mongodb 的js脚本或pymongodb脚本修改数据库的字段值

Posted 2020-06-13

使用

data$ mongo localhost:27017/jd_51job_raw updateName.js --shell

js 脚本: updateName.js

var cursor = db.getCollectionNames();
for(i=0;i<cursor.length;i++){
    if("system.indexes"==cursor[i]||"system.profile"==cursor[i]){
        print("Find system");
        continue;
    }
    else{
        curCol = db.getCollection(cursor[i]);
        print("now processing...",curCol);
        print(curCol.count());
        curCol.find().forEach(function(item){
            curCol.update({"_id":item._id},{"$set":{"jdJob.jobDiploma":item.jobDiploma}});
//            curCol.update({"_id":item._id},{"$unset":{"jobDiploma":0}});
        });
        printjson(curCol.findOne());

    }
        
}

updateName.py

#!/usr/bin/env python
# coding=utf-8


from pymongo import MongoClient
from bson.json_util import dumps

from multiprocessing import Pool

client = MongoClient("mongodb://localhost:xxxxx/")
db = client.jd_51job_raw

cols = db.collection_names(include_system_collections=False)
cols = sorted(cols)

def del_jobDiploma(colName):
    print "=="*20,colName
    col = db.get_collection(colName)
    col.update({},{"$unset":{"jobDiploma":""}},multi=True)
    print dumps(col.find_one(),ensure_ascii=False,indent=4)
    print ""


# 10个进程一起来！
pools = Pool(10)
pools.map(del_jobDiploma,cols)