entityLink.py 35 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753
  1. #coding:UTF8
  2. '''
  3. Created on 2019年5月21日
  4. @author: User
  5. '''
  6. import re
  7. import os
  8. import time
  9. import pandas as pd
  10. _time = time.time()
  11. from BiddingKG.dl.common.Utils import *
  12. from BiddingKG.dl.interface.Entitys import *
  13. import json
  14. from BiddingKG.dl.common.constDict import ConstDict
  15. def edit_distance(source,target):
  16. dp = [["" for i in range(len(source)+1)] for j in range(len(target)+1)]
  17. for i in range(len(dp)):
  18. for j in range(len(dp[i])):
  19. if i==0:
  20. dp[i][j] = j
  21. elif j==0:
  22. dp[i][j] = i
  23. else:
  24. if source[j-1]==target[i-1]:
  25. cost = 0
  26. else:
  27. cost = 2
  28. dp[i][j] = min([dp[i-1][j]+1,dp[i][j-1]+1,dp[i-1][j-1]+cost])
  29. return dp[-1][-1]
  30. def jaccard_score(source,target):
  31. source_set = set([s for s in source])
  32. target_set = set([s for s in target])
  33. if len(source_set)==0 or len(target_set)==0:
  34. return 0
  35. return max(len(source_set&target_set)/len(source_set),len(source_set&target_set)/len(target_set))
  36. def get_place_list():
  37. path = os.path.dirname(__file__) + '/../place_info.csv'
  38. place_df = pd.read_csv(path)
  39. place_list = []
  40. for index, row in place_df.iterrows():
  41. place_list.append(row[1])
  42. place_list.append('台湾')
  43. place_list.append('澳门')
  44. place_list.append('香港')
  45. # place_list.append('東莞')
  46. # place_list.append('廣州')
  47. # place_list.append('韩国')
  48. # place_list.append('德国')
  49. # place_list.append('英国')
  50. # place_list.append('日本')
  51. # place_list.append('意大利')
  52. # place_list.append('新加坡')
  53. # place_list.append('加拿大')
  54. # place_list.append('西班牙')
  55. # place_list.append('澳大利亚')
  56. # place_list.append('美国')
  57. place_list = list(set(place_list))
  58. return place_list
  59. place_list = get_place_list()
  60. place_pattern = "|".join(place_list)
  61. def is_short(shorter_cut, longer):
  62. '''
  63. 判断是否为简称
  64. :param shorter_cut: 简称
  65. :param longer: 全称
  66. :return:
  67. '''
  68. flag = 1
  69. for words in shorter_cut:
  70. if words in longer:
  71. longer = longer[longer.find(words) + len(words):]
  72. else:
  73. flag = 0
  74. break
  75. if flag:
  76. return 1
  77. else:
  78. return 0
  79. def get_business_data(enterprise_name):
  80. '''
  81. 获取指定公司名称是否有工商数据,有就返回True及相关招投标数据,没有返回False及{}
  82. :param enterprise_name: 公司名称
  83. :return:
  84. '''
  85. global ENTERPRISE_HUGE,SET_ENTERPRISE,POOL_REDIS
  86. # print("test",enterprise_name)
  87. if ENTERPRISE_HUGE:
  88. if POOL_REDIS is None:
  89. init_redis_pool()
  90. _db = POOL_REDIS.getConnector()
  91. try:
  92. _time = time.time()
  93. _v = _db.get(enterprise_name)
  94. POOL_REDIS.putConnector(_db)
  95. if _v is None:
  96. return False, {}
  97. else:
  98. _v = str(_v, 'utf-8')
  99. if 'have_business' in _v:
  100. # log("redis take %.5f of '%s' exists"%(time.time()-_time,enterprise_name))
  101. d = json.loads(_v)
  102. if d.get('have_business', '') == 1:
  103. return True, d
  104. return False, d
  105. else:
  106. return False, {}
  107. except Exception as e:
  108. traceback.print_exc()
  109. return False, {}
  110. else:
  111. if enterprise_name in SET_ENTERPRISE:
  112. return True, {}
  113. else:
  114. return False, {}
  115. def get_role(dic):
  116. '''
  117. 通过字典统计 招标、代理、中标公告数量 返回最大比例及对应类别
  118. :param dic: redics 获取实体的工商数据字典
