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_zip_files(path1, path2, source=None):
differences = []
try:
with ZipFile(path1, 'r') as zip1:
with ZipFile(path2, 'r') as zip2:
# look up differences in content
with make_temp_directory() as temp_dir1:
with make_temp_directory() as temp_dir2:
for name in sorted(set(zip1.namelist())
.intersection(zip2.namelist())):
# skip directories
if name.endswith('/'):
continue
logger.debug('extract member %s', name)
in_path1 = zip1.extract(name, temp_dir1)
in_path2 = zip2.extract(name, temp_dir2)
differences.append(
debbindiff.comparators.compare_files(
in_path1, in_path2,
source=name))
os.unlink(in_path1)
os.unlink(in_path2)
# look up differences in metadata
difference = Difference.from_command(Zipinfo, path1, path2)
if not difference:
# search harder
difference = Difference.from_command(ZipinfoVerbose, path1, path2)
differences.append(difference)
except BadZipfile:
logger.debug('Either %s or %s is not a zip file.' % (path1, path2))
# we'll fallback on binary comparison
return differences
def compare_cpio_files(path1, path2, source=None):
differences = []
difference = Difference.from_command(CpioContent,
path1,
path2,
source="file list")
if difference:
differences.append(difference)
# compare files contained in archive
content1 = get_cpio_names(path1)
content2 = get_cpio_names(path2)
with make_temp_directory() as temp_dir1:
with make_temp_directory() as temp_dir2:
extract_cpio_archive(path1, temp_dir1)
extract_cpio_archive(path2, temp_dir2)
files1 = content1.splitlines(1)
files2 = content2.splitlines(1)
for member in sorted(set(files1).intersection(set(files2))):
in_path1 = os.path.join(temp_dir1, member)
in_path2 = os.path.join(temp_dir2, member)
if not os.path.isfile(in_path1) or not os.path.isfile(
in_path2):
continue
differences.extend(
debbindiff.comparators.compare_files(in_path1,
in_path2,
source=member))
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_iso9660_files(path1, path2, source=None):
differences = []
# compare metadata
difference = Difference.from_command(ISO9660PVD, path1, path2)
if difference:
differences.append(difference)
for extension in (None, 'joliet', 'rockridge'):
difference = Difference.from_command(ISO9660Listing,
path1,
path2,
command_args=(extension, ))
if difference:
differences.append(difference)
# compare files contained in image
files1 = get_iso9660_names(path1)
files2 = get_iso9660_names(path2)
with make_temp_directory() as temp_dir1:
with make_temp_directory() as temp_dir2:
for name in sorted(set(files1).intersection(files2)):
logger.debug('extract file %s' % name)
in_path1 = os.path.join(temp_dir1, os.path.basename(name))
in_path2 = os.path.join(temp_dir2, os.path.basename(name))
with open(in_path1, 'w') as dest:
extract_from_iso9660(path1, name, dest)
with open(in_path2, 'w') as dest:
extract_from_iso9660(path2, name, dest)
differences.extend(
debbindiff.comparators.compare_files(in_path1,
in_path2,
source=name))
return differences
def compare_squashfs_files(path1, path2, source=None):
differences = []
# compare metadata
difference = Difference.from_command(SquashfsSuperblock, path1, path2)
if difference:
differences.append(difference)
difference = Difference.from_command(SquashfsListing, path1, path2)
if difference:
differences.append(difference)
# compare files contained in archive
files1 = get_squashfs_names(path1)
files2 = get_squashfs_names(path2)
with make_temp_directory() as temp_dir1:
with make_temp_directory() as temp_dir2:
extract_squashfs(path1, temp_dir1)
extract_squashfs(path2, temp_dir2)
for member in sorted(set(files1).intersection(set(files2))):
in_path1 = os.path.join(temp_dir1, member)
in_path2 = os.path.join(temp_dir2, member)
if not os.path.isfile(in_path1) or not os.path.isfile(in_path2):
continue
differences.extend(debbindiff.comparators.compare_files(
in_path1, in_path2, source=member))
return differences
def compare_iso9660_files(path1, path2, source=None):
differences = []
# compare metadata
differences.append(Difference.from_command(ISO9660PVD, path1, path2))
for extension in (None, 'joliet', 'rockridge'):
differences.append(Difference.from_command(ISO9660Listing, path1, path2, command_args=(extension,)))
