qingdan/preprocess/processQd.py

import pandas as pd
import os
import copy
import re
path = "/Users/xiaopengzhang/Downloads/qingdan/JD_QingDanXM.csv"

a = pd.read_csv(path)
print(a[a['qdbh']=='080904019'])
def transform(bh):
    result = re.findall(r'\d{9}', bh)
    bh = result[0]
    info = []
    if bh[0:2] == '01':
        info.append("建筑与装饰工程")
        if bh[2:4] == '01':
            info.append("土石方工程")
        elif bh[2:4] == '02':
            info.append("地基处理与边坡支护工程")
        elif bh[2:4] == '03':
            info.append("桩基工程")
        elif bh[2:4] == '04':
            info.append("砌筑工程")
        elif bh[2:4] == '05':
            info.append("混凝土及钢筋混凝土工程")
        elif bh[2:4] == '06':
            info.append("金属结构工程")
        elif bh[2:4] == '07':
            info.append("木结构工程")
        elif bh[2:4] == '08':
            info.append("门窗工程")
        elif bh[2:4] == '09':
            info.append("屋面及防水工程")
        elif bh[2:4] == '10':
            info.append("保温、隔热、防腐工程")
        elif bh[2:4] == '11':
            info.append("楼地面装饰工程")
        elif bh[2:4] == '12':
            info.append("墙、柱面装饰与隔断、幕墙工程")
        elif bh[2:4] == '13':
            info.append("天棚工程")
        elif bh[2:4] == '14':
            info.append("油漆、涂料、裱糊工程")
        elif bh[2:4] == '15':
            info.append("其他装饰工程")
        elif bh[2:4] == '16':
            info.append("拆除工程")
        elif bh[2:4] == '17':
            info.append("建筑工程措施项目")
    elif bh[0:2] == '03':
        info.append("安装工程")
    elif bh[0:2] == '04':
        info.append("市政工程")
    elif bh[0:2] == '05':
        info.append("园林绿化工程")
    else:
        info.append("")
    

parent = []
for i in range(len(a)):
    print(i)
    print(a.iloc[i])
    row = a.iloc[i]
    if row['fbcch'].item() == 4:
        series = []
        j = i - 1
        while j > -1 and a.iloc[j]['fbcch'].item() != 3:
            j = j - 1
        if j == -1:
            raise Exception("")
        series.append(a.iloc[j]['xmmc'])
        k = j - 1
        while k > -1 and a.iloc[k]['fbcch'].item() != 2:
            k = k - 1
        if k == -1:
            raise Exception("")
        series.append(a.iloc[k]['xmmc'])
        l = k - 1
        while l > -1 and a.iloc[l]['fbcch'].item() != 1:
            l = l - 1
        if l == -1:
            raise Exception("")
        series.append(a.iloc[l]['xmmc'])
        series.reverse()
        parent.append(" ".join(series))
    else:
        parent.append("")    
new_ = pd.Series(parent, name="parent")
df = pd.concat([a, new_.to_frame()], axis=1)
df.to_csv("JD_QingDanXM_parent.csv")