NEXT_FAILURE_HANDLE(failure); \ + } \ + DEBUG_PRINT (" Other string: %p\n", FAILURE_STR (failure)); \ +} while (false) + +/* Push the information about the state we will need + if we ever fail back to it. + + Requires variables fail_stack, regstart, regend and + num_regs be declared. GROW_FAIL_STACK requires 'destination' be + declared. + + Does 'return FAILURE_CODE' if runs out of memory. */ + +#define PUSH_FAILURE_POINT(pattern, string_place) \ +do { \ + char *destination; \ + DEBUG_STATEMENT (nfailure_points_pushed++); \ + DEBUG_PRINT ("\nPUSH_FAILURE_POINT:\n"); \ + DEBUG_PRINT (" Before push, next avail: %zu\n", (fail_stack).avail); \ + DEBUG_PRINT (" size: %zu\n", (fail_stack).size);\ + \ + ENSURE_FAIL_STACK (NUM_NONREG_ITEMS); \ + \ + DEBUG_PRINT ("\n"); \ + \ + DEBUG_PRINT (" Push frame index: %zu\n", fail_stack.frame); \ + PUSH_FAILURE_INT (fail_stack.frame); \ + \ + DEBUG_PRINT (" Push string %p: \"", string_place); \ + DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, size2);\ + DEBUG_PRINT ("\"\n"); \ + PUSH_FAILURE_POINTER (string_place); \ + \ + DEBUG_PRINT (" Push pattern %p: ", pattern); \ + DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern, pend); \ + PUSH_FAILURE_POINTER (pattern); \ + \ + /* Close the frame by moving the frame pointer past it. */ \ + fail_stack.frame = fail_stack.avail; \ +} while (false) + +/* Estimate the size of data pushed by a typical failure stack entry. + An estimate is all we need, because all we use this for + is to choose a limit for how big to make the failure stack. */ +/* BEWARE, the value `20' is hard-coded in emacs.c:main(). */ +#define TYPICAL_FAILURE_SIZE 20 + +/* How many items can still be added to the stack without overflowing it. */ +#define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail) + + +/* Pop what PUSH_FAIL_STACK pushes. + + Restore into the parameters, all of which should be lvalues: + STR -- the saved data position. + PAT -- the saved pattern position. + REGSTART, REGEND -- arrays of string positions. + + Also assume the variables FAIL_STACK and (if debugging) BUFP, PEND, + STRING1, SIZE1, STRING2, and SIZE2. */ + +#define POP_FAILURE_POINT(str, pat) \ +do { \ + eassert (!FAIL_STACK_EMPTY ()); \ + \ + /* Remove failure points and point to how many regs pushed. */ \ + DEBUG_PRINT ("POP_FAILURE_POINT:\n"); \ + DEBUG_PRINT (" Before pop, next avail: %zu\n", fail_stack.avail); \ + DEBUG_PRINT (" size: %zu\n", fail_stack.size); \ + \ + /* Pop the saved registers. */ \ + while (fail_stack.frame < fail_stack.avail) \ + POP_FAILURE_REG_OR_COUNT (); \ + \ + pat = POP_FAILURE_POINTER (); \ + DEBUG_PRINT (" Popping pattern %p: ", pat); \ + DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \ + \ + /* If the saved string location is NULL, it came from an \ + on_failure_keep_string_jump opcode, and we want to throw away the \ + saved NULL, thus retaining our current position in the string. */ \ + str = POP_FAILURE_POINTER (); \ + DEBUG_PRINT (" Popping string %p: \"", str); \ + DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \ + DEBUG_PRINT ("\"\n"); \ + \ + fail_stack.frame = POP_FAILURE_INT (); \ + DEBUG_PRINT (" Popping frame index: %zu\n", fail_stack.frame); \ + \ + eassert (fail_stack.avail >= 0); \ + eassert (fail_stack.frame <= fail_stack.avail); \ + \ + DEBUG_STATEMENT (nfailure_points_popped++); \ +} while (false) /* POP_FAILURE_POINT */ + + + +/* Registers are set to a sentinel when they haven't yet matched. */ +#define REG_UNSET(e) ((e) == NULL) + +/* Subroutine declarations and macros for regex_compile. */ + +static reg_errcode_t regex_compile (re_char *pattern, size_t size, + bool posix_backtracking, + const char *whitespace_regexp, + struct re_pattern_buffer *bufp); +static void store_op1 (re_opcode_t op, unsigned char *loc, int arg); +static void store_op2 (re_opcode_t op, unsigned char *loc, int arg1, int arg2); +static void insert_op1 (re_opcode_t op, unsigned char *loc, + int arg, unsigned char *end); +static void insert_op2 (re_opcode_t op, unsigned char *loc, + int arg1, int arg2, unsigned char *end); +static bool at_begline_loc_p (re_char *pattern, re_char *p); +static bool at_endline_loc_p (re_char *p, re_char *pend); +static re_char *skip_one_char (re_char *p); +static int analyze_first (re_char *p, re_char *pend, + char *fastmap, const int multibyte); + +/* Fetch the next character in the uncompiled pattern, with no + translation. */ +#define PATFETCH(c) \ + do { \ + int len; \ + if (p == pend) return REG_EEND; \ + c = RE_STRING_CHAR_AND_LENGTH (p, len, multibyte); \ + p += len; \ + } while (false) + + +#define RE_TRANSLATE(TBL, C) char_table_translate (TBL, C) +#define TRANSLATE(d) (!NILP (translate) ? RE_TRANSLATE (translate, d) : (d)) + +/* Macros for outputting the compiled pattern into 'buffer'. */ + +/* If the buffer isn't allocated when it comes in, use this. */ +#define INIT_BUF_SIZE 32 + +/* Ensure at least N more bytes of space in buffer. */ +#define GET_BUFFER_SPACE(n) \ + while ((size_t) (b - bufp->buffer + (n)) > bufp->allocated) \ + EXTEND_BUFFER () + +/* Ensure one more byte of buffer space and then add C to it. */ +#define BUF_PUSH(c) \ + do { \ + GET_BUFFER_SPACE (1); \ + *b++ = (unsigned char) (c); \ + } while (false) + + +/* Ensure we have two more bytes of buffer space and then append C1 and C2. */ +#define BUF_PUSH_2(c1, c2) \ + do { \ + GET_BUFFER_SPACE (2); \ + *b++ = (unsigned char) (c1); \ + *b++ = (unsigned char) (c2); \ + } while (false) + + +/* Store a jump with opcode OP at LOC to location TO. Store a + relative address offset by the three bytes the jump itself occupies. */ +#define STORE_JUMP(op, loc, to) \ + store_op1 (op, loc, (to) - (loc) - 3) + +/* Likewise, for a two-argument jump. */ +#define STORE_JUMP2(op, loc, to, arg) \ + store_op2 (op, loc, (to) - (loc) - 3, arg) + +/* Like 'STORE_JUMP', but for inserting. Assume B is the buffer end. */ +#define INSERT_JUMP(op, loc, to) \ + insert_op1 (op, loc, (to) - (loc) - 3, b) + +/* Like 'STORE_JUMP2', but for inserting. Assume B is the buffer end. */ +#define INSERT_JUMP2(op, loc, to, arg) \ + insert_op2 (op, loc, (to) - (loc) - 3, arg, b) + + +/* This is not an arbitrary limit: the arguments which represent offsets + into the pattern are two bytes long. So if 2^15 bytes turns out to + be too small, many things would have to change. */ +# define MAX_BUF_SIZE (1 << 15) + +/* Extend the buffer by twice its current size via realloc and + reset the pointers that pointed into the old block to point to the + correct places in the new one. If extending the buffer results in it + being larger than MAX_BUF_SIZE, then flag memory exhausted. */ +#define EXTEND_BUFFER() \ + do { \ + unsigned char *old_buffer = bufp->buffer; \ + if (bufp->allocated == MAX_BUF_SIZE) \ + return REG_ESIZE; \ + bufp->allocated <<= 1; \ + if (bufp->allocated > MAX_BUF_SIZE) \ + bufp->allocated = MAX_BUF_SIZE; \ + ptrdiff_t b_off = b - old_buffer; \ + ptrdiff_t begalt_off = begalt - old_buffer; \ + bool fixup_alt_jump_set = !!fixup_alt_jump; \ + bool laststart_set = !!laststart; \ + bool pending_exact_set = !!pending_exact; \ + ptrdiff_t fixup_alt_jump_off, laststart_off, pending_exact_off; \ + if (fixup_alt_jump_set) fixup_alt_jump_off = fixup_alt_jump - old_buffer; \ + if (laststart_set) laststart_off = laststart - old_buffer; \ + if (pending_exact_set) pending_exact_off = pending_exact - old_buffer; \ + RETALLOC (bufp->buffer, bufp->allocated, unsigned char); \ + unsigned char *new_buffer = bufp->buffer; \ + b = new_buffer + b_off; \ + begalt = new_buffer + begalt_off; \ + if (fixup_alt_jump_set) fixup_alt_jump = new_buffer + fixup_alt_jump_off; \ + if (laststart_set) laststart = new_buffer + laststart_off; \ + if (pending_exact_set) pending_exact = new_buffer + pending_exact_off; \ + } while (false) + + +/* Since we have one byte reserved for the register number argument to + {start,stop}_memory, the maximum number of groups we can report + things about is what fits in that byte. */ +#define MAX_REGNUM 255 + +/* But patterns can have more than 'MAX_REGNUM' registers. Just + ignore the excess. */ +typedef int regnum_t; + + +/* Macros for the compile stack. */ + +/* Since offsets can go either forwards or backwards, this type needs to + be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */ +typedef long pattern_offset_t; + +typedef struct +{ + pattern_offset_t begalt_offset; + pattern_offset_t fixup_alt_jump; + pattern_offset_t laststart_offset; + regnum_t regnum; +} compile_stack_elt_t; + + +typedef struct +{ + compile_stack_elt_t *stack; + size_t size; + size_t avail; /* Offset of next open position. */ +} compile_stack_type; + + +#define INIT_COMPILE_STACK_SIZE 32 + +#define COMPILE_STACK_EMPTY (compile_stack.avail == 0) +#define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size) + +/* The next available element. */ +#define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail]) + +/* Structure to manage work area for range table. */ +struct range_table_work_area +{ + int *table; /* actual work area. */ + int allocated; /* allocated size for work area in bytes. */ + int used; /* actually used size in words. */ + int bits; /* flag to record character classes */ +}; + +/* Make sure that WORK_AREA can hold more N multibyte characters. + This is used only in set_image_of_range and set_image_of_range_1. + It expects WORK_AREA to be a pointer. + If it can't get the space, it returns from the surrounding function. */ + +#define EXTEND_RANGE_TABLE(work_area, n) \ + do { \ + if (((work_area).used + (n)) * sizeof (int) > (work_area).allocated) \ + { \ + extend_range_table_work_area (&work_area); \ + if ((work_area).table == 0) \ + return (REG_ESPACE); \ + } \ + } while (false) + +#define SET_RANGE_TABLE_WORK_AREA_BIT(work_area, bit) \ + (work_area).bits |= (bit) + +/* Set a range (RANGE_START, RANGE_END) to WORK_AREA. */ +#define SET_RANGE_TABLE_WORK_AREA(work_area, range_start, range_end) \ + do { \ + EXTEND_RANGE_TABLE ((work_area), 2); \ + (work_area).table[(work_area).used++] = (range_start); \ + (work_area).table[(work_area).used++] = (range_end); \ + } while (false) + +/* Free allocated memory for WORK_AREA. */ +#define FREE_RANGE_TABLE_WORK_AREA(work_area) \ + do { \ + if ((work_area).table) \ + xfree ((work_area).table); \ + } while (false) + +#define CLEAR_RANGE_TABLE_WORK_USED(work_area) \ + ((work_area).used = 0, (work_area).bits = 0) +#define RANGE_TABLE_WORK_USED(work_area) ((work_area).used) +#define RANGE_TABLE_WORK_BITS(work_area) ((work_area).bits) +#define RANGE_TABLE_WORK_ELT(work_area, i) ((work_area).table[i]) + +/* Bits used to implement the multibyte-part of the various character classes + such as [:alnum:] in a charset's range table. The code currently assumes + that only the low 16 bits are used. */ +#define BIT_WORD 0x1 +#define BIT_LOWER 0x2 +#define BIT_PUNCT 0x4 +#define BIT_SPACE 0x8 +#define BIT_UPPER 0x10 +#define BIT_MULTIBYTE 0x20 +#define BIT_ALPHA 0x40 +#define BIT_ALNUM 0x80 +#define BIT_GRAPH 0x100 +#define BIT_PRINT 0x200 +#define BIT_BLANK 0x400 + + +/* Set the bit for character C in a list. */ +#define SET_LIST_BIT(c) (b[((c)) / BYTEWIDTH] |= 1 << ((c) % BYTEWIDTH)) + + +/* Store characters in the range FROM to TO in the bitmap at B (for + ASCII and unibyte characters) and WORK_AREA (for multibyte + characters) while translating them and paying attention to the + continuity of translated characters. + + Implementation note: It is better to implement these fairly big + macros by a function, but it's not that easy because macros called + in this macro assume various local variables already declared. */ + +/* Both FROM and TO are ASCII characters. */ + +#define SETUP_ASCII_RANGE(work_area, FROM, TO) \ + do { \ + int C0, C1; \ + \ + for (C0 = (FROM); C0 <= (TO); C0++) \ + { \ + C1 = TRANSLATE (C0); \ + if (! ASCII_CHAR_P (C1)) \ + { \ + SET_RANGE_TABLE_WORK_AREA ((work_area), C1, C1); \ + if ((C1 = RE_CHAR_TO_UNIBYTE (C1)) < 0) \ + C1 = C0; \ + } \ + SET_LIST_BIT (C1); \ + } \ + } while (false) + + +/* Both FROM and TO are unibyte characters (0x80..0xFF). */ + +#define SETUP_UNIBYTE_RANGE(work_area, FROM, TO) \ + do { \ + int C0, C1, C2, I; \ + int USED = RANGE_TABLE_WORK_USED (work_area); \ + \ + for (C0 = (FROM); C0 <= (TO); C0++) \ + { \ + C1 = RE_CHAR_TO_MULTIBYTE (C0); \ + if (CHAR_BYTE8_P (C1)) \ + SET_LIST_BIT (C0); \ + else \ + { \ + C2 = TRANSLATE (C1); \ + if (C2 == C1 \ + || (C1 = RE_CHAR_TO_UNIBYTE (C2)) < 0) \ + C1 = C0; \ + SET_LIST_BIT (C1); \ + for (I = RANGE_TABLE_WORK_USED (work_area) - 2; I >= USED; I -= 2) \ + { \ + int from = RANGE_TABLE_WORK_ELT (work_area, I); \ + int to = RANGE_TABLE_WORK_ELT (work_area, I + 1); \ + \ + if (C2 >= from - 1 && C2 <= to + 1) \ + { \ + if (C2 == from - 1) \ + RANGE_TABLE_WORK_ELT (work_area, I)--; \ + else if (C2 == to + 1) \ + RANGE_TABLE_WORK_ELT (work_area, I + 1)++; \ + break; \ + } \ + } \ + if (I < USED) \ + SET_RANGE_TABLE_WORK_AREA ((work_area), C2, C2); \ + } \ + } \ + } while (false) + + +/* Both FROM and TO are multibyte characters. */ + +#define SETUP_MULTIBYTE_RANGE(work_area, FROM, TO) \ + do { \ + int C0, C1, C2, I, USED = RANGE_TABLE_WORK_USED (work_area); \ + \ + SET_RANGE_TABLE_WORK_AREA ((work_area), (FROM), (TO)); \ + for (C0 = (FROM); C0 <= (TO); C0++) \ + { \ + C1 = TRANSLATE (C0); \ + if ((C2 = RE_CHAR_TO_UNIBYTE (C1)) >= 0 \ + || (C1 != C0 && (C2 = RE_CHAR_TO_UNIBYTE (C0)) >= 0)) \ + SET_LIST_BIT (C2); \ + if (C1 >= (FROM) && C1 <= (TO)) \ + continue; \ + for (I = RANGE_TABLE_WORK_USED (work_area) - 2; I >= USED; I -= 2) \ + { \ + int from = RANGE_TABLE_WORK_ELT (work_area, I); \ + int to = RANGE_TABLE_WORK_ELT (work_area, I + 1); \ + \ + if (C1 >= from - 1 && C1 <= to + 1) \ + { \ + if (C1 == from - 1) \ + RANGE_TABLE_WORK_ELT (work_area, I)--; \ + else if (C1 == to + 1) \ + RANGE_TABLE_WORK_ELT (work_area, I + 1)++; \ + break; \ + } \ + } \ + if (I < USED) \ + SET_RANGE_TABLE_WORK_AREA ((work_area), C1, C1); \ + } \ + } while (false) + +/* Get the next unsigned number in the uncompiled pattern. */ +#define GET_INTERVAL_COUNT(num) \ + do { \ + if (p == pend) \ + FREE_STACK_RETURN (REG_EBRACE); \ + else \ + { \ + PATFETCH (c); \ + while ('0' <= c && c <= '9') \ + { \ + if (num < 0) \ + num = 0; \ + if (RE_DUP_MAX / 10 - (RE_DUP_MAX % 10 < c - '0') < num) \ + FREE_STACK_RETURN (REG_ESIZEBR); \ + num = num * 10 + c - '0'; \ + if (p == pend) \ + FREE_STACK_RETURN (REG_EBRACE); \ + PATFETCH (c); \ + } \ + } \ + } while (false) + +/* Parse a character class, i.e. string such as "[:name:]". *strp + points to the string to be parsed and limit is length, in bytes, of + that string. + + If *strp point to a string that begins with "[:name:]", where name is + a non-empty sequence of lower case letters, *strp will be advanced past the + closing square bracket and RECC_* constant which maps to the name will be + returned. If name is not a valid character class name zero, or RECC_ERROR, + is returned. + + Otherwise, if *strp doesn't begin with "[:name:]", -1 is returned. + + The function can be used on ASCII and multibyte (UTF-8-encoded) strings. + */ +re_wctype_t +re_wctype_parse (const unsigned char **strp, unsigned limit) +{ + const char *beg = (const char *)*strp, *it; + + if (limit < 4 || beg[0] != '[' || beg[1] != ':') + return -1; + + beg += 2; /* skip opening "[:" */ + limit -= 3; /* opening "[:" and half of closing ":]"; --limit handles rest */ + for (it = beg; it[0] != ':' || it[1] != ']'; ++it) + if (!