diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2021-12-06 16:25:21 -0800 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2021-12-06 16:50:00 -0800 |
| commit | f60d37fb2f9079e6707f9a253983b6ea07964e18 (patch) | |
| tree | e36db11b2123a7a50d48dbf0f8449bcef9e13fcc /fatcat_scholar/work_pipeline.py | |
| parent | c1af67fc72c671c4dc40536960ab47e78195c881 (diff) | |
| download | fatcat-scholar-f60d37fb2f9079e6707f9a253983b6ea07964e18.tar.gz fatcat-scholar-f60d37fb2f9079e6707f9a253983b6ea07964e18.zip | |
fetch GROBID-parsed refs along with crossref metadata
Diffstat (limited to 'fatcat_scholar/work_pipeline.py')
| -rw-r--r-- | fatcat_scholar/work_pipeline.py | 3 |
1 files changed, 2 insertions, 1 deletions
diff --git a/fatcat_scholar/work_pipeline.py b/fatcat_scholar/work_pipeline.py index 9dce006..fa1a7bc 100644 --- a/fatcat_scholar/work_pipeline.py +++ b/fatcat_scholar/work_pipeline.py @@ -276,13 +276,14 @@ class WorkPipeline: # if this is definitely a Datacite or JALC DOI, can skip the Crossref cache lookup return None doi = re.ext_ids.doi.lower() - crossref_meta = self.sandcrawler_db_client.get_crossref(doi) + crossref_meta = self.sandcrawler_db_client.get_crossref_with_refs(doi) if not crossref_meta or not crossref_meta.get("record"): return None return dict( release_ident=re.ident, doi=doi, record=crossref_meta["record"], + grobid_refs=crossref_meta["refs_json"] or [], ) def lookup_sim(self, release: ReleaseEntity) -> Optional[SimIssueRow]: |