def compare_meta(path1, path2):
logger.debug('compare_meta(%s, %s)' % (path1, path2))
differences = []
try:
difference = Difference.from_command(Stat, path1, path2)
if difference:
differences.append(difference)
except RequiredToolNotFound:
logger.warn("'stat' not found! Is PATH wrong?")
try:
lsattr1 = lsattr(path1)
lsattr2 = lsattr(path2)
difference = Difference.from_unicode(
lsattr1, lsattr2, path1, path2, source="lattr")
if difference:
differences.append(difference)
except RequiredToolNotFound:
logger.info("Unable to find 'lsattr'.")
try:
difference = Difference.from_command(Getfacl, path1, path2)
if difference:
differences.append(difference)
except RequiredToolNotFound:
logger.info("Unable to find 'getfacl'.")
return differences
def compare_rpm_files(path1, path2, source=None):
try:
import rpm
except ImportError:
logger.info("Python module rpm not found.")
return []
differences = []
# compare headers
with make_temp_directory() as rpmdb_dir:
rpm.addMacro("_dbpath", rpmdb_dir)
ts = rpm.TransactionSet()
ts.setVSFlags(-1)
header1 = get_rpm_header(path1, ts)
header2 = get_rpm_header(path2, ts)
difference = Difference.from_unicode(
header1, header2, path1, path2, source="header")
if difference:
differences.append(difference)
# extract cpio archive
with extract_rpm_payload(path1) as archive1:
with extract_rpm_payload(path2) as archive2:
differences.extend(debbindiff.comparators.compare_files(
archive1, archive2, source=get_source(archive1, archive2)))
return differences
def compare_deb_files(path1, path2, source=None):
differences = []
# look up differences in content
ar1 = ArFile(filename=path1)
ar2 = ArFile(filename=path2)
with make_temp_directory() as temp_dir1:
with make_temp_directory() as temp_dir2:
logger.debug('content1 %s', ar1.getnames())
logger.debug('content2 %s', ar2.getnames())
for name in sorted(set(ar1.getnames())
.intersection(ar2.getnames())):
logger.debug('extract member %s', name)
member1 = ar1.getmember(name)
member2 = ar2.getmember(name)
in_path1 = os.path.join(temp_dir1, name)
in_path2 = os.path.join(temp_dir2, name)
with open(in_path1, 'w') as f1:
f1.write(member1.read())
with open(in_path2, 'w') as f2:
f2.write(member2.read())
differences.extend(
debbindiff.comparators.compare_files(
in_path1, in_path2, source=name))
os.unlink(in_path1)
os.unlink(in_path2)
# look up differences in file list and file metadata
content1 = get_ar_content(path1)
content2 = get_ar_content(path2)
difference = Difference.from_unicode(
content1, content2, path1, path2, source="metadata")
if difference:
differences.append(difference)
return differences
def compare_directories(path1, path2, source=None):
differences = []
logger.debug('path1 files: %s' % sorted(set(os.listdir(path1))))
logger.debug('path2 files: %s' % sorted(set(os.listdir(path2))))
for name in sorted(set(os.listdir(path1)).intersection(set(os.listdir(path2)))):
logger.debug('compare %s' % name)
in_path1 = os.path.join(path1, name)
in_path2 = os.path.join(path2, name)
in_differences = debbindiff.comparators.compare_files(
in_path1, in_path2, source=name)
if not os.path.isdir(in_path1):
if in_differences:
in_differences[0].add_details(compare_meta(in_path1, in_path2))
else:
d = Difference(None, path1, path2, source=name)
d.add_details(compare_meta(in_path1, in_path2))
in_differences = [d]
differences.extend(in_differences)
ls1 = ls(path1)
ls2 = ls(path2)
difference = Difference.from_unicode(ls1, ls2, path1, path2, source="ls")
if difference:
differences.append(difference)
differences.extend(compare_meta(path1, path2))
if differences:
d = Difference(None, path1, path2, source=source)
d.add_details(differences)
return [d]
return []
def compare_rpm_files(path1, path2, source=None):
try:
import rpm
except ImportError:
logger.info("Python module rpm not found.")
return []
differences = []
# compare headers
with make_temp_directory() as rpmdb_dir:
rpm.addMacro("_dbpath", rpmdb_dir)
ts = rpm.TransactionSet()
ts.setVSFlags(-1)
header1 = get_rpm_header(path1, ts)
header2 = get_rpm_header(path2, ts)
difference = Difference.from_unicode(header1,
header2,
path1,
path2,
source="header")
if difference:
differences.append(difference)
# extract cpio archive
with extract_rpm_payload(path1) as archive1:
with extract_rpm_payload(path2) as archive2:
differences.extend(
debbindiff.comparators.compare_files(archive1,
archive2,
source=get_source(
archive1, archive2)))
return differences
def compare_tar_files(path1, path2, source=None):
differences = []
with tarfile.open(path1, 'r') as tar1:
with tarfile.open(path2, 'r') as tar2:
# look up differences in content
with make_temp_directory() as temp_dir1:
with make_temp_directory() as temp_dir2:
logger.debug('content1 %s', tar1.getnames())
logger.debug('content2 %s', tar2.getnames())
for name in sorted(set(tar1.getnames())
.intersection(tar2.getnames())):
member1 = tar1.getmember(name)
member2 = tar2.getmember(name)
if not member1.isfile() or not member2.isfile():
continue
logger.debug('extract member %s', name)
tar1.extract(name, temp_dir1)
tar2.extract(name, temp_dir2)
in_path1 = os.path.join(temp_dir1, name).decode('utf-8')
in_path2 = os.path.join(temp_dir2, name).decode('utf-8')
differences.append(
debbindiff.comparators.compare_files(
in_path1, in_path2,
source=name.decode('utf-8')))
os.unlink(in_path1)
os.unlink(in_path2)
# look up differences in file list and file metadata
content1 = get_tar_content(tar1).decode('utf-8')
content2 = get_tar_content(tar2).decode('utf-8')
differences.append(Difference.from_unicode(
content1, content2, path1, path2, source="metadata"))
return differences
def compare_binary_files(path1, path2, source=None):
try:
with xxd(path1) as xxd1:
with xxd(path2) as xxd2:
return Difference.from_file(xxd1, xxd2, path1, path2, source)
except RequiredToolNotFound:
hexdump1 = hexdump_fallback(path1)
hexdump2 = hexdump_fallback(path2)
comment = 'xxd not available in path. Falling back to Python hexlify.\n'
return Difference.from_unicode(hexdump1, hexdump2, path1, path2, source, comment)
def compare_static_lib_files(path1, path2, source=None):
differences = []
# look up differences in metadata
content1 = get_ar_content(path1)
content2 = get_ar_content(path2)
difference = Difference.from_unicode(
content1, content2, path1, path2, source="metadata")
if difference:
differences.append(difference)
differences.extend(_compare_elf_data(path1, path2, source))
return differences
def compare_gzip_files(path1, path2, source=None):
differences = []
# check metadata
metadata1 = get_gzip_metadata(path1)
metadata2 = get_gzip_metadata(path2)
differences.append(Difference.from_unicode(
metadata1, metadata2, path1, path2, source='metadata'))
# check content
with decompress_gzip(path1) as new_path1:
with decompress_gzip(path2) as new_path2:
differences.append(debbindiff.comparators.compare_files(
new_path1, new_path2,
source=[os.path.basename(new_path1), os.path.basename(new_path2)]))
return differences
def compare_binary_files(path1, path2, source=None):
try:
with xxd(path1) as xxd1:
with xxd(path2) as xxd2:
difference = Difference.from_file(xxd1, xxd2, path1, path2,
source)
except RequiredToolNotFound:
hexdump1 = hexdump_fallback(path1)
hexdump2 = hexdump_fallback(path2)
comment = 'xxd not available in path. Falling back to Python hexlify.\n'
difference = Difference.from_unicode(hexdump1, hexdump2, path1, path2,
source, comment)
if not difference:
return []
return [difference]
