BIDI_ML_INFO_EXTRACTION/BiddingKG/dl/entityLink/entityLink.py

#coding:UTF8
'''
Created on 2019年5月21日

@author: User
'''
import re
import os
import time
import pandas as pd
_time = time.time()
from BiddingKG.dl.common.Utils import *
from BiddingKG.dl.interface.Entitys import *
import json

def edit_distance(source,target):
    dp = [["" for i in range(len(source)+1)] for j in range(len(target)+1)]
    for i in range(len(dp)):
        for j in range(len(dp[i])):
            if i==0:
                dp[i][j] = j
            elif j==0:
                dp[i][j] = i
            else:
                if source[j-1]==target[i-1]:
                    cost = 0
                else:
                    cost = 2
                dp[i][j] = min([dp[i-1][j]+1,dp[i][j-1]+1,dp[i-1][j-1]+cost])
    return dp[-1][-1]
    
def jaccard_score(source,target):
    source_set = set([s for s in source])
    target_set = set([s for s in target])
    if len(source_set)==0 or len(target_set)==0:
        return 0
    return max(len(source_set&target_set)/len(source_set),len(source_set&target_set)/len(target_set))


def get_place_list():
    path = os.path.dirname(__file__) + '/../place_info.csv'
    place_df = pd.read_csv(path)

    place_list = []
    for index, row in place_df.iterrows():
        place_list.append(row[1])

    place_list.append('台湾')
    place_list.append('澳门')
    place_list.append('香港')
    # place_list.append('東莞')
    # place_list.append('廣州')
    # place_list.append('韩国')
    # place_list.append('德国')
    # place_list.append('英国')
    # place_list.append('日本')
    # place_list.append('意大利')
    # place_list.append('新加坡')
    # place_list.append('加拿大')
    # place_list.append('西班牙')
    # place_list.append('澳大利亚')
    # place_list.append('美国')

    place_list = list(set(place_list))
    return place_list


place_list = get_place_list()
place_pattern = "|".join(place_list)


def link_entitys(list_entitys,on_value=0.81):
    for list_entity in list_entitys:
        range_entity = []
        for _entity in list_entity:
            if _entity.entity_type in ["org","company"]:
                range_entity.append(_entity)
        range_entity = range_entity[:1000]
        for first_i in range(len(range_entity)):
            _entity = range_entity[first_i]
            for second_i in range(first_i+1,len(range_entity)):
                _ent = range_entity[second_i]
                # 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
                if _entity.entity_text != _ent.entity_text and _entity.label != _ent.label and _entity.label in [0,1] and _ent.label in [0, 1]:
                    continue
                _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))
                if _entity.entity_text!=_ent.entity_text and _score>=on_value:
                    _entity.linked_entitys.append(_ent)
                    _ent.linked_entitys.append(_entity)
        #替换公司名称
        for _entity in range_entity:
            if re.search("公司",_entity.entity_text) is None:
                for _ent in _entity.linked_entitys:
                    if re.search("公司$",_ent.entity_text) is not None:
                        if len(_ent.entity_text)>len(_entity.entity_text):
                            _entity.entity_text = _ent.entity_text

        # 2021/12/21 替换通过字典识别到的取长度最大的相似实体
        for _entity in range_entity:
            used_linked_entitys = []
            if not _entity.linked_entitys:
                continue
            _entity.linked_entitys.sort(key=lambda x: len(x.entity_text), reverse=True)
            for _ent in _entity.linked_entitys:
                if _ent in used_linked_entitys:
                    break
                # print("_entity, _ent", _entity.entity_text, _ent.if_dict_match, _ent.entity_text)
                if _ent.if_dict_match == 1:
                    if len(_ent.entity_text) > len(_entity.entity_text):
                        # 判断两个公司地区相同
                        match_list_1, match_list_2 = [], []
                        for place in place_list:
                            if place in _entity.entity_text:
                                match_list_1.append(place)
                            if place in _ent.entity_text:
                                match_list_2.append(place)

