import pymysql
from loguru import logger

class MongoToMysql:
    def __init__(self, mongo_host, mongo_port, mongo_db, mongo_collection, mysql_host, mysql_port, mysql_user,
                 mysql_password, mysql_db,table_name=None,set_max_length=False,batch_size=10000,table_description=''):
        self.mongo_host = mongo_host
        self.mongo_port = mongo_port
        self.mongo_db = mongo_db
        self.mongo_collection = mongo_collection
        self.mysql_host = mysql_host
        self.mysql_port = mysql_port
        self.mysql_user = mysql_user
        self.mysql_password = mysql_password
        self.mysql_db = mysql_db
        self.table_name = table_name
        self.set_max_length = set_max_length
        self.batch_size = batch_size
        self.table_description = table_description
        self.data_types = self.get_mongo_data_types()
        self.create_mysql_table(self.data_types,set_max_length= self.set_max_length,table_description=self.table_description)
        self.push_data_to_mysql(self.batch_size)

    def get_mongo_data_types(self):
        logger.debug('正在獲取mongo中字段的類(lèi)型！')
        client = pymongo.MongoClient(host=self.mongo_host, port=self.mongo_port)
        db = client[self.mongo_db]
        collection = db[self.mongo_collection]
        data_types = {}
        for field in collection.find_one().keys():
            data_types[field] = type(collection.find_one()[field]).__name__
        return data_types

    def check_mysql_table_exists(self):
        logger.debug('檢查是否存在該表，有則刪之！')
        conn = pymysql.connect(host=self.mysql_host, port=self.mysql_port, user=self.mysql_user,
                               password=self.mysql_password, db=self.mysql_db)
        cursor = conn.cursor()
        sql = f"DROP TABLE IF EXISTS {self.mongo_collection}"
        cursor.execute(sql)
        conn.commit()
        conn.close()

    def get_max_length(self, field):
        logger.debug(f'正在獲取字段 {field} 最大長(zhǎng)度......')
        client = pymongo.MongoClient(host=self.mongo_host, port=self.mongo_port)
        db = client[self.mongo_db]
        collection = db[self.mongo_collection]
        max_length = 0
        for item in collection.find({},{field:1,'_id':0}):
            value = item.get(field)
            if isinstance(value, str) and len(value) > max_length:
                max_length = len(value)
        return max_length

    def create_mysql_table(self, data_types,set_max_length,table_description):
        logger.debug('正在mysql中創(chuàng)建表結(jié)構(gòu)！')
        self.check_mysql_table_exists()
        conn = pymysql.connect(host=self.mysql_host, port=self.mysql_port, user=self.mysql_user,
                               password=self.mysql_password, db=self.mysql_db)
        cursor = conn.cursor()
        if self.table_name:
            table_name = self.table_name
        else:
            table_name = self.mongo_collection
        fields = []
        for field, data_type in data_types.items():
            if data_type == 'int':
                fields.append(f"{field} INT")
            elif data_type == 'float':
                fields.append(f"{field} FLOAT")
            elif data_type == 'bool':
                fields.append(f"{field} BOOLEAN")
            else:
                if set_max_length:
                    fields.append(f"{field} TEXT)")
                else:
                    max_length = self.get_max_length(field)
                    fields.append(f"{field} VARCHAR({max_length + 200})")
        fields_str = ','.join(fields)
        sql = f"CREATE TABLE {table_name} (id INT PRIMARY KEY AUTO_INCREMENT,{fields_str}) COMMENT='{table_description}'"
        cursor.execute(sql)
        conn.commit()
        conn.close()


    def push_data_to_mysql(self, batch_size=10000):
        logger.debug('--- 正在準(zhǔn)備從mongo中每次推送10000條數(shù)據(jù)到mysql ----')
        client = pymongo.MongoClient(host=self.mongo_host, port=self.mongo_port)
        db = client[self.mongo_db]
        collection = db[self.mongo_collection]
        conn = pymysql.connect(host=self.mysql_host, port=self.mysql_port, user=self.mysql_user,
                               password=self.mysql_password, db=self.mysql_db)
        cursor = conn.cursor()
        if self.table_name:
            table_name = self.table_name
        else:
            table_name = self.mongo_collection
        # table_name = self.mongo_collection
        data = []
        count = 0
        for item in collection.find():
            count += 1
            row = []
            for field, data_type in self.data_types.items():
                value = item.get(field)
                if value is None:
                    row.append(None)
                elif data_type == 'int':
                    row.append(str(item.get(field, 0)))
                elif data_type == 'float':
                    row.append(str(item.get(field, 0.0)))
                elif data_type == 'bool':
                    row.append(str(item.get(field, False)))
                else:
                    row.append(str(item.get(field, '')))
            data.append(row)
            if count % batch_size == 0:
                placeholders = ','.join(['%s'] * len(data[0]))
                sql = f"INSERT INTO {table_name} VALUES (NULL,{placeholders})"
                cursor.executemany(sql, data)
                conn.commit()
                data = []
                logger.debug(f'--- 已完成推送：{count} 條數(shù)據(jù)！ ----')
        if data:
            placeholders = ','.join(['%s'] * len(data[0]))
            sql = f"INSERT INTO {table_name} VALUES (NULL,{placeholders})"
            cursor.executemany(sql, data)
            conn.commit()
            logger.debug(f'--- 已完成推送：{count} 條數(shù)據(jù)！ ----')
        conn.close()


if __name__ == '__main__':
    """MySQL"""
    mongo_host = '127.0.0.1'
    mongo_port = 27017
    mongo_db = 'db_name'
    mongo_collection = 'collection_name'

    """MongoDB"""
    mysql_host = '127.0.0.1'
    mysql_port = 3306
    mysql_user = 'root'
    mysql_password = '123456'
    mysql_db = 'mysql_db'

    table_description = ''  # 表描述

    mongo_to_mysql = MongoToMysql(mongo_host, mongo_port, mongo_db, mongo_collection, mysql_host, mysql_port,
                                  mysql_user, mysql_password, mysql_db,table_description=table_description)


    #
    # table_name = None  # 默認(rèn)為None 則MySQL中的表名跟Mongo保持一致
    # set_max_length = False # 默認(rèn)為False 根據(jù)mongo中字段最大長(zhǎng)度 加200 來(lái)設(shè)置字段的VARCHART長(zhǎng)度 , 否則定義TEXT類(lèi)型
    # batch_size = 10000   # 控制每次插入數(shù)據(jù)量的大小
    # table_description = '' # 表描述
    # mongo_to_mysql = MongoToMysql(mongo_host, mongo_port, mongo_db, mongo_collection, mysql_host, mysql_port,
    #                               mysql_user, mysql_password, mysql_db,table_name,set_max_length,batch_size,table_description)