--- /dev/null
+#!/usr/bin/python3
+"""A very simple test for the command-line zchunk tools."""
+
+import argparse
+import dataclasses
+import os
+import pathlib
+import re
+import subprocess
+import sys
+import tempfile
+
+from typing import Callable, Dict, List, Tuple
+
+
+MAGIC = bytes([0, ord("Z"), ord("C"), ord("K"), ord("1")])
+
+RE_DATA_SIZE = re.compile(
+ r""" ^
+ Data \s+ size \s* : \s*
+ (?P<size> 0 | [1-9][0-9]* )
+ \s*
+ $ """,
+ re.X,
+)
+
+RE_CHUNK_COUNT = re.compile(
+ r""" ^
+ Chunk \s+ count \s* : \s*
+ (?P<count> 0 | [1-9][0-9]* )
+ \s*
+ $ """,
+ re.X,
+)
+
+RE_CHUNKS = re.compile(
+ r""" ^
+ \s+
+ Chunk \s+
+ Checksum \s+
+ Start \s+
+ Comp \s size \s+
+ Size \s*
+ $ """,
+ re.X,
+)
+
+RE_CHUNK = re.compile(
+ r""" ^
+ \s+
+ (?P<idx> 0 | [1-9][0-9]* ) \s+
+ (?P<cksum> \S+ ) \s+
+ (?P<start> 0 | [1-9][0-9]* ) \s+
+ (?P<comp_size> 0 | [1-9][0-9]* ) \s+
+ (?P<size> 0 | [1-9][0-9]* ) \s*
+ $ """,
+ re.X,
+)
+
+
+@dataclasses.dataclass(frozen=True)
+class Config:
+ """Runtime configuration."""
+
+ tempd: pathlib.Path
+ bindir: pathlib.Path
+ env: Dict[str, str]
+
+ orig: pathlib.Path
+ compressed: pathlib.Path
+ uncompressed: pathlib.Path
+ recompressed: pathlib.Path
+
+
+def get_runenv() -> Dict[str, str]:
+ """Set up the environment for running the zchunk programs."""
+ env = dict(os.environ)
+ env["LC_ALL"] = "C.UTF-8"
+ env["LANGUAGE"] = ""
+ return env
+
+
+def parse_args(dirname: str) -> Config:
+ """Parse the command-line arguments, deduce some things."""
+ parser = argparse.ArgumentParser(prog="dictionary")
+ parser.add_argument(
+ "bindir",
+ type=str,
+ help="path to the directory containing the zchunk binaries",
+ )
+ parser.add_argument(
+ "filename", type=str, help="path to the filename to compress"
+ )
+
+ args = parser.parse_args()
+ bindir = pathlib.Path(args.bindir).absolute()
+ if not bindir.is_dir():
+ sys.exit(f"Not a directory: {bindir}")
+ zck = bindir / "zck"
+ if not zck.is_file() or not os.access(zck, os.X_OK):
+ sys.exit(f"Not an executable file: {zck}")
+
+ tempd = pathlib.Path(dirname).absolute()
+ return Config(
+ tempd=tempd,
+ bindir=bindir,
+ env=get_runenv(),
+ orig=pathlib.Path(args.filename).absolute(),
+ compressed=tempd / "words.txt.zck",
+ uncompressed=tempd / "un/words.txt",
+ recompressed=tempd / "re/words.txt.zck",
+ )
+
+
+def do_compress(cfg: Config, orig_size: int) -> int:
+ """Compress the original file."""
+ print(f"About to compress {cfg.orig} to {cfg.compressed}")
+ if cfg.compressed.exists():
+ sys.exit(f"Did not expect {cfg.compressed} to exist")
+ subprocess.check_call(
+ [cfg.bindir / "zck", "-o", cfg.compressed, "--", cfg.orig],
+ shell=False,
+ env=cfg.env,
+ )
+ if not cfg.compressed.is_file():
+ sys.exit(f"zck did not create the {cfg.compressed} file")
+ comp_size = cfg.compressed.stat().st_size
+ print(f"{cfg.compressed} size is {comp_size} bytes long")
+ if comp_size >= orig_size:
+ sys.exit(
+ f"sizeof({cfg.compressed}) == {comp_size} : "
+ f"sizeof({cfg.orig}) == {orig_size}"
+ )
+ start = cfg.compressed.open(mode="rb").read(5)
+ print(f"{cfg.compressed} starts with {start!r}")
+ if start != MAGIC:
+ sys.exit(f"{cfg.compressed} does not start with {MAGIC!r}: {start!r}")
+
+ return comp_size
+
+
+def read_chunks(cfg: Config, orig_size: int, comp_size: int) -> None:
+ """Parse the chunks of the compressed file."""