  119. :return:
  120. '''
  121. if 'zhao_biao_number' in dic:
  122. zhaobiao = dic.get('zhao_biao_number', 0)
  123. daili = dic.get('dai_li_number', 0)
  124. zhongbiao = dic.get('zhong_biao_number', 0)
  125. bid = zhaobiao+ daili+ zhongbiao
  126. if bid > 100: # 总数大于100的才统计
  127. if zhaobiao>=daili:
  128. if zhaobiao>=zhongbiao:
  129. return 0, zhaobiao/bid
  130. else:
  131. return 2, zhongbiao/bid
  132. elif daili >= zhongbiao:
  133. return 1, daili/bid
  134. else:
  135. return 2, zhongbiao/bid
  136. return 5, 0
  137. def link_entitys(list_entitys,on_value=1):#on_value=0.81
  138. for list_entity in list_entitys:
  139. range_entity = []
  140. short_entity = [] # 不包含工商数据实体
  141. long_entity = [] # 包含工商数据实体
  142. n = 0
  143. bus_dic = {} # 保存已查询包含工商数据实体 属于招标、代理、中标 何种类别及对应概率
  144. find_tenderee = False
  145. bus_tenderee = []
  146. for _entity in list_entity:
  147. if _entity.entity_type in ["org","company"]:
  148. range_entity.append(_entity)
  149. if _entity.entity_text in bus_dic:
  150. have_bus = True
  151. else:
  152. have_bus, dic = get_business_data(_entity.entity_text)
  153. if have_bus:
  154. lb, prob = get_role(dic)
  155. bus_dic[_entity.entity_text] = (lb, prob)
  156. if lb == 0 and prob > 0.9 and re.search('医院|学院|学校|中学|小学|大学|中心|幼儿园|保健院|党校|银行|研究院|血站|分校|红十字会|防治院|研究所', _entity.entity_text) and _entity.entity_text not in ['中华人民共和国', '营业执照', '人民法院','民办非企业单位','个体工商户','运输服务', '社会团体']:
  157. bus_tenderee.append(_entity)
  158. if have_bus: # 20231115 改为只判断是否有工商数据,没有就考虑替换
  159. long_entity.append(_entity)
  160. lb, prob = bus_dic[_entity.entity_text]
  161. if lb in [0,1] and prob>0.9 and _entity.label in [0, 1] and _entity.values[_entity.label]<0.55: # 如果工商统计概率较高,文中概率较低,换为统计类别,主要为标题及发布人等招标、代理划分不明确情况
  162. if _entity.label != lb:
  163. _entity.label = lb
  164. _entity.values[_entity.label] = 0.55
  165. else:
  166. _entity.values[_entity.label] += 0.05
  167. else:
  168. short_entity.append(_entity)
  169. if _entity.label == 0: # 找到招标人
  170. find_tenderee = True
  171. n += 1
  172. if n > 1000:
  173. break
  174. if find_tenderee == False and len(bus_tenderee)==1 and bus_tenderee[0].label==5: # 如果整篇都没招标人,工商统计只有一个高概率招标人把它作为招标人
  175. bus_tenderee[0].label = 0
  176. bus_tenderee[0].values[0] = 0.55
  177. range_entity = range_entity[:1000]
  178. #替换公司的逻辑有问题,先取消
  179. # for first_i in range(len(range_entity)):
  180. # _entity = range_entity[first_i]
  181. # for second_i in range(first_i+1,len(range_entity)):
  182. # _ent = range_entity[second_i]
  183. # # 2021/5/21 update: 两个实体标签互斥(一个是招标人、一个是代理人)且entity_text不相等时,跳过
  184. # if _entity.entity_text != _ent.entity_text and _entity.label != _ent.label and _entity.label in [0,1] and _ent.label in [0, 1]:
  185. # continue
  186. # _score = jaccard_score(re.sub("%s|%s"%("股份|责任|有限|公司",place_pattern),"",_entity.entity_text), re.sub("%s|%s"%("股份|责任|有限|公司",place_pattern),"",_ent.entity_text))
  187. # if _entity.entity_text!=_ent.entity_text and _score>=on_value:
  188. # _entity.linked_entitys.append(_ent)
  189. # _ent.linked_entitys.append(_entity)
  190. # print("=-===",_entity.entity_text,_ent.entity_text,_score)
  191. # #替换公司名称
  192. # for _entity in range_entity:
  193. # if re.search("公司",_entity.entity_text) is None:
  194. # for _ent in _entity.linked_entitys:
  195. # if re.search("公司$",_ent.entity_text) is not None:
  196. # if len(_ent.entity_text)>len(_entity.entity_text):
  197. # _entity.entity_text = _ent.entity_text