--limit) + return -1; + + *strp = (const unsigned char *)(it + 2); + + /* Sort tests in the length=five case by frequency the classes to minimize + number of times we fail the comparison. The frequencies of character class + names used in Emacs sources as of 2016-07-27: + + $ find \( -name \*.c -o -name \*.el \) -exec grep -h '\[:[a-z]*:]' {} + | + sed 's/]/]\n/g' |grep -o '\[:[a-z]*:]' |sort |uniq -c |sort -nr + 213 [:alnum:] + 104 [:alpha:] + 62 [:space:] + 39 [:digit:] + 36 [:blank:] + 26 [:word:] + 26 [:upper:] + 21 [:lower:] + 10 [:xdigit:] + 10 [:punct:] + 10 [:ascii:] + 4 [:nonascii:] + 4 [:graph:] + 2 [:print:] + 2 [:cntrl:] + 1 [:ff:] + + If you update this list, consider also updating chain of or'ed conditions + in execute_charset function. + */ + + switch (it - beg) { + case 4: + if (!memcmp (beg, "word", 4)) return RECC_WORD; + break; + case 5: + if (!memcmp (beg, "alnum", 5)) return RECC_ALNUM; + if (!memcmp (beg, "alpha", 5)) return RECC_ALPHA; + if (!memcmp (beg, "space", 5)) return RECC_SPACE; + if (!memcmp (beg, "digit", 5)) return RECC_DIGIT; + if (!memcmp (beg, "blank", 5)) return RECC_BLANK; + if (!memcmp (beg, "upper", 5)) return RECC_UPPER; + if (!memcmp (beg, "lower", 5)) return RECC_LOWER; + if (!memcmp (beg, "punct", 5)) return RECC_PUNCT; + if (!memcmp (beg, "ascii", 5)) return RECC_ASCII; + if (!memcmp (beg, "graph", 5)) return RECC_GRAPH; + if (!memcmp (beg, "print", 5)) return RECC_PRINT; + if (!memcmp (beg, "cntrl", 5)) return RECC_CNTRL; + break; + case 6: + if (!memcmp (beg, "xdigit", 6)) return RECC_XDIGIT; + break; + case 7: + if (!memcmp (beg, "unibyte", 7)) return RECC_UNIBYTE; + break; + case 8: + if (!memcmp (beg, "nonascii", 8)) return RECC_NONASCII; + break; + case 9: + if (!memcmp (beg, "multibyte", 9)) return RECC_MULTIBYTE; + break; + } + + return RECC_ERROR; +} + +/* True if CH is in the char class CC. */ +bool +re_iswctype (int ch, re_wctype_t cc) +{ + switch (cc) + { + case RECC_ALNUM: return ISALNUM (ch) != 0; + case RECC_ALPHA: return ISALPHA (ch) != 0; + case RECC_BLANK: return ISBLANK (ch) != 0; + case RECC_CNTRL: return ISCNTRL (ch) != 0; + case RECC_DIGIT: return ISDIGIT (ch) != 0; + case RECC_GRAPH: return ISGRAPH (ch) != 0; + case RECC_LOWER: return ISLOWER (ch) != 0; + case RECC_PRINT: return ISPRINT (ch) != 0; + case RECC_PUNCT: return ISPUNCT (ch) != 0; + case RECC_SPACE: return ISSPACE (ch) != 0; + case RECC_UPPER: return ISUPPER (ch) != 0; + case RECC_XDIGIT: return ISXDIGIT (ch) != 0; + case RECC_ASCII: return IS_REAL_ASCII (ch) != 0; + case RECC_NONASCII: return !IS_REAL_ASCII (ch); + case RECC_UNIBYTE: return ISUNIBYTE (ch) != 0; + case RECC_MULTIBYTE: return !ISUNIBYTE (ch); + case RECC_WORD: return ISWORD (ch) != 0; + case RECC_ERROR: return false; + default: + abort (); + } +} + +/* Return a bit-pattern to use in the range-table bits to match multibyte + chars of class CC. */ +static int +re_wctype_to_bit (re_wctype_t cc) +{ + switch (cc) + { + case RECC_NONASCII: + case RECC_MULTIBYTE: return BIT_MULTIBYTE; + case RECC_ALPHA: return BIT_ALPHA; + case RECC_ALNUM: return BIT_ALNUM; + case RECC_WORD: return BIT_WORD; + case RECC_LOWER: return BIT_LOWER; + case RECC_UPPER: return BIT_UPPER; + case RECC_PUNCT: return BIT_PUNCT; + case RECC_SPACE: return BIT_SPACE; + case RECC_GRAPH: return BIT_GRAPH; + case RECC_PRINT: return BIT_PRINT; + case RECC_BLANK: return BIT_BLANK; + case RECC_ASCII: case RECC_DIGIT: case RECC_XDIGIT: case RECC_CNTRL: + case RECC_UNIBYTE: case RECC_ERROR: return 0; + default: + abort (); + } +} + +/* Filling in the work area of a range. */ + +/* Actually extend the space in WORK_AREA. */ + +static void +extend_range_table_work_area (struct range_table_work_area *work_area) +{ + work_area->allocated += 16 * sizeof (int); + work_area->table = xrealloc (work_area->table, work_area->allocated); +} + +/* regex_compile and helpers. */ + +static bool group_in_compile_stack (compile_stack_type, regnum_t); + +/* Insert the 'jump' from the end of last alternative to "here". + The space for the jump has already been allocated. */ +#define FIXUP_ALT_JUMP() \ +do { \ + if (fixup_alt_jump) \ + STORE_JUMP (jump, fixup_alt_jump, b); \ +} while (false) + + +/* Return, freeing storage we allocated. */ +#define FREE_STACK_RETURN(value) \ + do { \ + FREE_RANGE_TABLE_WORK_AREA (range_table_work); \ + xfree (compile_stack.stack); \ + return value; \ + } while (false) + +/* Compile PATTERN (of length SIZE) according to SYNTAX. + Return a nonzero error code on failure, or zero for success. + + If WHITESPACE_REGEXP is given, use it instead of a space + character in PATTERN. + + Assume the 'allocated' (and perhaps 'buffer') and 'translate' + fields are set in BUFP on entry. + + If successful, put results in *BUFP (otherwise the + contents of *BUFP are undefined): + 'buffer' is the compiled pattern; + 'syntax' is set to SYNTAX; + 'used' is set to the length of the compiled pattern; + 'fastmap_accurate' is zero; + 're_nsub' is the number of subexpressions in PATTERN; + + The 'fastmap' field is neither examined nor set. */ + +static reg_errcode_t +regex_compile (re_char *pattern, size_t size, + bool posix_backtracking, + const char *whitespace_regexp, + struct re_pattern_buffer *bufp) +{ + /* Fetch characters from PATTERN here. */ + int c, c1; + + /* Points to the end of the buffer, where we should append. */ + unsigned char *b; + + /* Keeps track of unclosed groups. */ + compile_stack_type compile_stack; + + /* Points to the current (ending) position in the pattern. */ + re_char *p = pattern; + re_char *pend = pattern + size; + + /* How to translate the characters in the pattern. */ + Lisp_Object translate = bufp->translate; + + /* Address of the count-byte of the most recently inserted 'exactn' + command. This makes it possible to tell if a new exact-match + character can be added to that command or if the character requires + a new 'exactn' command. */ + unsigned char *pending_exact = 0; + + /* Address of start of the most recently finished expression. + This tells, e.g., postfix * where to find the start of its + operand. Reset at the beginning of groups and alternatives. */ + unsigned char *laststart = 0; + + /* Address of beginning of regexp, or inside of last group. */ + unsigned char *begalt; + + /* Place in the uncompiled pattern (i.e., the {) to + which to go back if the interval is invalid. */ + re_char *beg_interval; + + /* Address of the place where a forward jump should go to the end of + the containing expression. Each alternative of an 'or' -- except the + last -- ends with a forward jump of this sort. */ + unsigned char *fixup_alt_jump = 0; + + /* Work area for range table of charset. */ + struct range_table_work_area range_table_work; + + /* If the object matched can contain multibyte characters. */ + bool multibyte = RE_MULTIBYTE_P (bufp); + + /* Nonzero if we have pushed down into a subpattern. */ + int in_subpattern = 0; + + /* These hold the values of p, pattern, and pend from the main + pattern when we have pushed into a subpattern. */ + re_char *main_p; + re_char *main_pattern; + re_char *main_pend; + +#ifdef REGEX_EMACS_DEBUG + regex_emacs_debug++; + DEBUG_PRINT ("\nCompiling pattern: "); + if (regex_emacs_debug > 0) + { + size_t debug_count; + + for (debug_count = 0; debug_count < size; debug_count++) + putchar (pattern[debug_count]); + putchar ('\n'); + } +#endif + + /* Initialize the compile stack. */ + compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t); + compile_stack.size = INIT_COMPILE_STACK_SIZE; + compile_stack.avail = 0; + + range_table_work.table = 0; + range_table_work.allocated = 0; + + /* Initialize the pattern buffer. */ + bufp->fastmap_accurate = 0; + bufp->used_syntax = 0; + + /* Set 'used' to zero, so that if we return an error, the pattern + printer (for debugging) will think there's no pattern. We reset it + at the end. */ + bufp->used = 0; + + bufp->re_nsub = 0; + + if (bufp->allocated == 0) + { + if (bufp->buffer) + { /* If zero allocated, but buffer is non-null, try to realloc + enough space. This loses if buffer's address is bogus, but + that is the user's responsibility. */ + RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char); + } + else + { /* Caller did not allocate a buffer. Do it for them. */ + bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char); + } + bufp->allocated = INIT_BUF_SIZE; + } + + begalt = b = bufp->buffer; + + /* Loop through the uncompiled pattern until we're at the end. */ + while (1) + { + if (p == pend) + { + /* If this is the end of an included regexp, + pop back to the main regexp and try again. */ + if (in_subpattern) + { + in_subpattern = 0; + pattern = main_pattern; + p = main_p; + pend = main_pend; + continue; + } + /* If this is the end of the main regexp, we are done. */ + break; + } + + PATFETCH (c); + + switch (c) + { + case ' ': + { + re_char *p1 = p; + + /* If there's no special whitespace regexp, treat + spaces normally. And don't try to do this recursively. */ + if (!whitespace_regexp || in_subpattern) + goto normal_char; + + /* Peek past following spaces. */ + while (p1 != pend) + { + if (*p1 != ' ') + break; + p1++; + } + /* If the spaces are followed by a repetition op, + treat them normally. */ + if (p1 != pend + && (*p1 == '*' || *p1 == '+' || *p1 == '?' + || (*p1 == '\\' && p1 + 1 != pend && p1[1] == '{'))) + goto normal_char; + + /* Replace the spaces with the whitespace regexp. */ + in_subpattern = 1; + main_p = p1; + main_pend = pend; + main_pattern = pattern; + p = pattern = (re_char *) whitespace_regexp; + pend = p + strlen (whitespace_regexp); + break; + } + + case '^': + if (! (p == pattern + 1 || at_begline_loc_p (pattern, p))) + goto normal_char; + BUF_PUSH (begline); + break; + + case '$': + if (! (p == pend || at_endline_loc_p (p, pend))) + goto normal_char; + BUF_PUSH (endline); + break; + + + case '+': + case '?': + case '*': + /* If there is no previous pattern... */ + if (!laststart) + goto normal_char; + + { + /* 1 means zero (many) matches is allowed. */ + bool zero_times_ok = false, many_times_ok = false; + bool greedy = true; + + /* If there is a sequence of repetition chars, collapse it + down to just one (the right one). We can't combine + interval operators with these because of, e.g., 'a{2}*', + which should only match an even number of 'a's. */ + + for (;;) + { + if (c == '?' && (zero_times_ok || many_times_ok)) + greedy = false; + else + { + zero_times_ok |= c != '+'; + many_times_ok |= c != '?'; + } + + if (! (p < pend && (*p == '*' || *p == '+' || *p == '?'))) + break; + /* If we get here, we found another repeat character. */ + c = *p++; + } + + /* Star, etc. applied to an empty pattern is equivalent + to an empty pattern. */ + if (!laststart || laststart == b) + break; + + /* Now we know whether or not zero matches is allowed + and also whether or not two or more matches is allowed. */ + if (greedy) + { + if (many_times_ok) + { + bool simple = skip_one_char (laststart) == b; + size_t startoffset = 0; + re_opcode_t ofj = + /* Check if the loop can match the empty string. */ + (simple || !analyze_first (laststart, b, NULL, 0)) + ? on_failure_jump : on_failure_jump_loop; + eassert (skip_one_char (laststart) <= b); + + if (!zero_times_ok && simple) + { /* Since simple * loops can be made faster by using + on_failure_keep_string_jump, we turn simple P+ + into PP* if P is simple. */ + unsigned char *p1, *p2; + startoffset = b - laststart; + GET_BUFFER_SPACE (startoffset); + p1 = b; p2 = laststart; + while (p2 < p1) + *b++ = *p2++; + zero_times_ok = 1; + } + + GET_BUFFER_SPACE (6); + if (!zero_times_ok) + /* A + loop. */ + STORE_JUMP (ofj, b, b + 6); + else + /* Simple * loops can use on_failure_keep_string_jump + depending on what follows. But since we don't know + that yet, we leave the decision up to + on_failure_jump_smart. */ + INSERT_JUMP (simple ? on_failure_jump_smart : ofj, + laststart + startoffset, b + 6); + b += 3; + STORE_JUMP (jump, b, laststart + startoffset); + b += 3; + } + else + { + /* A simple ? pattern. */ + eassert (zero_times_ok); + GET_BUFFER_SPACE (3); + INSERT_JUMP (on_failure_jump, laststart, b + 3); + b += 3; + } + } + else /* not greedy */ + { /* I wish the greedy and non-greedy cases could be merged. */ + + GET_BUFFER_SPACE (7); /* We might use less. */ + if (many_times_ok) + { + bool emptyp = analyze_first (laststart, b, NULL, 0); + + /* The non-greedy multiple match looks like + a repeat..until: we only need a conditional jump + at the end of the loop. */ + if (emptyp) BUF_PUSH (no_op); + STORE_JUMP (emptyp ? on_failure_jump_nastyloop + : on_failure_jump, b, laststart); + b += 3; + if (zero_times_ok) + { + /* The repeat...until naturally matches one or more. + To also match zero times, we need to first jump to + the end of the loop (its conditional jump). */ + INSERT_JUMP (jump, laststart, b); + b += 3; + } + } + else + { + /* non-greedy a?? */ + INSERT_JUMP (jump, laststart, b + 3); + b += 3; + INSERT_JUMP (on_failure_jump, laststart, laststart + 6); + b += 3; + } + } + } + pending_exact = 0; + break; + + + case '.': + laststart = b; + BUF_PUSH (anychar); + break; + + + case '[': + { + re_char *p1; + + CLEAR_RANGE_TABLE_WORK_USED (range_table_work); + + if (p == pend) FREE_STACK_RETURN (REG_EBRACK); + + /* Ensure that we have enough space to push a charset: the + opcode, the length count, and the bitset; 34 bytes in all. */ + GET_BUFFER_SPACE (34); + + laststart = b; + + /* Test '*p == '^' twice, instead of using an if + statement, so we need only one BUF_PUSH. */ + BUF_PUSH (*p == '^' ? charset_not : charset); + if (*p == '^') + p++; + + /* Remember the first position in the bracket expression. */ + p1 = p; + + /* Push the number of bytes in the bitmap. */ + BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH); + + /* Clear the whole map. */ + memset (b, 0, (1 << BYTEWIDTH) / BYTEWIDTH); + + /* Read in characters and ranges, setting map bits. */ + for (;;) + { + const unsigned char *p2 = p; + int ch; + + if (p == pend) FREE_STACK_RETURN (REG_EBRACK); + + /* See if we're at the beginning of a possible character + class. */ + re_wctype_t cc = re_wctype_parse (&p, pend - p); + if (cc != -1) + { + if (cc == 0) + FREE_STACK_RETURN (REG_ECTYPE); + + if (p == pend) + FREE_STACK_RETURN (REG_EBRACK); + + /* Most character classes in a multibyte match just set + a flag. Exceptions are is_blank, is_digit, is_cntrl, and + is_xdigit, since they can only match ASCII characters. + We don't need to handle them for multibyte. */ + + /* Setup the gl_state object to its buffer-defined value. + This hardcodes the buffer-global syntax-table for ASCII + chars, while the other chars will obey syntax-table + properties. It's not ideal, but it's the way it's been + done until now. */ + SETUP_BUFFER_SYNTAX_TABLE (); + + for (c = 0; c < 0x80; ++c) + if (re_iswctype (c, cc)) + { + SET_LIST_BIT (c); + c1 = TRANSLATE (c); + if (c1 == c) + continue; + if (ASCII_CHAR_P (c1)) + SET_LIST_BIT (c1); + else if ((c1 = RE_CHAR_TO_UNIBYTE (c1)) >= 0) + SET_LIST_BIT (c1); + } + SET_RANGE_TABLE_WORK_AREA_BIT + (range_table_work, re_wctype_to_bit (cc)); + + /* In most cases the matching rule for char classes only + uses the syntax table for multibyte chars, so that the + content of the syntax-table is not hardcoded in the + range_table. SPACE and WORD are the two exceptions. */ + if ((1 << cc) & ((1 << RECC_SPACE) | (1 << RECC_WORD))) + bufp->used_syntax = 1; + + /* Repeat the loop. */ + continue; + } + + /* Don't translate yet. The range TRANSLATE(X..Y) cannot + always be determined from TRANSLATE(X) and TRANSLATE(Y) + So the translation is done later in a loop. Example: + (let ((case-fold-search t)) (string-match "[A-_]" "A")) */ + PATFETCH (c); + + /* Could be the end of the bracket expression. If it's + not (i.e., when the bracket expression is '[]' so + far), the ']' character bit gets set way below. */ + if (c == ']' && p2 != p1) + break; + + if (p < pend && p[0] == '-' && p[1] != ']') + { + + /* Discard the '-'. */ + PATFETCH (c1); + + /* Fetch the character which ends the range. */ + PATFETCH (c1); + + if (CHAR_BYTE8_P (c1) + && ! ASCII_CHAR_P (c) && ! CHAR_BYTE8_P (c)) + /* Treat the range from a multibyte character to + raw-byte character as empty. */ + c = c1 + 1; + } + else + /* Range from C to C. */ + c1 = c; + + if (c <= c1) + { + if (c < 128) + { + ch = min (127, c1); + SETUP_ASCII_RANGE (range_table_work, c, ch); + c = ch + 1; + if (CHAR_BYTE8_P (c1)) + c = BYTE8_TO_CHAR (128); + } + if (CHAR_BYTE8_P (c)) + { + c = CHAR_TO_BYTE8 (c); + c1 = CHAR_TO_BYTE8 (c1); + for (; c <= c1; c++) + SET_LIST_BIT (c); + } + else if (multibyte) + SETUP_MULTIBYTE_RANGE (range_table_work, c, c1); + else + SETUP_UNIBYTE_RANGE (range_table_work, c, c1); + } + } + + /* Discard any (non)matching list bytes that are all 0 at the + end of the map. Decrease the map-length byte too. */ + while ((int) b[-1] > 0 && b[b[-1] - 1] == 0) + b[-1]--; + b += b[-1]; + + /* Build real range table from work area. */ + if (RANGE_TABLE_WORK_USED (range_table_work) + || RANGE_TABLE_WORK_BITS (range_table_work)) + { + int i; + int used = RANGE_TABLE_WORK_USED (range_table_work); + + /* Allocate space for COUNT + RANGE_TABLE. Needs two + bytes for flags, two for COUNT, and three bytes for + each character. */ + GET_BUFFER_SPACE (4 + used * 3); + + /* Indicate the existence of range table. */ + laststart[1] |= 0x80; + + /* Store the character class flag bits into the range table. */ + *b++ = RANGE_TABLE_WORK_BITS (range_table_work) & 0xff; + *b++ = RANGE_TABLE_WORK_BITS (range_table_work) >> 8; + + STORE_NUMBER_AND_INCR (b, used / 2); + for (i = 0; i < used; i++) + STORE_CHARACTER_AND_INCR + (b, RANGE_TABLE_WORK_ELT (range_table_work, i)); + } + } + break; + + + case '\\': + if (p == pend) FREE_STACK_RETURN (REG_EESCAPE); + + /* Do not translate the character after the \, so that we can + distinguish, e.g., \B from \b, even if we normally would + translate, e.g., B to b. */ + PATFETCH (c); + + switch (c) + { + case '(': + { + int shy = 0; + regnum_t regnum = 0; + if (p+1 < pend) + { + /* Look for a special (?...) construct */ + if (*p == '?') + { + PATFETCH (c); /* Gobble up the '?'