def test_diff_manifest_add_file_last(test_data, testname, monkeypatch):
"""Diff two fileinfo lists, one having an additional file as last item.
The implementation of the corresponding command line tool used to
have a flaw in this particular case, ref. #55.
"""
monkeypatch.chdir(test_data)
with Archive().open(Path("archive.tar")) as archive:
manifest_ref = archive.manifest
base_dir = Path("base")
mtime_base = os.stat(base_dir).st_mtime
p = base_dir / "zzz.dat"
shutil.copy(gettestdata("rnd2.dat"), p)
os.utime(base_dir, times=(mtime_base, mtime_base))
fileinfos = get_fileinfos(base_dir)
diff = list(filter(non_match, diff_manifest(fileinfos, manifest_ref)))
assert len(diff) == 1
status, fi_a, fi_b = diff[0]
assert status == DiffStatus.MISSING_B
assert fi_a.type == 'f'
assert fi_a.path == p
assert fi_b is None
diff = list(filter(non_match, diff_manifest(manifest_ref, fileinfos)))
assert len(diff) == 1
status, fi_a, fi_b = diff[0]
assert status == DiffStatus.MISSING_A
assert fi_b.type == 'f'
assert fi_b.path == p
assert fi_a is None
def test_diff_manifest_mult(test_data, testname, monkeypatch):
"""Diff two fileinfo lists having multiple differences.
"""
monkeypatch.chdir(test_data)
with Archive().open(Path("archive.tar")) as archive:
manifest_ref = archive.manifest
base_dir = Path("base")
mtime_base = os.stat(base_dir).st_mtime
mtime_data = os.stat(base_dir / "data").st_mtime
pm = base_dir / "data" / "rnd.dat"
shutil.copy(gettestdata("rnd2.dat"), pm)
p1 = base_dir / "msg.txt"
p2 = base_dir / "o.txt"
p1.rename(p2)
os.utime(base_dir, times=(mtime_base, mtime_base))
os.utime(base_dir / "data", times=(mtime_data, mtime_data))
fileinfos = get_fileinfos(base_dir)
diff = list(filter(non_match, diff_manifest(fileinfos, manifest_ref)))
assert len(diff) == 3
status, fi_a, fi_b = diff[0]
assert status == DiffStatus.CONTENT
assert fi_a.type == fi_b.type == 'f'
assert fi_a.path == fi_b.path == pm
status, fi_a, fi_b = diff[1]
assert status == DiffStatus.MISSING_A
assert fi_b.type == 'f'
assert fi_b.path == p1
assert fi_a is None
status, fi_a, fi_b = diff[2]
assert status == DiffStatus.MISSING_B
assert fi_a.type == 'f'
assert fi_a.path == p2
assert fi_b is None
def test_diff_manifest_missing_files(test_data, testname, monkeypatch):
"""Diff two fileinfo lists having one file's name changed.
"""
monkeypatch.chdir(test_data)
with Archive().open(Path("archive.tar")) as archive:
manifest_ref = archive.manifest
base_dir = Path("base")
mtime_base = os.stat(base_dir).st_mtime
p1 = base_dir / "rnd.dat"
p2 = base_dir / "a.dat"
p1.rename(p2)
os.utime(base_dir, times=(mtime_base, mtime_base))
fileinfos = get_fileinfos(base_dir)
diff = list(filter(non_match, diff_manifest(fileinfos, manifest_ref)))
assert len(diff) == 2
status, fi_a, fi_b = diff[0]
assert status == DiffStatus.MISSING_B
assert fi_a.type == 'f'
assert fi_a.path == p2
assert fi_b is None
status, fi_a, fi_b = diff[1]
assert status == DiffStatus.MISSING_A
assert fi_b.type == 'f'
assert fi_b.path == p1
assert fi_a is None
def test_diff_manifest_equal(test_data, testname, monkeypatch):
"""Diff two fileinfo lists having equal content.
"""
monkeypatch.chdir(test_data)
with Archive().open(Path("archive.tar")) as archive:
manifest_ref = archive.manifest
base_dir = Path("base")
fileinfos = get_fileinfos(base_dir)
diff = list(filter(non_match, diff_manifest(fileinfos, manifest_ref)))
assert diff == []
def test_diff_manifest_metadata(test_data, testname, monkeypatch):
"""Diff two fileinfo lists having one file's metadata modified.
"""
monkeypatch.chdir(test_data)
with Archive().open(Path("archive.tar")) as archive:
manifest_ref = archive.manifest
base_dir = Path("base")
p = base_dir / "rnd.dat"
p.chmod(0o0444)
fileinfos = get_fileinfos(base_dir)
diff = list(filter(non_match, diff_manifest(fileinfos, manifest_ref)))
assert len(diff) == 1
status, fi_a, fi_b = diff[0]
assert status == DiffStatus.META
assert fi_a.type == fi_b.type == 'f'
assert fi_a.path == fi_b.path == p
def test_diff_manifest_modified_file(test_data, testname, monkeypatch):
"""Diff two fileinfo lists having one file's content modified.