  198. if short_entity and long_entity: #
  199. for first_i in range(len(short_entity)):
  200. _entity = short_entity[first_i]
  201. if _entity.label == 0:
  202. for second_i in range(len(long_entity)):
  203. _ent = long_entity[second_i]
  204. if _ent.label in [0,1,5]:
  205. if len(_entity.entity_text)<len(_ent.entity_text) and is_short(_entity.entity_text, _ent.entity_text): # 简称顺序包含在工商名称内的替换
  206. _entity.entity_text = _ent.entity_text
  207. lb, prob = bus_dic[_entity.entity_text]
  208. if lb in [0, 1] and prob > 0.9 and _entity.values[
  209. _entity.label] < 0.55: # 如果工商统计概率较高,文中概率较低,换为统计类别,主要为标题及发布人等招标、代理划分不明确情况
  210. if _entity.label != lb:
  211. _entity.label = lb
  212. _entity.values[_entity.label] = 0.55
  213. else:
  214. _entity.values[_entity.label] += 0.05
  215. break
  216. elif len(_entity.entity_text)>len(_ent.entity_text) and _ent.entity_text in _entity.entity_text: # 不包含工商数据实体完全包含工商数据实体名称的替换
  217. _entity.entity_text = _ent.entity_text
  218. lb, prob = bus_dic[_entity.entity_text]
  219. if lb in [0, 1] and prob > 0.9 and _entity.values[
  220. _entity.label] < 0.55: # 如果工商统计概率较高,文中概率较低,换为统计类别,主要为标题及发布人等招标、代理划分不明确情况
  221. if _entity.label != lb:
  222. _entity.label = lb
  223. _entity.values[_entity.label] = 0.55
  224. else:
  225. _entity.values[_entity.label] += 0.05
  226. break
  227. # 2021/12/21 替换通过字典识别到的取长度最大的相似实体
  228. for _entity in range_entity:
  229. used_linked_entitys = []
  230. if not _entity.linked_entitys:
  231. continue
  232. _entity.linked_entitys.sort(key=lambda x: len(x.entity_text), reverse=True)
  233. for _ent in _entity.linked_entitys:
  234. if _ent in used_linked_entitys:
  235. break
  236. # print("_entity, _ent", _entity.entity_text, _ent.if_dict_match, _ent.entity_text)
  237. if _ent.if_dict_match == 1:
  238. if len(_ent.entity_text) > len(_entity.entity_text):
  239. # 判断两个公司地区相同
  240. match_list_1, match_list_2 = [], []
  241. for place in place_list:
  242. if place in _entity.entity_text:
  243. match_list_1.append(place)
  244. if place in _ent.entity_text:
  245. match_list_2.append(place)
  246. if str(match_list_1) == str(match_list_2):
  247. # print("字典替换", _entity.entity_text, "->", _ent.entity_text)
  248. _entity.origin_entity_text = _entity.entity_text
  249. _entity.entity_text = _ent.entity_text
  250. used_linked_entitys.append(_ent)
  251. # print(_entity.entity_text, _entity.if_dict_match, _ent.entity_text, _ent.if_dict_match)
  252. # 用于去重的标题
  253. def doctitle_refine(doctitle):
  254. _doctitle_refine = re.sub(r'工程|服务|询价|比价|谈判|竞争性|磋商|结果|中标|招标|采购|的|公示|公开|成交|公告|评标|候选人|'
  255. r'交易|通知|废标|流标|终止|中止|一笔|预告|单一来源|竞价|合同', '', doctitle)
  256. return _doctitle_refine
  257. # 前100个公司实体
  258. def get_nlp_enterprise(list_entity):
  259. nlp_enterprise = []
  260. nlp_enterprise_attachment = []
  261. max_num = 100
  262. list_entity = sorted(list_entity,key=lambda x:(x.sentence_index,x.begin_index))
  263. for entity in list_entity:
  264. if entity.entity_type in ['org','company']:
  265. if not entity.in_attachment:
  266. if entity.entity_text not in nlp_enterprise:
  267. nlp_enterprise.append(entity.entity_text)
  268. else:
  269. if entity.entity_text not in nlp_enterprise_attachment:
  270. nlp_enterprise_attachment.append(entity.entity_text)
  271. return nlp_enterprise[:max_num],nlp_enterprise_attachment[:max_num]
  272. ENTERPRISE_HUGE = None
  273. def getEnterprisePath():
  274. global ENTERPRISE_HUGE
  275. filename_huge = "LEGAL_ENTERPRISE_HUGE.txt"
