diff options
| -rw-r--r-- | python/sandcrawler/ingest.py | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/python/sandcrawler/ingest.py b/python/sandcrawler/ingest.py index 6ec54f6..0d4e7c6 100644 --- a/python/sandcrawler/ingest.py +++ b/python/sandcrawler/ingest.py @@ -290,7 +290,7 @@ class IngestFileWorker(SandcrawlerWorker): return result file_meta = gen_file_metadata(resource.body) - if "html" in file_meta['mimetype'] or "xml" in file_meta['mimetype']: + if "html" in file_meta['mimetype'] or "xhtml" in file_meta['mimetype'] or "application/xml" in file_meta['mimetype']: # Got landing page or similar. Some XHTML detected as "application/xml" if resource.terminal_dt: result['terminal'] = { |