. */ + while (!shy) + { + PATFETCH (c); + switch (c) + { + case ':': shy = 1; break; + case '0': + /* An explicitly specified regnum must start + with non-0. */ + if (regnum == 0) + FREE_STACK_RETURN (REG_BADPAT); + FALLTHROUGH; + case '1': case '2': case '3': case '4': + case '5': case '6': case '7': case '8': case '9': + regnum = 10*regnum + (c - '0'); break; + default: + /* Only (?:...) is supported right now. */ + FREE_STACK_RETURN (REG_BADPAT); + } + } + } + } + + if (!shy) + regnum = ++bufp->re_nsub; + else if (regnum) + { /* It's actually not shy, but explicitly numbered. */ + shy = 0; + if (regnum > bufp->re_nsub) + bufp->re_nsub = regnum; + else if (regnum > bufp->re_nsub + /* Ideally, we'd want to check that the specified + group can't have matched (i.e. all subgroups + using the same regnum are in other branches of + OR patterns), but we don't currently keep track + of enough info to do that easily. */ + || group_in_compile_stack (compile_stack, regnum)) + FREE_STACK_RETURN (REG_BADPAT); + } + else + /* It's really shy. */ + regnum = - bufp->re_nsub; + + if (COMPILE_STACK_FULL) + { + RETALLOC (compile_stack.stack, compile_stack.size << 1, + compile_stack_elt_t); + compile_stack.size <<= 1; + } + + /* These are the values to restore when we hit end of this + group. They are all relative offsets, so that if the + whole pattern moves because of realloc, they will still + be valid. */ + COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer; + COMPILE_STACK_TOP.fixup_alt_jump + = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0; + COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer; + COMPILE_STACK_TOP.regnum = regnum; + + /* Do not push a start_memory for groups beyond the last one + we can represent in the compiled pattern. */ + if (regnum <= MAX_REGNUM && regnum > 0) + BUF_PUSH_2 (start_memory, regnum); + + compile_stack.avail++; + + fixup_alt_jump = 0; + laststart = 0; + begalt = b; + /* If we've reached MAX_REGNUM groups, then this open + won't actually generate any code, so we'll have to + clear pending_exact explicitly. */ + pending_exact = 0; + break; + } + + case ')': + if (COMPILE_STACK_EMPTY) + FREE_STACK_RETURN (REG_ERPAREN); + + FIXUP_ALT_JUMP (); + + /* See similar code for backslashed left paren above. */ + if (COMPILE_STACK_EMPTY) + FREE_STACK_RETURN (REG_ERPAREN); + + /* Since we just checked for an empty stack above, this + "can't happen". */ + eassert (compile_stack.avail != 0); + { + /* We don't just want to restore into 'regnum', because + later groups should continue to be numbered higher, + as in '(ab)c(de)' -- the second group is #2. */ + regnum_t regnum; + + compile_stack.avail--; + begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset; + fixup_alt_jump + = COMPILE_STACK_TOP.fixup_alt_jump + ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1 + : 0; + laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset; + regnum = COMPILE_STACK_TOP.regnum; + /* If we've reached MAX_REGNUM groups, then this open + won't actually generate any code, so we'll have to + clear pending_exact explicitly. */ + pending_exact = 0; + + /* We're at the end of the group, so now we know how many + groups were inside this one. */ + if (regnum <= MAX_REGNUM && regnum > 0) + BUF_PUSH_2 (stop_memory, regnum); + } + break; + + + case '|': /* '\|'. */ + /* Insert before the previous alternative a jump which + jumps to this alternative if the former fails. */ + GET_BUFFER_SPACE (3); + INSERT_JUMP (on_failure_jump, begalt, b + 6); + pending_exact = 0; + b += 3; + + /* The alternative before this one has a jump after it + which gets executed if it gets matched. Adjust that + jump so it will jump to this alternative's analogous + jump (put in below, which in turn will jump to the next + (if any) alternative's such jump, etc.). The last such + jump jumps to the correct final destination. A picture: + _____ _____ + | | | | + | v | v + A | B | C + + If we are at B, then fixup_alt_jump right now points to a + three-byte space after A. We'll put in the jump, set + fixup_alt_jump to right after B, and leave behind three + bytes which we'll fill in when we get to after C. */ + + FIXUP_ALT_JUMP (); + + /* Mark and leave space for a jump after this alternative, + to be filled in later either by next alternative or + when know we're at the end of a series of alternatives. */ + fixup_alt_jump = b; + GET_BUFFER_SPACE (3); + b += 3; + + laststart = 0; + begalt = b; + break; + + + case '{': + { + /* At least (most) this many matches must be made. */ + int lower_bound = 0, upper_bound = -1; + + beg_interval = p; + + GET_INTERVAL_COUNT (lower_bound); + + if (c == ',') + GET_INTERVAL_COUNT (upper_bound); + else + /* Interval such as '{1}' => match exactly once. */ + upper_bound = lower_bound; + + if (lower_bound < 0 + || (0 <= upper_bound && upper_bound < lower_bound) + || c != '\\') + FREE_STACK_RETURN (REG_BADBR); + if (p == pend) + FREE_STACK_RETURN (REG_EESCAPE); + if (*p++ != '}') + FREE_STACK_RETURN (REG_BADBR); + + /* We just parsed a valid interval. */ + + /* If it's invalid to have no preceding re. */ + if (!laststart) + goto unfetch_interval; + + if (upper_bound == 0) + /* If the upper bound is zero, just drop the sub pattern + altogether. */ + b = laststart; + else if (lower_bound == 1 && upper_bound == 1) + /* Just match it once: nothing to do here. */ + ; + + /* Otherwise, we have a nontrivial interval. When + we're all done, the pattern will look like: + set_number_at + set_number_at + succeed_n + + jump_n + (The upper bound and 'jump_n' are omitted if + 'upper_bound' is 1, though.) */ + else + { /* If the upper bound is > 1, we need to insert + more at the end of the loop. */ + unsigned int nbytes = (upper_bound < 0 ? 3 + : upper_bound > 1 ? 5 : 0); + unsigned int startoffset = 0; + + GET_BUFFER_SPACE (20); /* We might use less. */ + + if (lower_bound == 0) + { + /* A succeed_n that starts with 0 is really a + a simple on_failure_jump_loop. */ + INSERT_JUMP (on_failure_jump_loop, laststart, + b + 3 + nbytes); + b += 3; + } + else + { + /* Initialize lower bound of the 'succeed_n', even + though it will be set during matching by its + attendant 'set_number_at' (inserted next), + because 're_compile_fastmap' needs to know. + Jump to the 'jump_n' we might insert below. */ + INSERT_JUMP2 (succeed_n, laststart, + b + 5 + nbytes, + lower_bound); + b += 5; + + /* Code to initialize the lower bound. Insert + before the 'succeed_n'. The '5' is the last two + bytes of this 'set_number_at', plus 3 bytes of + the following 'succeed_n'. */ + insert_op2 (set_number_at, laststart, 5, + lower_bound, b); + b += 5; + startoffset += 5; + } + + if (upper_bound < 0) + { + /* A negative upper bound stands for infinity, + in which case it degenerates to a plain jump. */ + STORE_JUMP (jump, b, laststart + startoffset); + b += 3; + } + else if (upper_bound > 1) + { /* More than one repetition is allowed, so + append a backward jump to the 'succeed_n' + that starts this interval. + + When we've reached this during matching, + we'll have matched the interval once, so + jump back only 'upper_bound - 1' times. */ + STORE_JUMP2 (jump_n, b, laststart + startoffset, + upper_bound - 1); + b += 5; + + /* The location we want to set is the second + parameter of the 'jump_n'; that is 'b-2' as + an absolute address. 'laststart' will be + the 'set_number_at' we're about to insert; + 'laststart+3' the number to set, the source + for the relative address. But we are + inserting into the middle of the pattern -- + so everything is getting moved up by 5. + Conclusion: (b - 2) - (laststart + 3) + 5, + i.e., b - laststart. + + Insert this at the beginning of the loop + so that if we fail during matching, we'll + reinitialize the bounds. */ + insert_op2 (set_number_at, laststart, b - laststart, + upper_bound - 1, b); + b += 5; + } + } + pending_exact = 0; + beg_interval = NULL; + } + break; + + unfetch_interval: + /* If an invalid interval, match the characters as literals. */ + eassert (beg_interval); + p = beg_interval; + beg_interval = NULL; + eassert (p > pattern && p[-1] == '\\'); + c = '{'; + goto normal_char; + + case '=': + laststart = b; + BUF_PUSH (at_dot); + break; + + case 's': + laststart = b; + PATFETCH (c); + BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]); + break; + + case 'S': + laststart = b; + PATFETCH (c); + BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]); + break; + + case 'c': + laststart = b; + PATFETCH (c); + BUF_PUSH_2 (categoryspec, c); + break; + + case 'C': + laststart = b; + PATFETCH (c); + BUF_PUSH_2 (notcategoryspec, c); + break; + + case 'w': + laststart = b; + BUF_PUSH_2 (syntaxspec, Sword); + break; + + + case 'W': + laststart = b; + BUF_PUSH_2 (notsyntaxspec, Sword); + break; + + + case '<': + laststart = b; + BUF_PUSH (wordbeg); + break; + + case '>': + laststart = b; + BUF_PUSH (wordend); + break; + + case '_': + laststart = b; + PATFETCH (c); + if (c == '<') + BUF_PUSH (symbeg); + else if (c == '>') + BUF_PUSH (symend); + else + FREE_STACK_RETURN (REG_BADPAT); + break; + + case 'b': + BUF_PUSH (wordbound); + break; + + case 'B': + BUF_PUSH (notwordbound); + break; + + case '`': + BUF_PUSH (begbuf); + break; + + case '\'': + BUF_PUSH (endbuf); + break; + + case '1': case '2': case '3': case '4': case '5': + case '6': case '7': case '8': case '9': + { + regnum_t reg = c - '0'; + + if (reg > bufp->re_nsub || reg < 1 + /* Can't back reference to a subexp before its end. */ + || group_in_compile_stack (compile_stack, reg)) + FREE_STACK_RETURN (REG_ESUBREG); + + laststart = b; + BUF_PUSH_2 (duplicate, reg); + } + break; + + default: + /* You might think it would be useful for \ to mean + not to translate; but if we don't translate it + it will never match anything. */ + goto normal_char; + } + break; + + + default: + /* Expects the character in C. */ + normal_char: + /* If no exactn currently being built. */ + if (!pending_exact + + /* If last exactn not at current position. */ + || pending_exact + *pending_exact + 1 != b + + /* Only one byte follows the exactn for the count. */ + || *pending_exact >= (1 << BYTEWIDTH) - MAX_MULTIBYTE_LENGTH + + /* If followed by a repetition operator. */ + || (p != pend + && (*p == '*' || *p == '+' || *p == '?' || *p == '^')) + || (p + 1 < pend && p[0] == '\\' && p[1] == '{')) + { + /* Start building a new exactn. */ + + laststart = b; + + BUF_PUSH_2 (exactn, 0); + pending_exact = b - 1; + } + + GET_BUFFER_SPACE (MAX_MULTIBYTE_LENGTH); + { + int len; + + if (multibyte) + { + c = TRANSLATE (c); + len = CHAR_STRING (c, b); + b += len; + } + else + { + c1 = RE_CHAR_TO_MULTIBYTE (c); + if (! CHAR_BYTE8_P (c1)) + { + int c2 = TRANSLATE (c1); + + if (c1 != c2 && (c1 = RE_CHAR_TO_UNIBYTE (c2)) >= 0) + c = c1; + } + *b++ = c; + len = 1; + } + (*pending_exact) += len; + } + + break; + } /* switch (c) */ + } /* while p != pend */ + + + /* Through the pattern now. */ + + FIXUP_ALT_JUMP (); + + if (!COMPILE_STACK_EMPTY) + FREE_STACK_RETURN (REG_EPAREN); + + /* If we don't want backtracking, force success + the first time we reach the end of the compiled pattern. */ + if (!posix_backtracking) + BUF_PUSH (succeed); + + /* Success; set the length of the buffer. */ + bufp->used = b - bufp->buffer; + +#ifdef REGEX_EMACS_DEBUG + if (regex_emacs_debug > 0) + { + re_compile_fastmap (bufp); + DEBUG_PRINT ("\nCompiled pattern: \n"); + print_compiled_pattern (bufp); + } + regex_emacs_debug--; +#endif + + FREE_STACK_RETURN (REG_NOERROR); + +} /* regex_compile */ + +/* Subroutines for 'regex_compile'. */ + +/* Store OP at LOC followed by two-byte integer parameter ARG. */ + +static void +store_op1 (re_opcode_t op, unsigned char *loc, int arg) +{ + *loc = (unsigned char) op; + STORE_NUMBER (loc + 1, arg); +} + + +/* Like 'store_op1', but for two two-byte parameters ARG1 and ARG2. */ + +static void +store_op2 (re_opcode_t op, unsigned char *loc, int arg1, int arg2) +{ + *loc = (unsigned char) op; + STORE_NUMBER (loc + 1, arg1); + STORE_NUMBER (loc + 3, arg2); +} + + +/* Copy the bytes from LOC to END to open up three bytes of space at LOC + for OP followed by two-byte integer parameter ARG. */ + +static void +insert_op1 (re_opcode_t op, unsigned char *loc, int arg, unsigned char *end) +{ + register unsigned char *pfrom = end; + register unsigned char *pto = end + 3; + + while (pfrom != loc) + *--pto = *--pfrom; + + store_op1 (op, loc, arg); +} + + +/* Like 'insert_op1', but for two two-byte parameters ARG1 and ARG2. */ + +static void +insert_op2 (re_opcode_t op, unsigned char *loc, int arg1, int arg2, + unsigned char *end) +{ + register unsigned char *pfrom = end; + register unsigned char *pto = end + 5; + + while (pfrom != loc) + *--pto = *--pfrom; + + store_op2 (op, loc, arg1, arg2); +} + + +/* P points to just after a ^ in PATTERN. Return true if that ^ comes + after an alternative or a begin-subexpression. Assume there is at + least one character before the ^. */ + +static bool +at_begline_loc_p (re_char *pattern, re_char *p) +{ + re_char *prev = p - 2; + + switch (*prev) + { + case '(': /* After a subexpression. */ + case '|': /* After an alternative. */ + break; + + case ':': /* After a shy subexpression. */ + /* Skip over optional regnum. */ + while (prev > pattern && '0' <= prev[-1] && prev[-1] <= '9') + --prev; + + if (! (prev > pattern + 1 && prev[-1] == '?' && prev[-2] == '(')) + return false; + prev -= 2; + break; + + default: + return false; + } + + /* Count the number of preceding backslashes. */ + p = prev; + while (prev > pattern && prev[-1] == '\\') + --prev; + return (p - prev) & 1; +} + + +/* The dual of at_begline_loc_p. This one is for $. Assume there is + at least one character after the $, i.e., 'P < PEND'. */ + +static bool +at_endline_loc_p (re_char *p, re_char *pend) +{ + /* Before a subexpression or an alternative? */ + return *p == '\\' && p + 1 < pend && (p[1] == ')' || p[1] == '|'); +} + + +/* Returns true if REGNUM is in one of COMPILE_STACK's elements and + false if it's not. */ + +static bool +group_in_compile_stack (compile_stack_type compile_stack, regnum_t regnum) +{ + ptrdiff_t this_element; + + for (this_element = compile_stack.avail - 1; + this_element >= 0; + this_element--) + if (compile_stack.stack[this_element].regnum == regnum) + return true; + + return false; +} + +/* analyze_first. + If fastmap is non-NULL, go through the pattern and fill fastmap + with all the possible leading chars. If fastmap is NULL, don't + bother filling it up (obviously) and only return whether the + pattern could potentially match the empty string. + + Return 1 if p..pend might match the empty string. + Return 0 if p..pend matches at least one char. + Return -1 if fastmap was not updated accurately. */ + +static int +analyze_first (re_char *p, re_char *pend, char *fastmap, + const int multibyte) +{ + int j, k; + bool not; + + /* If all elements for base leading-codes in fastmap is set, this + flag is set true. */ + bool match_any_multibyte_characters = false; + + eassert (p); + + /* The loop below works as follows: + - It has a working-list kept in the PATTERN_STACK and which basically + starts by only containing a pointer to the first operation. + - If the opcode we're looking at is a match against some set of + chars, then we add those chars to the fastmap and go on to the + next work element from the worklist (done via 'break'). + - If the opcode is a control operator on the other hand, we either + ignore it (if it's meaningless at this point, such as 'start_memory') + or execute it (if it's a jump). If the jump has several destinations + (i.e. 'on_failure_jump'), then we push the other destination onto the + worklist. + We guarantee termination by ignoring backward jumps (more or less), + so that P is monotonically increasing. More to the point, we + never set P (or push) anything '<= p1'. */ + + while (p < pend) + { + /* P1 is used as a marker of how far back a 'on_failure_jump' + can go without being ignored. It is normally equal to P + (which prevents any backward 'on_failure_jump') except right + after a plain 'jump', to allow patterns such as: + 0: jump 10 + 3..9: + 10: on_failure_jump 3 + as used for the *? operator. */ + re_char *p1 = p; + + switch (*p++) + { + case succeed: + return 1; + + case duplicate: + /* If the first character has to match a backreference, that means + that the group was empty (since it already matched). Since this + is the only case that interests us here, we can assume that the + backreference must match the empty string. */ + p++; + continue; + + + /* Following are the cases which match a character. These end + with 'break'. */ + + case exactn: + if (fastmap) + { + /* If multibyte is nonzero, the first byte of each + character is an ASCII or a leading code. Otherwise, + each byte is a character. Thus, this works in both + cases. */ + fastmap[p[1]] = 1; + if (! multibyte) + { + /* For the case of matching this unibyte regex + against multibyte, we must set a leading code of + the corresponding multibyte character. */ + int c = RE_CHAR_TO_MULTIBYTE (p[1]); + + fastmap[CHAR_LEADING_CODE (c)] = 1; + } + } + break; + + + case anychar: + /* We could put all the chars except for \n (and maybe \0) + but we don't bother since it is generally not worth it. */ + if (!fastmap) break; + return -1; + + + case charset_not: + if (!fastmap) break; + { + /* Chars beyond end of bitmap are possible matches. */ + for (j = CHARSET_BITMAP_SIZE (&p[-1]) * BYTEWIDTH; + j < (1 << BYTEWIDTH); j++) + fastmap[j] = 1; + } + FALLTHROUGH; + case charset: + if (!fastmap) break; + not = (re_opcode_t) *(p - 1) == charset_not; + for (j = CHARSET_BITMAP_SIZE (&p[-1]) * BYTEWIDTH - 1, p++; + j >= 0; j--) + if (!!