  276. huge_path = getFileFromSysPath(filename_huge)
  277. if huge_path is None:
  278. if os.path.exists(filename_huge):
  279. log("enterprise path:%s"%(filename_huge))
  280. ENTERPRISE_HUGE = True
  281. return filename_huge,ENTERPRISE_HUGE
  282. else:
  283. log("enterprise path:%s"%(huge_path))
  284. ENTERPRISE_HUGE = True
  285. return huge_path,ENTERPRISE_HUGE
  286. filename = "LEGAL_ENTERPRISE.txt"
  287. real_path = getFileFromSysPath(filename)
  288. if real_path is None:
  289. real_path = filename
  290. log("ENTERPRISE path:%s"%(real_path))
  291. ENTERPRISE_HUGE = False
  292. return real_path,ENTERPRISE_HUGE
  293. DICT_ENTERPRISE_DONE = False
  294. POOL_REDIS = None
  295. ENTERPRISE_KEY_LEN = 3
  296. ENTERPRISE_PREFIX_LEN = 3
  297. ENTERPRISE_TAIL_LEN = 3
  298. SET_ENTERPRISE = set()
  299. SET_PREFIX_ENTERPRISE = set()
  300. SET_TAIL_ENTERPRISE = set()
  301. SET_PREFIX_ENTERPRISE_HUGE_FILE = "SET_PREFIX_ENTERPRISE_HUGE.pk"
  302. SET_TAIL_ENTERPRISE_HUGE_FILE = "SET_TAIL_ENTERPRISE_HUGE.pk"
  303. def getDict_enterprise():
  304. global DICT_ENTERPRISE_DONE,SET_ENTERPRISE,SET_PREFIX_ENTERPRISE,SET_TAIL_ENTERPRISE
  305. real_path,is_huge = getEnterprisePath()
  306. _ok = False
  307. if is_huge:
  308. if os.path.exists(SET_PREFIX_ENTERPRISE_HUGE_FILE) and os.path.exists(SET_TAIL_ENTERPRISE_HUGE_FILE):
  309. SET_PREFIX_ENTERPRISE = load(SET_PREFIX_ENTERPRISE_HUGE_FILE)
  310. SET_TAIL_ENTERPRISE = load(SET_TAIL_ENTERPRISE_HUGE_FILE)
  311. _ok = True
  312. if not _ok:
  313. with open(real_path,"r",encoding="UTF8") as f:
  314. for _e in f:
  315. if not _e:
  316. continue
  317. _e = _e.strip()
  318. if len(_e)>=4:
  319. key_enter = _e[:ENTERPRISE_KEY_LEN]
  320. SET_PREFIX_ENTERPRISE.add(key_enter)
  321. SET_TAIL_ENTERPRISE.add(_e[-ENTERPRISE_TAIL_LEN:])
  322. if not is_huge:
  323. SET_ENTERPRISE.add(_e)
  324. #仅在大文件情况下才使用缓存加载
  325. if is_huge:
  326. save(SET_PREFIX_ENTERPRISE,SET_PREFIX_ENTERPRISE_HUGE_FILE)
  327. save(SET_TAIL_ENTERPRISE,SET_TAIL_ENTERPRISE_HUGE_FILE)
  328. log("SET_PREFIX_ENTERPRISE takes memory:%.2fM size:%d"%(sys.getsizeof(SET_PREFIX_ENTERPRISE)/1024/1024,len(SET_PREFIX_ENTERPRISE)))
  329. log("SET_TAIL_ENTERPRISE takes memory:%.2fM size:%d"%(sys.getsizeof(SET_TAIL_ENTERPRISE)/1024/1024,len(SET_TAIL_ENTERPRISE)))
  330. log("SET_ENTERPRISE takes memory:%.2fM size:%d"%(sys.getsizeof(SET_ENTERPRISE)/1024/1024,len(SET_ENTERPRISE)))
  331. # for _e in ["河南省柘源","建筑工程有限公司"]:
  332. # if not _e:
  333. # continue
  334. # _e = _e.strip()
  335. # if len(_e)>=4:
  336. # key_enter = _e[:4]
  337. # if key_enter not in DICT_ENTERPRISE:
  338. # DICT_ENTERPRISE[key_enter] = set()
  339. # DICT_ENTERPRISE[key_enter].add(_e[4:])
  340. DICT_ENTERPRISE_DONE = True
  341. def init_redis_pool():
  342. from BiddingKG.dl.common.pool import ConnectorPool
  343. from BiddingKG.dl.common.source import getConnect_redis_baseline
  344. global POOL_REDIS
  345. if POOL_REDIS is None:
  346. POOL_REDIS = ConnectorPool(init_num=1,max_num=10,method_init=getConnect_redis_baseline)
  347. # 插入 Redis
  348. # def add_redis(company_list):
  349. # global ENTERPRISE_HUGE,POOL_REDIS
  350. # if ENTERPRISE_HUGE:
  351. # _db = POOL_REDIS.getConnector()
  352. # for enterprise_name in company_list:
  353. # _v = _db.get(enterprise_name)
  354. # if _v is None:
  355. # if isLegalNewName(enterprise_name):
  356. # _db.set(enterprise_name,1)
  357. # 新实体合法判断
  358. def isLegalNewName(enterprise_name):
  359. # head_character_list = ["[",'【',"(",'(']
