diff options
Diffstat (limited to 'python/sandcrawler/html_ingest.py')
-rw-r--r-- | python/sandcrawler/html_ingest.py | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/python/sandcrawler/html_ingest.py b/python/sandcrawler/html_ingest.py index 6b9497b..03ec6f4 100644 --- a/python/sandcrawler/html_ingest.py +++ b/python/sandcrawler/html_ingest.py @@ -156,7 +156,7 @@ def fetch_html_resources(resources: List[dict], wayback_client: WaybackClient, w wayback_resp = wayback_client.lookup_resource(resource['url'], closest=closest) if not wayback_resp or wayback_resp.status != 'success': raise NoCaptureError(f"HTML sub-resource not found: {resource['url']}") - file_meta = gen_file_metadata(wayback_resp.body) + file_meta = gen_file_metadata(wayback_resp.body, allow_empty=True) if file_meta['sha1hex'] != wayback_resp.cdx.sha1hex: raise WaybackContentError("wayback payload sha1hex mismatch: {wayback_resp.cdx.url}") full.append(WebResource( |