LongArticleTaskServer/applications/tasks/analysis_task/account_position_info.py

import asyncio
import traceback

import numpy as np
from collections import defaultdict
from typing import Dict, List
from pandas import DataFrame
from scipy import stats
from tqdm.asyncio import tqdm


from datetime import datetime, timedelta


class AccountPositionInfoConst:
    # 阅读率统计周期（秒）
    STATISTICS_PERIOD = 31 * 24 * 60 * 60
    # 一天的秒数
    ONE_DAY_IN_SECONDS = 60 * 60 * 24
    # 相对变化率阈值
    RELATIVE_VALUE_THRESHOLD = 0.1
    # 发文类型
    UNLIMITED_PUBLISH_TYPE = 10002
    BULK_PUBLISH_TYPE = 9
    # 文章位置
    ARTICLE_INDEX_LIST = [1, 2, 3, 4, 5, 6, 7, 8]

    # 默认粉丝
    DEFAULT_FANS = 0

    # 最低粉丝量
    MIN_FANS = 1000

    ARTICLES_DAILY = 1
    TOULIU = 2

    # 默认点赞
    DEFAULT_LIKE = 0

    # 状态
    USING_STATUS = 1
    NOT_USING_STATUS = 0

    # 服务号
    GROUP_ACCOUNT_SET = {
        "gh_9cf3b7ff486b",
        "gh_ecb21c0453af",
        "gh_45beb952dc74",
        # "gh_84e744b16b3a",
        "gh_b3ffc1ca3a04",
        "gh_b8baac4296cb",
        "gh_efaf7da157f5",
        # "gh_5855bed97938",
        "gh_b32125c73861",
        "gh_761976bb98a6",
        "gh_5e543853d8f0",
        # "gh_61a72b720de3",
    }

    # 违禁账号
    FORBIDDEN_GH_IDS = {
        "gh_4c058673c07e",
        "gh_de9f9ebc976b",
        "gh_7b4a5f86d68c",
        "gh_f902cea89e48",
        "gh_789a40fe7935",
        "gh_cd041ed721e6",
        "gh_62d7f423f382",
        "gh_043223059726",
        "gh_6cfd1132df94",
        "gh_7f5075624a50",
        "gh_d4dffc34ac39",
        "gh_c69776baf2cd",
        "gh_9877c8541764",
        "gh_ac43e43b253b",
        "gh_93e00e187787",
        "gh_080bb43aa0dc",
        "gh_b1c71a0e7a85",
        "gh_d5f935d0d1f2",
    }

    # 投流账号
    TOULIU_ACCOUNTS = {
        "小阳看天下",
        "趣味生活方式",
        "趣味生活漫时光",
        "史趣探秘",
        "暖心一隅",
        "趣味生活漫谈",
        "历史长河流淌",
        "美好意义时光",
        "银发生活畅谈",
        "美好时光阅读汇",
        "时光趣味生活",
    }


class AccountPositionReadRateAvg(AccountPositionInfoConst):
    """计算账号每个位置评价阅读率"""

    def __init__(self, pool, log_client, trace_id):
        self.pool = pool
        self.log_client = log_client
        self.trace_id = trace_id

    # 生成统计周期
    def generate_stat_duration(self, end_date: str) -> str:
        end_date_dt = datetime.strptime(end_date, "%Y-%m-%d")
        start_date_dt = end_date_dt - timedelta(seconds=self.STATISTICS_PERIOD)
        return start_date_dt.strftime("%Y-%m-%d")

    # 获取发文账号
    async def get_publishing_accounts(self):
        query = """
            select distinct
                t3.name as account_name,
                t3.gh_id as gh_id,
                group_concat(distinct t4.remark) as account_remark,
                t6.account_source_name as account_source,
                t6.mode_type as mode_type,
                t6.account_type as account_type,
                t6.`status` as status
            from
                publish_plan t1
                join publish_plan_account t2 on t1.id = t2.plan_id
                join publish_account t3 on t2.account_id = t3.id
                left join publish_account_remark t4 on t3.id = t4.publish_account_id
                left join wx_statistics_group_source_account t5 on t3.id = t5.account_id
                left join wx_statistics_group_source t6 on t5.group_source_name = t6.account_source_name
            where t1.plan_status = 1 and t1.content_modal = 3 and t3.channel = 5
            group by t3.id;
        """
        account_list = await self.pool.async_fetch(query, db_name="aigc")
        return [i for i in account_list if "自动回复" not in str(i["account_remark"])]