  360. # tail_character_list = ["]",'】',")",')']
  361. # 名称开头判断
  362. if re.search("^[\da-zA-Z][^\da-zA-Z]|"
  363. "^[^\da-zA-Z\u4e00-\u9fa5\[【((]|"
  364. "^[\[【((].{,1}[\]】))]|"
  365. "^[0〇]|"
  366. "^(20[0-2][0-9]|[0-2]?[0-9]年|[0-1]?[0-9]月|[0-3]?[0-9]日)",enterprise_name):
  367. return -1
  368. if len(re.findall("[\u4e00-\u9fa5]",enterprise_name))<2:
  369. return -1
  370. if re.search("╳|*|\*|×|xx|XX",enterprise_name):
  371. return -1
  372. if re.search("^(省|自治[县州区]|市|县|区|镇|乡|街道)",enterprise_name) and not re.search("^(镇江|乡宁|镇原|镇海|镇安|镇巴|镇坪|镇赉|镇康|镇沅|镇雄|镇远|镇宁|乡城|镇平|市中|市南|市北)",enterprise_name):
  373. return -1
  374. if re.search("\d{1,2}:\d{2}(:\d{2})?|(rar|xlsx|zip|png|jpg|swf|docx|txt|pdf|PDF|doc|xls|bmp|&?nbsp)",enterprise_name):
  375. return -1
  376. if re.search("(招标|代理)(人|机构)|联系(人|方式)|中标|候选|第.名",enterprise_name):
  377. return -1
  378. if re.search("[a-zA-Z\d]{1,2}(包|标段?)|第.批"):
  379. return 0
  380. return 1
  381. # 过滤掉Redis里值为0的错误实体
  382. def enterprise_filter(entity_list):
  383. global ENTERPRISE_HUGE,SET_ENTERPRISE,POOL_REDIS
  384. if ENTERPRISE_HUGE:
  385. if POOL_REDIS is None:
  386. init_redis_pool()
  387. _db = POOL_REDIS.getConnector()
  388. remove_list = []
  389. try:
  390. for entity in entity_list:
  391. if entity.entity_type in ['company','org']:
  392. _v = _db.get(entity.entity_text)
  393. if _v==0:
  394. remove_list.append(entity)
  395. except Exception as e:
  396. traceback.print_exc()
  397. POOL_REDIS.putConnector(_db)
  398. for _entity in remove_list:
  399. entity_list.remove(_entity)
  400. return entity_list
  401. def is_enterprise_exist(enterprise_name):
  402. global ENTERPRISE_HUGE,SET_ENTERPRISE,POOL_REDIS
  403. # print("test",enterprise_name)
  404. if ENTERPRISE_HUGE:
  405. if POOL_REDIS is None:
  406. init_redis_pool()
  407. _db = POOL_REDIS.getConnector()
  408. try:
  409. _time = time.time()
  410. _v = _db.get(enterprise_name)
  411. POOL_REDIS.putConnector(_db)
  412. if _v is None:
  413. return False
  414. else:
  415. if _v:
  416. # log("redis take %.5f of '%s' exists"%(time.time()-_time,enterprise_name))
  417. return True
  418. else:
  419. return False
  420. except Exception as e:
  421. traceback.print_exc()
  422. return False
  423. else:
  424. if enterprise_name in SET_ENTERPRISE:
  425. return True
  426. else:
  427. return False
  428. import threading
  429. import time
  430. load_enterprise_thread = threading.Thread(target=getDict_enterprise)
  431. load_enterprise_thread.start()
  432. MAX_ENTERPRISE_LEN = 30
  433. def match_enterprise_max_first(sentence):
  434. while True:
  435. if not DICT_ENTERPRISE_DONE:
  436. time.sleep(1)
  437. else:
  438. break
  439. list_match = []
  440. begin_index = 0
  441. if len(sentence)>4:
  442. while True:
  443. if begin_index+ENTERPRISE_KEY_LEN<len(sentence):
  444. key_enter = sentence[begin_index:begin_index+ENTERPRISE_KEY_LEN]
  445. # if key_enter in DICT_ENTERPRISE:
  446. # _len = min(MAX_ENTERPRISE_LEN-ENTERPRISE_KEY_LEN+1,len(sentence)-begin_index)
  447. # for _i in range(_len):
  448. # enter_name = sentence[begin_index+ENTERPRISE_KEY_LEN:begin_index+_len-_i]
  449. # if enter_name in DICT_ENTERPRISE[key_enter]:
  450. # match_item = {"entity_text":"%s%s"%(key_enter,enter_name),"begin_index":begin_index,"end_index":begin_index+len(key_enter)+len(enter_name)}
  451. # list_match.append(match_item)
  452. # begin_index += (len(key_enter)+len(enter_name))-1
  453. # break
  454. if key_enter in SET_PREFIX_ENTERPRISE:
