collect/wizard/wizard_sync_news_data.py

# -*- coding: UTF-8 -*-
from datetime import datetime

from odoo import models, fields, _
from pymongo import MongoClient

from odoo.exceptions import UserError


class SyncNewsData(models.TransientModel):
    _name = 'auto.news.sync'
    _description = 'Auto News Sync'

    message = fields.Text('Message', default='确认同步数据?', readonly=True)

    def btn_sync_data(self):
        try:
            client = MongoClient('mongodb://root:aaaAAA111!!!@erhe.top:38000/')

            # 指定数据库名称
            db_name = 'NEWS'  # 替换为你的数据库名称

            # 选择数据库
            db = client[db_name]

            # 列出数据库中的所有集合
            collections = db.list_collection_names()

        except Exception as e:
            raise UserError('连接数据库失败: {}'.format(e))

        all_data = []

        for collection_name in collections:
            # 选择集合
            collection = db[collection_name]

            # 读取集合中的所有数据
            for document in collection.find({}, {'_id': 0}):
                all_data.append(document)

        sorted_data = []

        if all_data:
            sorted_data = sorted(all_data, key=lambda x: x['create_time'], reverse=True)

        for doc in sorted_data:
            news_data_id = self.env['news.info'].search([('name', '=', doc['title'])], limit=1)

            if news_data_id:
                continue

            create_time_dt = None
            if doc.get('create_time'):
                create_time_dt = datetime.utcfromtimestamp(doc['create_time'])

            news_data = news_data_id.create({
                'name': doc.get('title'),
                'context': doc.get('context') or '',
                'source_url': doc.get('source_url') or '',
                'link': doc.get('line') or '',
                'article_type': doc.get('article_type') or '',
                'article_source': doc.get('article_source') or '',
                'img_url': doc.get('img_url') or '',
                'keyword': doc.get('keyword') or '',
                'posted_date': doc.get('posted_date') or '',
                'create_time_ts': doc.get('create_time') or '',
                'create_time': create_time_dt,
                'create_datetime': datetime.strptime(doc['create_datetime'], '%Y-%m-%d %H:%M:%S')
            })