def handle_all():
for (_, pat) in pp.read_events():
if not handle_patent(pat):
return False
clear(pat)
return True
def parse_gen3(fname_in):
global i, o, p
for (event, elem) in iterparse(fname_in,
tag='patent-assignment',
events=['end'],
recover=True):
# top-level section
record = elem.find('assignment-record')
assignor = elem.find('patent-assignors')[0]
assignee = elem.find('patent-assignees')[0]
patents = elem.find('patent-properties')
# conveyance
convey = get_text(record, 'conveyance-text')
# names
assignor_name = get_text(assignor, 'name')
assignee_name = get_text(assignee, 'name')
# dates
exec_sec = assignor.find('execution-date')
recd_sec = record.find('recorded-date')
exec_date = get_text(exec_sec, 'date') if exec_sec is not None else ''
recd_date = get_text(recd_sec, 'date') if recd_sec is not None else ''
# location
assignee_country = get_text(assignee,
'country-name',
default='UNITED STATES')
assignee_state = get_text(assignee, 'state')
# patent info
patnums = list(gen_patnums(patents))
npat = len(patnums)
if npat == 0:
continue
# code names
src_type = org_type(assignor_name)
dst_type = org_type(assignee_name)
ctype = convey_type(convey)
# throw out individuals
if src_type == ORG_INDV or dst_type == ORG_INDV or ctype == CONV_OTHER:
o += 1
continue
# output
for pn in patnums:
chunker.insert(None, pn, exec_date, recd_date, convey,
assignor_name, assignee_name, assignee_state,
assignee_country)
# free memory
clear(elem)
# stats
i += 1
p += npat
# logging
if i % 1000 == 0:
print('%4d: %40.40s -> %30.30s (%20.20s, %20.20s)' %
(npat, assignor_name, assignee_name, assignee_state,
assignee_country))
# break
if args.limit and i >= args.limit:
return False
return True
def handle_all():
for (_, pat) in pp.read_events():
if not handle_patent(pat):
return False
clear(pat)
return True