diff --git a/bin/swh-loader-tar-producer b/bin/swh-loader-tar-producer index 2ca6d82..c309a7f 100755 --- a/bin/swh-loader-tar-producer +++ b/bin/swh-loader-tar-producer @@ -1,221 +1,221 @@ #!/usr/bin/env python3 # Copyright (C) 2015 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import argparse import itertools import random import sys from swh.core import config from swh.loader.tar import build, file def compute_message_from(app, conf, root_dir, tarpath, filename, dry_run=False): """Compute and post the message to worker for the archive tarpath. Args: app: instance of the celery app conf: dictionary holding static metadata root_dir: root directory tarball: the archive's representation dry_run: will compute but not send messages Returns: None Raises: ValueError when release number computation error arise. """ origin = build.compute_origin(conf['url_scheme'], conf['type'], root_dir, tarpath) revision = build.compute_revision() occurrences = [build.occurrence_with_mtime(GNU_AUTHORITY, tarpath), build.occurrence_with_ctime(SWH_AUTHORITY, tarpath)] release = build.compute_release(filename, tarpath) if not dry_run: app.tasks['swh.loader.tar.tasks.LoadTarRepository'].delay(tarpath, origin, revision, release, occurrences) def grouper(iterable, n, fillvalue=None): """Collect data into fixed-length chunks or blocks. Args: iterable: an iterable n: size of block fillvalue: value to use for the last block Returns: fixed-length chunks of blocks as iterables """ args = [iter(iterable)] * n return itertools.zip_longest(*args, fillvalue=fillvalue) def random_blocks(iterable, block=100, fillvalue=None): """Given an iterable: - slice the iterable in data set of block-sized elements - randomized the data set - yield each element Args: iterable: iterable of data block: number of elements per block fillvalue: a fillvalue for the last block if not enough values in last block Returns: An iterable of element. """ for iterable in grouper(iterable, block, fillvalue=fillvalue): l = list(iterable) random.shuffle(l) for e in l: yield e def produce_archive_messages_from(app, conf, path, mirror_file=None, dry_run=False): """From path, produce archive tarball messages to celery. Will print error message when some computation arise on archive and continue. Args: app: instance of the celery app conf: dictionary holding static metadata path: top directory to list archives from. mirror_file: a filtering file of tarballs to load dry_run: will compute but not send messages Returns: None Raises: None """ LIMIT = conf['limit'] BLOCK = int(conf['block_messages']) count = 0 path_source_tarballs = mirror_file if mirror_file else path - archives = file.list_archives_from(path_source_tarballs) + archives = file.archives_from(path_source_tarballs) random_archives = random_blocks(archives, BLOCK, fillvalue=(None, None)) if LIMIT: random_archives = itertools.islice(random_archives, LIMIT) for tarpath, fname in [(t, f) for t, f in random_archives if t and f]: count += 1 try: compute_message_from(app, conf, path, tarpath, fname, dry_run) except ValueError: print('Problem with the following archive: %s' % tarpath) return count def load_config(conf_file): """Load the configuration from file. Args: conf_file: path to a configuration file with the following content: [main] # mirror's root directory holding tarballs to load into swh mirror_root_directory = /home/storage/space/mirrors/gnu.org/gnu/ # origin setup's possible scheme url url_scheme = rsync://ftp.gnu.org/gnu/ # origin type used for those tarballs type = ftp # For tryouts purposes (no limit if not specified) limit = 1 Returns: dictionary of data present in the configuration file. """ conf = config.read(conf_file, default_conf={'limit': ('int', None)}) url_scheme = conf['url_scheme'] mirror_dir = conf['mirror_root_directory'] # remove trailing / in configuration (to ease ulterior computation) if url_scheme[-1] == '/': conf.update({ 'url_scheme': url_scheme[0:-1] }) if mirror_dir[-1] == '/': conf.update({ 'mirror_root_directory': mirror_dir[0:-1] }) return conf def parse_args(): """Parse the configuration from the cli. """ cli = argparse.ArgumentParser( description='Tarball producer of local fs tarballs.') cli.add_argument('--dry-run', '-n', action='store_true', help='Dry run (print repo only)') cli.add_argument('--config', '-c', help='configuration file path') args = cli.parse_args() return args if __name__ == '__main__': args = parse_args() config_file = args.config if not config_file: print('Missing configuration file option.') sys.exit(1) # instantiate celery app with its configuration from swh.core.worker import app from swh.loader.tar import tasks # noqa conf = load_config(config_file) # state... SWH_AUTHORITY = conf['swh_authority'] GNU_AUTHORITY = conf['gnu_authority'] nb_tarballs = produce_archive_messages_from( app, conf, conf['mirror_root_directory'], conf.get('mirror_subset_archives'), args.dry_run) print('%s tarball(s) sent to worker.' % nb_tarballs) diff --git a/bin/swh-ls-tarball-size b/bin/swh-ls-tarball-size index df56eab..1eb8a77 100755 --- a/bin/swh-ls-tarball-size +++ b/bin/swh-ls-tarball-size @@ -1,32 +1,32 @@ #!/usr/bin/env python3 # Copyright (C) 2015 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import argparse import os from swh.loader.tar import file def parse_args(): """Parse the configuration from the cli. """ cli = argparse.ArgumentParser( description='Tarball listing tarballs size.') cli.add_argument('--mirror-root-dir', '-m', help='path to the root dir.') args = cli.parse_args() return args if __name__ == '__main__': args = parse_args() root_dir = args.mirror_root_dir - for tarpath, _ in file.list_archives_from(root_dir): + for tarpath, _ in file.archives_from(root_dir): print('%s %s' % (tarpath, os.path.getsize(tarpath))) diff --git a/swh/loader/tar/file.py b/swh/loader/tar/file.py index 24ac56c..332a2b2 100644 --- a/swh/loader/tar/file.py +++ b/swh/loader/tar/file.py @@ -1,67 +1,67 @@ # Copyright (C) 2015 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import os from swh.loader.tar import tarball -def list_archives_from_dir(path): +def archives_from_dir(path): """Given a path to a directory, walk such directory and yield tuple of tarpath, fname. Args: path: top level directory Returns: Generator of tuple tarpath, filename with tarpath a tarball. """ for dirpath, dirnames, filenames in os.walk(path): for fname in filenames: tarpath = os.path.join(dirpath, fname) if not os.path.exists(tarpath): continue if tarball.is_tarball(tarpath): yield tarpath, fname -def list_archives_from_file(mirror_file): +def archives_from_file(mirror_file): """Given a path to a file containing one tarball per line, yield a tuple of tarpath, fname. Args: mirror_file: path to the file containing list of tarpath. Returns: Generator of tuple tarpath, filename with tarpath a tarball. """ with open(mirror_file, 'r') as f: for tarpath in f.readlines(): tarpath = tarpath.strip() if not os.path.exists(tarpath): print('WARN: %s does not exist. Skipped.' % tarpath) continue if tarball.is_tarball(tarpath): yield tarpath, os.path.basename(tarpath) -def list_archives_from(path): +def archives_from(path): """From path, list tuple of tarpath, fname. Args: path: top directory to list archives from or custom file format. """ if os.path.isfile(path): - yield from list_archives_from_file(path) + yield from archives_from_file(path) elif os.path.isdir(path): - yield from list_archives_from_dir(path) + yield from archives_from_dir(path) else: raise ValueError( 'Input incorrect, %s must be a file or a directory.' % path)