# compare files contained in image
files1 = get_iso9660_names(path1)
files2 = get_iso9660_names(path2)
with make_temp_directory() as temp_dir1:
with make_temp_directory() as temp_dir2:
for name in sorted(set(files1).intersection(files2)):
logger.debug('extract file %s' % name)
in_path1 = os.path.join(temp_dir1, os.path.basename(name))
in_path2 = os.path.join(temp_dir2, os.path.basename(name))
with open(in_path1, 'w') as dest:
extract_from_iso9660(path1, name, dest)
with open(in_path2, 'w') as dest:
extract_from_iso9660(path2, name, dest)
differences.append(debbindiff.comparators.compare_files(
in_path1, in_path2, source=name))
return differences
def compare_zip_files(path1, path2, source=None):
differences = []
try:
with ZipFile(path1, 'r') as zip1:
with ZipFile(path2, 'r') as zip2:
# look up differences in content
with make_temp_directory() as temp_dir1:
with make_temp_directory() as temp_dir2:
for name in sorted(
set(zip1.namelist()).intersection(
zip2.namelist())):
# skip directories
if name.endswith('/'):
continue
logger.debug('extract member %s', name)
in_path1 = zip1.extract(name, temp_dir1)
in_path2 = zip2.extract(name, temp_dir2)
differences.extend(
debbindiff.comparators.compare_files(
in_path1, in_path2, source=name))
os.unlink(in_path1)
os.unlink(in_path2)
# look up differences in metadata
difference = Difference.from_command(Zipinfo, path1, path2)
if not difference:
# search harder
difference = Difference.from_command(
ZipinfoVerbose, path1, path2)
if difference:
differences.append(difference)
except BadZipfile:
logger.debug('Either %s or %s is not a zip file.' % (path1, path2))
# we'll fallback on binary comparison
return differences
def compare_cpio_files(path1, path2, source=None):
differences = []
difference = Difference.from_command(
CpioContent, path1, path2, source="file list")
if difference:
differences.append(difference)
# compare files contained in archive
content1 = get_cpio_names(path1)
content2 = get_cpio_names(path2)
with make_temp_directory() as temp_dir1:
with make_temp_directory() as temp_dir2:
extract_cpio_archive(path1, temp_dir1)
extract_cpio_archive(path2, temp_dir2)
files1 = content1.splitlines(1)
files2 = content2.splitlines(1)
for member in sorted(set(files1).intersection(set(files2))):
in_path1 = os.path.join(temp_dir1, member)
in_path2 = os.path.join(temp_dir2, member)
if not os.path.isfile(in_path1) or not os.path.isfile(in_path2):
continue
differences.extend(debbindiff.comparators.compare_files(
in_path1, in_path2, source=member))
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_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 decompress_gzip(path):
with make_temp_directory() as temp_dir:
if path.endswith('.gz'):
temp_path = os.path.join(temp_dir, os.path.basename(path[:-3]))
else:
temp_path = os.path.join(temp_dir, "%s-content" % path)
with open(temp_path, 'wb') as temp_file:
subprocess.check_call(
["gzip", "--decompress", "--stdout", path],
shell=False, stdout=temp_file, stderr=None)
yield temp_path
def extract_rpm_payload(path):
cmd = ["rpm2cpio", path]
with make_temp_directory() as temp_dir:
temp_path = os.path.join(temp_dir, "CONTENTS.cpio")
with open(temp_path, "wb") as temp_file:
p = subprocess.Popen(cmd, shell=False, stdout=temp_file, stderr=subprocess.PIPE)
p.wait()
if p.returncode != 0:
logger.error("rpm2cpio exited with error code %d", p.returncode)
yield temp_path
def decompress_xz(path):
with make_temp_directory() as temp_dir:
if path.endswith('.xz'):
temp_path = os.path.join(temp_dir, os.path.basename(path[:-3]))
else:
temp_path = os.path.join(temp_dir, "%s-content" % path)
with open(temp_path, 'wb') as temp_file:
subprocess.check_call(["xz", "--decompress", "--stdout", path],
shell=False,
stdout=temp_file,
stderr=None)
yield temp_path
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 extract_rpm_payload(path):
cmd = ['rpm2cpio', path]
with make_temp_directory() as temp_dir:
temp_path = os.path.join(temp_dir, "CONTENTS.cpio")
with open(temp_path, 'wb') as temp_file:
p = subprocess.Popen(cmd,
shell=False,
stdout=temp_file,
stderr=subprocess.PIPE)
p.wait()
if p.returncode != 0:
logger.error("rpm2cpio exited with error code %d",
p.returncode)
yield temp_path
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
