mongo2csv

#!/usr/bin/env python3import pymongo
import datetime
import osyesterday &＃61; str(datetime.date.today() - datetime.timedelta(days&＃61;1))def mongo2csv(db_name, ts_string&＃61;yesterday, column_handle&＃61;None, column_delete&＃61;None):&＃39;&＃39;&＃39;:param db_name: 表名:param ts_string: 日期:param column_handle: 需要处理的列&＃xff0c;格式{k:v} ,k:需要处理的列&＃xff0c;v&＃xff1a;1.数字&＃xff0c;即列表索引&＃xff0c;取列表的第几个元素2.字符串&＃xff0c;即字典key&＃xff0c;取子字典的某个value3.join &＃xff0c;即需要将列表转为字符串:param column_delete: 需要删除的列:return:&＃39;&＃39;&＃39;print(&＃39;starting...&＃39;)host &＃61; &＃39;127.0.0.1&＃39;port &＃61; 5600user &＃61; &＃39;root&＃39;password &＃61; &＃39;&＃39;url &＃61; &＃39;mongodb://&＃39; &＃43; user &＃43; &＃39;:&＃39; &＃43; password &＃43; &＃39;&＃64;&＃39; &＃43; host &＃43; &＃39;:&＃39; &＃43; str(port) &＃43; &＃39;/&＃39;client &＃61; pymongo.MongoClient(url)db &＃61; client.rentalcollection &＃61; db[db_name]select_condition &＃61; {&＃39;_id&＃39;: 0, &＃39;ts&＃39;: 0}if not column_delete:column_delete &＃61; []if not column_handle:column_handle &＃61; {}if len(column_delete) > 0:for temp in column_delete:select_condition[temp] &＃61; 0title_dic &＃61; collection.find_one({&＃39;ts_string&＃39;: ts_string}, select_condition)title &＃61; sorted(title_dic.keys())con &＃61; collection.find({&＃39;ts_string&＃39;: ts_string}, select_condition)file_name &＃61; db_name &＃43; &＃39;_&＃39; &＃43; yesterday &＃43; &＃39;.csv&＃39;with open(file_name, &＃39;a&＃39;, encoding&＃61;&＃39;utf8&＃39;) as f:f.write(&＃39;,&＃39;.join(title) &＃43; &＃39;\n&＃39;)for item in con:for k, v in column_handle.items():item[k] &＃61; item[k][v] if v !&＃61; &＃39;join&＃39; else (&＃39;|&＃39;.join([str(x) for x in item[k]]) if isinstance(item[k], list) else item[k])f.write(&＃39;,&＃39;.join([str(item[x]) for x in title]) &＃43; &＃39;\n&＃39;)if __name__ &＃61;&＃61; &＃39;__main__&＃39;: mongo2csv(&＃39;lianjia_detail&＃39;) mongo2csv(&＃39;mogu_detail&＃39;, column_handle&＃61;{&＃39;metroInfo&＃39;: 0, &＃39;rentType&＃39;: &＃39;value&＃39;}) mongo2csv(&＃39;qingke_detail&＃39;, column_handle&＃61;{}) mongo2csv(&＃39;xiangyu_detail&＃39;, column_delete&＃61;[&＃39;endDate&＃39;,&＃39;vacantStartDate&＃39;,&＃39;tabList&＃39;,&＃39;vacantEndDate&＃39;]) mongo2csv(&＃39;ziru_detail&＃39;, column_handle&＃61;{&＃39;subway_line_code&＃39;: &＃39;join&＃39;, &＃39;subway_station_code&＃39;: &＃39;join&＃39;}) print(os.system(&＃39;wc -l *.csv&＃39;))