def normalize_patent_au(patent):
"""
Normalizes "Australian" format, e.g. AU2003212220A1, AU200042655B2, AU00784257B2
Patent Application Number:
old: 4+5 digits (Patadmin, before 5 July 2002)
new: 4+6 digits (PAMS, after 5 July 2002)
http://apa.hpa.com.au:8080/ipapa/intro
http://pericles.ipaustralia.gov.au/aub/aub_pages_1.intro
Patent Number:
6 digits
http://pericles.ipaustralia.gov.au/aub/aub_pages_1.intro
"""
assert patent['country'] == 'AU'
patched = copy(patent)
length = len(patent['number'])
# convert from 4+5 to 4+6 (old to new format)
if length == 9:
fullyear = patched['number'][0:4]
seqnumber = patched['number'][4:]
# pad sequential number to 6 characters
patched['number'] = fullyear + pad_left(seqnumber, '0', 6)
else:
patched['number'] = trim_leading_zeros(patched['number'])
if len(patched['number']) < 6:
patched['number'] = pad_left(patched['number'], '0', 6)
return patched
def normalize_patent_wo(patent):
"""
Normalizes to "WIPO Publication Number" format, e.g. WO2005092324
see "Pub. No.":
http://www.wipo.int/pctdb/en/wo.jsp?IA=WO/2005/092324
http://www.wipo.int/pctdb/en/wo.jsp?IA=WO0067640
"""
assert patent['country'] == 'WO'
patched = copy(patent)
# filter: leave special documents untouched (with alphanumeric prefix)
pattern = '^\D+'
r = re.compile(pattern)
if r.match(patched['number']):
return patched
length = len(patent['number'])
# convert from 2+5 or 2+6 to 4+6
if length == 7 or length == 8:
year = patched['number'][0:2]
seqnumber = patched['number'][2:]
# assume for century: 78-99 => 19, otherwise => 20
# build fullyear from (2-digit) year
fullyear = fullyear_from_year(year)
"""
# try different decoding: 1 zero + 2 year + 5 seqnumber
# (wrong format due to "pad everything to 8 characters" logic of Bestellsystem)
# so strip off first leading zero before decoding again
# TODO: what about WO09802618A2, WO00202618A2, WO00402618A2, WO09201000A1, WO09901000A3, WO00101000A1?
if length == 8:
# 1. numbers like WO00101000A1 are ambiguous, could be WO2000101000A1 or WO2001001000A1
ambiguous_2000_2003 = ( 2000 <= int(fullyear) and int(fullyear) <= 2003 and patched['number'].startswith('00') )
# 2. same with 8 digit numbers >= 2004, starting with "WO004..."
# hint: WO00402618A2 can not be WO2000402618A2 (due to format 2+6 and release date), so must be WO2004002618A2
ambiguous_2004_bis = ( int(fullyear) >= 2004 )
if ambiguous_2000_2003: # or ambiguous_2004_bis:
patched['number'] = patched['number'][1:]
year = patched['number'][0:2]
seqnumber = patched['number'][2:]
fullyear = fullyear_from_year(year)
"""
#if length == 8 and patched['number'].startswith('0') and int(fullyear) < 2003:
# return
# pad sequential number to 6 characters
patched['number'] = fullyear + pad_left(seqnumber, '0', 6)
# convert from 4+5 to 4+6 (wrong format)
elif length == 9:
fullyear = patched['number'][0:4]
seqnumber = patched['number'][4:]
# pad sequential number to 6 characters
patched['number'] = fullyear + pad_left(seqnumber, '0', 6)
patched['number'] = trim_leading_zeros(patched['number'])
return patched
def normalize_patent_us(patent, provider=None):
# USPTO number formats
# PATFT - Issued Patents:
# http://patft.uspto.gov/netahtml/PTO/srchnum.htm
#
# Utility -- 5,146,634 6923014 0000001
# Design -- D339,456 D321987 D000152
# Plant -- PP08,901 PP07514 PP00003
# Reissue -- RE35,312 RE12345 RE00007
# Defensive Publication -- T109,201 T855019 T100001
# Statutory Invention Registration -- H001,523 H001234 H000001
# Re-examination -- RX12
# Additional Improvement -- AI00,002 AI000318 AI00007
subtype_prefixes = ['D', 'PP', 'RD', 'RE', 'T', 'H', 'AI']
# AppFT - Patent Applications
# http://appft.uspto.gov/netahtml/PTO/srchnum.html
#
# Utility: 20010000044
assert patent['country'] == 'US'
patched = copy(patent)
length = len(patched['number'])
if provider == 'ops' or provider == 'espacenet':
