def extract_references(filepath):
"""Extract references from PDF and return in INSPIRE format."""
references = extract_references_from_file(
filepath,
reference_format="{title},{volume},{page}",
override_kbs_files={'journals': get_mappings_from_kbname(cfg['REFEXTRACT_KB_NAME'])}
)
mapped_references = []
if references.get('references'):
for ref in references.get('references'):
reference = {}
reference["journal_pubnote"] = ref.get('journal_reference')
reference["year"] = ref.get('year')
reference["collaboration"] = ref.get('collaboration')
reference["title"] = ref.get('title')
reference["misc"] = ref.get('misc')
reference["number"] = ref.get('linemarker')
reference["authors"] = ref.get('author')
reference["isbn"] = ref.get('isbn')
reference["doi"] = ref.get('doi')
reference["report_number"] = ref.get('reportnumber')
reference["publisher"] = ref.get('publisher')
reference["recid"] = ref.get('recid')
for key, value in reference.items():
if value and isinstance(value, list):
reference[key] = ",".join(value)
elif not value:
del reference[key]
mapped_references.append(reference)
return mapped_references
def extract_journal_info(obj, eng):
"""Extract journal, volume etc. from any freetext publication info."""
model = eng.workflow_definition.model(obj)
record = get_record_from_model(model)
publication_info = record.get("publication_info")
if not publication_info:
return
new_publication_info = []
for pubnote in publication_info:
freetext = pubnote.get("pubinfo_freetext")
if freetext:
extracted_publication_info = extract_journal_reference(
freetext,
override_kbs_files={'journals': get_mappings_from_kbname(cfg['REFEXTRACT_KB_NAME'])}
)
if extracted_publication_info:
if "volume" in extracted_publication_info:
pubnote["journal_volume"] = extracted_publication_info.get(
"volume"
)
if "title" in extracted_publication_info:
pubnote["journal_title"] = extracted_publication_info.get(
"title"
)
if "year" in extracted_publication_info:
pubnote["year"] = extracted_publication_info.get(
"year"
)
if "page" in extracted_publication_info:
pubnote["page_artid"] = extracted_publication_info.get(
"page"
)
new_publication_info.append(pubnote)
record["publication_info"] = new_publication_info
model.update()
