def parse_references(reference_lines, recid=1, kbs_files=None):
"""Parse a list of references
Given a list of raw reference lines (list of strings),
output the MARC-XML content extracted version
"""
# RefExtract knowledge bases
kbs = get_kbs(custom_kbs_files=kbs_files)
# Identify journal titles, report numbers, URLs, DOIs, and authors...
(processed_references, counts, dummy_bad_titles_count) = \
parse_references_elements(reference_lines, kbs)
# Generate marc xml using the elements list
xml_out = build_xml_references(processed_references)
# Generate the xml string to be outputted
return create_xml_record(counts, recid, xml_out)
def cli_main(options, args):
if options.help or not args:
usage()
return
if options.kb_journals:
kbs_files = {'journals': options.kb_journals}
else:
kbs_files = {}
kb = get_kbs(custom_kbs_files=kbs_files)['journals']
out_records = []
for path in args:
f = open(path)
try:
xml = f.read()
finally:
f.close()
out_records += convert_journals_list(kb, create_records(xml))
write_records(options, out_records)
def cli_main(options, args):
if options.help or not args:
usage()
return
if options.kb_journals:
kbs_files = {'journals': options.kb_journals}
else:
kbs_files = {}
kb = get_kbs(custom_kbs_files=kbs_files)['journals']
out_records = []
for path in args:
f = open(path)
try:
xml = f.read()
finally:
f.close()
out_records += convert_journals_list(kb, create_records(xml))
write_records(options, out_records)
def setUp(self):
from invenio.legacy.refextract.kbs import get_kbs
kb = [("TEST JOURNAL NAME", "Converted")]
kbs_files = {'journals': kb}
self.kb = get_kbs(custom_kbs_files=kbs_files)['journals']
def setUp(self):
from invenio.legacy.refextract.kbs import get_kbs
kb = [("TEST JOURNAL NAME", "Converted")]
kbs_files = {'journals': kb}
self.kb = get_kbs(custom_kbs_files=kbs_files)['journals']