pkgdiff

#!/usr/bin/env python

from collections import defaultdict
import itertools
import pathlib
import re
import subprocess
import sys
import tempfile

def usage():
    print(f'usage: {sys.argv[0]} PACKAGE.src.tar.zst PACKAGE_DIRECTORY/', file=sys.stderr)
    sys.exit(1)

# The following fields may appear only once in each .SRCINFO file, in the pkgbase section
SRCINFO_SINGULAR_PKGBASE = {'pkgver', 'pkgrel', 'epoch'}

# The following fields may appear up to once in any section
SRCINFO_SINGULAR_ANY = {'pkgdesc', 'url', 'install', 'changelog'}

# The following fields may be repeated within a section to specify multiple values
SRCINFO_MULTIPLE_VALUES = {
    'arch',
    'groups',
    'license',
    'noextract',
    'options',
    'backup',
    'validpgpkeys', # can only be in pkgbase
}

# The following fields may, additionally, specify multiple architectures as shown below
SRCINFO_MULTIARCH = {
    'source',
    'depends', 'checkdepends', 'makedepends', 'optdepends',
    'provides', 'conflicts', 'replaces',
    'md5sums', 'sha1sums', 'sha224sums', 'sha256sums', 'sha384sums', 'sha512sums',
}

def parse_srcinfo(f):
    pkgbase_section = defaultdict(list)
    package_sections = defaultdict(lambda: defaultdict(list))

    current_section = None

    for line in f:
        line = line.rstrip()
        line = re.sub(r'#.*', '', line)
        line = re.sub(r'\s+$', '', line)
        if line == '':
            continue

        m = re.match(r'^\s*(\w+)\s*=\s*(.*)', line)
        assert m, f'line {line!r} did not match!'
        key, value = m.groups()

        match key:
            case 'pkgbase':
                current_section = pkgbase_section
            case 'pkgname':
                assert value not in package_sections
                package_sections[value] = current_section = defaultdict(list)
            case _:
                current_section[key].append(value)

    return { name:dict(pkgbase_section | section) for name, section in package_sections.items() }

def identify_remote_files(pkg_dir):
    with pathlib.Path(pkg_dir, '.SRCINFO').open() as f:
        src_info = parse_srcinfo(f)
    sources = list(itertools.chain.from_iterable(package.get('source', []) for package in src_info.values()))

    for source in sources:
        if '::' in source:
            filename, source = source.split('::')
        else:
            filename = pathlib.Path(source).name

        # XXX other protocols
        if source.startswith('http://') or source.startswith('https://'):
            yield filename

def gather_files(pkg_dir, name_blocklist):
    files = set()

    git_subdir = pathlib.Path(pkg_dir) / '.git'

    for child in pathlib.Path(pkg_dir).glob('**/*'):
        if child.is_dir():
            continue

        if child.name in name_blocklist:
            continue

        if child.name == '.SRCINFO':
            continue

        if child.is_relative_to(git_subdir):
            continue

        files.add(child.relative_to(pkg_dir))

    return files

if len(sys.argv) < 3:
    usage()

_, older_src_pkg, pkg_dir, *_ = sys.argv

with tempfile.TemporaryDirectory() as tempdir:
    subprocess.check_call(['tar', 'xf', older_src_pkg, '--zstd', '--strip-components=1', '--directory=' + tempdir])

    old_remote_files = set(identify_remote_files(tempdir))
    new_remote_files = set(identify_remote_files(pkg_dir))

    old_files = gather_files(tempdir, old_remote_files)
    new_files = gather_files(pkg_dir, new_remote_files)

    compare_me = []

    for rel_path in old_files | new_files:
        old_path = tempdir / rel_path
        new_path = pkg_dir / rel_path

        if rel_path in old_files and rel_path in new_files:
            if old_path.read_bytes() != new_path.read_bytes():
                compare_me.extend( (old_path, new_path) )
        elif rel_path in old_files:
            compare_me.extend( (old_path, pathlib.Path('/dev/null')) )
        else: # rel_path in new_files
            compare_me.extend( (pathlib.Path('/dev/null'), new_path) )

    subprocess.check_call(['vimdiff-pairs'] + compare_me)