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))) ^ not) + fastmap[j] = 1; + + if (/* Any leading code can possibly start a character + which doesn't match the specified set of characters. */ + not + || + /* If we can match a character class, we can match any + multibyte characters. */ + (CHARSET_RANGE_TABLE_EXISTS_P (&p[-2]) + && CHARSET_RANGE_TABLE_BITS (&p[-2]) != 0)) + + { + if (match_any_multibyte_characters == false) + { + for (j = MIN_MULTIBYTE_LEADING_CODE; + j <= MAX_MULTIBYTE_LEADING_CODE; j++) + fastmap[j] = 1; + match_any_multibyte_characters = true; + } + } + + else if (!not && CHARSET_RANGE_TABLE_EXISTS_P (&p[-2]) + && match_any_multibyte_characters == false) + { + /* Set fastmap[I] to 1 where I is a leading code of each + multibyte character in the range table. */ + int c, count; + unsigned char lc1, lc2; + + /* Make P points the range table. '+ 2' is to skip flag + bits for a character class. */ + p += CHARSET_BITMAP_SIZE (&p[-2]) + 2; + + /* Extract the number of ranges in range table into COUNT. */ + EXTRACT_NUMBER_AND_INCR (count, p); + for (; count > 0; count--, p += 3) + { + /* Extract the start and end of each range. */ + EXTRACT_CHARACTER (c, p); + lc1 = CHAR_LEADING_CODE (c); + p += 3; + EXTRACT_CHARACTER (c, p); + lc2 = CHAR_LEADING_CODE (c); + for (j = lc1; j <= lc2; j++) + fastmap[j] = 1; + } + } + break; + + case syntaxspec: + case notsyntaxspec: + if (!fastmap) break; + /* This match depends on text properties. These end with + aborting optimizations. */ + return -1; + + case categoryspec: + case notcategoryspec: + if (!fastmap) break; + not = (re_opcode_t)p[-1] == notcategoryspec; + k = *p++; + for (j = (1 << BYTEWIDTH); j >= 0; j--) + if ((CHAR_HAS_CATEGORY (j, k)) ^ not) + fastmap[j] = 1; + + /* Any leading code can possibly start a character which + has or doesn't has the specified category. */ + if (match_any_multibyte_characters == false) + { + for (j = MIN_MULTIBYTE_LEADING_CODE; + j <= MAX_MULTIBYTE_LEADING_CODE; j++) + fastmap[j] = 1; + match_any_multibyte_characters = true; + } + break; + + /* All cases after this match the empty string. These end with + 'continue'. */ + + case at_dot: + case no_op: + case begline: + case endline: + case begbuf: + case endbuf: + case wordbound: + case notwordbound: + case wordbeg: + case wordend: + case symbeg: + case symend: + continue; + + + case jump: + EXTRACT_NUMBER_AND_INCR (j, p); + if (j < 0) + /* Backward jumps can only go back to code that we've already + visited. 're_compile' should make sure this is true. */ + break; + p += j; + switch (*p) + { + case on_failure_jump: + case on_failure_keep_string_jump: + case on_failure_jump_loop: + case on_failure_jump_nastyloop: + case on_failure_jump_smart: + p++; + break; + default: + continue; + }; + /* Keep P1 to allow the 'on_failure_jump' we are jumping to + to jump back to "just after here". */ + FALLTHROUGH; + case on_failure_jump: + case on_failure_keep_string_jump: + case on_failure_jump_nastyloop: + case on_failure_jump_loop: + case on_failure_jump_smart: + EXTRACT_NUMBER_AND_INCR (j, p); + if (p + j <= p1) + ; /* Backward jump to be ignored. */ + else + { /* We have to look down both arms. + We first go down the "straight" path so as to minimize + stack usage when going through alternatives. */ + int r = analyze_first (p, pend, fastmap, multibyte); + if (r) return r; + p += j; + } + continue; + + + case jump_n: + /* This code simply does not properly handle forward jump_n. */ + DEBUG_STATEMENT (EXTRACT_NUMBER (j, p); eassert (j < 0)); + p += 4; + /* jump_n can either jump or fall through. The (backward) jump + case has already been handled, so we only need to look at the + fallthrough case. */ + continue; + + case succeed_n: + /* If N == 0, it should be an on_failure_jump_loop instead. */ + DEBUG_STATEMENT (EXTRACT_NUMBER (j, p + 2); eassert (j > 0)); + p += 4; + /* We only care about one iteration of the loop, so we don't + need to consider the case where this behaves like an + on_failure_jump. */ + continue; + + + case set_number_at: + p += 4; + continue; + + + case start_memory: + case stop_memory: + p += 1; + continue; + + + default: + abort (); /* We have listed all the cases. */ + } /* switch *p++ */ + + /* Getting here means we have found the possible starting + characters for one path of the pattern -- and that the empty + string does not match. We need not follow this path further. */ + return 0; + } /* while p */ + + /* We reached the end without matching anything. */ + return 1; + +} /* analyze_first */ + +/* Compute a fastmap for the compiled pattern in BUFP. + A fastmap records which of the (1 << BYTEWIDTH) possible + characters can start a string that matches the pattern. This fastmap + is used by re_search to skip quickly over impossible starting points. + + Character codes above (1 << BYTEWIDTH) are not represented in the + fastmap, but the leading codes are represented. Thus, the fastmap + indicates which character sets could start a match. + + The caller must supply the address of a (1 << BYTEWIDTH)-byte data + area as BUFP->fastmap. + + Set the 'fastmap', 'fastmap_accurate', and 'can_be_null' fields in + the pattern buffer. */ + +static void +re_compile_fastmap (struct re_pattern_buffer *bufp) +{ + char *fastmap = bufp->fastmap; + int analysis; + + eassert (fastmap && bufp->buffer); + + memset (fastmap, 0, 1 << BYTEWIDTH); /* Assume nothing's valid. */ + bufp->fastmap_accurate = 1; /* It will be when we're done. */ + + analysis = analyze_first (bufp->buffer, bufp->buffer + bufp->used, + fastmap, RE_MULTIBYTE_P (bufp)); + bufp->can_be_null = (analysis != 0); +} /* re_compile_fastmap */ + +/* Set REGS to hold NUM_REGS registers, storing them in STARTS and + ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use + this memory for recording register information. STARTS and ENDS + must be allocated using the malloc library routine, and must each + be at least NUM_REGS * sizeof (ptrdiff_t) bytes long. + + If NUM_REGS == 0, then subsequent matches should allocate their own + register data. + + Unless this function is called, the first search or match using + PATTERN_BUFFER will allocate its own register data, without + freeing the old data. */ + +void +re_set_registers (struct re_pattern_buffer *bufp, struct re_registers *regs, + unsigned int num_regs, ptrdiff_t *starts, ptrdiff_t *ends) +{ + if (num_regs) + { + bufp->regs_allocated = REGS_REALLOCATE; + regs->num_regs = num_regs; + regs->start = starts; + regs->end = ends; + } + else + { + bufp->regs_allocated = REGS_UNALLOCATED; + regs->num_regs = 0; + regs->start = regs->end = 0; + } +} + +/* Searching routines. */ + +/* Like re_search_2, below, but only one string is specified, and + doesn't let you say where to stop matching. */ + +ptrdiff_t +re_search (struct re_pattern_buffer *bufp, const char *string, size_t size, + ptrdiff_t startpos, ptrdiff_t range, struct re_registers *regs) +{ + return re_search_2 (bufp, NULL, 0, string, size, startpos, range, + regs, size); +} + +/* Head address of virtual concatenation of string. */ +#define HEAD_ADDR_VSTRING(P) \ + (((P) >= size1 ? string2 : string1)) + +/* Address of POS in the concatenation of virtual string. */ +#define POS_ADDR_VSTRING(POS) \ + (((POS) >= size1 ? string2 - size1 : string1) + (POS)) + +/* Using the compiled pattern in BUFP->buffer, first tries to match the + virtual concatenation of STRING1 and STRING2, starting first at index + STARTPOS, then at STARTPOS + 1, and so on. + + STRING1 and STRING2 have length SIZE1 and SIZE2, respectively. + + RANGE is how far to scan while trying to match. RANGE = 0 means try + only at STARTPOS; in general, the last start tried is STARTPOS + + RANGE. + + In REGS, return the indices of the virtual concatenation of STRING1 + and STRING2 that matched the entire BUFP->buffer and its contained + subexpressions. + + Do not consider matching one past the index STOP in the virtual + concatenation of STRING1 and STRING2. + + Return either the position in the strings at which the match was + found, -1 if no match, or -2 if error (such as failure + stack overflow). */ + +ptrdiff_t +re_search_2 (struct re_pattern_buffer *bufp, const char *str1, size_t size1, + const char *str2, size_t size2, + ptrdiff_t startpos, ptrdiff_t range, + struct re_registers *regs, ptrdiff_t stop) +{ + ptrdiff_t val; + re_char *string1 = (re_char *) str1; + re_char *string2 = (re_char *) str2; + char *fastmap = bufp->fastmap; + Lisp_Object translate = bufp->translate; + size_t total_size = size1 + size2; + ptrdiff_t endpos = startpos + range; + bool anchored_start; + /* Nonzero if we are searching multibyte string. */ + bool multibyte = RE_TARGET_MULTIBYTE_P (bufp); + + /* Check for out-of-range STARTPOS. */ + if (startpos < 0 || startpos > total_size) + return -1; + + /* Fix up RANGE if it might eventually take us outside + the virtual concatenation of STRING1 and STRING2. + Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */ + if (endpos < 0) + range = 0 - startpos; + else if (endpos > total_size) + range = total_size - startpos; + + /* If the search isn't to be a backwards one, don't waste time in a + search for a pattern anchored at beginning of buffer. */ + if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0) + { + if (startpos > 0) + return -1; + else + range = 0; + } + + /* In a forward search for something that starts with \=. + don't keep searching past point. */ + if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0) + { + range = PT_BYTE - BEGV_BYTE - startpos; + if (range < 0) + return -1; + } + + /* Update the fastmap now if not correct already. */ + if (fastmap && !bufp->fastmap_accurate) + re_compile_fastmap (bufp); + + /* See whether the pattern is anchored. */ + anchored_start = (bufp->buffer[0] == begline); + + gl_state.object = re_match_object; /* Used by SYNTAX_TABLE_BYTE_TO_CHAR. */ + { + ptrdiff_t charpos = SYNTAX_TABLE_BYTE_TO_CHAR (POS_AS_IN_BUFFER (startpos)); + + SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object, charpos, 1); + } + + /* Loop through the string, looking for a place to start matching. */ + for (;;) + { + /* If the pattern is anchored, + skip quickly past places we cannot match. + Don't bother to treat startpos == 0 specially + because that case doesn't repeat. */ + if (anchored_start && startpos > 0) + { + if (! ((startpos <= size1 ? string1[startpos - 1] + : string2[startpos - size1 - 1]) + == '\n')) + goto advance; + } + + /* If a fastmap is supplied, skip quickly over characters that + cannot be the start of a match. If the pattern can match the + null string, however, we don't need to skip characters; we want + the first null string. */ + if (fastmap && startpos < total_size && !bufp->can_be_null) + { + re_char *d; + int buf_ch; + + d = POS_ADDR_VSTRING (startpos); + + if (range > 0) /* Searching forwards. */ + { + ptrdiff_t irange = range, lim = 0; + + if (startpos < size1 && startpos + range >= size1) + lim = range - (size1 - startpos); + + /* Written out as an if-else to avoid testing 'translate' + inside the loop. */ + if (!NILP (translate)) + { + if (multibyte) + while (range > lim) + { + int buf_charlen; + + buf_ch = STRING_CHAR_AND_LENGTH (d, buf_charlen); + buf_ch = RE_TRANSLATE (translate, buf_ch); + if (fastmap[CHAR_LEADING_CODE (buf_ch)]) + break; + + range -= buf_charlen; + d += buf_charlen; + } + else + while (range > lim) + { + buf_ch = *d; + int ch = RE_CHAR_TO_MULTIBYTE (buf_ch); + int translated = RE_TRANSLATE (translate, ch); + if (translated != ch + && (ch = RE_CHAR_TO_UNIBYTE (translated)) >= 0) + buf_ch = ch; + if (fastmap[buf_ch]) + break; + d++; + range--; + } + } + else + { + if (multibyte) + while (range > lim) + { + int buf_charlen; + + buf_ch = STRING_CHAR_AND_LENGTH (d, buf_charlen); + if (fastmap[CHAR_LEADING_CODE (buf_ch)]) + break; + range -= buf_charlen; + d += buf_charlen; + } + else + while (range > lim && !fastmap[*d]) + { + d++; + range--; + } + } + startpos += irange - range; + } + else /* Searching backwards. */ + { + if (multibyte) + { + buf_ch = STRING_CHAR (d); + buf_ch = TRANSLATE (buf_ch); + if (! fastmap[CHAR_LEADING_CODE (buf_ch)]) + goto advance; + } + else + { + buf_ch = *d; + int ch = RE_CHAR_TO_MULTIBYTE (buf_ch); + int translated = TRANSLATE (ch); + if (translated != ch + && (ch = RE_CHAR_TO_UNIBYTE (translated)) >= 0) + buf_ch = ch; + if (! fastmap[TRANSLATE (buf_ch)]) + goto advance; + } + } + } + + /* If can't match the null string, and that's all we have left, fail. */ + if (range >= 0 && startpos == total_size && fastmap + && !bufp->can_be_null) + return -1; + + val = re_match_2_internal (bufp, string1, size1, string2, size2, + startpos, regs, stop); + + if (val >= 0) + return startpos; + + if (val == -2) + return -2; + + advance: + if (!range) + break; + else if (range > 0) + { + /* Update STARTPOS to the next character boundary. */ + if (multibyte) + { + re_char *p = POS_ADDR_VSTRING (startpos); + int len = BYTES_BY_CHAR_HEAD (*p); + + range -= len; + if (range < 0) + break; + startpos += len; + } + else + { + range--; + startpos++; + } + } + else + { + range++; + startpos--; + + /* Update STARTPOS to the previous character boundary. */ + if (multibyte) + { + re_char *p = POS_ADDR_VSTRING (startpos) + 1; + re_char *p0 = p; + re_char *phead = HEAD_ADDR_VSTRING (startpos); + + /* Find the head of multibyte form. */ + PREV_CHAR_BOUNDARY (p, phead); + range += p0 - 1 - p; + if (range > 0) + break; + + startpos -= p0 - 1 - p; + } + } + } + return -1; +} /* re_search_2 */ + +/* Declarations and macros for re_match_2. */ + +static int bcmp_translate (re_char *s1, re_char *s2, + ptrdiff_t len, + Lisp_Object translate, + const int multibyte); + +/* This converts PTR, a pointer into one of the search strings 'string1' + and 'string2' into an offset from the beginning of that string. */ +#define POINTER_TO_OFFSET(ptr) \ + (FIRST_STRING_P (ptr) \ + ? (ptr) - string1 \ + : (ptr) - string2 + (ptrdiff_t) size1) + +/* Call before fetching a character with *d. This switches over to + string2 if necessary. + Check re_match_2_internal for a discussion of why end_match_2 might + not be within string2 (but be equal to end_match_1 instead). */ +#define PREFETCH() \ + while (d == dend) \ + { \ + /* End of string2 => fail. */ \ + if (dend == end_match_2) \ + goto fail; \ + /* End of string1 => advance to string2. */ \ + d = string2; \ + dend = end_match_2; \ + } + +/* Call before fetching a char with *d if you already checked other limits. + This is meant for use in lookahead operations like wordend, etc.. + where we might need to look at parts of the string that might be + outside of the LIMITs (i.e past 'stop'). */ +#define PREFETCH_NOLIMIT() \ + if (d == end1) \ + { \ + d = string2; \ + dend = end_match_2; \ + } \ + +/* Test if at very beginning or at very end of the virtual concatenation + of STRING1 and STRING2. If only one string, it's STRING2. */ +#define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2) +#define AT_STRINGS_END(d) ((d) == end2) + +/* Disabled due to a compiler bug -- see comment at case wordbound */ + +/* The comment at case wordbound is following one, but we don't use + AT_WORD_BOUNDARY anymore to support multibyte form. + + The DEC Alpha C compiler 3.x generates incorrect code for the + test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of + AT_WORD_BOUNDARY, so this code is disabled. Expanding the + macro and introducing temporary variables works around the bug. */ + +#if 0 +/* Test if D points to a character which is word-constituent. We have + two special cases to check for: if past the end of string1, look at + the first character in string2; and if before the beginning of + string2, look at the last character in string1. */ +#define WORDCHAR_P(d) \ + (SYNTAX ((d) == end1 ? *string2 \ + : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \ + == Sword) + +/* Test if the character before D and the one at D differ with respect + to being word-constituent. */ +#define AT_WORD_BOUNDARY(d) \ + (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \ + || WORDCHAR_P (d - 1) != WORDCHAR_P (d)) +#endif + + +/* Optimization routines. */ + +/* If the operation is a match against one or more chars, + return a pointer to the next operation, else return NULL. */ +static re_char * +skip_one_char (re_char *p) +{ + switch (*p++) + { + case anychar: + break; + + case exactn: + p += *p + 1; + break; + + case charset_not: + case charset: + if (CHARSET_RANGE_TABLE_EXISTS_P (p - 1)) + { + int mcnt; + p = CHARSET_RANGE_TABLE (p - 1); + EXTRACT_NUMBER_AND_INCR (mcnt, p); + p = CHARSET_RANGE_TABLE_END (p, mcnt); + } + else + p += 1 + CHARSET_BITMAP_SIZE (p - 1); + break; + + case syntaxspec: + case notsyntaxspec: + case categoryspec: + case notcategoryspec: + p++; + break; + + default: + p = NULL; + } + return p; +} + + +/* Jump over non-matching operations. */ +static re_char * +skip_noops (re_char *p, re_char *pend) +{ + int mcnt; + while (p < pend) + { + switch (*p) + { + case start_memory: + case stop_memory: + p += 2; break; + case no_op: + p += 1; break; + case jump: + p += 1; + EXTRACT_NUMBER_AND_INCR (mcnt, p); + p += mcnt; + break; + default: + return p; + } + } + eassert (p == pend); + return p; +} + +/* Test if C matches charset op. *PP points to the charset or charset_not + opcode. When the function finishes, *PP will be advanced past that opcode. + C is character to test (possibly after translations) and CORIG is original + character (i.e. without any translations). UNIBYTE denotes whether c is + unibyte or multibyte character. */ +static bool +execute_charset (re_char **pp, unsigned c, unsigned corig, bool unibyte) +{ + re_char *p = *pp, *rtp = NULL; + bool not = (re_opcode_t) *p == charset_not; + + if (CHARSET_RANGE_TABLE_EXISTS_P (p)) + { + int count; + rtp = CHARSET_RANGE_TABLE (p); + EXTRACT_NUMBER_AND_INCR (count, rtp); + *pp = CHARSET_RANGE_TABLE_END ((rtp), (count)); + } + else + *pp += 2 + CHARSET_BITMAP_SIZE (p); + + if (unibyte && c < (1 << BYTEWIDTH)) + { /* Lookup bitmap. */ + /* Cast to 'unsigned' instead of 'unsigned char' in + case the bit list is a full 32 bytes long. */ + if (c < (unsigned) (CHARSET_BITMAP_SIZE (p) * BYTEWIDTH) + && p[2 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) + return !not; + } + else if (rtp) + { + int class_bits = CHARSET_RANGE_TABLE_BITS (p); + int range_start, range_end; + + /* Sort tests by the most commonly used classes with some adjustment to which + tests are easiest to perform. Take a look at comment in re_wctype_parse + for table with frequencies of character class names. */ + + if ((class_bits & BIT_MULTIBYTE) || + (class_bits & BIT_ALNUM && ISALNUM (c)) || + (class_bits & BIT_ALPHA && ISALPHA (c)) || + (class_bits & BIT_SPACE && ISSPACE (c)) || + (class_bits & BIT_BLANK && ISBLANK (c)) || + (class_bits & BIT_WORD && ISWORD (c)) || + ((class_bits & BIT_UPPER) && + (ISUPPER (c) || (corig != c && + c == downcase (corig) && ISLOWER (c)))) || + ((class_bits & BIT_LOWER) && + (ISLOWER (c) || (corig != c && + c == upcase (corig) && ISUPPER(c)))) || + (class_bits & BIT_PUNCT && ISPUNCT (c)) || + (class_bits & BIT_GRAPH && ISGRAPH (c)) || + (class_bits & BIT_PRINT && ISPRINT (c))) + return !not; + + for (p = *pp; rtp < p; rtp += 2 * 3) + { + EXTRACT_CHARACTER (range_start, rtp); + EXTRACT_CHARACTER (range_end, rtp + 3); + if (range_start <= c && c <= range_end) + return !not; + } + } + + return not; +} + +/* Non-zero if "p1 matches something" implies "p2 fails". */ +static int +mutually_exclusive_p (struct re_pattern_buffer *bufp, re_char *p1, + re_char *p2) +{ + re_opcode_t op2; + bool multibyte = RE_MULTIBYTE_P (bufp); + unsigned char *pend = bufp->buffer + bufp->used; + + eassert (p1 >= bufp->buffer && p1 < pend + && p2 >= bufp->buffer && p2 <= pend); + + /* Skip over open/close-group commands. + If what follows this loop is a ...+ construct, + look at what begins its body, since we will have to + match at least one of that. */ + p2 = skip_noops (p2, pend); + /* The same skip can be done for p1, except that this function + is only used in the case where p1 is a simple match operator. */ + /* p1 = skip_noops (p1, pend); */ + + eassert (p1 >= bufp->buffer && p1 < pend + && p2 >= bufp->buffer && p2 <= pend); + + op2 = p2 == pend ? succeed : *p2; + + switch (op2) + { + case succeed: + case endbuf: + /* If we're at the end of the pattern, we can change. */ + if (skip_one_char (p1)) + { + DEBUG_PRINT (" End of pattern: fast loop.\n"); + return 1; + } + break; + + case endline: + case exactn: + { + int c + = (re_opcode_t) *p2 == endline ? '\n' + : RE_STRING_CHAR (p2 + 2, multibyte); + + if ((re_opcode_t) *p1 == exactn) + { + if (c != RE_STRING_CHAR (p1 + 2, multibyte)) + { + DEBUG_PRINT (" '%c' != '%c' => fast loop.\n", c, p1[2]); + return 1; + } + } + + else if ((re_opcode_t) *p1 == charset + || (re_opcode_t) *p1 == charset_not) + { + if (!execute_charset (&p1, c, c, !multibyte || ASCII_CHAR_P (c))) + { + DEBUG_PRINT (" No match => fast loop.\n"); + return 1; + } + } + else if ((re_opcode_t) *p1 == anychar + && c == '\n') + { + DEBUG_PRINT (" . != \\n => fast loop.\n"); + return 1; + } + } + break; + + case charset: + { + if ((re_opcode_t) *p1 == exactn) + /* Reuse the code above. */ + return mutually_exclusive_p (bufp, p2, p1); + + /* It is hard to list up all the character in charset + P2 if it includes multibyte character. Give up in + such case. */ + else if (!multibyte || !CHARSET_RANGE_TABLE_EXISTS_P (p2)) + { + /* Now, we are sure that P2 has no range table. + So, for the size of bitmap in P2, 'p2[1]' is + enough. But P1 may have range table, so the + size of bitmap table of P1 is extracted by + using macro 'CHARSET_BITMAP_SIZE'. + + In a multibyte case, we know that all the character + listed in P2 is ASCII. In a unibyte case, P1 has only a + bitmap table. So, in both cases, it is enough to test + only the bitmap table of P1. */ + + if ((re_opcode_t) *p1 == charset) + { + int idx; + /* We win if the charset inside the loop + has no overlap with the one after the loop. */ + for (idx = 0; + (idx < (int) p2[1] + && idx < CHARSET_BITMAP_SIZE (p1)); + idx++) + if ((p2[2 + idx] & p1[2 + idx]) != 0) + break; + + if (idx == p2[1] + || idx == CHARSET_BITMAP_SIZE (p1)) + { + DEBUG_PRINT (" No match => fast loop.\n"); + return 1; + } + } + else if ((re_opcode_t) *p1 == charset_not) + { + int idx; + /* We win if the charset_not inside the loop lists + every character listed in the charset after. */ + for (idx = 0; idx < (int) p2[1]; idx++) + if (! (p2[2 + idx] == 0 + || (idx < CHARSET_BITMAP_SIZE (p1) + && ((p2[2 + idx] & ~ p1[2 + idx]) == 0)))) + break; + + if (idx == p2[1]) + { + DEBUG_PRINT (" No match => fast loop.\n"); + return 1; + } + } + } + } + break; + + case charset_not: + switch (*p1) + { + case exactn: + case charset: + /* Reuse the code above. */ + return mutually_exclusive_p (bufp, p2, p1); + case charset_not: + /* When we have two charset_not, it's very unlikely that + they don't overlap. The union of the two sets of excluded + chars should cover all possible chars, which, as a matter of + fact, is virtually impossible in multibyte buffers. */ + break; + } + break; + + case wordend: + return ((re_opcode_t) *p1 == syntaxspec && p1[1] == Sword); + case symend: + return ((re_opcode_t) *p1 == syntaxspec + && (p1[1] == Ssymbol || p1[1] == Sword)); + case notsyntaxspec: + return ((re_opcode_t) *p1 == syntaxspec && p1[1] == p2[1]); + + case wordbeg: + return ((re_opcode_t) *p1 == notsyntaxspec && p1[1] == Sword); + case symbeg: + return ((re_opcode_t) *p1 == notsyntaxspec + && (p1[1] == Ssymbol || p1[1] == Sword)); + case syntaxspec: + return ((re_opcode_t) *p1 == notsyntaxspec && p1[1] == p2[1]); + + case wordbound: + return (((re_opcode_t) *p1 == notsyntaxspec + || (re_opcode_t) *p1 == syntaxspec) + && p1[1] == Sword); + + case categoryspec: + return ((re_opcode_t) *p1 == notcategoryspec && p1[1] == p2[1]); + case notcategoryspec: + return ((re_opcode_t) *p1 == categoryspec && p1[1] == p2[1]); + + default: + ; + } + + /* Safe default. */ + return 0; +} + + +/* Matching routines. */ + +/* re_match_2 matches the compiled pattern in BUFP against the + the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1 + and SIZE2, respectively). We start matching at POS, and stop + matching at STOP. + + If REGS is non-null, store offsets for the substring each group + matched in REGS. + + We return -1 if no match, -2 if an internal error (such as the + failure stack overflowing). Otherwise, we return the length of the + matched substring. */ + +ptrdiff_t +re_match_2 (struct re_pattern_buffer *bufp, const char *string1, + size_t size1, const char *string2, size_t size2, ptrdiff_t pos, + struct re_registers *regs, ptrdiff_t stop) +{ + ptrdiff_t result; + + ptrdiff_t charpos; + gl_state.object = re_match_object; /* Used by SYNTAX_TABLE_BYTE_TO_CHAR. */ + charpos = SYNTAX_TABLE_BYTE_TO_CHAR (POS_AS_IN_BUFFER (pos)); + SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object, charpos, 1); + + result = re_match_2_internal (bufp, (re_char *) string1, size1, + (re_char *) string2, size2, + pos, regs, stop); + return result; +} + + +/* This is a separate function so that we can force an alloca cleanup + afterwards. */ +static ptrdiff_t +re_match_2_internal (struct re_pattern_buffer *bufp, re_char *string1, + size_t size1, re_char *string2, size_t size2, + ptrdiff_t pos, struct re_registers *regs, ptrdiff_t stop) +{ + /* General temporaries. */ + int mcnt; + size_t reg; + + /* Just past the end of the corresponding string. */ + re_char *end1, *end2; + + /* Pointers into string1 and string2, just past the last characters in + each to consider matching. */ + re_char *end_match_1, *end_match_2; + + /* Where we are in the data, and the end of the current string. */ + re_char *d, *dend; + + /* Used sometimes to remember where we were before starting matching + an operator so that we can go back in case of failure. This "atomic" + behavior of matching opcodes is indispensable to the correctness + of the on_failure_keep_string_jump optimization. */ + re_char *dfail; + + /* Where we are in the pattern, and the end of the pattern. */ + re_char *p = bufp->buffer; + re_char *pend = p + bufp->used; + + /* We use this to map every character in the string. */ + Lisp_Object translate = bufp->translate; + + /* True if BUFP is setup from a multibyte regex. */ + bool multibyte = RE_MULTIBYTE_P (bufp); + + /* True if STRING1/STRING2 are multibyte. */ + bool target_multibyte = RE_TARGET_MULTIBYTE_P (bufp); + + /* Failure point stack. Each place that can handle a failure further + down the line pushes a failure point on this stack. It consists of + regstart, and regend for all registers corresponding to + the subexpressions we're currently inside, plus the number of such + registers, and, finally, two char *'s. The first char * is where + to resume scanning the pattern; the second one is where to resume + scanning the strings. */ + fail_stack_type fail_stack; +#ifdef DEBUG_COMPILES_ARGUMENTS + unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0; +#endif + + /* We fill all the registers internally, independent of what we + return, for use in backreferences. The number here includes + an element for register zero. */ + size_t num_regs = bufp->re_nsub + 1; + + /* Information on the contents of registers. These are pointers into + the input strings; they record just what was matched (on this + attempt) by a subexpression part of the pattern, that is, the + regnum-th regstart pointer points to where in the pattern we began + matching and the regnum-th regend points to right after where we + stopped matching the regnum-th subexpression. (The zeroth register + keeps track of what the whole pattern matches.) */ + re_char **regstart UNINIT, **regend UNINIT; + + /* The following record the register info as found in the above + variables when we find a match better than any we've seen before. + This happens as we backtrack through the failure points, which in + turn happens only if we have not yet matched the entire string. */ + unsigned best_regs_set = false; + re_char **best_regstart UNINIT, **best_regend UNINIT; + + /* Logically, this is 'best_regend[0]'. But we don't want to have to + allocate space for that if we're not allocating space for anything + else (see below). Also, we never need info about register 0 for + any of the other register vectors, and it seems rather a kludge to + treat 'best_regend' differently than the rest. So we keep track of + the end of the best match so far in a separate variable. We + initialize this to NULL so that when we backtrack the first time + and need to test it, it's not garbage. */ + re_char *match_end = NULL; + +#ifdef DEBUG_COMPILES_ARGUMENTS + /* Counts the total number of registers pushed. */ + unsigned num_regs_pushed = 0; +#endif + + DEBUG_PRINT ("\n\nEntering re_match_2.\n"); + + REGEX_USE_SAFE_ALLOCA; + + INIT_FAIL_STACK (); + + /* Do not bother to initialize all the register variables if there are + no groups in the pattern, as it takes a fair amount of time. If + there are groups, we include space for register 0 (the whole + pattern), even though we never use it, since it simplifies the + array indexing. We should fix this. */ + if (bufp->re_nsub) + { + regstart = SAFE_ALLOCA (num_regs * 4 * sizeof *regstart); + regend = regstart + num_regs; + best_regstart = regend + num_regs; + best_regend = best_regstart + num_regs; + } + + /* The starting position is bogus. */ + if (pos < 0 || pos > size1 + size2) + { + SAFE_FREE (); + return -1; + } + + /* Initialize subexpression text positions to -1 to mark ones that no + start_memory/stop_memory has been seen for. Also initialize the + register information struct. */ + for (reg = 1; reg < num_regs; reg++) + regstart[reg] = regend[reg] = NULL; + + /* We move 'string1' into 'string2' if the latter's empty -- but not if + 'string1' is null. */ + if (size2 == 0 && string1 != NULL) + { + string2 = string1; + size2 = size1; + string1 = 0; + size1 = 0; + } + end1 = string1 + size1; + end2 = string2 + size2; + + /* P scans through the pattern as D scans through the data. + DEND is the end of the input string that D points within. + Advance D into the following input string whenever necessary, but + this happens before fetching; therefore, at the beginning of the + loop, D can be pointing at the end of a string, but it cannot + equal STRING2. */ + if (pos >= size1) + { + /* Only match within string2. */ + d = string2 + pos - size1; + dend = end_match_2 = string2 + stop - size1; + end_match_1 = end1; /* Just to give it a value. */ + } + else + { + if (stop < size1) + { + /* Only match within string1. */ + end_match_1 = string1 + stop; + /* BEWARE! + When we reach end_match_1, PREFETCH normally switches to string2. + But in the present case, this means that just doing a PREFETCH + makes us jump from 'stop' to 'gap' within the string. + What we really want here is for the search to stop as + soon as we hit end_match_1. That's why we set end_match_2 + to end_match_1 (since PREFETCH fails as soon as we hit + end_match_2). */ + end_match_2 = end_match_1; + } + else + { /* It's important to use this code when STOP == SIZE so that + moving D from end1 to string2 will not prevent the D == DEND + check from catching the end of string. */ + end_match_1 = end1; + end_match_2 = string2 + stop - size1; + } + d = string1 + pos; + dend = end_match_1; + } + + DEBUG_PRINT ("The compiled pattern is: "); + DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend); + DEBUG_PRINT ("The string to match is: \""); + DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2); + DEBUG_PRINT ("\"\n"); + + /* This loops over pattern commands. It exits by returning from the + function if the match is complete, or it drops through if the match + fails at this starting point in the input data. */ + for (;;) + { + DEBUG_PRINT ("\n%p: ", p); + + if (p == pend) + { + /* End of pattern means we might have succeeded. */ + DEBUG_PRINT ("end of pattern ... "); + + /* If we haven't matched the entire string, and we want the + longest match, try backtracking. */ + if (d != end_match_2) + { + /* True if this match is the best seen so far. */ + bool best_match_p; + + { + /* True if this match ends in the same string (string1 + or string2) as the best previous match. */ + bool same_str_p = (FIRST_STRING_P (match_end) + == FIRST_STRING_P (d)); + + /* AIX compiler got confused when this was combined + with the previous declaration. */ + if (same_str_p) + best_match_p = d > match_end; + else + best_match_p = !FIRST_STRING_P (d); + } + + DEBUG_PRINT ("backtracking.\n"); + + if (!FAIL_STACK_EMPTY ()) + { /* More failure points to try. */ + + /* If exceeds best match so far, save it. */ + if (!best_regs_set || best_match_p) + { + best_regs_set = true; + match_end = d; + + DEBUG_PRINT ("\nSAVING match as best so far.\n"); + + for (reg = 1; reg < num_regs; reg++) + { + best_regstart[reg] = regstart[reg]; + best_regend[reg] = regend[reg]; + } + } + goto fail; + } + + /* If no failure points, don't restore garbage. And if + last match is real best match, don't restore second + best one. */ + else if (best_regs_set && !best_match_p) + { + restore_best_regs: + /* Restore best match. It may happen that 'dend == + end_match_1' while the restored d is in string2. + For example, the pattern 'x.