From 214c35db713ee5f34e9c6e8aa24df6456addcece Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Mon, 2 Mar 2020 15:52:47 -0800 Subject: remove protocols.io octet-stream hack --- python/sandcrawler/ingest.py | 8 ++------ 1 file changed, 2 insertions(+), 6 deletions(-) (limited to 'python') 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) -- cgit v1.2.3