diff options
author | Bryan Newbold <bnewbold@archive.org> | 2021-10-06 15:13:46 -0700 |
---|---|---|
committer | Bryan Newbold <bnewbold@archive.org> | 2021-10-15 18:15:29 -0700 |
commit | 07e8a199766be77f4e89561d03e9b4e995ab7396 (patch) | |
tree | f4882b3fd32e0ed46e2359900a01e1413287d53e /python/sandcrawler/ingest_fileset.py | |
parent | 206969ccebb5007b6c687edd6e09b5c4910e0152 (diff) | |
download | sandcrawler-07e8a199766be77f4e89561d03e9b4e995ab7396.tar.gz sandcrawler-07e8a199766be77f4e89561d03e9b4e995ab7396.zip |
fileset ingest progress for dataverse
Diffstat (limited to 'python/sandcrawler/ingest_fileset.py')
-rw-r--r-- | python/sandcrawler/ingest_fileset.py | 14 |
1 files changed, 8 insertions, 6 deletions
diff --git a/python/sandcrawler/ingest_fileset.py b/python/sandcrawler/ingest_fileset.py index 50e3c05..3b55793 100644 --- a/python/sandcrawler/ingest_fileset.py +++ b/python/sandcrawler/ingest_fileset.py @@ -262,19 +262,21 @@ class IngestFilesetWorker(IngestFileWorker): if resource: terminal_url = resource.terminal_url dataset_meta = platform_helper.process_request(request, terminal_url, html_biblio) - print(dataset_meta, file=sys.stderr) + #print(dataset_meta, file=sys.stderr) platform = dataset_meta.platform_name result['platform'] = dataset_meta.platform_name result['platform_id'] = dataset_meta.platform_id result['item_name'] = dataset_meta.archiveorg_item_name - if not dataset_meta.manifest: + result['item_meta'] = dataset_meta.archiveorg_item_meta + + if dataset_meta.manifest: + result['manifest'] = [m.dict() for m in dataset_meta.manifest] + result['file_count'] = len(dataset_meta.manifest) + result['total_size'] = sum([m.size for m in dataset_meta.manifest if m.size]) + else: result['status'] = 'no-manifest' return result - result['manifest'] = dataset_meta.manifest or None - result['file_count'] = len(dataset_meta.manifest) or None - result['total_size'] = sum([m.size for m in dataset_meta.manifest if m.size]) or None - ingest_strategy = platform_helper.chose_strategy(dataset_meta) result['ingest_strategy'] = ingest_strategy |