feat: add archive plugin for zip, tar and gzip comments

tests/test_archive.py

@@ -0,0 +1,139 @@
+import gzip
+import io
+import struct
+import tarfile
+import zipfile
+import zlib
+
+import pytest
+
+import orvanitag.orvani as orvani
+from orvanitag.entry import Entry
+from orvanitag.orvani import OrvaniTag
+from orvanitag.errors import UnsupportedFile
+
+
+@pytest.fixture
+def cwd(tmp_path, monkeypatch):
+    monkeypatch.chdir(tmp_path)
+    return tmp_path
+
+
+def make_tar(path, **kwargs):
+    with tarfile.open(path, 'w', **kwargs) as archive:
+        data = b'packed'
+        info = tarfile.TarInfo('a.txt')
+        info.size = len(data)
+        archive.addfile(info, io.BytesIO(data))
+
+
+def members(path):
+    with tarfile.open(path) as archive:
+        return {member.name: archive.extractfile(member).read() for member in archive.getmembers()}
+
+
+def roundtrip(cwd, name, *tags):
+    """Writes the tags, reads them back into a fresh entry and clears them again"""
+    OrvaniTag().add(name, *tags)
+    OrvaniTag().write_metadata(name, 'archive')
+    expected = (cwd / orvani.CONTEXT_FILE).read_text()
+    (cwd / orvani.CONTEXT_FILE).unlink()
+    OrvaniTag().read_metadata(name)
+    assert (cwd / orvani.CONTEXT_FILE).read_text() == expected
+
+
+def test_zip_comment(cwd):
+    with zipfile.ZipFile(cwd / 'a.zip', 'w') as z:
+        z.writestr('a.txt', 'packed')
+    roundtrip(cwd, 'a.zip', '+backup', '#old')
+    with zipfile.ZipFile(cwd / 'a.zip') as z:
+        assert z.comment == b'otag: +backup #old'
+        assert z.read('a.txt') == b'packed'
+
+
+@pytest.mark.parametrize('name', ['a.gz', 'a.tar.gz', 'a.tgz'])
+def test_gzip_comment_and_clear_restores_the_file(cwd, name):
+    with gzip.GzipFile(cwd / name, 'wb', mtime=0) as file:
+        file.write(b'packed')
+    before = (cwd / name).read_bytes()
+
+    roundtrip(cwd, name, '#фантастика', 'a note')
+    assert b'otag: #\xd1\x84' in (cwd / name).read_bytes()
+    assert gzip.decompress((cwd / name).read_bytes()) == b'packed'
+
+    OrvaniTag().write_metadata(name, 'archive', clear=True)
+    assert (cwd / name).read_bytes() == before
+
+
+def test_gzip_keeps_extra_field_and_drops_header_crc(cwd):
+    header = b'\x1f\x8b\x08' + bytes([2 | 4 | 8]) + bytes(6) + struct.pack('<H', 4) + b'ab\0\0' + b'a.txt\0'
+    deflate = zlib.compressobj(wbits=-15)
+    body = deflate.compress(b'packed') + deflate.flush() + struct.pack('<II', zlib.crc32(b'packed'), 6)
+    (cwd / 'a.gz').write_bytes(header + struct.pack('<H', zlib.crc32(header) & 0xffff) + body)
+
+    roundtrip(cwd, 'a.gz', '#lofi')
+    data = (cwd / 'a.gz').read_bytes()
+    assert data == b'\x1f\x8b\x08' + bytes([4 | 8 | 16]) + header[4:] + b'otag: #lofi\0' + body
+    assert gzip.decompress(data) == b'packed'
+
+
+def test_tar_global_header_and_clear_restores_the_file(cwd):
+    make_tar(cwd / 'a.tar')
+    before = (cwd / 'a.tar').read_bytes()
+
+    roundtrip(cwd, 'a.tar', '+backup', 'a note')
+    with tarfile.open(cwd / 'a.tar') as archive:
+        assert archive.pax_headers == {'comment': 'otag: +backup a note'}
+    assert members(cwd / 'a.tar') == {'a.txt': b'packed'}
+
+    OrvaniTag().add('a.tar', '#more')
+    OrvaniTag().write_metadata('a.tar', 'archive')
+    with tarfile.open(cwd / 'a.tar') as archive:
+        assert archive.pax_headers == {'comment': 'otag: +backup #more a note'}
+
+    OrvaniTag().write_metadata('a.tar', 'archive', clear=True)
+    assert (cwd / 'a.tar').read_bytes() == before
+
+
+def test_tar_keeps_other_global_records(cwd):
+    make_tar(cwd / 'a.tar', pax_headers={'mtime': '1'})
+    OrvaniTag().add('a.tar', '#lofi')
+    OrvaniTag().write_metadata('a.tar', 'archive')
+    OrvaniTag().write_metadata('a.tar', 'archive', clear=True)
+    with tarfile.open(cwd / 'a.tar') as archive:
+        assert archive.pax_headers == {'mtime': '1'}
+    assert members(cwd / 'a.tar') == {'a.txt': b'packed'}
+
+
+def foreign_zip(path):
+    with zipfile.ZipFile(path, 'w') as z:
+        z.comment = b'foreign'
+
+
+def foreign_gzip(path):
+    path.write_bytes(b'\x1f\x8b\x08' + bytes([16]) + bytes(6) + b'foreign\0' + gzip.compress(b'packed', mtime=0)[10:])
+
+
+@pytest.mark.parametrize('name, make', [
+    ('a.tar', lambda path: make_tar(path, pax_headers={'comment': '0123abcd'})), # like git archive
+    ('a.gz', foreign_gzip),
+    ('a.zip', foreign_zip),
+])
+def test_foreign_comments_are_kept(cwd, name, make):
+    make(cwd / name)
+    before = (cwd / name).read_bytes()
+    assert OrvaniTag().read_metadata(name) == Entry()
+    OrvaniTag().add(name, '#lofi')
+    with pytest.raises(UnsupportedFile, match='used by another tool'):
+        OrvaniTag().write_metadata(name, 'archive')
+    OrvaniTag().write_metadata(name, 'archive', clear=True)
+    assert (cwd / name).read_bytes() == before
+
+
+@pytest.mark.parametrize('name', ['a.tar', 'a.gz'])
+def test_other_files_are_refused_unchanged(cwd, name):
+    (cwd / name).write_bytes(gzip.compress(b'packed') if name == 'a.tar' else b'plain text')
+    OrvaniTag().add(name, '#lofi')
+    with pytest.raises(UnsupportedFile):
+        OrvaniTag().write_metadata(name, 'archive')
+    assert OrvaniTag().scan('.', False)[1]