                        if str(match_list_1) == str(match_list_2):
                            # print("字典替换", _entity.entity_text, "->", _ent.entity_text)
                            _entity.origin_entity_text = _entity.entity_text
                            _entity.entity_text = _ent.entity_text
                            used_linked_entitys.append(_ent)
                            # print(_entity.entity_text, _entity.if_dict_match, _ent.entity_text, _ent.if_dict_match)


def getEnterprisePath():
    filename = "LEGAL_ENTERPRISE.txt"
    real_path = getFileFromSysPath(filename)
    if real_path is None:
        real_path = filename
    return real_path

DICT_ENTERPRISE = {}
DICT_ENTERPRISE_DONE = False
def getDict_enterprise():
    global DICT_ENTERPRISE,DICT_ENTERPRISE_DONE
    real_path = getEnterprisePath()
    with open(real_path,"r",encoding="UTF8") as f:
        for _e in f:
            if not _e:
                continue
            _e = _e.strip()
            if len(_e)>=4:
                key_enter = _e[:4]
                if key_enter not in DICT_ENTERPRISE:
                    DICT_ENTERPRISE[key_enter] = set()
                DICT_ENTERPRISE[key_enter].add(_e[4:])

    # for _e in ["河南省柘源","建筑工程有限公司"]:
    #     if not _e:
    #         continue
    #     _e = _e.strip()
    #     if len(_e)>=4:
    #         key_enter = _e[:4]
    #         if key_enter not in DICT_ENTERPRISE:
    #             DICT_ENTERPRISE[key_enter] = set()
    #         DICT_ENTERPRISE[key_enter].add(_e[4:])
    DICT_ENTERPRISE_DONE = True
    return DICT_ENTERPRISE

import threading
import time

load_enterprise_thread = threading.Thread(target=getDict_enterprise)
load_enterprise_thread.start()


MAX_ENTERPRISE_LEN = 30

def match_enterprise_max_first(sentence):
    while True:
        if not DICT_ENTERPRISE_DONE:
            time.sleep(1)
        else:
            break
    list_match = []
    begin_index = 0
    if len(sentence)>4:
        while True:
            if begin_index+4<len(sentence):
                key_enter = sentence[begin_index:begin_index+4]
                if key_enter in DICT_ENTERPRISE:
                    for _i in range(MAX_ENTERPRISE_LEN-4+1):
                        enter_name = sentence[begin_index+4:begin_index+MAX_ENTERPRISE_LEN-_i]
                        if enter_name in DICT_ENTERPRISE[key_enter]:
                            match_item = {"entity_text":"%s%s"%(key_enter,enter_name),"begin_index":begin_index,"end_index":begin_index+len(key_enter)+len(enter_name)}
                            list_match.append(match_item)
                            begin_index += (len(key_enter)+len(enter_name))-1
                            break
                begin_index += 1
            else:
                break
    return list_match

def calibrateEnterprise(list_articles,list_sentences,list_entitys):
    for _article,list_sentence,list_entity in zip(list_articles,list_sentences,list_entitys):
        list_calibrate = []
        match_add = False
        match_replace = False
        range_entity = []
        for p_entity in list_entity:
            if p_entity.entity_type in ("org","company","location"):
                range_entity.append(p_entity)
            if len(range_entity)>1000:
                break
        for p_sentence in list_sentence:
            sentence = p_sentence.sentence_text
            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']]
            list_match = match_enterprise_max_first(sentence)
            # print("list_match", list_match)

            doc_id = p_sentence.doc_id
            sentence_index = p_sentence.sentence_index
            tokens = p_sentence.tokens

            list_match.sort(key=lambda x:x["begin_index"])


            for _match_index in range(len(list_match)):
                _match = list_match[_match_index]
                find_flag = False
                for p_entity in range_entity:
                    if p_entity.sentence_index!=p_sentence.sentence_index:
                        continue

                    if p_entity.entity_type=="location" and p_entity.entity_text==_match["entity_text"]:
                        find_flag = True
                        p_entity.entity_type = "company"
                        p_entity.if_dict_match = 1

                    if p_entity.entity_type not in ["location","org","company"]:
                        continue

                    if _match["entity_text"] == p_entity.entity_text:
                        p_entity.if_dict_match = 1