+ # pylint: disable=too-many-statements
+ output = subprocess.check_output(
+ [cfg.bindir / "zck_read_header", "-c", "--", cfg.compressed],
+ shell=False,
+ env=cfg.env,
+ ).decode("UTF-8")
+
+ params: Dict[str, int] = {}
+ chunks: List[Tuple[int, int, int, int, int]] = []
+
+ def ignore_till_end(line: str) -> str:
+ """Ignore anything until EOF."""
+ raise NotImplementedError(line)
+
+ def parse_chunk(line: str) -> str:
+ """Parse a single chunk line."""
+ # pylint: disable=too-many-branches
+ data = RE_CHUNK.match(line)
+ if not data:
+ sys.exit(f"Unexpected line for chunk {len(chunks)}: {line!r}")
+ idx = int(data.group("idx"))
+ start = int(data.group("start"))
+ csize = int(data.group("comp_size"))
+ size = int(data.group("size"))
+
+ if idx != len(chunks):
+ sys.exit(f"Expected index {len(chunks)}: {line!r}")
+ if chunks:
+ last_chunk = chunks[-1]
+ if start != last_chunk[3]:
+ sys.exit(f"Expected start {last_chunk[3]}: {line!r}")
+ else:
+ if start != params["size_diff"]:
+ sys.exit(f"Expected start {params['size_diff']}: {line!r}")
+ last_chunk = (0, 0, 0, params["size_diff"], 0)
+
+ next_chunk = (
+ start,
+ csize,
+ size,
+ last_chunk[3] + csize,
+ last_chunk[4] + size,
+ )
+ if next_chunk[3] > comp_size:
+ sys.exit(
+ f"Compressed size overflow: {next_chunk[3]} > {comp_size}"
+ )
+
+ more = idx + 1 != params["chunk_count"]
+ if more:
+ if next_chunk[4] >= orig_size:
+ sys.exit(
+ f"Original size overflow: {next_chunk[4]} >= {orig_size}"
+ )
+ else:
+ if next_chunk[3] != comp_size:
+ sys.exit(
+ f"Compressed size mismatch: {next_chunk[3]} != {comp_size}"
+ )
+ if next_chunk[4] != orig_size:
+ sys.exit(
+ f"Original size mismatch: {next_chunk[4]} != {orig_size}"
+ )
+
+ print(f"- appending {next_chunk!r}")
+ chunks.append(next_chunk)
+
+ if more:
+ return "parse_chunk"
+ return "ignore_till_end"
+
+ def wait_for_chunks(line: str) -> str:
+ """Wait for the 'Chunks:' line."""
+ if not RE_CHUNKS.match(line):
+ return "wait_for_chunks"
+
+ return "parse_chunk"
+
+ def wait_for_chunk_count(line: str) -> str:
+ """Wait for the 'chunk count' line."""
+ data = RE_CHUNK_COUNT.match(line)
+ if not data:
+ return "wait_for_chunk_count"
+ print(f"- got a chunk count: {data.groupdict()!r}")
+
+ count = int(data.group("count"))
+ if count < 1:
+ sys.exit(f"zck_read_header said chunk count {count}")
+ params["chunk_count"] = count
+
+ return "wait_for_chunks"
+
+ def wait_for_total_size(line: str) -> str:
+ """Wait for the 'data size' line."""