*y.*z' against the + strings 'x-' and 'y-z-', if the two strings are + not consecutive in memory. */ + DEBUG_PRINT ("Restoring best registers.\n"); + + d = match_end; + dend = ((d >= string1 && d <= end1) + ? end_match_1 : end_match_2); + + for (reg = 1; reg < num_regs; reg++) + { + regstart[reg] = best_regstart[reg]; + regend[reg] = best_regend[reg]; + } + } + } /* d != end_match_2 */ + + succeed_label: + DEBUG_PRINT ("Accepting match.\n"); + + /* If caller wants register contents data back, do it. */ + if (regs) + { + /* Have the register data arrays been allocated? */ + if (bufp->regs_allocated == REGS_UNALLOCATED) + { /* No. So allocate them with malloc. We need one + extra element beyond 'num_regs' for the '-1' marker + GNU code uses. */ + regs->num_regs = max (RE_NREGS, num_regs + 1); + regs->start = TALLOC (regs->num_regs, ptrdiff_t); + regs->end = TALLOC (regs->num_regs, ptrdiff_t); + bufp->regs_allocated = REGS_REALLOCATE; + } + else if (bufp->regs_allocated == REGS_REALLOCATE) + { /* Yes. If we need more elements than were already + allocated, reallocate them. If we need fewer, just + leave it alone. */ + if (regs->num_regs < num_regs + 1) + { + regs->num_regs = num_regs + 1; + RETALLOC (regs->start, regs->num_regs, ptrdiff_t); + RETALLOC (regs->end, regs->num_regs, ptrdiff_t); + } + } + else + eassert (bufp->regs_allocated == REGS_FIXED); + + /* Convert the pointer data in 'regstart' and 'regend' to + indices. Register zero has to be set differently, + since we haven't kept track of any info for it. */ + if (regs->num_regs > 0) + { + regs->start[0] = pos; + regs->end[0] = POINTER_TO_OFFSET (d); + } + + /* Go through the first 'min (num_regs, regs->num_regs)' + registers, since that is all we initialized. */ + for (reg = 1; reg < min (num_regs, regs->num_regs); reg++) + { + if (REG_UNSET (regstart[reg]) || REG_UNSET (regend[reg])) + regs->start[reg] = regs->end[reg] = -1; + else + { + regs->start[reg] = POINTER_TO_OFFSET (regstart[reg]); + regs->end[reg] = POINTER_TO_OFFSET (regend[reg]); + } + } + + /* If the regs structure we return has more elements than + were in the pattern, set the extra elements to -1. If + we (re)allocated the registers, this is the case, + because we always allocate enough to have at least one + -1 at the end. */ + for (reg = num_regs; reg < regs->num_regs; reg++) + regs->start[reg] = regs->end[reg] = -1; + } + + DEBUG_PRINT ("%u failure points pushed, %u popped (%u remain).\n", + nfailure_points_pushed, nfailure_points_popped, + nfailure_points_pushed - nfailure_points_popped); + DEBUG_PRINT ("%u registers pushed.\n", num_regs_pushed); + + ptrdiff_t dcnt = POINTER_TO_OFFSET (d) - pos; + + DEBUG_PRINT ("Returning %td from re_match_2.\n", dcnt); + + SAFE_FREE (); + return dcnt; + } + + /* Otherwise match next pattern command. */ + switch (*p++) + { + /* Ignore these. Used to ignore the n of succeed_n's which + currently have n == 0. */ + case no_op: + DEBUG_PRINT ("EXECUTING no_op.\n"); + break; + + case succeed: + DEBUG_PRINT ("EXECUTING succeed.\n"); + goto succeed_label; + + /* Match the next n pattern characters exactly. The following + byte in the pattern defines n, and the n bytes after that + are the characters to match. */ + case exactn: + mcnt = *p++; + DEBUG_PRINT ("EXECUTING exactn %d.\n", mcnt); + + /* Remember the start point to rollback upon failure. */ + dfail = d; + + /* The cost of testing 'translate' is comparatively small. */ + if (target_multibyte) + do + { + int pat_charlen, buf_charlen; + int pat_ch, buf_ch; + + PREFETCH (); + if (multibyte) + pat_ch = STRING_CHAR_AND_LENGTH (p, pat_charlen); + else + { + pat_ch = RE_CHAR_TO_MULTIBYTE (*p); + pat_charlen = 1; + } + buf_ch = STRING_CHAR_AND_LENGTH (d, buf_charlen); + + if (TRANSLATE (buf_ch) != pat_ch) + { + d = dfail; + goto fail; + } + + p += pat_charlen; + d += buf_charlen; + mcnt -= pat_charlen; + } + while (mcnt > 0); + else + do + { + int pat_charlen; + int pat_ch, buf_ch; + + PREFETCH (); + if (multibyte) + { + pat_ch = STRING_CHAR_AND_LENGTH (p, pat_charlen); + pat_ch = RE_CHAR_TO_UNIBYTE (pat_ch); + } + else + { + pat_ch = *p; + pat_charlen = 1; + } + buf_ch = RE_CHAR_TO_MULTIBYTE (*d); + if (! CHAR_BYTE8_P (buf_ch)) + { + buf_ch = TRANSLATE (buf_ch); + buf_ch = RE_CHAR_TO_UNIBYTE (buf_ch); + if (buf_ch < 0) + buf_ch = *d; + } + else + buf_ch = *d; + if (buf_ch != pat_ch) + { + d = dfail; + goto fail; + } + p += pat_charlen; + d++; + } + while (--mcnt); + + break; + + + /* Match any character except newline. */ + case anychar: + { + int buf_charlen; + int buf_ch; + + DEBUG_PRINT ("EXECUTING anychar.\n"); + + PREFETCH (); + buf_ch = RE_STRING_CHAR_AND_LENGTH (d, buf_charlen, + target_multibyte); + buf_ch = TRANSLATE (buf_ch); + if (buf_ch == '\n') + goto fail; + + DEBUG_PRINT (" Matched \"%d\".\n", *d); + d += buf_charlen; + } + break; + + + case charset: + case charset_not: + { + register unsigned int c, corig; + int len; + + /* Whether matching against a unibyte character. */ + bool unibyte_char = false; + + DEBUG_PRINT ("EXECUTING charset%s.\n", + (re_opcode_t) *(p - 1) == charset_not ? "_not" : ""); + + PREFETCH (); + corig = c = RE_STRING_CHAR_AND_LENGTH (d, len, target_multibyte); + if (target_multibyte) + { + int c1; + + c = TRANSLATE (c); + c1 = RE_CHAR_TO_UNIBYTE (c); + if (c1 >= 0) + { + unibyte_char = true; + c = c1; + } + } + else + { + int c1 = RE_CHAR_TO_MULTIBYTE (c); + + if (! CHAR_BYTE8_P (c1)) + { + c1 = TRANSLATE (c1); + c1 = RE_CHAR_TO_UNIBYTE (c1); + if (c1 >= 0) + { + unibyte_char = true; + c = c1; + } + } + else + unibyte_char = true; + } + + p -= 1; + if (!execute_charset (&p, c, corig, unibyte_char)) + goto fail; + + d += len; + } + break; + + + /* The beginning of a group is represented by start_memory. + The argument is the register number. The text + matched within the group is recorded (in the internal + registers data structure) under the register number. */ + case start_memory: + DEBUG_PRINT ("EXECUTING start_memory %d:\n", *p); + + /* In case we need to undo this operation (via backtracking). */ + PUSH_FAILURE_REG (*p); + + regstart[*p] = d; + regend[*p] = NULL; /* probably unnecessary. -sm */ + DEBUG_PRINT (" regstart: %td\n", POINTER_TO_OFFSET (regstart[*p])); + + /* Move past the register number and inner group count. */ + p += 1; + break; + + + /* The stop_memory opcode represents the end of a group. Its + argument is the same as start_memory's: the register number. */ + case stop_memory: + DEBUG_PRINT ("EXECUTING stop_memory %d:\n", *p); + + eassert (!REG_UNSET (regstart[*p])); + /* Strictly speaking, there should be code such as: + + eassert (REG_UNSET (regend[*p])); + PUSH_FAILURE_REGSTOP ((unsigned int)*p); + + But the only info to be pushed is regend[*p] and it is known to + be UNSET, so there really isn't anything to push. + Not pushing anything, on the other hand deprives us from the + guarantee that regend[*p] is UNSET since undoing this operation + will not reset its value properly. This is not important since + the value will only be read on the next start_memory or at + the very end and both events can only happen if this stop_memory + is *not* undone. */ + + regend[*p] = d; + DEBUG_PRINT (" regend: %td\n", POINTER_TO_OFFSET (regend[*p])); + + /* Move past the register number and the inner group count. */ + p += 1; + break; + + + /* \ has been turned into a 'duplicate' command which is + followed by the numeric value of as the register number. */ + case duplicate: + { + re_char *d2, *dend2; + int regno = *p++; /* Get which register to match against. */ + DEBUG_PRINT ("EXECUTING duplicate %d.\n", regno); + + /* Can't back reference a group which we've never matched. */ + if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno])) + goto fail; + + /* Where in input to try to start matching. */ + d2 = regstart[regno]; + + /* Remember the start point to rollback upon failure. */ + dfail = d; + + /* Where to stop matching; if both the place to start and + the place to stop matching are in the same string, then + set to the place to stop, otherwise, for now have to use + the end of the first string. */ + + dend2 = ((FIRST_STRING_P (regstart[regno]) + == FIRST_STRING_P (regend[regno])) + ? regend[regno] : end_match_1); + for (;;) + { + ptrdiff_t dcnt; + + /* If necessary, advance to next segment in register + contents. */ + while (d2 == dend2) + { + if (dend2 == end_match_2) break; + if (dend2 == regend[regno]) break; + + /* End of string1 => advance to string2. */ + d2 = string2; + dend2 = regend[regno]; + } + /* At end of register contents => success */ + if (d2 == dend2) break; + + /* If necessary, advance to next segment in data. */ + PREFETCH (); + + /* How many characters left in this segment to match. */ + dcnt = dend - d; + + /* Want how many consecutive characters we can match in + one shot, so, if necessary, adjust the count. */ + if (dcnt > dend2 - d2) + dcnt = dend2 - d2; + + /* Compare that many; failure if mismatch, else move + past them. */ + if (!NILP (translate) + ? bcmp_translate (d, d2, dcnt, translate, target_multibyte) + : memcmp (d, d2, dcnt)) + { + d = dfail; + goto fail; + } + d += dcnt, d2 += dcnt; + } + } + break; + + + /* begline matches the empty string at the beginning of the string, + and after newlines. */ + case begline: + DEBUG_PRINT ("EXECUTING begline.\n"); + + if (AT_STRINGS_BEG (d)) + break; + else + { + unsigned c; + GET_CHAR_BEFORE_2 (c, d, string1, end1, string2, end2); + if (c == '\n') + break; + } + goto fail; + + + /* endline is the dual of begline. */ + case endline: + DEBUG_PRINT ("EXECUTING endline.\n"); + + if (AT_STRINGS_END (d)) + break; + PREFETCH_NOLIMIT (); + if (*d == '\n') + break; + goto fail; + + + /* Match at the very beginning of the data. */ + case begbuf: + DEBUG_PRINT ("EXECUTING begbuf.\n"); + if (AT_STRINGS_BEG (d)) + break; + goto fail; + + + /* Match at the very end of the data. */ + case endbuf: + DEBUG_PRINT ("EXECUTING endbuf.\n"); + if (AT_STRINGS_END (d)) + break; + goto fail; + + + /* on_failure_keep_string_jump is used to optimize '.*\n'. It + pushes NULL as the value for the string on the stack. Then + 'POP_FAILURE_POINT' will keep the current value for the + string, instead of restoring it. To see why, consider + matching 'foo\nbar' against '.*\n'. The .* matches the foo; + then the . fails against the \n. But the next thing we want + to do is match the \n against the \n; if we restored the + string value, we would be back at the foo. + + Because this is used only in specific cases, we don't need to + check all the things that 'on_failure_jump' does, to make + sure the right things get saved on the stack. Hence we don't + share its code. The only reason to push anything on the + stack at all is that otherwise we would have to change + 'anychar's code to do something besides goto fail in this + case; that seems worse than this. */ + case on_failure_keep_string_jump: + EXTRACT_NUMBER_AND_INCR (mcnt, p); + DEBUG_PRINT ("EXECUTING on_failure_keep_string_jump %d (to %p):\n", + mcnt, p + mcnt); + + PUSH_FAILURE_POINT (p - 3, NULL); + break; + + /* A nasty loop is introduced by the non-greedy *? and +?. + With such loops, the stack only ever contains one failure point + at a time, so that a plain on_failure_jump_loop kind of + cycle detection cannot work. Worse yet, such a detection + can not only fail to detect a cycle, but it can also wrongly + detect a cycle (between different instantiations of the same + loop). + So the method used for those nasty loops is a little different: + We use a special cycle-detection-stack-frame which is pushed + when the on_failure_jump_nastyloop failure-point is *popped*. + This special frame thus marks the beginning of one iteration + through the loop and we can hence easily check right here + whether something matched between the beginning and the end of + the loop. */ + case on_failure_jump_nastyloop: + EXTRACT_NUMBER_AND_INCR (mcnt, p); + DEBUG_PRINT ("EXECUTING on_failure_jump_nastyloop %d (to %p):\n", + mcnt, p + mcnt); + + eassert ((re_opcode_t)p[-4] == no_op); + { + int cycle = 0; + CHECK_INFINITE_LOOP (p - 4, d); + if (!cycle) + /* If there's a cycle, just continue without pushing + this failure point. The failure point is the "try again" + option, which shouldn't be tried. + We want (x?)*?y\1z to match both xxyz and xxyxz. */ + PUSH_FAILURE_POINT (p - 3, d); + } + break; + + /* Simple loop detecting on_failure_jump: just check on the + failure stack if the same spot was already hit earlier. */ + case on_failure_jump_loop: + on_failure: + EXTRACT_NUMBER_AND_INCR (mcnt, p); + DEBUG_PRINT ("EXECUTING on_failure_jump_loop %d (to %p):\n", + mcnt, p + mcnt); + { + int cycle = 0; + CHECK_INFINITE_LOOP (p - 3, d); + if (cycle) + /* If there's a cycle, get out of the loop, as if the matching + had failed. We used to just 'goto fail' here, but that was + aborting the search a bit too early: we want to keep the + empty-loop-match and keep matching after the loop. + We want (x?)*y\1z to match both xxyz and xxyxz. */ + p += mcnt; + else + PUSH_FAILURE_POINT (p - 3, d); + } + break; + + + /* Uses of on_failure_jump: + + Each alternative starts with an on_failure_jump that points + to the beginning of the next alternative. Each alternative + except the last ends with a jump that in effect jumps past + the rest of the alternatives. (They really jump to the + ending jump of the following alternative, because tensioning + these jumps is a hassle.) + + Repeats start with an on_failure_jump that points past both + the repetition text and either the following jump or + pop_failure_jump back to this on_failure_jump. */ + case on_failure_jump: + EXTRACT_NUMBER_AND_INCR (mcnt, p); + DEBUG_PRINT ("EXECUTING on_failure_jump %d (to %p):\n", + mcnt, p + mcnt); + + PUSH_FAILURE_POINT (p -3, d); + break; + + /* This operation is used for greedy *. + Compare the beginning of the repeat with what in the + pattern follows its end. If we can establish that there + is nothing that they would both match, i.e., that we + would have to backtrack because of (as in, e.g., 'a*a') + then we can use a non-backtracking loop based on + on_failure_keep_string_jump instead of on_failure_jump. */ + case on_failure_jump_smart: + EXTRACT_NUMBER_AND_INCR (mcnt, p); + DEBUG_PRINT ("EXECUTING on_failure_jump_smart %d (to %p).\n", + mcnt, p + mcnt); + { + re_char *p1 = p; /* Next operation. */ + /* Discard 'const', making re_search non-reentrant. */ + unsigned char *p2 = (unsigned char *) p + mcnt; /* Jump dest. */ + unsigned char *p3 = (unsigned char *) p - 3; /* opcode location. */ + + p -= 3; /* Reset so that we will re-execute the + instruction once it's been changed. */ + + EXTRACT_NUMBER (mcnt, p2 - 2); + + /* Ensure this is indeed the trivial kind of loop + we are expecting. */ + eassert (skip_one_char (p1) == p2 - 3); + eassert ((re_opcode_t) p2[-3] == jump && p2 + mcnt == p); + DEBUG_STATEMENT (regex_emacs_debug += 2); + if (mutually_exclusive_p (bufp, p1, p2)) + { + /* Use a fast 'on_failure_keep_string_jump' loop. */ + DEBUG_PRINT (" smart exclusive => fast loop.\n"); + *p3 = (unsigned char) on_failure_keep_string_jump; + STORE_NUMBER (p2 - 2, mcnt + 3); + } + else + { + /* Default to a safe 'on_failure_jump' loop. */ + DEBUG_PRINT (" smart default => slow loop.\n"); + *p3 = (unsigned char) on_failure_jump; + } + DEBUG_STATEMENT (regex_emacs_debug -= 2); + } + break; + + /* Unconditionally jump (without popping any failure points). */ + case jump: + unconditional_jump: + maybe_quit (); + EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */ + DEBUG_PRINT ("EXECUTING jump %d ", mcnt); + p += mcnt; /* Do the jump. */ + DEBUG_PRINT ("(to %p).\n", p); + break; + + + /* Have to succeed matching what follows at least n times. + After that, handle like 'on_failure_jump'. */ + case succeed_n: + /* Signedness doesn't matter since we only compare MCNT to 0. */ + EXTRACT_NUMBER (mcnt, p + 2); + DEBUG_PRINT ("EXECUTING succeed_n %d.\n", mcnt); + + /* Originally, mcnt is how many times we HAVE to succeed. */ + if (mcnt != 0) + { + /* Discard 'const', making re_search non-reentrant. */ + unsigned char *p2 = (unsigned char *) p + 2; /* counter loc. */ + mcnt--; + p += 4; + PUSH_NUMBER (p2, mcnt); + } + else + /* The two bytes encoding mcnt == 0 are two no_op opcodes. */ + goto on_failure; + break; + + case jump_n: + /* Signedness doesn't matter since we only compare MCNT to 0. */ + EXTRACT_NUMBER (mcnt, p + 2); + DEBUG_PRINT ("EXECUTING jump_n %d.\n", mcnt); + + /* Originally, this is how many times we CAN jump. */ + if (mcnt != 0) + { + /* Discard 'const', making re_search non-reentrant. */ + unsigned char *p2 = (unsigned char *) p + 2; /* counter loc. */ + mcnt--; + PUSH_NUMBER (p2, mcnt); + goto unconditional_jump; + } + /* If don't have to jump any more, skip over the rest of command. */ + else + p += 4; + break; + + case set_number_at: + { + unsigned char *p2; /* Location of the counter. */ + DEBUG_PRINT ("EXECUTING set_number_at.