From 6a2630f236115b8c15b895f1506d02f1f110aa28 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Mon, 25 Jul 2022 13:39:15 -0700 Subject: CDX: skip sha-256 digests --- python/sandcrawler/ia.py | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) (limited to 'python') diff --git a/python/sandcrawler/ia.py b/python/sandcrawler/ia.py index 65f30e7..8c8e59e 100644 --- a/python/sandcrawler/ia.py +++ b/python/sandcrawler/ia.py @@ -211,10 +211,14 @@ class CdxApiClient: else: status_code = int(raw[4]) - # CDX rows with no WARC records? + # remove CDX rows with no WARC records (?) if raw[8] == "-" or raw[9] == "-" or raw[10] == "-": continue + # remove CDX rows with SHA256 (not SHA1) digests + if raw[5].startswith("sha-256"): + continue + row = CdxRow( surt=raw[0], datetime=raw[1], -- cgit v1.2.3