BIDI_EXTRACTION_AI/BiddingKG/dl/person/general_data.py

# DEPRECATED(Phase1): 本文件中的 PostgreSQL 硬编码连接（host=192.168.*,
# password=postgres 等）将在后续 training/ Phase 迁移到 BiddingKG.dl.infra.db。
# 迁移完成前可临时使用：from BiddingKG.dl.infra.db import get_connection;
# conn = get_connection("<dbname>")
# 详见 ARCHITECTURE.md 第 11 章 Phase 1 与 REFACTOR_LOG.md。

import pandas as pd
import psycopg2
import pickle
import re

def get_data():
    '''
    @summary: 取出待标注的数据到excel中
    '''
    conn = psycopg2.connect(dbname='BidiPro', user='postgres',password='postgres',host='192.168.2.101')
    cursor = conn.cursor()    
    #sql = '''SELECT e.doc_id,e.entity_id,e.sentence_index,e.entity_text,e.entity_type,e.begin_index,e.end_index,s.tokens from entity_mention e,sentences s 
    #WHERE s.doc_id=e.doc_id AND s.sentence_index=e.sentence_index AND entity_type in ('person') ORDER BY doc_id,sentence_index,begin_index LIMIT 20000
    #'''
    sql = '''SELECT e.doc_id,e.entity_id,e.sentence_index,e.entity_text,e.entity_type,e.begin_index,e.end_index,s.tokens from entity_mention e,sentences s 
    WHERE s.doc_id=e.doc_id AND s.sentence_index=e.sentence_index AND entity_type in ('person') 
    and e.doc_id in (select id from articles_processed order by id desc limit 4000) ORDER BY doc_id,sentence_index,begin_index limit 20000
    '''    
    cursor.execute(sql)
    rows  = cursor.fetchmany(5000)
    new_df = pd.DataFrame()
    i = 0 
    while(rows):
        df = pd.DataFrame(rows, columns=['doc_id','entity_id','sentence_index','entity_text','entity_type','begin_index','end_index','tokens']) 
        i += 1
        new_df = pd.concat([new_df, df],ignore_index=True)
        #df.to_excel('data/person_'+str(i)+'.xls', encoding='utf-8',index=False)
        rows = cursor.fetchmany(5000)

    with open('data/person.pk', 'wb') as f:
        pickle.dump(new_df, f)
    #new_df.to_excel('data/person_total.xls', encoding='utf-8', index=False)    
    #print(rows)
    cursor.close()
    conn.close()   
    
def label_data():
    '''
    @summary: 先通过规则预标注
    '''
    file2 = 'data/person_label.xls'
    with open('data/person.pk', 'rb') as f:
        data = pickle.load(f)
    # 分类：未知0 招标1 代理2 中标3 监督4 施工员5 联系人6
    zhaobiao = re.compile('采购中心|采购单位|采购人|采购经办人|招标单位|建设单位|招标人|项目单位|比选人|发包人|项目业主')
    daili = re.compile('代理机构|招标代理|采购代理|采购机构|招标代理机构|招标代理人')
#     zhongbiao = re.compile('供应商|法人代表|法定代表|中标人|中标单位|中标候选人|第[一|二|三|1|2|3]名|中标项目|项目负责人|项目经理')
    jiandu = re.compile('评标|评审|审批|审查|评委|监标|专家|小组|成员|名单|监督|监管|监察|监审|主管|受理|处室|反映|异议|质疑|(\d{2}\.\d{2})[^\d]')
    shigong = re.compile('甲方代表|管理人员|管理机构人员|施工员|安全员|质检员|质量员|材料员|预算员|建造师|造价员|监理员|监理人员|项目总监')
    lianxi = re.compile('经办人|联系人|联系方式|联系电话|法人代表|法定代表|中标供应商|中标人|中标单位|中标候选人|第[一|二|三|1|2|3]名|中标项目|项目负责人|项目经理')
    
    pattern_pos = re.compile('联系方式|联系人|项目负责人|项目经理|法人|法定代表|级别及证书|采购人|第一|第二|第三|第1|第2|第3')
    #pattern = re.compile('采购代理|采购机构|采购人|代理机构|项目负责人|联系人|技术负责人|第一|第二|第三|中标人|中标供应商|中标机构|中标候选人|招标|代理|资质|法人代表')
    pattern_neg = re.compile('监管|监察|监督|主管|受理|处室|异议|反映|评委|评审|评标|监标|委员会|磋商|专家|小组|人员类别|管理人员|人员配备|成员|名单')
    count = 0
    span = 10
    
    tokens =  data['tokens']         
    ben = data['begin_index']
    end = data['end_index']
    ent_id = data['entity_id']
    ent = data['entity_text']
    ent_type = data['entity_type']
    sen_index = data['sentence_index']
    
    pre_ent = []
    cur_ent = []
    label = [] # 标签列表
    ent_idl = []
    shiti = []
    s_list = []
    b_list = []
    
    for i in range(len(tokens)):  
        
        if ent_type[i] == 'person':   
            begin1 = ben[i] - span if ben[i] > span else 0
            end1 = end[i] + span if  end[i] + span < len(tokens[i]) else len(tokens[i])
            pre_ent.append(tokens[i][begin1:ben[i]])
            cur_ent.append(tokens[i][end[i]:end1])
            ent_idl.append(ent_id[i])
            shiti.append(ent[i])
            s_list.append(sen_index[i])
            b_list.append(ben[i])
            str_tok = ''.join(tokens[i][begin1:ben[i]]) 
            str_tok = re.sub('，|\s','',str_tok)
            cur_tok = ''.join(tokens[i][begin1:ben[i]]) 
            cur_tok = re.sub('，|\s','',cur_tok) 
            if re.findall(jiandu, str_tok):
                flag = 0               
            elif re.findall(zhaobiao, str_tok):
                flag = 1    
            elif re.findall(daili, str_tok):
                flag = 2  
#             elif re.findall(zhongbiao, str_tok):
#                 flag = 3   
 
            elif re.findall(shigong, str_tok):
                flag = 0
            elif re.findall(lianxi, str_tok):
                flag = 3                              
            else:
                flag = 0
            count += 1
            label.append(flag)
        else:
            pass    
     
    
    new_data = {'pre_ent':pre_ent, 'label':label,'cur_ent':cur_ent,  'entity_id':ent_idl, 'shiti':shiti, 'sentence_index':s_list, 'begin_index':b_list}   
    data_label = pd.DataFrame(new_data)
    data_label.to_excel(file2, encoding='utf-8', index=False, columns=['entity_id','sentence_index','begin_index','pre_ent','label','cur_ent','shiti'])
    with open('data/person_label.pk', 'wb') as f:
        pickle.dump(data_label, f)    

def post_data():
    '''
    @summary: 将标注好的数据推送到数据库
    '''
    conn = psycopg2.connect(dbname='BidiPro', user='postgres',password='postgres',host='192.168.2.101')
    cursor = conn.cursor()    
    table = 'person_label'
    cursor.execute(" select to_regclass('"+table+"') is null ")
    notExists = cursor.fetchall()[0][0]
    if notExists:
        cursor.execute(" create table "+table+" (entity_id text,label int)")
    else:
        cursor.execute(" delete from "+table)
    
    df3 = pd.read_excel('data/person_label.xls', header=0)
    df3.head(3)
    
    entity_id = df3['entity_id']
    label = df3['label']   
    for i in range(len(entity_id)):
                                                                  
        sql = " insert into "+table+"(entity_id,label) values('"+str(df3['entity_id'][i])+"',"+str(int(label[i]))+")"
        #print(sql)
        cursor.execute(sql)
    conn.commit()
    cursor.close()
    conn.close()    

if __name__ == '__main__':
    #get_data()
    label_data()
    post_data()