    # 获取统计周期内，每个账号的粉丝量
    async def get_fans_for_each_date(self, start_date: str):
        # 获取订阅号粉丝量
        query = """
            SELECT t1.date_str as dt, 
                   CASE 
                        WHEN t1.fans_count IS NULL OR t1.fans_count = 0 THEN t2.follower_count
                        ELSE t1.fans_count
                    END	AS fans, 
                   t2.gh_id as gh_id
            FROM datastat_wx t1 JOIN publish_account t2 ON t1.account_id = t2.id
            WHERE t2.channel = 5 AND t2.status = 1 AND t1.date_str >= %s;
        """
        task1 = self.pool.async_fetch(query=query, db_name="aigc", params=(start_date,))

        if self.GROUP_ACCOUNT_SET:
            gh_ids = tuple(self.GROUP_ACCOUNT_SET)
            placeholders = ",".join(["%s"] * len(gh_ids))
            query_group = f"""
                    SELECT gh_id, publish_date AS dt, CAST(SUM(sent_count) / 8 AS SIGNED) AS fans
                    FROM long_articles_group_send_result
                    WHERE publish_date >= %s AND gh_id IN ({placeholders})
                    GROUP BY publish_date, gh_id;
            """
            params_group = (start_date, *gh_ids)
            task2 = self.pool.async_fetch(query=query_group, params=params_group)
        else:
            # 没有 group 账号，返回空列表
            task2 = asyncio.sleep(0, result=[])

        account_with_fans, group_account_with_fans = await asyncio.gather(task1, task2)

        # 合并粉丝数据
        account_dt_fans_mapper: Dict[str, Dict[str, int]] = defaultdict(dict)

        # 订阅号
        for item in account_with_fans or []:
            gh_id = item["gh_id"]
            dt = item["dt"]
            fans = int(item.get("fans") or 0)
            account_dt_fans_mapper[gh_id][dt] = fans

        # 服务号（覆盖相同 gh_id + dt）
        for item in group_account_with_fans or []:
            gh_id = item["gh_id"]
            dt = item["dt"]
            fans = int(item.get("fans") or 0)
            account_dt_fans_mapper[gh_id][dt] = fans

        return account_dt_fans_mapper

    # 从数据库获取账号群发文章 && 群发数据
    async def get_single_account_published_articles(
        self, gh_id: str, start_timestamp: int
    ):
        query = """
            SELECT 
                ghId as gh_id, accountName as account_name,
                ItemIndex as position, 
                CAST(AVG(show_view_count) AS SIGNED) as read_count,
                FROM_UNIXTIME(publish_timestamp, '%%Y-%%m-%%d') AS pub_dt
            FROM 
                official_articles_v2
            WHERE 
                ghId = %s and Type = %s and publish_timestamp >= %s
            GROUP BY ghId, accountName, ItemIndex, pub_dt;
        """
        return await self.pool.async_fetch(
            query=query,
            db_name="piaoquan_crawler",
            params=(gh_id, self.BULK_PUBLISH_TYPE, start_timestamp),
        )

    # 计算单个账号的每篇文章的阅读率
    async def cal_read_rate_for_single_account(
        self,
        publish_details: List[Dict],
        gh_id: str,
        fans_mapper: Dict[str, Dict[str, int]],
    ) -> DataFrame | None:
        if not publish_details:
            return None

        article_list_with_fans = []
        for article in publish_details:
            fans = fans_mapper.get(gh_id, {}).get(article["pub_dt"], self.DEFAULT_FANS)

            if not fans:
                print(
                    f"账号 {article['account_name']} 在 {article['pub_dt']} 没有粉丝数据"
                )
                continue

            article["fans"] = fans
            if fans > self.MIN_FANS:
                article["read_rate"] = article["read_count"] / fans if fans else 0
                article_list_with_fans.append(article)

        # 转化为 DataFrame 方便后续处理
        return DataFrame(
            article_list_with_fans,
            columns=[
                "gh_id",
                "account_name",
                "position",
                "read_count",
                "pub_dt",
                "fans",
                "read_rate",
            ],
        )

    # 更新账号阅读率均值并且更新数据库
    async def update_read_rate_avg_for_each_account(
        self,
        account: dict,
        start_date: str,
        end_dt: str,
        df: DataFrame,
        fans_dict: Dict[str, int],
    ):
        avg_date = (datetime.strptime(end_dt, "%Y-%m-%d") - timedelta(days=1)).strftime(
            "%Y-%m-%d"
        )

        insert_error_list = []
        for index in self.ARTICLE_INDEX_LIST:
            # 过滤
            filter_df = df[
                (df["position"] == index)
                & (df["pub_dt"] < end_dt)
                & (df["pub_dt"] >= start_date)
            ]
            read_average = filter_df["read_count"].mean()
            read_std = filter_df["read_count"].std()

            output_df = filter_df[
                (filter_df["read_count"] > read_average - 2 * read_std)
                & (filter_df["read_count"] < read_average + 2 * read_std)
            ]

