From 4d0224f3e73315ef4db39643e6d4851e4a466658 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Wed, 15 Jan 2020 13:52:42 -0800 Subject: pass through revisit_cdx --- python/sandcrawler/ingest.py | 3 +++ 1 file changed, 3 insertions(+) (limited to 'python/sandcrawler/ingest.py') diff --git a/python/sandcrawler/ingest.py b/python/sandcrawler/ingest.py index e5eb6e8..de5e957 100644 --- a/python/sandcrawler/ingest.py +++ b/python/sandcrawler/ingest.py @@ -239,6 +239,7 @@ class IngestFileWorker(SandcrawlerWorker): if not resource.hit: result['status'] = resource.status return result + file_meta = gen_file_metadata(resource.body) if "html" in file_meta['mimetype'] or "xml" in file_meta['mimetype']: @@ -289,6 +290,8 @@ class IngestFileWorker(SandcrawlerWorker): result['file_meta'] = file_meta result['cdx'] = cdx_to_dict(resource.cdx) + if resource.revisit_cdx: + result['revisit_cdx'] = cdx_to_dict(resource.revisit_cdx) # other failure cases if not resource.body or file_meta['size_bytes'] == 0: -- cgit v1.2.3