def compare_ipk_files(path1, path2, source=None):
differences = []
# metadata
metadata1 = get_gzip_metadata(path1)
metadata2 = get_gzip_metadata(path2)
differences.append(Difference.from_unicode(
metadata1, metadata2, path1, path2, source='metadata'))
# content
with decompress_gzip(path1) as tar1:
with decompress_gzip(path2) as tar2:
differences.append(compare_tar_files(tar1, tar2,
source=[os.path.basename(tar1), os.path.basename(tar2)]))
return differences
def compare_files(path1, path2, source=None):
if os.path.islink(path1) or os.path.islink(path2):
dest1, dest2 = None, None
try:
dest1 = os.readlink(path1)
text1 = "%s -> %s" % (path1, dest1)
except OSError:
text1 = "[ No symlink ]"
try:
dest2 = os.readlink(path2)
text2 = "%s -> %s" % (path2, dest2)
except OSError:
text2 = "[ No symlink ]"
if dest1 and dest2 and dest1 == dest2:
return None
return Difference.from_unicode(text1, text2, path1, path2, source=source, comment="symlink")
if os.path.isdir(path1) and os.path.isdir(path2):
return compare_directories(path1, path2, source)
if not os.path.isfile(path1):
logger.critical("%s is not a file", path1)
sys.exit(2)
if not os.path.isfile(path2):
logger.critical("%s is not a file", path2)
sys.exit(2)
# try comparing small files directly first
size1 = os.path.getsize(path1)
size2 = os.path.getsize(path2)
if size1 == size2 and size1 <= SMALL_FILE_THRESHOLD:
if file(path1).read() == file(path2).read():
return None
# ok, let's do the full thing
mime_type1 = guess_mime_type(path1)
mime_type2 = guess_mime_type(path2)
for mime_type_regex, filename_regex, comparator in COMPARATORS:
if filename_regex and re.search(filename_regex, path1) \
and re.search(filename_regex, path2):
return comparator(path1, path2, source=source)
if mime_type_regex:
match1 = re.search(mime_type_regex, mime_type1)
match2 = re.search(mime_type_regex, mime_type2)
if match1 and match2 and match1.groupdict() == match2.groupdict():
return comparator(path1, path2, source=source, **match1.groupdict())
return compare_unknown(path1, path2, source)
def compare_rpm_files(path1, path2, source=None):
differences = []
# compare headers
with make_temp_directory() as rpmdb_dir:
rpm.addMacro("_dbpath", rpmdb_dir)
ts = rpm.TransactionSet()
ts.setVSFlags(-1)
header1 = get_rpm_header(path1, ts)
header2 = get_rpm_header(path2, ts)
differences.append(Difference.from_unicode(header1, header2, path1, path2, source="header"))
# extract cpio archive
with extract_rpm_payload(path1) as archive1:
with extract_rpm_payload(path2) as archive2:
differences.append(
debbindiff.comparators.compare_files(archive1, archive2, source=get_source(archive1, archive2))
)
return differences
def compare_tar_files(path1, path2, source=None):
differences = []
with tarfile.open(path1, 'r') as tar1:
with tarfile.open(path2, 'r') as tar2:
# look up differences in content
with make_temp_directory() as temp_dir1:
with make_temp_directory() as temp_dir2:
logger.debug('content1 %s', tar1.getnames())
logger.debug('content2 %s', tar2.getnames())
for name in sorted(
set(tar1.getnames()).intersection(
tar2.getnames())):
member1 = tar1.getmember(name)
member2 = tar2.getmember(name)
if not member1.isfile() or not member2.isfile():
continue
logger.debug('extract member %s', name)
tar1.extract(name, temp_dir1)
tar2.extract(name, temp_dir2)
in_path1 = os.path.join(temp_dir1, name)
in_path2 = os.path.join(temp_dir2, name)
differences.extend(
debbindiff.comparators.compare_files(in_path1,
in_path2,
source=name))
os.unlink(in_path1)
os.unlink(in_path2)
# look up differences in file list and file metadata
content1 = get_tar_content(tar1).decode('utf-8')
content2 = get_tar_content(tar2).decode('utf-8')