  455. _len = min(MAX_ENTERPRISE_LEN-ENTERPRISE_KEY_LEN+1,len(sentence)-begin_index)
  456. for _i in range(_len):
  457. enter_name = sentence[begin_index:begin_index+_len-_i]
  458. enter_tail = enter_name[-ENTERPRISE_TAIL_LEN:]
  459. if re.search('[\u4e00-\u9fa5]', enter_tail) == None: # 20240111不包含中文后缀不要
  460. continue
  461. if enter_tail in SET_TAIL_ENTERPRISE or re.search('(中心|中学|小学|医院|学院|大学|学校|监狱|大队|支队|林场|海关|分局|商行)$', enter_tail):
  462. have_bus, dic = get_business_data(enter_name) # 20210124 改为有工商数据的实体才添加
  463. if have_bus:
  464. # if is_enterprise_exist(enter_name):
  465. match_item = {"entity_text":"%s"%(enter_name),"begin_index":begin_index,"end_index":begin_index+len(enter_name)}
  466. # print("match_item",key_enter,enter_name)
  467. list_match.append(match_item)
  468. begin_index += len(enter_name)-1
  469. break
  470. begin_index += 1
  471. else:
  472. break
  473. # print("======",list_match)
  474. return list_match
  475. def calibrateEnterprise(list_articles,list_sentences,list_entitys):
  476. for _article,list_sentence,list_entity in zip(list_articles,list_sentences,list_entitys):
  477. list_calibrate = []
  478. match_add = False
  479. match_replace = False
  480. range_entity = []
  481. for p_entity in list_entity:
  482. if p_entity.entity_type in ("org","company","location"):
  483. range_entity.append(p_entity)
  484. if len(range_entity)>1000:
  485. break
  486. for p_sentence in list_sentence:
  487. sentence = p_sentence.sentence_text
  488. sentence_entitys = [(ent.entity_text,ent.wordOffset_begin,ent.wordOffset_end) for ent in list_entity if ent.sentence_index==p_sentence.sentence_index and ent.entity_type in ['org','company']]
  489. list_match = match_enterprise_max_first(sentence)
  490. # print("list_match", list_match)
  491. doc_id = p_sentence.doc_id
  492. sentence_index = p_sentence.sentence_index
  493. tokens = p_sentence.tokens
  494. list_match.sort(key=lambda x:x["begin_index"])
  495. for _match_index in range(len(list_match)):
  496. _match = list_match[_match_index]
  497. find_flag = False
  498. for p_entity in range_entity:
  499. if p_entity.sentence_index!=p_sentence.sentence_index:
  500. continue
  501. if p_entity.entity_type=="location" and p_entity.entity_text==_match["entity_text"]:
  502. find_flag = True
  503. p_entity.entity_type = "company"
  504. p_entity.if_dict_match = 1
  505. if p_entity.entity_type not in ["location","org","company"]:
  506. continue
  507. if _match["entity_text"] == p_entity.entity_text:
  508. p_entity.if_dict_match = 1
  509. #有重叠
  510. #match部分被包含则不处理
  511. if _match["begin_index"]>=p_entity.wordOffset_begin and _match["end_index"]<=p_entity.wordOffset_end:
  512. find_flag = True
  513. #判断是否是多个公司
  514. if re.search('[分支](公司|中心|部|行)', p_entity.entity_text):
  515. continue
  516. for _match_j in range(_match_index,len(list_match)):
  517. if not list_match[_match_j]["end_index"]<=p_entity.wordOffset_end:
  518. _match_j -= 1
  519. break
  520. if _match_j>_match_index:
  521. match_replace = True
  522. match_add = True
  523. begin_index = changeIndexFromWordToWords(tokens,_match["begin_index"])
  524. end_index = changeIndexFromWordToWords(tokens,_match["end_index"]-1)
  525. list_calibrate.append({"type":"update","from":p_entity.entity_text,"to":_match["entity_text"]})
  526. p_entity.entity_text = _match["entity_text"]
  527. p_entity.wordOffset_begin = _match["begin_index"]
  528. p_entity.wordOffset_end = _match["end_index"]