                    #有重叠
                    #match部分被包含则不处理
                    if _match["begin_index"]>=p_entity.wordOffset_begin and _match["end_index"]<=p_entity.wordOffset_end:
                        find_flag = True
                        #判断是否是多个公司
                        for _match_j in range(_match_index,len(list_match)):
                            if not list_match[_match_j]["end_index"]<=p_entity.wordOffset_end:
                                _match_j -= 1
                                break
                        if _match_j>_match_index:

                            match_replace = True
                            match_add = True
                            begin_index = changeIndexFromWordToWords(tokens,_match["begin_index"])
                            end_index = changeIndexFromWordToWords(tokens,_match["end_index"]-1)
                            list_calibrate.append({"type":"update","from":p_entity.entity_text,"to":_match["entity_text"]})
                            p_entity.entity_text = _match["entity_text"]
                            p_entity.wordOffset_begin = _match["begin_index"]
                            p_entity.wordOffset_end = _match["end_index"]
                            p_entity.begin_index = begin_index
                            p_entity.end_index = end_index
                            # 该公司实体是字典识别的
                            p_entity.if_dict_match = 1

                            for _match_h in range(_match_index+1,_match_j+1):
                                entity_text = list_match[_match_h]["entity_text"]
                                entity_type = "company"

                                begin_index = changeIndexFromWordToWords(tokens,list_match[_match_h]["begin_index"])
                                end_index = changeIndexFromWordToWords(tokens,list_match[_match_h]["end_index"]-1)
                                entity_id = "%s_%d_%d_%d"%(doc_id,sentence_index,begin_index,end_index)
                                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)
                                add_entity.if_dict_match = 1
                                list_entity.append(add_entity)

                                range_entity.append(add_entity)

                                list_calibrate.append({"type":"add","from":"","to":entity_text})
                            _match_index = _match_j
                            break

                        continue
                    elif _match["begin_index"]<=p_entity.wordOffset_begin and _match["end_index"]>p_entity.wordOffset_begin:
                        find_flag = True

                        if _match["begin_index"]<p_entity.wordOffset_begin and _match["end_index"]<=p_entity.wordOffset_end:
                            if p_entity.entity_type in ("org","company"):
                                _diff_text = sentence[p_entity.wordOffset_end:_match["end_index"]]
                                if re.search("分",_diff_text) is not None:
                                    pass
                                else:
                                    match_replace = True
                                    begin_index = changeIndexFromWordToWords(tokens,_match["begin_index"])
                                    end_index = changeIndexFromWordToWords(tokens,_match["end_index"]-1)
                                    list_calibrate.append({"type":"update","from":p_entity.entity_text,"to":_match["entity_text"]})
                                    p_entity.entity_text = _match["entity_text"]
                                    p_entity.wordOffset_begin = _match["begin_index"]
                                    p_entity.wordOffset_end = _match["end_index"]
                                    p_entity.begin_index = begin_index
                                    p_entity.end_index = end_index
                                    p_entity.if_dict_match = 1
                        elif _match["end_index"]>=p_entity.wordOffset_end:
                            # 原entity列表已有实体，则不重复添加
                            if (_match["entity_text"],_match["begin_index"],_match["end_index"]) not in sentence_entitys:
                                match_replace = True
                                begin_index = changeIndexFromWordToWords(tokens,_match["begin_index"])
                                end_index = changeIndexFromWordToWords(tokens,_match["end_index"]-1)
                                list_calibrate.append({"type":"update","from":p_entity.entity_text,"to":_match["entity_text"]})
                                p_entity.entity_text = _match["entity_text"]
                                p_entity.wordOffset_begin = _match["begin_index"]
                                p_entity.wordOffset_end = _match["end_index"]
                                p_entity.begin_index = begin_index
                                p_entity.end_index = end_index
                                p_entity.entity_type = "company"
                                p_entity.if_dict_match = 1
                    elif _match["begin_index"]<p_entity.wordOffset_end and _match["end_index"]>p_entity.wordOffset_end:
                        find_flag = True
                        if p_entity.entity_type in ("org","company"):
                            match_replace = True
                            begin_index = changeIndexFromWordToWords(tokens,_match["begin_index"])
                            end_index = changeIndexFromWordToWords(tokens,_match["end_index"]-1)
                            list_calibrate.append({"type":"update","from":p_entity.entity_text,"to":_match["entity_text"]})
                            p_entity.entity_text = _match["entity_text"]
                            p_entity.wordOffset_begin = _match["begin_index"]
                            p_entity.wordOffset_end = _match["end_index"]
                            p_entity.begin_index = begin_index
                            p_entity.end_index = end_index
                            p_entity.if_dict_match = 1
                if not find_flag:
                    match_add = True
                    entity_text = _match["entity_text"]
                    entity_type = "company"

