LongArticlesJob/coldStartTasks/crawler/weixin_article_association.py

"""
@author: luojunhui
"""
import json
import time
from typing import AnyStr, List, Dict

from tqdm import tqdm

from applications import longArticlesMySQL, Functions, WeixinSpider
from applications.const import ArticleAssociationTaskConst

functions = Functions()
db_client = longArticlesMySQL()
spider = WeixinSpider()
const = ArticleAssociationTaskConst()


def get_good_articles() -> List[Dict]:
    """
    获取表现好的文章
    :return: List[Dict] 查询到到的文章列表
    """
    sql = f"""
        SELECT account_name, gh_id, view_count, read_rate, link, title
        FROM datastat_sort_strategy
        WHERE 
            type = 9 
            and position = 1 
            and date_str > '20241101' 
            and fans > 300000 
            and view_count > 5000
            and read_rate > 1.1
            and status = 1;
    """
    article_list = db_client.select_json(sql)
    return article_list


def get_recommend_article_list_for_each_article(account_name: AnyStr, article_url: AnyStr, title: AnyStr) -> List[Dict]:
    """
    获取推荐文章
    :param title: 种子标题
    :param account_name: 种子账号
    :param article_url: 种子文章链接
    :return:
    """
    recommend_response = spider.get_recommend_articles(content_link=article_url)
    if recommend_response['code'] == const.SPIDER_API_SUCCESS_CODE:
        recommend_article_list = recommend_response['data']['data']['list']
        filter_recommend_article_list = [
            {
                "seed_account_name": account_name,
                "seed_url": article_url,
                "seed_title": title,
                "recommend_title": recommend_article['title'],
                "recommend_account_name": recommend_article['nickname'],
                "recommend_gh_id": recommend_article['username'],
                "recommend_url": recommend_article['url'],
                "recommend_send_timestamp": recommend_article['send_time'],
                "recommend_read": recommend_article['read_num'],
                "recommend_like": recommend_article['old_like_num'],
                "recommend_index": recommend_article['idx'],
                "recommend_time": int(time.time())
            }
            for recommend_article in recommend_article_list if recommend_article['nickname'] != account_name
        ]
        return filter_recommend_article_list
    else:
        return []


def insert_recommend_list_into_meta(recommend_article_list: List[Dict]) -> None:
    """
    插入数据
    :param recommend_article_list:
    :return:
    """
    if not recommend_article_list:
        return
    for recommend_obj in recommend_article_list:
        try:
            insert_sql = f"""
                INSERT INTO crawler_meta_article
                (platform, mode, category, out_account_id, article_index, title, link, read_cnt, like_cnt, publish_time, crawler_time, status, unique_index, source_article_title, source_account)
                VALUES
                (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s)
            """
            db_client.update(
                insert_sql,
                params=(
                    "weixin",
                    "association",
                    "article_association",
                    recommend_obj['recommend_gh_id'],
                    recommend_obj['recommend_index'],
                    recommend_obj['recommend_title'],
                    recommend_obj['recommend_url'],
                    recommend_obj['recommend_read'],
                    recommend_obj['recommend_like'],
                    recommend_obj['recommend_send_timestamp'],
                    int(time.time()),
                    1,
                    functions.generateGzhId(url=recommend_obj['recommend_url']),
                    recommend_obj['seed_title'],
                    recommend_obj['seed_account_name'],
                )
            )
        except Exception as e:
            print("insert error", e)
            update_sql = f"""
                UPDATE crawler_meta_article
                SET
                    read_cnt = %s,
                    like_cnt = %s,
                    source_article_title = %s,
                    source_account = %s
                WHERE
                    unique_index = %s and category = %s;
            """
            try:
                db_client.update(
                    update_sql,
                    params=(
                        recommend_obj['recommend_read'],
                        recommend_obj['recommend_like'],
                        recommend_obj['seed_title'],
                        recommend_obj['seed_account_name'],
                        functions.generateGzhId(url=recommend_obj['recommend_url']),
                        "article_association",
                    )
                )
            except Exception as e:
                print("update error", e)


def do_i2i_crawler_task() -> None:
    """
    获取推荐文章
    :return:
    """
    article_list = get_good_articles()
    for article_detail_tuple in tqdm(article_list[:1], desc="article list"):
        account_name = article_detail_tuple['account_name']
        url = article_detail_tuple['link']
        title = article_detail_tuple['title']
        recommend_article_list = get_recommend_article_list_for_each_article(
            account_name=account_name,
            article_url=url,
            title=title
        )
        insert_recommend_list_into_meta(recommend_article_list)


def main():
    """
    主函数
    :return:
    """
    do_i2i_crawler_task()