diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2020-09-14 00:16:22 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2020-09-14 00:16:22 -0700 |
| commit | d52cb39476aad977ffe8b73b16e831f78d3ab8fe (patch) | |
| tree | 36902f6a0a4ff8b9d20ff4342320178d7c192396 /fatcat_scholar/grobid2json.py | |
| parent | 71ce30e7547871cb6fe02fa4237af735bd6b9c49 (diff) | |
| download | fatcat-scholar-d52cb39476aad977ffe8b73b16e831f78d3ab8fe.tar.gz fatcat-scholar-d52cb39476aad977ffe8b73b16e831f78d3ab8fe.zip | |
refs and grobid2json bugfixes from testing
Diffstat (limited to 'fatcat_scholar/grobid2json.py')
| -rwxr-xr-x | fatcat_scholar/grobid2json.py | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/fatcat_scholar/grobid2json.py b/fatcat_scholar/grobid2json.py index dcf9ce8..898275b 100755 --- a/fatcat_scholar/grobid2json.py +++ b/fatcat_scholar/grobid2json.py @@ -123,7 +123,7 @@ def biblio_info(elem: ET.Element) -> Dict[str, Any]: ref["arxiv_id"] = elem.findtext('.//{%s}idno[@type="arXiv"]' % ns) ref["pmcid"] = elem.findtext('.//{%s}idno[@type="PMCID"]' % ns) ref["pmid"] = elem.findtext('.//{%s}idno[@type="PMID"]' % ns) - el = elem.find(".//{%s}biblScope[@page]" % ns) + el = elem.find('.//{%s}biblScope[@unit="page"]' % ns) if el is not None: if el.attrib.get("from") and el.attrib.get("to"): ref["pages"] = "{}-{}".format(el.attrib["from"], el.attrib["to"]) |