def gen_patnums(patents):
for pat in patents:
for doc in pat.findall('document-id'):
kind = get_text(doc, 'kind')
pnum = get_text(doc, 'doc-number')
if not kind.startswith('B'):
continue
yield pnum
def gen_patnums(patents):
for pat in patents:
for doc in pat.findall('document-id'):
kind = get_text(doc, 'kind')
pnum = get_text(doc, 'doc-number')
if not kind.startswith('B'):
continue
yield pnum
def parse_grants_gen2(elem):
pat = copy(default)
# top-level section
bib = elem.find('subdoc-bibliographic-information')
# publication data
pub = bib.find('document-id')
if pub is not None:
pat['pubnum'] = get_text(pub, 'doc-number')
pat['pubdate'] = get_text(pub, 'document-date')
# application data
app = bib.find('domestic-filing-data')
if app is not None:
pat['appnum'] = get_text(app, 'application-number/doc-number')
pat['appdate'] = get_text(app, 'filing-date')
pat['appname'] = get_text(bib, 'assignee/organization-name')
# title
tech = bib.find('technical-information')
pat['title'] = get_text(tech, 'title-of-invention')
# ipc code
ipcsec = tech.find('classification-ipc')
pat['ipcver'] = get_text(ipcsec, 'classification-ipc-edition')
if ipcsec is not None:
ipclist = list(gen2_ipc(ipcsec))
if len(ipclist) > 0:
pat['ipc1'] = ipclist[0]
pat['ipc2'] = ';'.join(ipclist)
# applicant info
address = bib.find('correspondence-address/address')
if address is not None:
pat['city'] = get_text(address, 'city')
pat['state'] = get_text(address, 'state')
pat['country'] = get_text(address, 'country/country-code')
# abstract
abst = elem.find('subdoc-abstract')
if abst is not None:
pat['abstract'] = raw_text(abst, sep=' ')
# roll it in
return store_patent(pat)
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)
# stats
i += 1
p += npat
if i % 100 == 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 parse_grants_gen3(elem):
pat = copy(default)
# top-level section
bib = elem.find('us-bibliographic-data-application')
pubref = bib.find('publication-reference')
appref = bib.find('application-reference')
# published patent
pubinfo = pubref.find('document-id')
pat['pubnum'] = get_text(pubinfo, 'doc-number')
pat['pubdate'] = get_text(pubinfo, 'date')
# filing date
pat['appnum'] = get_text(appref, 'document-id/doc-number')
pat['appdate'] = get_text(appref, 'document-id/date')
pat['appname'] = get_text(bib, 'assignees/assignee/orgname')
# title
pat['title'] = get_text(bib, 'invention-title')
# ipc code
ipcsec = bib.find('classifications-ipcr')
if ipcsec is not None:
ipclist = list(gen3_ipcr(ipcsec))
pat['ipc1'], pat['ipcver'] = ipclist[0]
pat['ipc2'] = ';'.join([i for i, _ in ipclist])
ipcsec = bib.find('classification-ipc')
if ipcsec is not None:
ipclist = list(gen3_ipc(ipcsec))
pat['ipc1'], pat['ipcver'] = ipclist[0]
pat['ipc2'] = ';'.join([i for i, _ in ipclist])
# applicant name and address
address = bib.find('parties/applicants/applicant/addressbook/address')
if address is not None:
pat['city'] = get_text(address, 'city')
pat['state'] = get_text(address, 'state')
pat['country'] = get_text(address, 'country')
# abstract
abspar = elem.find('abstract')
if abspar is not None:
pat['abstract'] = raw_text(abspar, sep=' ')
# roll it in
return store_patent(pat)
def gen3_ipcr(ipcsec):
for ipc in ipcsec.findall('classification-ipcr'):
yield (
'%s%s%s%s%s' % (
get_text(ipc, 'section'),
get_text(ipc, 'class'),
get_text(ipc, 'subclass'),
get_text(ipc, 'main-group'),
get_text(ipc, 'subgroup')
),
get_text(ipc, 'ipc-version-indicator/date')
)
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_patent(elem):
pat = defaultdict(str)
pat['gen'] = 3
# top-level section
bib = elem.find('us-bibliographic-data-grant')
pubref = bib.find('publication-reference')
appref = bib.find('application-reference')
# published patent
pubinfo = pubref.find('document-id')
pat['patnum'] = get_text(pubinfo, 'doc-number')
