def extract_journal_info(obj, eng):
"""Extract journal, volume etc. from any freetext publication info."""
publication_info = obj.data.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)
obj.data["publication_info"] = new_publication_info
def extract_journal_info(obj, eng):
"""Extract the journal information from ``pubinfo_freetext``.
Runs ``extract_journal_reference`` on the ``pubinfo_freetext`` key of each
``publication_info``, if it exists, and uses the extracted information to
populate the other keys.
Args:
obj: a workflow object.
eng: a workflow engine.
Returns:
None
"""
if not obj.data.get('publication_info'):
return
for publication_info in obj.data['publication_info']:
try:
with local_refextract_kbs_path() as kbs_path:
extracted_publication_info = extract_journal_reference(
publication_info['pubinfo_freetext'],
override_kbs_files=kbs_path,
)
if not extracted_publication_info:
continue
if extracted_publication_info.get('title'):
publication_info['journal_title'] = extracted_publication_info[
'title']
if extracted_publication_info.get('volume'):
publication_info[
'journal_volume'] = extracted_publication_info['volume']
if extracted_publication_info.get('page'):
page_start, page_end, artid = split_page_artid(
extracted_publication_info['page'])
if page_start:
publication_info['page_start'] = page_start
if page_end:
publication_info['page_end'] = page_end
if artid:
publication_info['artid'] = artid
if extracted_publication_info.get('year'):
year = maybe_int(extracted_publication_info['year'])
if year:
publication_info['year'] = year
except KeyError:
pass
obj.data['publication_info'] = convert_old_publication_info_to_new(
obj.data['publication_info'])
def extract_journal_info(obj, eng):
"""Extract the journal information from ``pubinfo_freetext``.
Runs ``extract_journal_reference`` on the ``pubinfo_freetext`` key of each
``publication_info``, if it exists, and uses the extracted information to
populate the other keys.
Args:
obj: a workflow object.
eng: a workflow engine.
Returns:
None
"""
if not obj.data.get('publication_info'):
return
for publication_info in obj.data['publication_info']:
try:
with local_refextract_kbs_path() as kbs_path:
extracted_publication_info = extract_journal_reference(
publication_info['pubinfo_freetext'],
override_kbs_files=kbs_path,
)
if not extracted_publication_info:
continue
if extracted_publication_info.get('title'):
publication_info['journal_title'] = extracted_publication_info['title']
if extracted_publication_info.get('volume'):
publication_info['journal_volume'] = extracted_publication_info['volume']
if extracted_publication_info.get('page'):
page_start, page_end, artid = split_page_artid(extracted_publication_info['page'])
if page_start:
publication_info['page_start'] = page_start
if page_end:
publication_info['page_end'] = page_end
if artid:
publication_info['artid'] = artid
if extracted_publication_info.get('year'):
year = maybe_int(extracted_publication_info['year'])
if year:
publication_info['year'] = year
except KeyError:
pass
obj.data['publication_info'] = convert_old_publication_info_to_new(obj.data['publication_info'])
def extract_journal_info(obj, eng):
"""Extract journal, volume etc. from any freetext publication info."""
publication_info = get_value(obj.data, "publication_info")
if not publication_info:
return
new_publication_info = []
for pubnote in publication_info:
if not pubnote:
continue
freetext = pubnote.get("pubinfo_freetext")
if freetext:
if isinstance(freetext, (list, tuple)):
freetext = ". ".join(freetext)
extracted_publication_info = extract_journal_reference(
freetext,
# override_kbs_files={
# 'journals': get_mappings_from_kbname(['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:
year = maybe_int(extracted_publication_info.get('year'))
if year is not None:
pubnote['year'] = year
if "page" in extracted_publication_info:
page_start, page_end, artid = split_page_artid(
extracted_publication_info.get("page"))
if page_start:
pubnote["page_start"] = page_start
if page_end:
pubnote["page_end"] = page_end
if artid:
pubnote["artid"] = artid
if any(value for value in pubnote.values()):
new_publication_info.append(pubnote)
obj.data["publication_info"] = new_publication_info
def extract_journal_info(obj, eng):
"""Extract journal, volume etc. from any freetext publication info."""
publication_info = get_value(obj.data, "publication_info")
if not publication_info:
return
new_publication_info = []
for pubnote in publication_info:
if not pubnote:
continue
freetext = pubnote.get("pubinfo_freetext")
if not freetext:
continue
if isinstance(freetext, (list, tuple)):
freetext = ". ".join(freetext)
extracted_publication_info = extract_journal_reference(
freetext,
# override_kbs_files={
# 'journals': get_mappings_from_kbname(['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:
page_start, page_end, artid = split_page_artid(
extracted_publication_info.get("page"))
if page_start:
pubnote["page_start"] = page_start
if page_end:
pubnote["page_end"] = page_end
if artid:
pubnote["artid"] = artid
new_publication_info.append(pubnote)
obj.data["publication_info"] = new_publication_info
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()