\n"); + + EXTRACT_NUMBER_AND_INCR (mcnt, p); + /* Discard 'const', making re_search non-reentrant. */ + p2 = (unsigned char *) p + mcnt; + /* Signedness doesn't matter since we only copy MCNT's bits. */ + EXTRACT_NUMBER_AND_INCR (mcnt, p); + DEBUG_PRINT (" Setting %p to %d.\n", p2, mcnt); + PUSH_NUMBER (p2, mcnt); + break; + } + + case wordbound: + case notwordbound: + { + bool not = (re_opcode_t) *(p - 1) == notwordbound; + DEBUG_PRINT ("EXECUTING %swordbound.\n", not ? "not" : ""); + + /* We SUCCEED (or FAIL) in one of the following cases: */ + + /* Case 1: D is at the beginning or the end of string. */ + if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d)) + not = !not; + else + { + /* C1 is the character before D, S1 is the syntax of C1, C2 + is the character at D, and S2 is the syntax of C2. */ + int c1, c2; + int s1, s2; + int dummy; + ptrdiff_t offset = PTR_TO_OFFSET (d - 1); + ptrdiff_t charpos = SYNTAX_TABLE_BYTE_TO_CHAR (offset); + UPDATE_SYNTAX_TABLE (charpos); + GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2); + s1 = SYNTAX (c1); + UPDATE_SYNTAX_TABLE_FORWARD (charpos + 1); + PREFETCH_NOLIMIT (); + GET_CHAR_AFTER (c2, d, dummy); + s2 = SYNTAX (c2); + + if (/* Case 2: Only one of S1 and S2 is Sword. */ + ((s1 == Sword) != (s2 == Sword)) + /* Case 3: Both of S1 and S2 are Sword, and macro + WORD_BOUNDARY_P (C1, C2) returns nonzero. */ + || ((s1 == Sword) && WORD_BOUNDARY_P (c1, c2))) + not = !not; + } + if (not) + break; + else + goto fail; + } + + case wordbeg: + DEBUG_PRINT ("EXECUTING wordbeg.\n"); + + /* We FAIL in one of the following cases: */ + + /* Case 1: D is at the end of string. */ + if (AT_STRINGS_END (d)) + goto fail; + else + { + /* C1 is the character before D, S1 is the syntax of C1, C2 + is the character at D, and S2 is the syntax of C2. */ + int c1, c2; + int s1, s2; + int dummy; + ptrdiff_t offset = PTR_TO_OFFSET (d); + ptrdiff_t charpos = SYNTAX_TABLE_BYTE_TO_CHAR (offset); + UPDATE_SYNTAX_TABLE (charpos); + PREFETCH (); + GET_CHAR_AFTER (c2, d, dummy); + s2 = SYNTAX (c2); + + /* Case 2: S2 is not Sword. */ + if (s2 != Sword) + goto fail; + + /* Case 3: D is not at the beginning of string ... */ + if (!AT_STRINGS_BEG (d)) + { + GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2); + UPDATE_SYNTAX_TABLE_BACKWARD (charpos - 1); + s1 = SYNTAX (c1); + + /* ... and S1 is Sword, and WORD_BOUNDARY_P (C1, C2) + returns 0. */ + if ((s1 == Sword) && !WORD_BOUNDARY_P (c1, c2)) + goto fail; + } + } + break; + + case wordend: + DEBUG_PRINT ("EXECUTING wordend.\n"); + + /* We FAIL in one of the following cases: */ + + /* Case 1: D is at the beginning of string. */ + if (AT_STRINGS_BEG (d)) + goto fail; + else + { + /* C1 is the character before D, S1 is the syntax of C1, C2 + is the character at D, and S2 is the syntax of C2. */ + int c1, c2; + int s1, s2; + int dummy; + ptrdiff_t offset = PTR_TO_OFFSET (d) - 1; + ptrdiff_t charpos = SYNTAX_TABLE_BYTE_TO_CHAR (offset); + UPDATE_SYNTAX_TABLE (charpos); + GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2); + s1 = SYNTAX (c1); + + /* Case 2: S1 is not Sword. */ + if (s1 != Sword) + goto fail; + + /* Case 3: D is not at the end of string ... */ + if (!AT_STRINGS_END (d)) + { + PREFETCH_NOLIMIT (); + GET_CHAR_AFTER (c2, d, dummy); + UPDATE_SYNTAX_TABLE_FORWARD (charpos); + s2 = SYNTAX (c2); + + /* ... and S2 is Sword, and WORD_BOUNDARY_P (C1, C2) + returns 0. */ + if ((s2 == Sword) && !WORD_BOUNDARY_P (c1, c2)) + goto fail; + } + } + break; + + case symbeg: + DEBUG_PRINT ("EXECUTING symbeg.\n"); + + /* We FAIL in one of the following cases: */ + + /* Case 1: D is at the end of string. */ + if (AT_STRINGS_END (d)) + goto fail; + else + { + /* C1 is the character before D, S1 is the syntax of C1, C2 + is the character at D, and S2 is the syntax of C2. */ + int c1, c2; + int s1, s2; + ptrdiff_t offset = PTR_TO_OFFSET (d); + ptrdiff_t charpos = SYNTAX_TABLE_BYTE_TO_CHAR (offset); + UPDATE_SYNTAX_TABLE (charpos); + PREFETCH (); + c2 = RE_STRING_CHAR (d, target_multibyte); + s2 = SYNTAX (c2); + + /* Case 2: S2 is neither Sword nor Ssymbol. */ + if (s2 != Sword && s2 != Ssymbol) + goto fail; + + /* Case 3: D is not at the beginning of string ... */ + if (!AT_STRINGS_BEG (d)) + { + GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2); + UPDATE_SYNTAX_TABLE_BACKWARD (charpos - 1); + s1 = SYNTAX (c1); + + /* ... and S1 is Sword or Ssymbol. */ + if (s1 == Sword || s1 == Ssymbol) + goto fail; + } + } + break; + + case symend: + DEBUG_PRINT ("EXECUTING symend.\n"); + + /* We FAIL in one of the following cases: */ + + /* Case 1: D is at the beginning of string. */ + if (AT_STRINGS_BEG (d)) + goto fail; + else + { + /* C1 is the character before D, S1 is the syntax of C1, C2 + is the character at D, and S2 is the syntax of C2. */ + int c1, c2; + int s1, s2; + ptrdiff_t offset = PTR_TO_OFFSET (d) - 1; + ptrdiff_t charpos = SYNTAX_TABLE_BYTE_TO_CHAR (offset); + UPDATE_SYNTAX_TABLE (charpos); + GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2); + s1 = SYNTAX (c1); + + /* Case 2: S1 is neither Ssymbol nor Sword. */ + if (s1 != Sword && s1 != Ssymbol) + goto fail; + + /* Case 3: D is not at the end of string ... */ + if (!AT_STRINGS_END (d)) + { + PREFETCH_NOLIMIT (); + c2 = RE_STRING_CHAR (d, target_multibyte); + UPDATE_SYNTAX_TABLE_FORWARD (charpos + 1); + s2 = SYNTAX (c2); + + /* ... and S2 is Sword or Ssymbol. */ + if (s2 == Sword || s2 == Ssymbol) + goto fail; + } + } + break; + + case syntaxspec: + case notsyntaxspec: + { + bool not = (re_opcode_t) *(p - 1) == notsyntaxspec; + mcnt = *p++; + DEBUG_PRINT ("EXECUTING %ssyntaxspec %d.\n", not ? "not" : "", + mcnt); + PREFETCH (); + { + ptrdiff_t offset = PTR_TO_OFFSET (d); + ptrdiff_t pos1 = SYNTAX_TABLE_BYTE_TO_CHAR (offset); + UPDATE_SYNTAX_TABLE (pos1); + } + { + int len; + int c; + + GET_CHAR_AFTER (c, d, len); + if ((SYNTAX (c) != (enum syntaxcode) mcnt) ^ not) + goto fail; + d += len; + } + } + break; + + case at_dot: + DEBUG_PRINT ("EXECUTING at_dot.\n"); + if (PTR_BYTE_POS (d) != PT_BYTE) + goto fail; + break; + + case categoryspec: + case notcategoryspec: + { + bool not = (re_opcode_t) *(p - 1) == notcategoryspec; + mcnt = *p++; + DEBUG_PRINT ("EXECUTING %scategoryspec %d.\n", + not ? "not" : "", mcnt); + PREFETCH (); + + { + int len; + int c; + GET_CHAR_AFTER (c, d, len); + if ((!CHAR_HAS_CATEGORY (c, mcnt)) ^ not) + goto fail; + d += len; + } + } + break; + + default: + abort (); + } + continue; /* Successfully executed one pattern command; keep going. */ + + + /* We goto here if a matching operation fails. */ + fail: + maybe_quit (); + if (!FAIL_STACK_EMPTY ()) + { + re_char *str, *pat; + /* A restart point is known. Restore to that state. */ + DEBUG_PRINT ("\nFAIL:\n"); + POP_FAILURE_POINT (str, pat); + switch (*pat++) + { + case on_failure_keep_string_jump: + eassert (str == NULL); + goto continue_failure_jump; + + case on_failure_jump_nastyloop: + eassert ((re_opcode_t)pat[-2] == no_op); + PUSH_FAILURE_POINT (pat - 2, str); + FALLTHROUGH; + case on_failure_jump_loop: + case on_failure_jump: + case succeed_n: + d = str; + continue_failure_jump: + EXTRACT_NUMBER_AND_INCR (mcnt, pat); + p = pat + mcnt; + break; + + case no_op: + /* A special frame used for nastyloops. */ + goto fail; + + default: + abort (); + } + + eassert (p >= bufp->buffer && p <= pend); + + if (d >= string1 && d <= end1) + dend = end_match_1; + } + else + break; /* Matching at this starting point really fails. */ + } /* for (;;) */ + + if (best_regs_set) + goto restore_best_regs; + + SAFE_FREE (); + + return -1; /* Failure to match. */ +} + +/* Subroutine definitions for re_match_2. */ + +/* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN + bytes; nonzero otherwise. */ + +static int +bcmp_translate (re_char *s1, re_char *s2, ptrdiff_t len, + Lisp_Object translate, int target_multibyte) +{ + re_char *p1 = s1, *p2 = s2; + re_char *p1_end = s1 + len; + re_char *p2_end = s2 + len; + + /* FIXME: Checking both p1 and p2 presumes that the two strings might have + different lengths, but relying on a single LEN would break this. -sm */ + while (p1 < p1_end && p2 < p2_end) + { + int p1_charlen, p2_charlen; + int p1_ch, p2_ch; + + GET_CHAR_AFTER (p1_ch, p1, p1_charlen); + GET_CHAR_AFTER (p2_ch, p2, p2_charlen); + + if (RE_TRANSLATE (translate, p1_ch) + != RE_TRANSLATE (translate, p2_ch)) + return 1; + + p1 += p1_charlen, p2 += p2_charlen; + } + + if (p1 != p1_end || p2 != p2_end) + return 1; + + return 0; +} + +/* Entry points for GNU code. */ + +/* re_compile_pattern is the GNU regular expression compiler: it + compiles PATTERN (of length SIZE) and puts the result in BUFP. + Returns 0 if the pattern was valid, otherwise an error string. + + Assumes the 'allocated' (and perhaps 'buffer') and 'translate' fields + are set in BUFP on entry. + + We call regex_compile to do the actual compilation. */ + +const char * +re_compile_pattern (const char *pattern, size_t length, + bool posix_backtracking, const char *whitespace_regexp, + struct re_pattern_buffer *bufp) +{ + reg_errcode_t ret; + + /* GNU code is written to assume at least RE_NREGS registers will be set + (and at least one extra will be -1). */ + bufp->regs_allocated = REGS_UNALLOCATED; + + ret = regex_compile ((re_char *) pattern, length, + posix_backtracking, + whitespace_regexp, + bufp); + + if (!ret) + return NULL; + return re_error_msgid[ret]; +} diff --cc src/sound.c index 6f15f5dab6d,c1f869045f5..2b8715010e7 --- a/src/sound.c +++ b/src/sound.c @@@ -1,9 -1,7 +1,9 @@@ /* sound.c -- sound support. - Copyright (C) 1998-1999, 2001-2018 Free Software Foundation, Inc. + Copyright (C) 1998-1999, 2001-2019 Free Software Foundation, Inc. +Author: Gerd Moellmann + This file is part of GNU Emacs. GNU Emacs is free software: you can redistribute it and/or modify diff --cc src/w16select.c index a5f07578671,fb8161b61fa..b935b9f4f54 --- a/src/w16select.c +++ b/src/w16select.c @@@ -1,9 -1,7 +1,9 @@@ /* 16-bit Windows Selection processing for emacs on MS-Windows - Copyright (C) 1996-1997, 2001-2018 Free Software Foundation, Inc. + Copyright (C) 1996-1997, 2001-2019 Free Software Foundation, Inc. +Author: Dale P. Smith + This file is part of GNU Emacs. GNU Emacs is free software: you can redistribute it and/or modify diff --cc src/w32notify.c index 67385b80a81,e03650f0fd3..53787fd45db --- a/src/w32notify.c +++ b/src/w32notify.c @@@ -1,8 -1,5 +1,8 @@@ /* Filesystem notifications support for GNU Emacs on the Microsoft Windows API. - Copyright (C) 2012-2019 Free Software Foundation, Inc. + - Copyright (C) 2012-2018 Free Software Foundation, Inc. ++Copyright (C) 2012-2019 Free Software Foundation, Inc. + +Author: Eli Zaretskii This file is part of GNU Emacs. diff --cc src/w32reg.c index 4ddbaa3f268,e2aebbb1b76..aff131dd37d --- a/src/w32reg.c +++ b/src/w32reg.c @@@ -1,8 -1,6 +1,8 @@@ /* Emulate the X Resource Manager through the registry. - Copyright (C) 1990, 1993-1994, 2001-2019 Free Software Foundation, - Inc. + - Copyright (C) 1990, 1993-1994, 2001-2018 Free Software Foundation, Inc. ++Copyright (C) 1990, 1993-1994, 2001-2019 Free Software Foundation, Inc. + +Author: Kevin Gallo This file is part of GNU Emacs. diff --cc src/w32select.c index dc568d47f27,6c7808d9813..1c84cb47eb1 --- a/src/w32select.c +++ b/src/w32select.c @@@ -1,10 -1,7 +1,10 @@@ /* Selection processing for Emacs on the Microsoft Windows API. - Copyright (C) 1993-1994, 2001-2018 Free Software Foundation, Inc. + Copyright (C) 1993-1994, 2001-2019 Free Software Foundation, Inc. +Author: Kevin Gallo + Benjamin Riefenstahl + This file is part of GNU Emacs. GNU Emacs is free software: you can redistribute it and/or modify diff --cc src/xmenu.c index 31034f71122,49cd5940eae..96c278d42d0 --- a/src/xmenu.c +++ b/src/xmenu.c @@@ -1,12 -1,8 +1,12 @@@ /* X Communication module for terminals which understand the X protocol. - Copyright (C) 1986, 1988, 1993-1994, 1996, 1999-2018 Free Software + Copyright (C) 1986, 1988, 1993-1994, 1996, 1999-2019 Free Software Foundation, Inc. +Author: Jon Arnold + Roman Budzianowski + Robert Krawitz + This file is part of GNU Emacs. GNU Emacs is free software: you can redistribute it and/or modify diff --cc test/src/regex-emacs-tests.el index 7a075908a6b,00000000000..e84af6b131b mode 100644,000000..100644 --- a/test/src/regex-emacs-tests.el +++ b/test/src/regex-emacs-tests.el @@@ -1,686 -1,0 +1,686 @@@ +;;; regex-emacs-tests.el --- tests for regex-emacs.c -*- lexical-binding: t -*- + - ;; Copyright (C) 2015-2018 Free Software Foundation, Inc. ++;; Copyright (C) 2015-2019 Free Software Foundation, Inc. + +;; This file is part of GNU Emacs. + +;; GNU Emacs is free software: you can redistribute it and/or modify +;; it under the terms of the GNU General Public License as published by +;; the Free Software Foundation, either version 3 of the License, or +;; (at your option) any later version. + +;; GNU Emacs is distributed in the hope that it will be useful, +;; but WITHOUT ANY WARRANTY; without even the implied warranty of +;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +;; GNU General Public License for more details. + +;; You should have received a copy of the GNU General Public License +;; along with GNU Emacs. If not, see . + +;;; Code: + +(require 'ert) + +(defvar regex-tests--resources-dir + (concat (concat (file-name-directory (or load-file-name buffer-file-name)) + "/regex-resources/")) + "Path to regex-resources directory next to the \"regex-emacs-tests.el\" file.") + +(ert-deftest regex-word-cc-fallback-test () + "Test that \"[[:cc:]]*x\" matches \"x\" (bug#24020). + +Test that a regex of the form \"[[:cc:]]*x\" where CC is +a character class which matches a multibyte character X, matches +string \"x\". + +For example, \"[[:word:]]*\u2620\" regex (note: \u2620 is a word +character) must match a string \"\u2420\"." + (dolist (class '("[[:word:]]" "\\sw")) + (dolist (repeat '("*" "+")) + (dolist (suffix '("" "b" "bar" "\u2620")) + (dolist (string '("" "foo")) + (when (not (and (string-equal repeat "+") + (string-equal string ""))) + (should (string-match (concat "^" class repeat suffix "$") + (concat string suffix))))))))) + +(defun regex--test-cc (name matching not-matching) + (let (case-fold-search) + (should (string-match-p (concat "^[[:" name ":]]*$") matching)) + (should (string-match-p (concat "^[[:" name ":]]*?\u2622$") + (concat matching "\u2622"))) + (should (string-match-p (concat "^[^[:" name ":]]*$") not-matching)) + (should (string-match-p (concat "^[^[:" name ":]]*\u2622$") + (concat not-matching "\u2622"))) + (with-temp-buffer + (insert matching) + (let ((p (point))) + (insert not-matching) + (goto-char (point-min)) + (skip-chars-forward (concat "[:" name ":]")) + (should (equal (point) p)) + (skip-chars-forward (concat "^[:" name ":]")) + (should (equal (point) (point-max))) + (goto-char (point-min)) + (skip-chars-forward (concat "[:" name ":]\u2622")) + (should (or (equal (point) p) (equal (point) (1+ p)))))))) + +(dolist (test '(("alnum" "abcABC012łąka" "-, \t\n") + ("alpha" "abcABCłąka" "-,012 \t\n") + ("digit" "012" "abcABCłąka-, \t\n") + ("xdigit" "0123aBc" "łąk-, \t\n") + ("upper" "ABCŁĄKA" "abc012-, \t\n") + ("lower" "abcłąka" "ABC012-, \t\n") + + ("word" "abcABC012\u2620" "-, \t\n") + + ("punct" ".,-" "abcABC012\u2620 \t\n") + ("cntrl" "\1\2\t\n" ".,-abcABC012\u2620 ") + ("graph" "abcłąka\u2620-," " \t\n\1") + ("print" "abcłąka\u2620-, " "\t\n\1") + + ("space" " \t\n\u2001" "abcABCł0123") + ("blank" " \t\u2001" "\n") + + ("ascii" "abcABC012 \t\n\1" "łą\u2620") + ("nonascii" "łą\u2622" "abcABC012 \t\n\1") + ("unibyte" "abcABC012 \t\n\1" "łą\u2622") + ("multibyte" "łą\u2622" "abcABC012 \t\n\1"))) + (let ((name (intern (concat "regex-tests-" (car test) "-character-class"))) + (doc (concat "Perform sanity test of regexes using " (car test) + " character class. + +Go over all the supported character classes and test whether the +classes and their inversions match what they are supposed to +match. The test is done using `string-match-p' as well as +`skip-chars-forward'."))) + (eval `(ert-deftest ,name () ,doc ,(cons 'regex--test-cc test)) t))) + + +(defmacro regex-tests-generic-line (comment-char test-file whitelist &rest body) + "Reads a line of the test file TEST-FILE, skipping +comments (defined by COMMENT-CHAR), and evaluates the tests in +this line as defined in the BODY. Line numbers in the WHITELIST +are known failures, and are skipped." + + `(with-temp-buffer + (modify-syntax-entry ?