                    begin_index = changeIndexFromWordToWords(tokens,_match["begin_index"])
                    end_index = changeIndexFromWordToWords(tokens,_match["end_index"]-1)
                    entity_id = "%s_%d_%d_%d"%(doc_id,sentence_index,begin_index,end_index)
                    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)
                    list_entity.append(add_entity)
                    range_entity.append(add_entity)

                    list_calibrate.append({"type":"add","from":"","to":entity_text})

        #去重
        set_calibrate = set()
        list_match_enterprise = []
        for _calibrate in list_calibrate:
            _from = _calibrate.get("from","")
            _to = _calibrate.get("to","")
            _key = _from+_to
            if _key not in set_calibrate:
                list_match_enterprise.append(_calibrate)
            set_calibrate.add(_key)
        match_enterprise_type = 0
        if match_add:
            match_enterprise_type += 1
        if match_replace:
            match_enterprise_type += 2
        _article.match_enterprise = list_match_enterprise
        _article.match_enterprise_type = match_enterprise_type

def isLegalEnterprise(name):
    is_legal = True
    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)$",name) or re.search("标段|标包|名称",name) is not None:
        is_legal = False
    return is_legal

def fix_LEGAL_ENTERPRISE():
    unlegal_enterprise = []
    _path = getEnterprisePath()
    _sum = 0
    set_enter = set()
    paths = [_path]
    for _p in paths:
        with open(_p,"r",encoding="utf8") as f:
            while True:
                line = f.readline()
                if not line:
                    break
                line = line.strip()
                if isLegalEnterprise(line):
                    set_enter.add(line)
                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):
                    print(line)
                    if line in set_enter:
                        set_enter.remove(line)

    with open("enter.txt","w",encoding="utf8") as fwrite:
        for line in list(set_enter):

            fwrite.write(line.replace("(","（").replace(")","）"))
            fwrite.write("\n")
                # if re.search("标段|地址|标包|名称",line) is not None:#\(|\)||
                #     _count += 1
                #     print("=",line)
                #     print("%d/%d"%(_count,_sum))
    # a_list = []
    # with open("电信分公司.txt","r",encoding="utf8") as f:
    #     while True:
    #         _line = f.readline()
    #         if not _line:
    #             break
    #         if _line.strip()!="":
    #             a_list.append(_line.strip())
    # with open("enter.txt","a",encoding="utf8") as f:
    #     for _line in a_list:
    #         f.write(_line)
    #         f.write("\n")



                                
if __name__=="__main__":
    # edit_distance("GUMBO","GAMBOL")
    # 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"))
    #
    # 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"
    # print(match_enterprise_max_first(sentences))
    #
    # print("takes %d s"%(time.time()-_time))
    fix_LEGAL_ENTERPRISE()
    # print(jaccard_score("中国南方航空股份有限公司上海分公司","南方航空上海分公司"))
    # print(match_enterprise_max_first("中国南方航空股份有限公司黑龙江分公司"))