summaryrefslogtreecommitdiffstats
path: root/python/fatcat_tools
diff options
context:
space:
mode:
authorBryan Newbold <bnewbold@robocracy.org>2019-01-29 16:58:04 -0800
committerBryan Newbold <bnewbold@robocracy.org>2019-01-29 16:58:06 -0800
commit1305a503b116e8f8081cbc48dc14c92b4b64b17d (patch)
tree0f69fba9f48470037b221174c42085b2e1c89272 /python/fatcat_tools
parent26d30331174087a57591c9860c1a84d7010b3f20 (diff)
downloadfatcat-1305a503b116e8f8081cbc48dc14c92b4b64b17d.tar.gz
fatcat-1305a503b116e8f8081cbc48dc14c92b4b64b17d.zip
crossref import tweaks/fixes
- refs: article-title not title; save unstructured; authors not author - save 'language' field (already an ISO code)
Diffstat (limited to 'python/fatcat_tools')
-rw-r--r--python/fatcat_tools/importers/crossref.py16
1 files changed, 9 insertions, 7 deletions
diff --git a/python/fatcat_tools/importers/crossref.py b/python/fatcat_tools/importers/crossref.py
index c4e55962..f9a4fe46 100644
--- a/python/fatcat_tools/importers/crossref.py
+++ b/python/fatcat_tools/importers/crossref.py
@@ -239,11 +239,13 @@ class CrossrefImporter(EntityImporter):
ref_extra['journal-title'] = rm['journal-title']
if rm.get('DOI'):
ref_extra['doi'] = rm.get('DOI').lower()
- # TODO: what fields here? CSL citation stuff
- for k in ('author', 'editor', 'edition', 'authority', 'version',
- 'genre', 'url', 'event', 'issue', 'volume', 'date',
- 'accessed_date', 'issued', 'page', 'medium',
- 'collection_title', 'chapter_number'):
+ author = clean(rm.get('author'))
+ if author:
+ ref_extra['authors'] = [author]
+ for k in ('editor', 'edition', 'authority', 'version', 'genre',
+ 'url', 'event', 'issue', 'volume', 'date', 'accessed_date',
+ 'issued', 'page', 'medium', 'collection_title', 'chapter_number',
+ 'unstructured', 'series-title', 'volume-title'):
if clean(rm.get(k)):
ref_extra[k] = clean(rm[k])
if not ref_extra:
@@ -255,7 +257,7 @@ class CrossrefImporter(EntityImporter):
key=key,
year=year,
container_name=clean(container_name),
- title=clean(rm.get('title')),
+ title=clean(rm.get('article-title')),
locator=clean(rm.get('first-page')),
# TODO: just dump JSON somewhere here?
extra=ref_extra))
@@ -378,7 +380,7 @@ class CrossrefImporter(EntityImporter):
volume=clean(obj.get('volume')),
issue=clean(obj.get('issue')),
pages=clean(obj.get('page')),
- language=None, # crossref doesn't supply language info
+ language=clean(obj.get('language')),
license_slug=license_slug,
extra=extra,
abstracts=abstracts,