def get_journal_id(record, created_by):
journal_abbr = record.get('TA', '')
journal_full_name = record.get('JT', '')
# 1469-221X (Print) 1469-221X (Linking)
# 1573-6881 (Electronic) 0145-479X (Linking)
issn_list = record.get('IS', '').split(') ')
issn_print = ''
issn_electronic = ''
for issn in issn_list:
if "Print" in issn or "Linking" in issn:
issn_print = issn.split(' ')[0]
if "Electronic" in issn:
issn_electronic = issn.split(' ')[0]
if issn_print:
journals = DBSession.query(Journal).filter_by(
issn_print=issn_print).all()
if len(journals) > 0:
return journals[0].journal_id, journals[
0].med_abbr, journal_full_name, issn_print
if journal_abbr == '':
return None, '', '', ''
if journal_abbr:
journals = DBSession.query(Journal).filter_by(
med_abbr=journal_abbr).all()
if len(journals) > 0:
return journals[0].journal_id, journals[
0].med_abbr, journal_full_name, issn_print
source_id = 824 # 'PubMed'
shortened_full_name = (
journal_full_name[:197] +
'...') if len(journal_full_name) > 200 else journal_full_name
format_name = journal_full_name.replace(' ', '_') + journal_abbr.replace(
' ', '_')
j = Journal(issn_print=issn_print,
issn_electronic=issn_electronic,
display_name=shortened_full_name,
format_name=(format_name[:97] +
'...') if len(format_name) > 100 else format_name,
title=shortened_full_name,
med_abbr=journal_abbr,
source_id=source_id,
obj_url='/journal/' + format_name,
created_by=created_by)
DBSession.add(j)
DBSession.flush()
DBSession.refresh(j)
return j.journal_id, j.med_abbr, journal_full_name, issn_print
def get_journal_id(nex_session, record, source_id=None):
journal_abbr = record.get('TA', '')
journal_full_name = record.get('JT', '')
# 1469-221X (Print) 1469-221X (Linking)
# 1573-6881 (Electronic) 0145-479X (Linking)
issn_list = record.get('IS', '').split(') ')
issn_print = ''
issn_electronic = ''
for issn in issn_list:
if "Print" in issn or "Linking" in issn:
issn_print = issn.split(' ')[0]
if "Electronic" in issn:
issn_electronic = issn.split(' ')[0]
if issn_print:
journals = nex_session.query(Journal).filter_by(issn_print=issn_print).all()
if len(journals) > 0:
return journals[0].journal_id, journals[0].med_abbr, journal_full_name, issn_print
if journal_abbr:
journals = nex_session.query(Journal).filter_by(med_abbr=journal_abbr).all()
if len(journals) > 0:
return journals[0].journal_id, journals[0].med_abbr, journal_full_name, issn_print
if source_id is None:
source_id = get_source_id(nex_session, 'PubMed')
format_name = journal_full_name.replace(' ', '_') + journal_abbr.replace(' ', '_')
j = Journal(issn_print = issn_print,
issn_electronic = issn_electronic,
display_name = journal_full_name,
format_name = format_name,
title = journal_full_name,
med_abbr = journal,
source_id = source_id,
obj_url = '/journal/'+format_name,
created_by = CREATED_BY)
nex_session.add(j)
nex_session.flush()
nex_session.refresh(x)
return j.journal_id, j.med_abbr, journal_full_name, issn_print
def get_journal_id(record, created_by):
journal_abbr = record.get('journal_abbrev', '')
journal_full_name = record.get('journal_title', '')
issn_print = record.get('issn_print')
issn_electronic = record.get('issn_electronic')
if issn_print:
journals = DBSession.query(Journal).filter_by(
issn_print=issn_print).all()
if len(journals) > 0:
return journals[0].journal_id, journals[
0].med_abbr, journal_full_name, issn_print
if journal_abbr:
journals = DBSession.query(Journal).filter_by(
med_abbr=journal_abbr).all()
if len(journals) > 0:
return journals[0].journal_id, journals[
0].med_abbr, journal_full_name, issn_print
source_id = 824 # 'PubMed'
# format_name = journal_full_name.replace(' ', '_') + journal_abbr.replace(' ', '_')
journal_full_name = journal_full_name.split(" : ")[0]
format_name = journal_full_name.replace(' ', '_')
j = Journal(issn_print=issn_print,
issn_electronic=issn_electronic,
display_name=journal_full_name,
format_name=format_name,
title=journal_full_name,
med_abbr=journal_abbr,
source_id=source_id,
obj_url='/journal/' + format_name,
created_by=created_by)
DBSession.add(j)
DBSession.flush()
DBSession.refresh(j)
return j.journal_id, j.med_abbr, journal_full_name, issn_print
def get_words(x):
return ' '.join(sample(['Hello', 'World', 'Great', 'Neo4j', 'Project', '.']*5, x))
def get_date():
return date(1990+randint(1, 20), randint(1, 12), randint(1, 28))
print('Creating sample data...')
for article_index in range(randint(8, 12)):
pmid = get_pmid()
print(f'Create relationship with article {article_index+1}:{pmid}...')
journal = Journal(issn=get_words(3)).merge()
article = Paper(
pmid=pmid,
# Some data may be missing
doi=get_pmid() if random() > 0.8 else '',
title=get_words(randint(5, 8)),
date=get_date(),
date_type="emmm"
).merge()
article.publish_on.connect(journal)
for refer_index in range(randint(10, 20)):
reference = Paper(pmid=get_pmid(), date=get_date(), date_type="emmm").merge()
reference.cited_by.connect(article)
for refer_index in range(randint(10, 20)):
reference = Paper(pmid=get_pmid(), date=get_date(), date_type="emmm").merge()