  529. p_entity.begin_index = begin_index
  530. p_entity.end_index = end_index
  531. # 该公司实体是字典识别的
  532. p_entity.if_dict_match = 1
  533. for _match_h in range(_match_index+1,_match_j+1):
  534. entity_text = list_match[_match_h]["entity_text"]
  535. entity_type = "company"
  536. begin_index = changeIndexFromWordToWords(tokens,list_match[_match_h]["begin_index"])
  537. end_index = changeIndexFromWordToWords(tokens,list_match[_match_h]["end_index"]-1)
  538. entity_id = "%s_%d_%d_%d"%(doc_id,sentence_index,begin_index,end_index)
  539. add_entity = Entity(p_sentence.doc_id,entity_id,entity_text,entity_type,sentence_index,begin_index,end_index,list_match[_match_h]["begin_index"],list_match[_match_h]["end_index"],in_attachment=p_sentence.in_attachment)
  540. add_entity.if_dict_match = 1
  541. list_entity.append(add_entity)
  542. range_entity.append(add_entity)
  543. list_calibrate.append({"type":"add","from":"","to":entity_text})
  544. _match_index = _match_j
  545. break
  546. continue
  547. elif _match["begin_index"]<=p_entity.wordOffset_begin and _match["end_index"]>p_entity.wordOffset_begin:
  548. find_flag = True
  549. if _match["begin_index"]<p_entity.wordOffset_begin and _match["end_index"]<=p_entity.wordOffset_end:
  550. if p_entity.entity_type in ("org","company"):
  551. _diff_text = sentence[p_entity.wordOffset_end:_match["end_index"]]
  552. if re.search("分",_diff_text) is not None:
  553. pass
  554. else:
  555. match_replace = True
  556. begin_index = changeIndexFromWordToWords(tokens,_match["begin_index"])
  557. end_index = changeIndexFromWordToWords(tokens,_match["end_index"]-1)
  558. list_calibrate.append({"type":"update","from":p_entity.entity_text,"to":_match["entity_text"]})
  559. p_entity.entity_text = _match["entity_text"]
  560. p_entity.wordOffset_begin = _match["begin_index"]
  561. p_entity.wordOffset_end = _match["end_index"]
  562. p_entity.begin_index = begin_index
  563. p_entity.end_index = end_index
  564. p_entity.if_dict_match = 1
  565. elif _match["end_index"]>=p_entity.wordOffset_end:
  566. # 原entity列表已有实体,则不重复添加
  567. if (_match["entity_text"],_match["begin_index"],_match["end_index"]) not in sentence_entitys:
  568. match_replace = True
  569. begin_index = changeIndexFromWordToWords(tokens,_match["begin_index"])
  570. end_index = changeIndexFromWordToWords(tokens,_match["end_index"]-1)
  571. list_calibrate.append({"type":"update","from":p_entity.entity_text,"to":_match["entity_text"]})
  572. p_entity.entity_text = _match["entity_text"]
  573. p_entity.wordOffset_begin = _match["begin_index"]
  574. p_entity.wordOffset_end = _match["end_index"]
  575. p_entity.begin_index = begin_index
  576. p_entity.end_index = end_index
  577. p_entity.entity_type = "company"
  578. p_entity.if_dict_match = 1
  579. elif _match["begin_index"]<p_entity.wordOffset_end and _match["end_index"]>p_entity.wordOffset_end:
  580. find_flag = True
  581. if p_entity.entity_type in ("org","company"):
  582. match_replace = True
  583. begin_index = changeIndexFromWordToWords(tokens,_match["begin_index"])
  584. end_index = changeIndexFromWordToWords(tokens,_match["end_index"]-1)
  585. list_calibrate.append({"type":"update","from":p_entity.entity_text,"to":_match["entity_text"]})
  586. p_entity.entity_text = _match["entity_text"]
  587. p_entity.wordOffset_begin = _match["begin_index"]
  588. p_entity.wordOffset_end = _match["end_index"]
  589. p_entity.begin_index = begin_index
  590. p_entity.end_index = end_index
