checksum: Checksum each image only once

There is no point in reading the same image multiple times. This happens
for at least source ISOs.

Signed-off-by: Lubomír Sedlář <lsedlar@redhat.com>
This commit is contained in:
Lubomír Sedlář 2017-07-13 14:10:02 +02:00
parent 81cb0952ca
commit 085a8ef7c7

View File

@ -73,7 +73,8 @@ class ImageChecksumPhase(PhaseBase):
make_checksums(topdir, self.compose.im, self.checksums, self.one_file, self._get_base_filename) make_checksums(topdir, self.compose.im, self.checksums, self.one_file, self._get_base_filename)
def _compute_checksums(results, variant, arch, path, images, checksum_types, base_checksum_name, one_file): def _compute_checksums(results, cache, variant, arch, path, images,
checksum_types, base_checksum_name, one_file):
for image in images: for image in images:
filename = os.path.basename(image.path) filename = os.path.basename(image.path)
full_path = os.path.join(path, filename) full_path = os.path.join(path, filename)
@ -82,7 +83,12 @@ def _compute_checksums(results, variant, arch, path, images, checksum_types, bas
filesize = image.size or get_file_size(full_path) filesize = image.size or get_file_size(full_path)
digests = shortcuts.compute_file_checksums(full_path, checksum_types) if full_path not in cache:
# Source ISO is listed under each binary architecture. There's no
# point in checksumming it twice, so we can just remember the
# digest from first run..
cache[full_path] = shortcuts.compute_file_checksums(full_path, checksum_types)
digests = cache[full_path]
for checksum, digest in digests.iteritems(): for checksum, digest in digests.iteritems():
# Update metadata with the checksum # Update metadata with the checksum
image.add_checksum(None, checksum, digest) image.add_checksum(None, checksum, digest)
@ -101,9 +107,10 @@ def _compute_checksums(results, variant, arch, path, images, checksum_types, bas
def make_checksums(topdir, im, checksum_types, one_file, base_checksum_name_gen): def make_checksums(topdir, im, checksum_types, one_file, base_checksum_name_gen):
results = defaultdict(set) results = defaultdict(set)
cache = {}
for (variant, arch, path), images in get_images(topdir, im).iteritems(): for (variant, arch, path), images in get_images(topdir, im).iteritems():
base_checksum_name = base_checksum_name_gen(variant, arch) base_checksum_name = base_checksum_name_gen(variant, arch)
_compute_checksums(results, variant, arch, path, images, _compute_checksums(results, cache, variant, arch, path, images,
checksum_types, base_checksum_name, one_file) checksum_types, base_checksum_name, one_file)
for file in results: for file in results: