diff options
| author | Bryan Newbold <bnewbold@robocracy.org> | 2018-09-14 19:00:09 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@robocracy.org> | 2018-09-14 19:00:09 -0700 |
| commit | a11ee8d8c0fd1f924d7ba7feec58cd3fcf28752b (patch) | |
| tree | 3bae0ed3ffad2d41418a328416b6021f89acf55d /python | |
| parent | 2076c3cebce3384811d147a7546665d3172f4562 (diff) | |
| download | fatcat-a11ee8d8c0fd1f924d7ba7feec58cd3fcf28752b.tar.gz fatcat-a11ee8d8c0fd1f924d7ba7feec58cd3fcf28752b.zip | |
fix matched_importer
Diffstat (limited to 'python')
| -rw-r--r-- | python/fatcat/matched_importer.py | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/python/fatcat/matched_importer.py b/python/fatcat/matched_importer.py index 44735d52..7f55369b 100644 --- a/python/fatcat/matched_importer.py +++ b/python/fatcat/matched_importer.py @@ -42,10 +42,10 @@ class FatcatMatchedImporter(FatcatImporter): rel = self.default_link_rel # TODO: this is where we could map specific domains to rel types, # and also filter out bad domains, invalid URLs, etc - if "//archive.org/" in url or "//arxiv.org/" in url: + if "//archive.org/" in raw or "//arxiv.org/" in raw: # TODO: special-case the arxiv.org bulk mirror? rel = "repository" - elif "//web.archive.org/" in url or "//archive.is/" in url: + elif "//web.archive.org/" in raw or "//archive.is/" in raw: rel = "webarchive" return fatcat_client.FileEntityUrls(url=raw, rel=rel) |