  591. p_entity.if_dict_match = 1
  592. if not find_flag:
  593. match_add = True
  594. entity_text = _match["entity_text"]
  595. entity_type = "company"
  596. begin_index = changeIndexFromWordToWords(tokens,_match["begin_index"])
  597. end_index = changeIndexFromWordToWords(tokens,_match["end_index"]-1)
  598. entity_id = "%s_%d_%d_%d"%(doc_id,sentence_index,begin_index,end_index)
  599. add_entity = Entity(p_sentence.doc_id,entity_id,entity_text,entity_type,sentence_index,begin_index,end_index,_match["begin_index"],_match["end_index"],in_attachment=p_sentence.in_attachment)
  600. list_entity.append(add_entity)
  601. range_entity.append(add_entity)
  602. list_calibrate.append({"type":"add","from":"","to":entity_text})
  603. #去重
  604. set_calibrate = set()
  605. list_match_enterprise = []
  606. for _calibrate in list_calibrate:
  607. _from = _calibrate.get("from","")
  608. _to = _calibrate.get("to","")
  609. _key = _from+_to
  610. if _key not in set_calibrate:
  611. list_match_enterprise.append(_calibrate)
  612. set_calibrate.add(_key)
  613. match_enterprise_type = 0
  614. if match_add:
  615. match_enterprise_type += 1
  616. if match_replace:
  617. match_enterprise_type += 2
  618. _article.match_enterprise = list_match_enterprise
  619. _article.match_enterprise_type = match_enterprise_type
  620. def isLegalEnterprise(name):
  621. is_legal = True
  622. if re.search("^[省市区县]",name) is not None or re.search("^\**.{,3}(分(公司|行|支)|街道|中心|办事处|经营部|委员会|有限公司)$",name) or re.search("标段|标包|名称|联系人|联系方式|中标单位|中标人|测试单位|采购单位|采购人|代理人|代理机构|盖章|(主)",name) is not None:
  623. is_legal = False
  624. return is_legal
  625. def fix_LEGAL_ENTERPRISE():
  626. unlegal_enterprise = []
  627. _path = getEnterprisePath()
  628. _sum = 0
  629. set_enter = set()
  630. paths = [_path]
  631. for _p in paths:
  632. with open(_p,"r",encoding="utf8") as f:
  633. while True:
  634. line = f.readline()
  635. if not line:
  636. break
  637. line = line.strip()
  638. if isLegalEnterprise(line):
  639. set_enter.add(line)
  640. if line=="有限责任公司" or line=='设计研究院' or line=='限责任公司' or (re.search("^.{,4}(分公司|支行|分行)$",line) is not None and re.search("电信|移动|联通|建行|工行|农行|中行|交行",line) is None):
  641. print(line)
  642. if line in set_enter:
  643. set_enter.remove(line)
  644. with open("enter.txt","w",encoding="utf8") as fwrite:
  645. for line in list(set_enter):
  646. fwrite.write(line.replace("(","(").replace(")",")"))
  647. fwrite.write("\n")
  648. # if re.search("标段|地址|标包|名称",line) is not None:#\(|\)||
  649. # _count += 1
  650. # print("=",line)
  651. # print("%d/%d"%(_count,_sum))
  652. # a_list = []
  653. # with open("电信分公司.txt","r",encoding="utf8") as f:
  654. # while True:
  655. # _line = f.readline()
  656. # if not _line:
  657. # break
  658. # if _line.strip()!="":
  659. # a_list.append(_line.strip())
  660. # with open("enter.txt","a",encoding="utf8") as f:
  661. # for _line in a_list:
  662. # f.write(_line)
  663. # f.write("\n")
  664. if __name__=="__main__":
  665. # edit_distance("GUMBO","GAMBOL")
  666. # print(jaccard_score("周口经济开发区陈营运粮河两岸拆迁工地土工布覆盖项目竞争性谈判公告","周口经济开发区陈营运粮河两岸拆迁工地土工布覆盖项目-成交公告"))
  667. #
  668. # sentences = "广州比地数据科技有限公司比地数据科技有限公司1111111123沈阳南光工贸有限公司"
  669. # print(match_enterprise_max_first(sentences))
  670. #
  671. # print("takes %d s"%(time.time()-_time))
  672. # fix_LEGAL_ENTERPRISE()
  673. # print(jaccard_score("吉林省九台","吉林省建苑设计集团有限公司"))
  674. print(match_enterprise_max_first("中国南方航空股份有限公司黑龙江分公司"))