+ data = RE_DATA_SIZE.match(line)
+ if not data:
+ return "wait_for_total_size"
+ print(f"- got a size line: {data.groupdict()!r}")
+
+ size = int(data.group("size"))
+ if size < 1 or size > comp_size:
+ sys.exit(
+ f"zck_read_header said data size {size} (comp {comp_size})"
+ )
+ params["size_diff"] = comp_size - size
+
+ return "wait_for_chunk_count"
+
+ handlers: Dict[str, Callable[[str], str]] = {
+ func.__name__: func
+ for func in (
+ wait_for_total_size,
+ wait_for_chunk_count,
+ wait_for_chunks,
+ parse_chunk,
+ ignore_till_end,
+ )
+ }
+
+ handler: Callable[[str], str] = wait_for_total_size
+
+ for line in output.splitlines():
+ print(f"- read a line: {line}")
+ new_handler = handler(line)
+ assert new_handler in handlers, new_handler
+ handler = handlers[new_handler]
+
+ if handler != ignore_till_end: # pylint: disable=comparison-with-callable
+ sys.exit(f"handler is {handler!r} instead of {ignore_till_end!r}")
+
+
+def do_uncompress(cfg: Config, orig_size: int) -> None:
+ """Uncompress and compare."""
+ # OK, so unzck's behavior is... weird.
+ cfg.uncompressed.parent.mkdir(mode=0o755)
+
+ print(f"Extracting {cfg.compressed} to {cfg.uncompressed}")
+ if cfg.uncompressed.exists():
+ sys.exit(f"Did not expect {cfg.uncompressed} to exist")
+ subprocess.check_call(
+ [cfg.bindir / "unzck", "--", cfg.compressed],
+ shell=False,
+ env=cfg.env,
+ cwd=cfg.uncompressed.parent,
+ )
+ if not cfg.uncompressed.is_file():
+ subprocess.check_call(["ls", "-lt", "--", cfg.tempd], shell=False)
+ sys.exit(f"unzck did not create the {cfg.uncompressed} file")
+
+ new_size = cfg.uncompressed.stat().st_size
+ print(f"Uncompressed size {new_size}")
+ if new_size != orig_size:
+ sys.exit(f"Uncompressed size {new_size} != original size {orig_size}")
+
+ print(f"Comparing {cfg.orig} to {cfg.uncompressed}")
+ subprocess.check_call(
+ ["cmp", "--", cfg.orig, cfg.uncompressed], shell=False, env=cfg.env
+ )
+
+
+def do_recompress(cfg: Config, comp_size: int) -> None:
+ """Recompress the file and compare."""
+ # OK, so zck's behavior is also weird...
+ cfg.recompressed.parent.mkdir(mode=0o755)
+
+ print(f"Recompressing {cfg.uncompressed} to {cfg.recompressed}")
+ if cfg.recompressed.exists():
+ sys.exit(f"Did not expect {cfg.recompressed} to exist")
+ subprocess.check_call(
+ [cfg.bindir / "zck", "--", cfg.uncompressed],
+ shell=False,
+ env=cfg.env,
+ cwd=cfg.recompressed.parent,
+ )
+ if not cfg.recompressed.is_file():
+ sys.exit(f"zck did not create the {cfg.recompressed} file")
+
+ new_size = cfg.recompressed.stat().st_size
+ print(f"Recompressed size {new_size}")
+ if new_size != comp_size:
+ sys.exit(
+ f"Recompressed size {new_size} != compressed size {comp_size}"
+ )
+
+ print(f"Comparing {cfg.compressed} to {cfg.recompressed}")
+ subprocess.check_call(
+ ["cmp", "--", cfg.compressed, cfg.recompressed],
+ shell=False,
+ env=cfg.env,
+ )
+
+
+def main() -> None:
+ """Create a temporary directory, compress a file, analyze it."""
+ with tempfile.TemporaryDirectory() as dirname:
+ print(f"Using temporary directory {dirname}")
+ cfg = parse_args(dirname)
+ orig_size = cfg.orig.stat().st_size
+ print(f"{cfg.orig} is {orig_size} bytes long")
+
+ comp_size = do_compress(cfg, orig_size)
+ read_chunks(cfg, orig_size, comp_size)
+ do_uncompress(cfg, orig_size)
+ do_recompress(cfg, comp_size)
+ print("Seems fine!")
+
+
+if __name__ == "__main__":
+ main()
projects / zchunk.git / commitdiff