pat['grantdate'] = get_text(pubinfo, 'date')
# filing date
pat['filedate'] = get_text(appref, 'document-id/date')
# title
pat['title'] = get_text(bib, 'invention-title')
# ipc code
ipclist = []
ipcsec = bib.find('classifications-ipcr')
if ipcsec is not None:
for ipc in ipcsec.findall('classification-ipcr'):
ipclist.append(
('%s%s%s%3s%s' %
(get_text(ipc, 'section'), get_text(ipc, 'class'),
get_text(ipc, 'subclass'), get_text(
ipc, 'main-group'), get_text(ipc, 'subgroup')),
get_text(ipc, 'ipc-version-indicator/date')))
ipcsec = bib.find('classification-ipc')
if ipcsec is not None:
ipcver = get_text(ipcsec, 'edition')
ipc0 = ipcsec.find('main-classification')
for ipc in chain([ipc0], ipcsec.findall('further-classification')):
itxt = ipc.text
itxt = itxt[:4] + itxt[4:7].replace(
'0', ' ') + itxt[7:].replace('/', '')
ipclist.append((itxt, ipcver))
pat['ipclist'] = ipclist
# us class
oclsec = bib.find('classification-national')
if oclsec is not None:
pat['class'] = get_text(oclsec, 'main-classification')
# claims
pat['claims'] = get_text(bib, 'number-of-claims')
# citations
refs = bib.find('references-cited')
prefix = ''
if refs is None:
refs = bib.find('us-references-cited')
prefix = 'us-'
cites = []
if refs is not None:
for cite in refs.findall(prefix + 'citation'):
pcite = cite.find('patcit')
if pcite is not None:
docid = pcite.find('document-id')
pnum = get_text(docid, 'doc-number')
kind = get_text(docid, 'kind')
if kind == 'A' or kind.startswith('B'):
cites.append(pnum)
pat['citlist'] = cites
# applicant name and address
assignee = bib.find('assignees/assignee/addressbook')
if assignee is not None:
pat['owner'] = get_text(assignee, 'orgname').upper()
address = assignee.find('address')
pat['city'] = get_text(address, 'city').upper()
pat['state'] = get_text(address, 'state')
pat['country'] = get_text(address, 'country')
# abstract
abspar = elem.find('abstract')
if abspar is not None:
pat['abstract'] = raw_text(abspar, sep=' ')
# roll it in
return store_patent(pat)
def handle_patent(elem):
pat = defaultdict(str)
pat['gen'] = 2
# top-level section
bib = elem.find('SDOBI')
# published patent
pubref = bib.find('B100')
pat['patnum'] = get_text(pubref, 'B110/DNUM/PDAT')
pat['grantdate'] = get_text(pubref, 'B140/DATE/PDAT')
# filing date
appref = bib.find('B200')
pat['filedate'] = get_text(appref, 'B220/DATE/PDAT')
# ipc code
patref = bib.find('B500')
ipcsec = patref.find('B510')
ipcver = get_text(ipcsec, 'B516/PDAT')
ipclist = []
ipc1 = get_text(ipcsec, 'B511/PDAT')
if ipc1 is not None:
ipclist.append((ipc1, ipcver))
for child in ipcsec.findall('B512'):
ipc = get_text(child, 'PDAT')
ipclist.append((ipc, ipcver))
pat['ipclist'] = ipclist
# us class
pat['class'] = get_text(patref, 'B520/B521/PDAT')
# citations
cites = []
refs = patref.find('B560')
if refs is not None:
for cite in refs.findall('B561'):
pcit = get_text(cite, 'PCIT/DOC/DNUM/PDAT')
cites.append(pcit)
pat['citlist'] = cites
# title
pat['title'] = get_text(patref, 'B540/STEXT/PDAT')
# claims
pat['claims'] = get_text(patref, 'B570/B577/PDAT')
# applicant name and address
ownref = bib.find('B700/B730/B731/PARTY-US')
if ownref is not None:
pat['owner'] = get_text(ownref, 'NAM/ONM/STEXT/PDAT').upper()
address = ownref.find('ADR')
if address is not None:
pat['city'] = get_text(address, 'CITY/PDAT').upper()
pat['state'] = get_text(address, 'STATE/PDAT')
pat['country'] = get_text(address, 'CTRY/PDAT', default='US')
# abstract
abspars = elem.findall('SDOAB/BTEXT/PARA')
if len(abspars) > 0:
pat['abstract'] = '\n'.join([raw_text(e) for e in abspars])
# roll it in
return store_patent(pat)
def handle_patent(elem):
pat = defaultdict(str)
pat['gen'] = 2
# top-level section
bib = elem.find('SDOBI')
# published patent
pubref = bib.find('B100')
pat['patnum'] = get_text(pubref, 'B110/DNUM/PDAT')
