def compare_root_paths(path1, path2):
from ..directory import (
FilesystemDirectory,
FilesystemFile,
compare_directories,
compare_meta,
)
if not Config().new_file:
bail_if_non_existing(path1, path2)
if any_excluded(path1, path2):
return None
if os.path.isdir(path1) and os.path.isdir(path2):
return compare_directories(path1, path2)
container1 = FilesystemDirectory(os.path.dirname(path1)).as_container
file1 = specialize(FilesystemFile(path1, container=container1))
container2 = FilesystemDirectory(os.path.dirname(path2)).as_container
file2 = specialize(FilesystemFile(path2, container=container2))
difference = compare_files(file1, file2)
if Config().exclude_directory_metadata in ('no', 'recursive'):
meta = compare_meta(path1, path2)
if meta:
# Create an "empty" difference so we have something to attach file
# metadata to.
if difference is None:
difference = Difference(None, file1.name, file2.name)
difference.add_details(meta)
return difference
def compare_files(file1, file2, source=None, diff_content_only=False):
logger.debug(
"Comparing %s (%s) and %s (%s)",
file1.name,
file1.__class__.__name__,
file2.name,
file2.__class__.__name__,
)
if any_excluded(file1.name, file2.name):
return None
with profile('has_same_content_as', file1):
if file1.has_same_content_as(file2):
logger.debug(
"has_same_content_as returned True; skipping further comparisons"
)
return None
if diff_content_only:
difference = Difference(None, file1.name, file2.name)
difference.add_comment("Files differ")
return difference
specialize(file1)
specialize(file2)
if isinstance(file1, MissingFile):
file1.other_file = file2
elif isinstance(file2, MissingFile):
file2.other_file = file1
elif file1.__class__.__name__ != file2.__class__.__name__:
return file1.compare_bytes(file2, source)
with profile('compare_files (cumulative)', file1):
return file1.compare(file2, source)
def get_filtered_members(self):
try:
with libarchive.file_reader(self.source.path) as archive:
for entry in archive:
if any_excluded(entry.pathname):
continue
yield entry.pathname, self.get_subclass(entry)
except libarchive.exception.ArchiveError:
pass
def compare_root_paths(path1, path2):
from ..directory import FilesystemDirectory, FilesystemFile, compare_directories
if not Config().new_file:
bail_if_non_existing(path1, path2)
if any_excluded(path1, path2):
return None
if os.path.isdir(path1) and os.path.isdir(path2):
return compare_directories(path1, path2)
container1 = FilesystemDirectory(os.path.dirname(path1)).as_container
file1 = specialize(FilesystemFile(path1, container=container1))
container2 = FilesystemDirectory(os.path.dirname(path2)).as_container
file2 = specialize(FilesystemFile(path2, container=container2))
return compare_files(file1, file2)
def ensure_unpacked(self):
if hasattr(self, '_members'):
return
tmpdir = get_temporary_directory().name
self._members = collections.OrderedDict()
logger.debug("Extracting %s to %s", self.source.path, tmpdir)
with libarchive.file_reader(self.source.path) as archive:
for idx, entry in enumerate(archive):
# Always skip directories
if entry.isdir:
continue
# Save extracting excluded files
if any_excluded(entry.pathname):
continue
# Keep directory sizes small. could be improved but should be
# good enough for "ordinary" large archives.
dst = os.path.join(tmpdir, str(idx // 4096), str(idx % 4096))
root, ext = os.path.splitext(entry.pathname)
dst += ext
# Maintain a mapping of archive path to the extracted path,
# avoiding the need to sanitise filenames.
self._members[entry.pathname] = dst
logger.debug("Extracting %s to %s", entry.pathname, dst)
os.makedirs(os.path.dirname(dst), exist_ok=True)
try:
with open(dst, 'wb') as f:
for block in entry.get_blocks():
f.write(block)
except Exception as exc:
raise ContainerExtractionError(entry.pathname, exc)
logger.debug(
"Extracted %d entries from %s to %s",
len(self._members),
self.source.path,
tmpdir,
)
def compare_files(file1, file2, source=None, diff_content_only=False):
logger.debug(
"Comparing %s (%s) and %s (%s)",
file1.name,
file1.__class__.__name__,
file2.name,
file2.__class__.__name__,
)
if any_excluded(file1.name, file2.name):
return None
force_details = Config().force_details
with profile('has_same_content_as', file1):
has_same_content = file1.has_same_content_as(file2)
if has_same_content:
if not force_details:
logger.debug(
"has_same_content_as returned True; skipping further comparisons"
)
return None
if diff_content_only:
return None
elif diff_content_only:
assert not has_same_content
return Difference(None, file1.name, file2.name, comment="Files differ")
specialize(file1)
specialize(file2)
if isinstance(file1, MissingFile):
file1.other_file = file2
elif isinstance(file2, MissingFile):
file2.other_file = file1
elif (file1.__class__.__name__ != file2.__class__.__name__) and (
file1.as_container is None or file2.as_container is None):
return file1.compare_bytes(file2, source)
with profile('compare_files (cumulative)', file1):
return file1.compare(file2, source)