difference = Difference.from_unicode(content1,
content2,
path1,
path2,
source="metadata")
if difference:
differences.append(difference)
return differences
def compare_deb_files(path1, path2, source=None):
differences = []
# look up differences in content
ar1 = ArFile(filename=path1)
ar2 = ArFile(filename=path2)
with make_temp_directory() as temp_dir1:
with make_temp_directory() as temp_dir2:
logger.debug('content1 %s', ar1.getnames())
logger.debug('content2 %s', ar2.getnames())
for name in sorted(
set(ar1.getnames()).intersection(ar2.getnames())):
logger.debug('extract member %s', name)
member1 = ar1.getmember(name)
member2 = ar2.getmember(name)
in_path1 = os.path.join(temp_dir1, name)
in_path2 = os.path.join(temp_dir2, name)
with open(in_path1, 'w') as f1:
f1.write(member1.read())
with open(in_path2, 'w') as f2:
f2.write(member2.read())
differences.extend(
debbindiff.comparators.compare_files(in_path1,
in_path2,
source=name))
os.unlink(in_path1)
os.unlink(in_path2)
# look up differences in file list and file metadata
content1 = get_ar_content(path1)
content2 = get_ar_content(path2)
difference = Difference.from_unicode(content1,
content2,
path1,
path2,
source="metadata")
if difference:
differences.append(difference)
return differences
def compare_gzip_files(path1, path2, source=None):
differences = []
# check metadata
metadata1 = get_gzip_metadata(path1)
metadata2 = get_gzip_metadata(path2)
difference = Difference.from_unicode(metadata1,
metadata2,
path1,
path2,
source='metadata')
if difference:
differences.append(difference)
# check content
with decompress_gzip(path1) as new_path1:
with decompress_gzip(path2) as new_path2:
differences.extend(
debbindiff.comparators.compare_files(
new_path1,
new_path2,
source=[
os.path.basename(new_path1),
os.path.basename(new_path2)
]))
return differences
dot_changes1 = Changes(filename=path1)
dot_changes1.validate(check_signature=False)
dot_changes2 = Changes(filename=path2)
dot_changes2.validate(check_signature=False)
except IOError, e:
logger.critical(e)
sys.exit(2)
differences = []
for field in DOT_CHANGES_FIELDS:
if dot_changes1[field] != dot_changes2[field]:
content1 = "%s: %s" % (field, dot_changes1[field])
content2 = "%s: %s" % (field, dot_changes2[field])
difference = Difference.from_unicode(
content1, content2,
dot_changes1.get_changes_file(),
dot_changes2.get_changes_file(),
source=source)
if difference:
differences.append(difference)
# This will handle differences in the list of files, checksums, priority
# and section
files1 = dot_changes1.get('Files')
files2 = dot_changes2.get('Files')
logger.debug(dot_changes1.get_as_string('Files'))
files_difference = Difference.from_unicode(
dot_changes1.get_as_string('Files'),
dot_changes2.get_as_string('Files'),
dot_changes1.get_changes_file(),
@binary_fallback
@returns_details
def compare_dot_changes_files(path1, path2, source=None):
try:
dot_changes1 = Changes(filename=path1)
dot_changes1.validate(check_signature=False)
dot_changes2 = Changes(filename=path2)
dot_changes2.validate(check_signature=False)
except IOError, e:
logger.critical(e)
sys.exit(2)
differences = []
for field in DOT_CHANGES_FIELDS:
differences.append(Difference.from_unicode(
dot_changes1[field].lstrip(),
dot_changes2[field].lstrip(),
path1, path2, source=field))
files_difference = Difference.from_unicode(
dot_changes1.get_as_string('Files'),
dot_changes2.get_as_string('Files'),
path1, path2,
source='Files')
if not files_difference:
return differences
differences.append(files_difference)
# we are only interested in file names
files1 = dict([(d['name'], d) for d in dot_changes1.get('Files')])