"""
monkeypatch.chdir(test_data)
with Archive().open(Path("archive.tar")) as archive:
manifest_ref = archive.manifest
base_dir = Path("base")
mtime_base = os.stat(base_dir).st_mtime
p = base_dir / "rnd.dat"
shutil.copy(gettestdata("rnd2.dat"), p)
os.utime(base_dir, times=(mtime_base, mtime_base))
fileinfos = get_fileinfos(base_dir)
diff = list(filter(non_match, diff_manifest(fileinfos, manifest_ref)))
assert len(diff) == 1
status, fi_a, fi_b = diff[0]
assert status == DiffStatus.CONTENT
assert fi_a.type == fi_b.type == 'f'
assert fi_a.path == fi_b.path == p
def test_diff_manifest_symlink_target(test_data, testname, monkeypatch):
"""Diff two fileinfo lists having one symlink's target modified.
"""
monkeypatch.chdir(test_data)
with Archive().open(Path("archive.tar")) as archive:
manifest_ref = archive.manifest
base_dir = Path("base")
mtime_base = os.stat(base_dir).st_mtime
p = base_dir / "s.dat"
p.unlink()
p.symlink_to(Path("msg.txt"))
os.utime(base_dir, times=(mtime_base, mtime_base))
fileinfos = get_fileinfos(base_dir)
diff = list(filter(non_match, diff_manifest(fileinfos, manifest_ref)))
assert len(diff) == 1
status, fi_a, fi_b = diff[0]
assert status == DiffStatus.SYMLNK_TARGET
assert fi_a.type == fi_b.type == 'l'
assert fi_a.path == fi_b.path == p
def test_diff_manifest_wrong_type(test_data, testname, monkeypatch):
"""Diff two fileinfo lists with one entry having a wrong type.
"""
monkeypatch.chdir(test_data)
with Archive().open(Path("archive.tar")) as archive:
manifest_ref = archive.manifest
base_dir = Path("base")
mtime_base = os.stat(base_dir).st_mtime
p = base_dir / "rnd.dat"
p.unlink()
p.symlink_to(Path("data", "rnd.dat"))
os.utime(base_dir, times=(mtime_base, mtime_base))
fileinfos = get_fileinfos(base_dir)
diff = list(filter(non_match, diff_manifest(fileinfos, manifest_ref)))
assert len(diff) == 1
status, fi_a, fi_b = diff[0]
assert status == DiffStatus.TYPE
assert fi_a.type == 'l'
assert fi_b.type == 'f'
assert fi_a.path == fi_b.path == p
def test_diff_manifest_dircontent(test_data, testname, monkeypatch):
"""Diff two fileinfo lists with one subdirectory missing.
"""
monkeypatch.chdir(test_data)
with Archive().open(Path("archive.tar")) as archive:
manifest_ref = archive.manifest
base_dir = Path("base")
mtime_base = os.stat(base_dir).st_mtime
pd = base_dir / "data"
shutil.rmtree(pd)
os.utime(base_dir, times=(mtime_base, mtime_base))
fileinfos = get_fileinfos(base_dir)
diff = list(filter(non_match, diff_manifest(fileinfos, manifest_ref)))
assert len(diff) == 2
status, fi_a, fi_b = diff[0]
assert status == DiffStatus.MISSING_A
assert fi_b.type == 'd'
assert fi_b.path == pd
assert fi_a is None
status, fi_a, fi_b = diff[1]
assert status == DiffStatus.MISSING_A
assert fi_b.type == 'f'
assert fi_b.path == pd / "rnd.dat"
assert fi_a is None
def diff(args):
archive1 = Archive().open(args.archive1)
manifest1 = archive1.manifest
archive1.close()
archive2 = Archive().open(args.archive2)
manifest2 = archive2.manifest
archive2.close()
algorithm = _common_checksum(manifest1, manifest2)
diff = diff_manifest(manifest1, manifest2, algorithm)
if args.skip_dir_content:
diff = _skip_dir_filter(diff)
status = 0
for diff_stat, fi1, fi2 in diff:
if diff_stat == DiffStatus.MISSING_A:
print("Only in %s: %s" % (args.archive2, fi2.path))
status = max(status, 102)
elif diff_stat == DiffStatus.MISSING_B:
print("Only in %s: %s" % (args.archive1, fi1.path))
status = max(status, 102)
elif diff_stat == DiffStatus.TYPE:
print("Entries %s:%s and %s:%s have different type"
% (args.archive1, fi1.path, args.archive2, fi2.path))
status = max(status, 102)
elif diff_stat == DiffStatus.SYMLNK_TARGET:
print("Symbol links %s:%s and %s:%s have different target"
% (args.archive1, fi1.path, args.archive2, fi2.path))
status = max(status, 101)
elif diff_stat == DiffStatus.CONTENT:
print("Files %s:%s and %s:%s differ"
% (args.archive1, fi1.path, args.archive2, fi2.path))
status = max(status, 101)
elif diff_stat == DiffStatus.META and args.report_meta:
print("File system metadata for %s:%s and %s:%s differ"
% (args.archive1, fi1.path, args.archive2, fi2.path))
status = max(status, 100)
return status
def filter_fileinfos(base, fileinfos, algorithm):
for stat, fi1, fi2 in diff_manifest(base, fileinfos, algorithm):
if stat == DiffStatus.MISSING_B or stat == DiffStatus.MATCH:
continue
yield fi2