aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
-rw-r--r--python/sandcrawler/ingest.py8
1 files changed, 2 insertions, 6 deletions
diff --git a/python/sandcrawler/ingest.py b/python/sandcrawler/ingest.py
index 0d4e7c6..8e0efeb 100644
--- a/python/sandcrawler/ingest.py
+++ b/python/sandcrawler/ingest.py
@@ -348,12 +348,8 @@ class IngestFileWorker(SandcrawlerWorker):
return result
if not (resource.hit and file_meta['mimetype'] == "application/pdf"):
- # protocols.io PDFs are "application/octet-stream"
- if (file_meta['mimetype'] == "application/octet-stream" and "://protocols.io/" in resource.terminal_url):
- pass
- else:
- result['status'] = "wrong-mimetype" # formerly: "other-mimetype"
- return result
+ result['status'] = "wrong-mimetype" # formerly: "other-mimetype"
+ return result
info = self.process_hit(resource, file_meta)
result.update(info)