diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2021-10-01 19:11:07 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2021-10-01 19:11:07 -0700 |
| commit | cc5f6f72294eee3f4310dcc38ba5b00d5d98fe69 (patch) | |
| tree | cd6d2c40b1e848f21d15c6714ddbb783c489373c | |
| parent | 1b5ee74818da93fd80201a60a18632ff28692d91 (diff) | |
| download | sandcrawler-cc5f6f72294eee3f4310dcc38ba5b00d5d98fe69.tar.gz sandcrawler-cc5f6f72294eee3f4310dcc38ba5b00d5d98fe69.zip | |
html: fix logging of broken CDX URL
| -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 3e57a04..e81a16d 100644 --- a/python/sandcrawler/html_ingest.py +++ b/python/sandcrawler/html_ingest.py @@ -168,7 +168,7 @@ def fetch_html_resources(resources: List[dict], wayback_client: WaybackClient, w raise NoCaptureError(f"HTML sub-resource not found: {resource['url']}") 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}") + raise WaybackContentError(f"wayback payload sha1hex mismatch: {wayback_resp.cdx.url}") full.append(WebResource( surt=wayback_resp.cdx.surt, timestamp=parse_cdx_datetime(wayback_resp.cdx.datetime), |