# OPS accepts US patent application publication numbers in 4+6=10 format
# Examples: US2015322651A1, US2017250417A1, US2017285092A1
# 2017-10-25
# DEPATISnet started delivering application publication numbers in 5+7 format
# with a leading zero after the country, e.g. US000006166174A, US020170285092A1
# around October 2017. Account for that.
if length == 12:
patched['number'] = patched['number'].lstrip('0')
length = len(patched['number'])
# US application publication numbers: Convert from 4+5=9 to 4+6=10
if length == 9:
padding = '0' * (10 - length)
patched['number'] = patched['number'][0:4] + padding + patched[
'number'][4:]
# US application publication numbers: Convert from 4+7=11 to 4+6=10
# 2015-12-20: Normalize responses from SIP like "US20150322651A1" to "US2015322651A1"
elif length == 11:
if patched['number'][4] == '0':
patched[
'number'] = patched['number'][0:4] + patched['number'][5:]
# US patents: Handle document numbers with character prefixes
# Trim leading zeros for OPS
elif 'number-type' in patched and 'number-real' in patched:
subtype = patched['number-type']
seqnumber = patched['number-real']
if subtype in subtype_prefixes:
patched['number'] = subtype + trim_leading_zeros(seqnumber)
# US patents: Strip leading zeros
else:
patched['number'] = patched['number'].lstrip('0')
else:
# US patents: Handle document numbers with character prefixes
# Pad patent number with zeros to get total length of 7 characters
if 'number-type' in patched and 'number-real' in patched:
subtype = patched['number-type']
seqnumber = patched['number-real']
if subtype in subtype_prefixes:
patched['number'] = subtype + seqnumber.zfill(7)
# Convert from 4+5=9 or 4+6=10 to 4+7=11
# US20170000054A1
elif length == 9 or length == 10:
padding = '0' * (11 - length)
patched['number'] = patched['number'][0:4] + padding + patched[
'number'][4:]
# 2018-04-23: Espacenet changed behavior, handle edge case for
# USD813591S to yield https://worldwide.espacenet.com/publicationDetails/claims?CC=US&NR=D813591S&KC=S
if provider == 'espacenet':
if 'number-type' in patched:
if patched['number-type'] == 'D' and patched['kind'] == 'S':
patched['number'] += patched['kind']
return patched
def patch_patent(patent, provider=None):
if not patent:
return
number_length = len(patent['number'])
patched = copy(patent)
#print 'patched:', patched
# strip leading zeros of *publication* to 6 digits, if seqnumber is longer than 6 digits
# examples: publication: AT401234; application: AT 967/1994 => AT96794
if patched['country'] == 'AT':
"""
if len(patched['number']) > 6 and not '/' in patched['number']:
patched['number'] = trim_leading_zeros(patched['number'])
patched['number'] = pad_left(patched['number'], '0', 6)
"""
patched['number'] = trim_leading_zeros(patched['number'])
# pad to 6 characters with leading zeros
elif patched['country'] == 'AR':
patched['number'] = patched['number'].lstrip('0').rjust(6, '0')
elif patched['country'] == 'AU':
patched = normalize_patent_au(patched)
elif patched['country'] == 'BR':
patched['number'] = patched['number'].lstrip('0')
# strip leading zeros with exception of kindcode == T1, then pad to 7 digits like EP
# "Veröffentlichung der europäischen Patentanmeldung"
elif patched['country'] == 'DE':
patched['number'] = trim_leading_zeros(patched['number'])
#if patched.get('kind') == 'T1':
# patched['number'] = pad_left(patched['number'], '0', 7)
# The Eurasian Patent Organization (EAPO)
# Pad to 6 characters with leading zeros
elif patched['country'] == 'EA' and number_length < 9:
patched['number'] = trim_leading_zeros(patched['number'])
patched['number'] = pad_left(patched['number'], '0', 6)
# pad to 7 characters with leading zeros
elif patched['country'] == 'EP':
patched['number'] = trim_leading_zeros(patched['number'])
patched['number'] = pad_left(patched['number'], '0', 7)
elif patched['country'] == 'GE':
patched['number'] = patched['number'].lstrip('0')
# e.g.
# GE00U200501210Y = GEU20051210Y
# GE00P200503700B = GEP20053700B
#print '77777777777:', patched['number'][5]
if patched['number'][5] == '0':
patched['number'] = patched['number'][:5] + patched['number'][6:]
elif patched['country'] == 'IT':
patched['number'] = patched['number'].lstrip('0')
patched = normalize_patent_it(patched)
# 2017-09-06: KR numbers
# e.g. KR1020150124192A => KR20150124192A
elif patched['country'] == 'KR':
patched['number'] = trim_leading_zeros(patched['number'])
if len(patched['number']) > 11 and patched['number'][:2] == '10':
patched['number'] = patched['number'][2:]
# 2009-11-09: JP numbers
elif patched['country'] == 'JP':
patched = normalize_patent_jp(patched)
# 2015-09-01: SE numbers
elif patched['country'] == 'SE':
patched = normalize_patent_se(patched)
patched['number'] = trim_leading_zeros(patched['number'])
# 2007-07-26: US applications are 4+7
elif patched['country'] == 'US':
patched = normalize_patent_us(patched, provider=provider)
# normalize wo numbers to 4+6 format
elif patched['country'] == 'WO':
# WOPCT/US86/01765 or WOEP/2004/008531
if patched['number'].startswith('PCT'):
patched = normalize_patent_wo_pct(patched)
else:
patched = normalize_patent_wo(patched)
#patched = denormalize_patent_wo(patched)
# strip leading zeros
else:
patched['number'] = trim_leading_zeros(patched['number'])
#print "patched (regular):", patent, patched
return patched
