g_set_error (error, G_IO_ERROR, G_IO_ERROR_FAILED, "%s", archive_error_string (a));
}
+/*
+ * Get pathname from archive entry as UTF-8.
+ *
+ * libarchive attempts to convert filenames to the current locale's charset,
+ * which fails in POSIX/C locale for non-ASCII UTF-8 characters. This function
+ * uses archive_entry_pathname_utf8() to bypass locale conversion, falling back
+ * to archive_entry_pathname() with explicit UTF-8 validation.
+ *
+ * Returns NULL and sets error if the pathname is missing or not valid UTF-8.
+ */
+static const char *
+archive_entry_require_pathname_utf8 (struct archive_entry *entry, GError **error)
+{
+ /* Try the UTF-8 accessor first - this returns the UTF-8 form directly
+ * without locale conversion. */
+ const char *pathname = archive_entry_pathname_utf8 (entry);
+ if (pathname != NULL)
+ return pathname;
+
+ /* Fall back to regular accessor. When libarchive's charset conversion
+ * fails (e.g., in POSIX locale), it falls back to copying raw bytes,
+ * which for OCI/Docker tarballs should be valid UTF-8. */
+ pathname = archive_entry_pathname (entry);
+ if (pathname == NULL)
+ {
+ g_set_error (error, G_IO_ERROR, G_IO_ERROR_FAILED, "Archive entry has no pathname");
+ return NULL;
+ }
+
+ if (!g_utf8_validate (pathname, -1, NULL))
+ {
+ g_set_error (error, G_IO_ERROR, G_IO_ERROR_INVALID_DATA,
+ "Archive entry pathname is not valid UTF-8");
+ return NULL;
+ }
+
+ return pathname;
+}
+
+/*
+ * Get symlink target from archive entry, validating it is UTF-8.
+ * Returns NULL (without error) if entry is not a symlink.
+ * Returns NULL with error set if symlink target is not valid UTF-8.
+ */
+static const char *
+archive_entry_require_symlink_utf8 (struct archive_entry *entry, GError **error)
+{
+ /* Try the UTF-8 accessor first - this returns the UTF-8 form directly
+ * without locale conversion. */
+ const char *target = archive_entry_symlink_utf8 (entry);
+ if (target != NULL)
+ return target;
+
+ /* Fall back to regular accessor with explicit UTF-8 validation */
+ target = archive_entry_symlink (entry);
+ if (target == NULL)
+ return NULL;
+
+ if (!g_utf8_validate (target, -1, NULL))
+ {
+ g_set_error (error, G_IO_ERROR, G_IO_ERROR_INVALID_DATA,
+ "Archive entry symlink target is not valid UTF-8");
+ return NULL;
+ }
+
+ return target;
+}
+
static const char *
path_relative (const char *src, GError **error)
{
stbuf->st_mode |= S_IFREG;
}
-/* Create a GFileInfo from archive_entry_stat() */
+/* Create a GFileInfo from archive_entry_stat().
+ *
+ * For symlinks, validates that the target is valid UTF-8.
+ * Returns NULL with error set on failure.
+ */
static GFileInfo *
-file_info_from_archive_entry (struct archive_entry *entry)
+file_info_from_archive_entry (struct archive_entry *entry, GError **error)
{
struct stat stbuf;
read_archive_entry_stat (entry, &stbuf);
- g_autoptr (GFileInfo) info = _ostree_stbuf_to_gfileinfo (&stbuf);
+ /* For symlinks, validate and get the UTF-8 target */
+ const char *symlink_target = NULL;
if (S_ISLNK (stbuf.st_mode))
{
- const char *target = archive_entry_symlink (entry);
- if (target != NULL)
- g_file_info_set_attribute_byte_string (info, "standard::symlink-target", target);
+ symlink_target = archive_entry_require_symlink_utf8 (entry, error);
+ /* A symlink without a target is an error */
+ if (symlink_target == NULL)
+ {
+ if (error != NULL && *error == NULL)
+ g_set_error (error, G_IO_ERROR, G_IO_ERROR_FAILED, "Symlink entry has no target");
+ return NULL;
+ }
}
+ g_autoptr (GFileInfo) info = _ostree_stbuf_to_gfileinfo (&stbuf);
+ if (symlink_target != NULL)
+ g_file_info_set_attribute_byte_string (info, "standard::symlink-target", symlink_target);
+
return g_steal_pointer (&info);
}
OstreeMutableTree *root;
struct archive *archive;
struct archive_entry *entry;
+ GFileInfo *file_info; /* Cached file info for current entry, set by aic_import_entry */
GHashTable *deferred_hardlinks;
OstreeRepoCommitModifier *modifier;
} OstreeRepoArchiveImportContext;
static inline char *
aic_get_final_entry_pathname (OstreeRepoArchiveImportContext *ctx, GError **error)
{
- const char *pathname = archive_entry_pathname (ctx->entry);
+ const char *pathname = archive_entry_require_pathname_utf8 (ctx->entry, error);
+ if (pathname == NULL)
+ return NULL;
+
g_autofree char *final = aic_get_final_path (ctx, pathname, error);
if (final == NULL)
return NULL;
aic_apply_modifier_filter (OstreeRepoArchiveImportContext *ctx, const char *relpath,
GFileInfo **out_file_info)
{
- g_autoptr (GFileInfo) file_info = NULL;
g_autofree char *abspath = NULL;
const char *cb_path = NULL;
if (ctx->opts->callback_with_entry_pathname)
- cb_path = archive_entry_pathname (ctx->entry);
+ cb_path = archive_entry_pathname_utf8 (ctx->entry);
else
{
/* the user expects an abspath (where the dir to commit represents /) */
cb_path = abspath;
}
- file_info = file_info_from_archive_entry (ctx->entry);
+ /* Use the pre-validated file_info from ctx, computed by aic_import_entry
+ * which has proper error propagation for UTF-8 validation failures. */
+ g_assert (ctx->file_info != NULL);
- return _ostree_repo_commit_modifier_apply (ctx->repo, ctx->modifier, cb_path, file_info,
+ return _ostree_repo_commit_modifier_apply (ctx->repo, ctx->modifier, cb_path, ctx->file_info,
out_file_info);
}
}
if (ctx->opts->callback_with_entry_pathname)
- cb_path = archive_entry_pathname (ctx->entry);
+ cb_path = archive_entry_pathname_utf8 (ctx->entry);
if (ctx->modifier && ctx->modifier->xattr_callback)
{
if (path == NULL)
return FALSE;
+ /* Compute file info early while we have error propagation. This validates
+ * symlink targets are UTF-8 and caches the result in ctx->file_info for
+ * use by aic_apply_modifier_filter which cannot propagate errors.
