From b1b34d44ce1a416ee70be665b71b99ba9f98d9a3 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Fri, 6 Nov 2020 19:16:31 -0800 Subject: ingest tool: support for setting ingest type --- python/fatcat_ingest.py | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'python/fatcat_ingest.py') diff --git a/python/fatcat_ingest.py b/python/fatcat_ingest.py index 68676ad2..b9d71a7c 100755 --- a/python/fatcat_ingest.py +++ b/python/fatcat_ingest.py @@ -87,6 +87,7 @@ def _run_search_dump(args, search): ingest_request = release_ingest_request( release, ingest_request_source="fatcat-ingest", + ingest_type=args.ingest_type, ) if not ingest_request: continue @@ -214,6 +215,9 @@ def main(): parser.add_argument('--force-recrawl', action='store_true', help="Tell ingest worker to skip GWB history lookup and do SPNv2 crawl") + parser.add_argument('--ingest-type', + default="pdf", + help="What medium to ingest (pdf, xml, html)") subparsers = parser.add_subparsers() sub_container = subparsers.add_parser('container', -- cgit v1.2.3