diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2020-09-04 18:16:36 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2020-09-04 18:16:36 -0700 |
| commit | 6955d603734440835e28f7ab218b24e569523780 (patch) | |
| tree | a00d3bc685bcfd3036532e88a8a3b1625703466e | |
| parent | 3bf706f07d68aaacebcbf08d21c7b4c4f91856ea (diff) | |
| download | fatcat-scholar-6955d603734440835e28f7ab218b24e569523780.tar.gz fatcat-scholar-6955d603734440835e28f7ab218b24e569523780.zip | |
fixes to refs transform (for non-str author fields)
| -rw-r--r-- | fatcat_scholar/transform.py | 8 |
1 files changed, 6 insertions, 2 deletions
diff --git a/fatcat_scholar/transform.py b/fatcat_scholar/transform.py index 158ed6f..a21abf9 100644 --- a/fatcat_scholar/transform.py +++ b/fatcat_scholar/transform.py @@ -454,12 +454,14 @@ def refs_from_grobid(release: ReleaseEntity, tei_dict: dict) -> Sequence[RefStru ref_year: Optional[int] = None if ref_date and len(ref_date) > 4 and ref_date[:4].isdigit(): ref_year = int(ref_date[:4]) + authors = ref.get("authors") or [] + authors = [a for a in authors if type(a) == str] output.append( RefStructured( biblio=RefBiblio( title=ref.get("title"), # subtitle - contrib_raw_names=ref.get("authors") or [], + contrib_raw_names=authors, year=ref_year, container_name=ref.get("journal"), volume=ref.get("volume"), @@ -495,12 +497,14 @@ def refs_from_release_refs(release: ReleaseEntity) -> Sequence[RefStructured]: elif release.extra and release.extra.get("datacite"): ref_source = "datacite" extra = ref.extra or dict() + authors = extra.get("authors") or [] + authors = [a for a in authors if type(a) == str] output.append( RefStructured( biblio=RefBiblio( title=ref.title, subtitle=extra.get("subtitle"), - contrib_raw_names=extra.get("authors") or [], + contrib_raw_names=authors, year=ref.year, container_name=ref.container_name, volume=extra.get("volume"), |