diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2020-06-29 20:35:32 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2020-06-29 20:35:32 -0700 |
| commit | 06adf39b0ab38e08d1977127606b784d1a9aca4f (patch) | |
| tree | a9d3fd4bcdbff89c3849a4cc0a7467d121df293e /fatcat_scholar/schema.py | |
| parent | a0babf64426da4b5c19da224f17c9c8f683ed57d (diff) | |
| download | fatcat-scholar-06adf39b0ab38e08d1977127606b784d1a9aca4f.tar.gz fatcat-scholar-06adf39b0ab38e08d1977127606b784d1a9aca4f.zip | |
include GROBID-extracted abstracts in search documents
Diffstat (limited to 'fatcat_scholar/schema.py')
| -rw-r--r-- | fatcat_scholar/schema.py | 8 |
1 files changed, 8 insertions, 0 deletions
diff --git a/fatcat_scholar/schema.py b/fatcat_scholar/schema.py index cf88011..35cf9a1 100644 --- a/fatcat_scholar/schema.py +++ b/fatcat_scholar/schema.py @@ -255,6 +255,14 @@ def contrib_affiliation(contrib: ReleaseContrib) -> Optional[str]: return None +def es_abstracts_from_grobid(tei_dict: dict) -> List[ScholarAbstract]: + + if tei_dict.get("abstract"): + return [ScholarAbstract(lang_code=tei_dict.get("lang"), body=scrub_text(tei_dict["abstract"]))] + else: + return [] + + def es_abstracts_from_release(release: ReleaseEntity) -> List[ScholarAbstract]: d = dict() |