import requests
from bs4 import BeautifulSoup
from cnsipo.utils import retry, JobQueue, threaded
from cnsipo.shared import get_logger, ContentError, FORGIVEN_ERROR, \
DETAIL_KINDS
KINDS = ['fmgb', 'fmsq', 'syxx', 'wgsq']
STR_SRC = ['fmmost', 'fmmost', 'xxmost', 'wgmost']
STR_WHERE = ['GB', 'SQ', 'GB', 'SQ']
DELAY = 3
RETRIES = 1000
logger = get_logger()
def detail_params(patent_id, kind):
params = {
'strSources': STR_SRC[kind],
'strWhere': "申请号='{}' and {}INDEX=1".format(
patent_id, STR_WHERE[kind]), 'strLicenseCode': "", 'pageNow': 1
}
return "http://epub.sipo.gov.cn/patentdetail.action", params
def detail_parse(bs, kind):
# TODO: not work for kind 'wgsq'
details = {}
tbl = bs.table.table
# -*- coding: utf-8 -*-
"""
Create a UIG(University/Industry/Government) database
"""
from __future__ import print_function
import sys
from optparse import OptionParser
import psycopg2
from cnsipo.shared import get_logger
from cnsipo.patent_parser import PatentParser
logger = get_logger()
APP_NO = 'app_no'
APP_YEAR = 'app_year'
STATE = 'state'
ADDRESS = 'address'
APPLICANT = 'applicant'
COLLAB = 'collab'
KIND = 'kind'
ORG = 'org'
ORG2 = 'org2'
patent_parser = None
def gen_uig_data(conn, table, aux_tbl, year, batch_size):
stmt = "SELECT {}, {}, {} FROM {} WHERE {} IN "\