+ * The struct owns the reference until we clear it. */
+ g_assert (ctx->file_info == NULL);
+ ctx->file_info = file_info_from_archive_entry (ctx->entry, error);
+ if (ctx->file_info == NULL)
+ return FALSE;
+
g_autoptr (GFileInfo) fi = NULL;
- if (aic_apply_modifier_filter (ctx, path, &fi) == OSTREE_REPO_COMMIT_FILTER_SKIP)
+ OstreeRepoCommitFilterResult filter_result = aic_apply_modifier_filter (ctx, path, &fi);
+
+ g_clear_object (&ctx->file_info);
+
+ if (filter_result == OSTREE_REPO_COMMIT_FILTER_SKIP)
return TRUE;
g_autoptr (OstreeMutableTree) parent = NULL;
int r = archive_read_next_header (a, &aictx.entry);
if (r == ARCHIVE_EOF)
break;
- if (r != ARCHIVE_OK)
+ /* Accept ARCHIVE_WARN: libarchive returns this for "partial success"
+ * conditions like charset conversion failures (e.g., UTF-8 to ASCII
+ * in POSIX locale). The entry is still fully populated; we validate
+ * filenames as UTF-8 ourselves. Fail only on ARCHIVE_FATAL/FAILED. */
+ if (r != ARCHIVE_OK && r != ARCHIVE_WARN)
{
propagate_libarchive_error (error, a);
goto out;
}
+ if (r == ARCHIVE_WARN)
+ g_debug ("libarchive warning: %s", archive_error_string (a));
if (g_cancellable_set_error_if_cancelled (cancellable, error))
goto out;
skip_without_ostree_feature libarchive
-echo "1..18"
+echo "1..19"
setup_test_repository "bare"
assert_file_has_content sizes.txt 'Unpacked size (needed/total): 0[ ย ]bytes/921[ ย ]bytes'
assert_file_has_content sizes.txt 'Number of objects (needed/total): 0/14'
echo "ok tar sizes metadata"
+
+# Test UTF-8 filenames work in POSIX/C locale (where libarchive's charset
+# conversion fails). This reproduces the issue from
+# https://github.com/ostreedev/ostree/issues/3431 where Python 3.14's
+# venv creates a symlink named "๐thon" (U+1D70B, Mathematical Italic Small Pi).
+cd ${test_tmpdir}
+rm -rf utf8-test
+mkdir utf8-test
+cd utf8-test
+mkdir -p usr/bin
+echo "#!/bin/sh" > usr/bin/python3
+chmod +x usr/bin/python3
+# Create symlink with non-ASCII UTF-8 name (๐ = 4-byte UTF-8: F0 9D 9C 8B)
+# and symlink target with non-ASCII UTF-8
+ln -s python3 'usr/bin/๐thon'
+ln -s '๐thon' 'usr/bin/๐-link'
+tar -c -f ../utf8.tar .
+cd ..
+
+# Import with POSIX locale - this previously failed with:
+# "Pathname can't be converted from UTF-8 to current locale"
+LC_ALL=C $OSTREE commit -s "from tar with utf8" -b test-tar-utf8 \
+ --tar-autocreate-parents \
+ --tree=tar=utf8.tar
+# Verify the files exist with correct names
+$OSTREE ls test-tar-utf8 /usr/bin/๐thon >/dev/null
+$OSTREE ls test-tar-utf8 /usr/bin/๐-link >/dev/null
+# Verify symlink targets are correct
+rm -rf utf8-checkout
+$OSTREE checkout test-tar-utf8 utf8-checkout
+test "$(readlink utf8-checkout/usr/bin/๐thon)" = "python3"
+test "$(readlink utf8-checkout/usr/bin/๐-link)" = "๐thon"
+echo "ok tar commit with utf8 filenames in POSIX locale"
projects / ostree.git / commitdiff