pat['grantdate'] = get_text(pubref, 'B140/DATE/PDAT')
# filing date
appref = bib.find('B200')
pat['filedate'] = get_text(appref, 'B220/DATE/PDAT')
# ipc code
patref = bib.find('B500')
ipcsec = patref.find('B510')
ipcver = get_text(ipcsec, 'B516/PDAT')
ipclist = []
ipc1 = get_text(ipcsec, 'B511/PDAT')
if ipc1 is not None:
ipclist.append((ipc1, ipcver))
for child in ipcsec.findall('B512'):
ipc = get_text(child, 'PDAT')
ipclist.append((ipc, ipcver))
pat['ipclist'] = ipclist
# citations
cites = []
refs = patref.find('B560')
if refs is not None:
for cite in refs.findall('B561'):
pcit = get_text(cite, 'PCIT/DOC/DNUM/PDAT')
cites.append(pcit)
pat['citlist'] = cites
# title
pat['title'] = get_text(patref, 'B540/STEXT/PDAT')
# claims
pat['claims'] = get_text(patref, 'B570/B577/PDAT')
# applicant name and address
ownref = bib.find('B700/B730/B731/PARTY-US')
if ownref is not None:
pat['owner'] = get_text(ownref, 'NAM/ONM/STEXT/PDAT').upper()
address = ownref.find('ADR')
if address is not None:
pat['state'] = get_text(address, 'STATE/PDAT')
pat['country'] = get_text(address, 'CTRY/PDAT', default='US')
# abstract
abspars = elem.findall('SDOAB/BTEXT/PARA')
if len(abspars) > 0:
pat['abstract'] = '\n'.join([raw_text(e) for e in abspars])
# roll it in
return store_patent(pat)
def handle_patent(elem):
pat = defaultdict(str)
pat['gen'] = 3
# top-level section
bib = elem.find('us-bibliographic-data-grant')
pubref = bib.find('publication-reference')
appref = bib.find('application-reference')
# published patent
pubinfo = pubref.find('document-id')
pat['patnum'] = get_text(pubinfo, 'doc-number')
pat['grantdate'] = get_text(pubinfo, 'date')
# filing date
pat['filedate'] = get_text(appref, 'document-id/date')
# title
pat['title'] = get_text(bib, 'invention-title')
# ipc code
ipclist = []
ipcsec = bib.find('classifications-ipcr')
if ipcsec is not None:
for ipc in ipcsec.findall('classification-ipcr'):
ipclist.append(('%s%s%s%3s%s' % (get_text(ipc, 'section'),
get_text(ipc, 'class'),
get_text(ipc, 'subclass'),
get_text(ipc, 'main-group'),
get_text(ipc, 'subgroup')),
get_text(ipc, 'ipc-version-indicator/date')))
ipcsec = bib.find('classification-ipc')
if ipcsec is not None:
ipcver = get_text(ipcsec, 'edition')
ipc0 = ipcsec.find('main-classification')
for ipc in chain([ipc0], ipcsec.findall('further-classification')):
itxt = ipc.text
itxt = itxt[:4] + itxt[4:7].replace('0',' ') + itxt[7:].replace('/','')
ipclist.append((itxt, ipcver))
pat['ipclist'] = ipclist
# claims
pat['claims'] = get_text(bib, 'number-of-claims')
# citations
refs = bib.find('references-cited')
prefix = ''
if refs is None:
refs = bib.find('us-references-cited')
prefix = 'us-'
cites = []
if refs is not None:
for cite in refs.findall(prefix + 'citation'):
pcite = cite.find('patcit')
if pcite is not None:
docid = pcite.find('document-id')
pnum = get_text(docid, 'doc-number')
kind = get_text(docid, 'kind')
if kind == 'A' or kind.startswith('B'):
cites.append(pnum)
pat['citlist'] = cites
# applicant name and address
assignee = bib.find('assignees/assignee/addressbook')
if assignee is not None:
pat['owner'] = get_text(assignee, 'orgname').upper()
address = assignee.find('address')
pat['state'] = get_text(address, 'state')
pat['country'] = get_text(address, 'country')
# abstract
abspar = elem.find('abstract')
if abspar is not None:
pat['abstract'] = raw_text(abspar, sep=' ')
# roll it in
return store_patent(pat)
def gen3_ipc(ipcsec):
ipcver = get_text(ipcsec, 'edition')
ipc0 = get_text(ipcsec, 'main-classification')
yield ipc0, ipcver
for ipc in ipcsec.findall('further-classification'):
yield (ipc.text or ''), ipcver
def gen2_ipc(ipcsec):
ipc0 = ipcsec.find('classification-ipc-primary')
if ipc0 is not None:
yield get_text(ipc0, 'ipc')
for ipc in ipcsec.findall('classification-ipc-secondary'):
yield get_text(ipc, 'ipc')