_ "w;; ") ; tests expect _ to be a word + (insert-file-contents (concat regex-tests--resources-dir ,test-file)) + (let ((case-fold-search nil) + (line-number 1) + (whitelist-idx 0)) + + (goto-char (point-min)) + + (while (not (eobp)) + (let ((start (point))) + (end-of-line) + (narrow-to-region start (point)) + + (goto-char (point-min)) + + (when + (and + ;; ignore comments + (save-excursion + (re-search-forward ,(concat "^[^" (string comment-char) "]") nil t)) + + ;; skip lines in the whitelist + (let ((whitelist-next + (condition-case nil + (aref ,whitelist whitelist-idx) (args-out-of-range nil)))) + (cond + ;; whitelist exhausted. do process this line + ((null whitelist-next) t) + + ;; we're not yet at the next whitelist element. do + ;; process this line + ((< line-number whitelist-next) t) + + ;; we're past the next whitelist element. This + ;; shouldn't happen + ((> line-number whitelist-next) + (error + (format + "We somehow skipped the next whitelist element: line %d" whitelist-next))) + + ;; we're at the next whitelist element. Skip this + ;; line, and advance the whitelist index + (t + (setq whitelist-idx (1+ whitelist-idx)) nil)))) + ,@body) + + (widen) + (forward-line) + (beginning-of-line) + (setq line-number (1+ line-number))))))) + +(defun regex-tests-compare (string what-failed bounds-ref &optional substring-ref) + "I just ran a search, looking at STRING. WHAT-FAILED describes +what failed, if anything; valid values are 'search-failed, +'compilation-failed and nil. I compare the beginning/end of each +group with their expected values. This is done with either +BOUNDS-REF or SUBSTRING-REF; one of those should be non-nil. +BOUNDS-REF is a sequence \[start-ref0 end-ref0 start-ref1 +end-ref1 ....] while SUBSTRING-REF is the expected substring +obtained by indexing the input string by start/end-ref. + +If the search was supposed to fail then start-ref0/substring-ref0 +is 'search-failed. If the search wasn't even supposed to compile +successfully, then start-ref0/substring-ref0 is +'compilation-failed. If I only care about a match succeeding, +this can be set to t. + +This function returns a string that describes the failure, or nil +on success" + + (when (or + (and bounds-ref substring-ref) + (not (or bounds-ref substring-ref))) + (error "Exactly one of bounds-ref and bounds-ref should be non-nil")) + + (let ((what-failed-ref (car (or bounds-ref substring-ref)))) + + (cond + ((eq what-failed 'search-failed) + (cond + ((eq what-failed-ref 'search-failed) + nil) + ((eq what-failed-ref 'compilation-failed) + "Expected pattern failure; but no match") + (t + "Expected match; but no match"))) + + ((eq what-failed 'compilation-failed) + (cond + ((eq what-failed-ref 'search-failed) + "Expected no match; but pattern failure") + ((eq what-failed-ref 'compilation-failed) + nil) + (t + "Expected match; but pattern failure"))) + + ;; The regex match succeeded + ((eq what-failed-ref 'search-failed) + "Expected no match; but match") + ((eq what-failed-ref 'compilation-failed) + "Expected pattern failure; but match") + + ;; The regex match succeeded, as expected. I now check all the + ;; bounds + (t + (let ((idx 0) + msg + ref next-ref-function compare-ref-function mismatched-ref-function) + + (if bounds-ref + (setq ref bounds-ref + next-ref-function (lambda (x) (cddr x)) + compare-ref-function (lambda (ref start-pos end-pos) + (or (eq (car ref) t) + (and (eq start-pos (car ref)) + (eq end-pos (cadr ref))))) + mismatched-ref-function (lambda (ref start-pos end-pos) + (format + "beginning/end positions: %d/%s and %d/%s" + start-pos (car ref) end-pos (cadr ref)))) + (setq ref substring-ref + next-ref-function (lambda (x) (cdr x)) + compare-ref-function (lambda (ref start-pos end-pos) + (or (eq (car ref) t) + (string= (substring string start-pos end-pos) (car ref)))) + mismatched-ref-function (lambda (ref start-pos end-pos) + (format + "beginning/end positions: %d/%s and %d/%s" + start-pos (car ref) end-pos (cadr ref))))) + + (while (not (or (null ref) msg)) + + (let ((start (match-beginning idx)) + (end (match-end idx))) + + (when (not (funcall compare-ref-function ref start end)) + (setq msg + (format + "Have expected match, but mismatch in group %d: %s" idx (funcall mismatched-ref-function ref start end)))) + + (setq ref (funcall next-ref-function ref) + idx (1+ idx)))) + + (or msg + nil)))))) + + + +(defun regex-tests-match (pattern string bounds-ref &optional substring-ref) + "I match the given STRING against PATTERN. I compare the +beginning/end of each group with their expected values. +BOUNDS-REF is a sequence [start-ref0 end-ref0 start-ref1 end-ref1 +....]. + +If the search was supposed to fail then start-ref0 is +'search-failed. If the search wasn't even supposed to compile +successfully, then start-ref0 is 'compilation-failed. + +This function returns a string that describes the failure, or nil +on success" + + (if (string-match "\\[\\([\\.=]\\)..?\\1\\]" pattern) + ;; Skipping test: [.x.] and [=x=] forms not supported by emacs + nil + + (regex-tests-compare + string + (condition-case nil + (if (string-match pattern string) nil 'search-failed) + ('invalid-regexp 'compilation-failed)) + bounds-ref substring-ref))) + + +(defconst regex-tests-re-even-escapes + "\\(?:^\\|[^\\\\]\\)\\(?:\\\\\\\\\\)*" + "Regex that matches an even number of \\ characters") + +(defconst regex-tests-re-odd-escapes + (concat regex-tests-re-even-escapes "\\\\") + "Regex that matches an odd number of \\ characters") + + +(defun regex-tests-unextend (pattern) + "Basic conversion from extended regexes to emacs ones. This is +mostly a hack that adds \\ to () and | and {}, and removes it if +it already exists. We also change \\S (and \\s) to \\S- (and +\\s-) because extended regexes see the former as whitespace, but +emacs requires an extra symbol character" + + (with-temp-buffer + (insert pattern) + (goto-char (point-min)) + + (while (re-search-forward "[()|{}]" nil t) + ;; point is past special character. If it is escaped, unescape + ;; it + + (if (save-excursion + (re-search-backward (concat regex-tests-re-odd-escapes ".\\=") nil t)) + + ;; This special character is preceded by an odd number of \, + ;; so I unescape it by removing the last one + (progn + (forward-char -2) + (delete-char 1) + (forward-char 1)) + + ;; This special character is preceded by an even (possibly 0) + ;; number of \. I add an escape + (forward-char -1) + (insert "\\") + (forward-char 1))) + + ;; convert \s to \s- + (goto-char (point-min)) + (while (re-search-forward (concat regex-tests-re-odd-escapes "[Ss]") nil t) + (insert "-")) + + (buffer-string))) + +(defun regex-tests-BOOST-frob-escapes (s ispattern) + "Mangle \\ the way it is done in frob_escapes() in +regex-tests-BOOST.c in glibc: \\t, \\n, \\r are interpreted; +\\\\, \\^, \{, \\|, \} are unescaped for the string (not +pattern)" + + ;; this is all similar to (regex-tests-unextend) + (with-temp-buffer + (insert s) + + (let ((interpret-list (list "t" "n" "r"))) + (while interpret-list + (goto-char (point-min)) + (while (re-search-forward + (concat "\\(" regex-tests-re-even-escapes "\\)" + "\\\\" (car interpret-list)) + nil t) + (replace-match (concat "\\1" (car (read-from-string + (concat "\"\\" (car interpret-list) "\"")))))) + + (setq interpret-list (cdr interpret-list)))) + + (when (not ispattern) + ;; unescape \\, \^, \{, \|, \} + (let ((unescape-list (list "\\\\" "^" "{" "|" "}"))) + (while unescape-list + (goto-char (point-min)) + (while (re-search-forward + (concat "\\(" regex-tests-re-even-escapes "\\)" + "\\\\" (car unescape-list)) + nil t) + (replace-match (concat "\\1" (car unescape-list)))) + + (setq unescape-list (cdr unescape-list)))) + ) + (buffer-string))) + + + + +(defconst regex-tests-BOOST-whitelist + [ + ;; emacs is more stringent with regexes involving unbalanced ) + 63 65 69 + + ;; in emacs, regex . doesn't match \n + 91 + + ;; emacs is more forgiving with * and ? that don't apply to + ;; characters + 107 108 109 122 123 124 140 141 142 + + ;; emacs accepts regexes with {} + 161 + + ;; emacs doesn't fail on bogus ranges such as [3-1] or [1-3-5] + 222 223 + + ;; emacs doesn't match (ab*)[ab]*\1 greedily: only 4 chars of + ;; ababaaa match + 284 294 + + ;; ambiguous groupings are ambiguous + 443 444 445 446 448 449 450 + + ;; emacs doesn't know how to handle weird ranges such as [a-Z] and + ;; [[:alpha:]-a] + 539 580 581 + + ;; emacs matches non-greedy regex ab.*? non-greedily + 639 677 712 + ] + "Line numbers in the boost test that should be skipped. These +are false-positive test failures that represent known/benign +differences in behavior.") + +;; - Format +;; - Comments are lines starting with ; +;; - Lines starting with - set options passed to regcomp() and regexec(): +;; - if no "REG_BASIC" is found, with have an extended regex +;; - These set a flag: +;; - REG_ICASE +;; - REG_NEWLINE (ignored by this function) +;; - REG_NOTBOL +;; - REG_NOTEOL +;; +;; - Test lines are +;; pattern string start0 end0 start1 end1 ... +;; +;; - pattern, string can have escapes +;; - string can have whitespace if enclosed in "" +;; - if string is "!", then the pattern is supposed to fail compilation +;; - start/end are of group0, group1, etc. group 0 is the full match +;; - start<0 indicates "no match" +;; - start is the 0-based index of the first character +;; - end is the 0-based index of the first character past the group +(defun regex-tests-BOOST () + (let (failures + basic icase notbol noteol) + (regex-tests-generic-line + ?\; "BOOST.tests" regex-tests-BOOST-whitelist + (if (save-excursion (re-search-forward "^-" nil t)) + (setq basic (save-excursion (re-search-forward "REG_BASIC" nil t)) + icase (save-excursion (re-search-forward "REG_ICASE" nil t)) + notbol (save-excursion (re-search-forward "REG_NOTBOL" nil t)) + noteol (save-excursion (re-search-forward "REG_NOTEOL" nil t))) + + (save-excursion + (or (re-search-forward "\\(\\S-+\\)\\s-+\"\\(.*\\)\"\\s-+?\\(.+\\)" nil t) + (re-search-forward "\\(\\S-+\\)\\s-+\\(\\S-+\\)\\s-+?\\(.+\\)" nil t) + (re-search-forward "\\(\\S-+\\)\\s-+\\(!\\)" nil t))) + + (let* ((pattern-raw (match-string 1)) + (string-raw (match-string 2)) + (positions-raw (match-string 3)) + (pattern (regex-tests-BOOST-frob-escapes pattern-raw t)) + (string (regex-tests-BOOST-frob-escapes string-raw nil)) + (positions + (if (string= string "!") + (list 'compilation-failed 0) + (mapcar + (lambda (x) + (let ((x (string-to-number x))) + (if (< x 0) nil x))) + (split-string positions-raw))))) + + (when (null (car positions)) + (setcar positions 'search-failed)) + + (when (not basic) + (setq pattern (regex-tests-unextend pattern))) + + ;; great. I now have all the data parsed. Let's use it to do + ;; stuff + (let* ((case-fold-search icase) + (msg (regex-tests-match pattern string positions))) + + (if (and + ;; Skipping test: notbol/noteol not supported + (not notbol) (not noteol) + + msg) + + ;; store failure + (setq failures + (cons (format "line number %d: Regex '%s': %s" + line-number pattern msg) + failures))))))) + + failures)) + +(defconst regex-tests-PCRE-whitelist + [ + ;; ambiguous groupings are ambiguous + 610 611 1154 1157 1160 1168 1171 1176 1179 1182 1185 1188 1193 1196 1203 + ] + "Line numbers in the PCRE test that should be skipped. These +are false-positive test failures that represent known/benign +differences in behavior.") + +;; - Format +;; +;; regex +;; input_string +;; group_num: group_match | "No match" +;; input_string +;; group_num: group_match | "No match" +;; input_string +;; group_num: group_match | "No match" +;; input_string +;; group_num: group_match | "No match" +;; ... +(defun regex-tests-PCRE () + (let (failures + pattern icase string what-failed matches-observed) + (regex-tests-generic-line + ?# "PCRE.tests" regex-tests-PCRE-whitelist + + (cond + + ;; pattern + ((save-excursion (re-search-forward "^/\\(.*\\)/\\(.*i?\\)$" nil t)) + (setq icase (string= "i" (match-string 2)) + pattern (regex-tests-unextend (match-string 1)))) + + ;; string. read it in, match against pattern, and save all the results + ((save-excursion (re-search-forward "^ \\(.*\\)" nil t)) + (let ((case-fold-search icase)) + (setq string (match-string 1) + + ;; the regex match under test + what-failed + (condition-case nil + (if (string-match pattern string) nil 'search-failed) + ('invalid-regexp 'compilation-failed)) + + matches-observed + (cl-loop for x from 0 to 20 + collect (and (not what-failed) + (or (match-string x string) ""))))) + nil) + + ;; verification line: failed match + ((save-excursion (re-search-forward "^No match" nil t)) + (unless what-failed + (setq failures + (cons (format "line number %d: Regex '%s': Expected no match; but match" + line-number pattern) + failures)))) + + ;; verification line: succeeded match + ((save-excursion (re-search-forward "^ *\\([0-9]+\\): \\(.*\\)" nil t)) + (let* ((match-ref (match-string 2)) + (idx (string-to-number (match-string 1)))) + + (if what-failed + "Expected match; but no match" + (unless (string= match-ref (elt matches-observed idx)) + (setq failures + (cons (format "line number %d: Regex '%s': Have expected match, but group %d is wrong: '%s'/'%s'" + line-number pattern + idx match-ref (elt matches-observed idx)) + failures)))))) + + ;; reset + (t (setq pattern nil) nil))) + + failures)) + +(defconst regex-tests-PTESTS-whitelist + [ + ;; emacs doesn't barf on weird ranges such as [b-a], but simply + ;; fails to match + 138 + + ;; emacs doesn't see DEL (0x78) as a [:cntrl:] character + 168 + ] + "Line numbers in the PTESTS test that should be skipped. These +are false-positive test failures that represent known/benign +differences in behavior.") + +;; - Format +;; - fields separated by ¦ (note: this is not a |) +;; - start¦end¦pattern¦string +;; - start is the 1-based index of the first character +;; - end is the 1-based index of the last character +(defun regex-tests-PTESTS () + (let (failures) + (regex-tests-generic-line + ?# "PTESTS" regex-tests-PTESTS-whitelist + (let* ((fields (split-string (buffer-string) "¦")) + + ;; string has 1-based index of first char in the + ;; match. -1 means "no match". -2 means "invalid + ;; regex". + ;; + ;; start-ref is 0-based index of first char in the + ;; match + ;; + ;; string==0 is a special case, and I have to treat + ;; it as start-ref = 0 + (start-ref (let ((raw (string-to-number (elt fields 0)))) + (cond + ((= raw -2) 'compilation-failed) + ((= raw -1) 'search-failed) + ((= raw 0) 0) + (t (1- raw))))) + + ;; string has 1-based index of last char in the + ;; match. end-ref is 0-based index of first char past + ;; the match + (end-ref (string-to-number (elt fields 1))) + (pattern (elt fields 2)) + (string (elt fields 3))) + + (let ((msg (regex-tests-match pattern string (list start-ref end-ref)))) + (when msg + (setq failures + (cons (format "line number %d: Regex '%s': %s" + line-number pattern msg) + failures)))))) + failures)) + +(defconst regex-tests-TESTS-whitelist + [ + ;; emacs doesn't barf on weird ranges such as [b-a], but simply + ;; fails to match + 42 + + ;; emacs is more forgiving with * and ? that don't apply to + ;; characters + 57 58 59 60 + + ;; emacs is more stringent with regexes involving unbalanced ) + 67 + ] + "Line numbers in the TESTS test that should be skipped. These +are false-positive test failures that represent known/benign +differences in behavior.") + +;; - Format +;; - fields separated by :. Watch for [\[:xxx:]] +;; - expected:pattern:string +;; +;; expected: +;; | 0 | successful match | +;; | 1 | failed match | +;; | 2 | regcomp() should fail | +(defun regex-tests-TESTS () + (let (failures) + (regex-tests-generic-line + ?# "TESTS" regex-tests-TESTS-whitelist + (if (save-excursion (re-search-forward "^\\([^:]+\\):\\(.*\\):\\([^:]*\\)$" nil t)) + (let* ((what-failed + (let ((raw (string-to-number (match-string 1)))) + (cond + ((= raw 2) 'compilation-failed) + ((= raw 1) 'search-failed) + (t t)))) + (string (match-string 3)) + (pattern (regex-tests-unextend (match-string 2)))) + + (let ((msg (regex-tests-match pattern string nil (list what-failed)))) + (when msg + (setq failures + (cons (format "line number %d: Regex '%s': %s" + line-number pattern msg) + failures))))) + + (error "Error parsing TESTS file line: '%s'" (buffer-string)))) + failures)) + +(ert-deftest regex-tests-BOOST () + "Tests of the regular expression engine. +This evaluates the BOOST test cases from glibc." + (should-not (regex-tests-BOOST))) + +(ert-deftest regex-tests-PCRE () + "Tests of the regular expression engine. +This evaluates the PCRE test cases from glibc." + (should-not (regex-tests-PCRE))) + +(ert-deftest regex-tests-PTESTS () + "Tests of the regular expression engine. +This evaluates the PTESTS test cases from glibc." + (should-not (regex-tests-PTESTS))) + +(ert-deftest regex-tests-TESTS () + "Tests of the regular expression engine. +This evaluates the TESTS test cases from glibc." + (should-not (regex-tests-TESTS))) + +(ert-deftest regex-repeat-limit () + "Test the #xFFFF repeat limit." + (should (string-match "\\`x\\{65535\\}" (make-string 65535 ?x))) + (should-not (string-match "\\`x\\{65535\\}" (make-string 65534 ?x))) + (should-error (string-match "\\`x\\{65536\\}" "X") :type 'invalid-regexp)) + +;;; regex-emacs-tests.el ends here