            records = len(output_df)
            if records:
                # todo: 需要检查波动
                # if index <= 2:
                #     print("position need to be checked")
                # insert
                try:
                    insert_query = """
                        INSERT INTO long_articles_read_rate
                            (account_name, gh_id, position, read_rate_avg, remark, articles_count, earliest_publish_time, latest_publish_time, dt_version, is_delete, fans)
                        VALUES 
                        (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s);
                    """
                    await self.pool.async_save(
                        query=insert_query,
                        params=(
                            account["account_name"],
                            account["gh_id"],
                            index,
                            output_df["read_rate"].mean(),
                            "从 {} 开始往前计算 31  天".format(start_date),
                            records,
                            output_df["pub_dt"].min(),
                            output_df["pub_dt"].max(),
                            avg_date.replace("-", ""),
                            0,
                            fans_dict.get(avg_date, 0),
                        ),
                    )

                except Exception as e:
                    insert_error_list.append(str(e))

    # 入口函数
    async def deal(self, end_date: str | None):
        if not end_date:
            end_date = datetime.now().strftime("%Y-%m-%d")

        start_dt = self.generate_stat_duration(end_date)

        fans_mapper = await self.get_fans_for_each_date(start_date=start_dt)

        accounts = await self.get_publishing_accounts()
        for account in tqdm(accounts, desc="计算单个账号阅读率均值"):
            if account["gh_id"] in self.FORBIDDEN_GH_IDS:
                continue

            published_articles = await self.get_single_account_published_articles(
                gh_id=account["gh_id"],
                start_timestamp=int(
                    datetime.strptime(start_dt, "%Y-%m-%d").timestamp()
                ),
            )
            article_dataframe = await self.cal_read_rate_for_single_account(
                publish_details=published_articles,
                gh_id=account["gh_id"],
                fans_mapper=fans_mapper,
            )
            if article_dataframe is None:
                continue

            if article_dataframe.empty:
                continue

            await self.update_read_rate_avg_for_each_account(
                account=account,
                start_date=start_dt,
                end_dt=end_date,
                df=article_dataframe,
                fans_dict=fans_mapper.get(account["gh_id"], {}),
            )


class AccountPositionReadAvg(AccountPositionReadRateAvg):
    # 计算阅读均值置信区间上限
    async def cal_read_avg_ci_upper(self, gh_id: str, index: int):
        fetch_query = f"""
            select read_avg, update_time
            from account_avg_info_v3
            where gh_id = %s and position = %s 
            order by update_time desc limit 30;
        """
        fetch_response_list = await self.pool.async_fetch(
            query=fetch_query, db_name="piaoquan_crawler", params=(gh_id, index)
        )
        read_avg_list = [i["read_avg"] for i in fetch_response_list]
        n = len(read_avg_list)
        mean = np.mean(read_avg_list)
        std = np.std(read_avg_list, ddof=1)
        se = std / np.sqrt(n)
        t = stats.t.ppf(0.975, df=n - 1)
        upper_t = mean + t * se
        return upper_t

    # 获取账号的阅读率均值信息
    async def get_accounts_read_avg(self, dt):
        query = """
            select gh_id, position, fans, read_rate_avg, fans * read_rate_avg as read_avg
            from long_articles_read_rate
            where dt_version = %s
        """
        fetch_result = await self.pool.async_fetch(query=query, params=(dt.replace("-", ""),))
        response = {}
        for item in fetch_result:
            key = f"{item['gh_id']}_{item['position']}"
            response[key] = {
                "read_rate_avg": item["read_rate_avg"],
                "read_avg": item["read_avg"],
                "fans": item["fans"],
            }
        return response

    # 计算阅读均值置信区间上限
    async def cal_read_avg_detail(
        self, account: Dict, dt: str, account_with_read_rate_avg: Dict
    ):
        for index in self.ARTICLE_INDEX_LIST:
            key = f"{account['gh_id']}_{index}"
            print(key)
            if account_with_read_rate_avg.get(key) is None:
                continue

            read_avg = account_with_read_rate_avg[key]["read_avg"]

            # 计算阅读均值置信区间上限
            read_avg_ci_upper = await self.cal_read_avg_ci_upper(
                gh_id=account["gh_id"], index=index
            )

            await self.process_each_record(
                account=account,
                index=index,
                fans=account_with_read_rate_avg[key]["fans"],
                read_rate_avg=account_with_read_rate_avg[key]["read_rate_avg"],
                read_avg=read_avg,
                read_avg_ci_upper=read_avg_ci_upper,
                dt=dt,
            )

    async def process_each_record(
        self, account, index, fans, read_rate_avg, read_avg, read_avg_ci_upper, dt
    ):
        gh_id = account["gh_id"]
        account_name = account["account_name"]
        business_type = (
            self.TOULIU if account_name in self.TOULIU_ACCOUNTS else self.ARTICLES_DAILY
        )
        # insert into database
        insert_sql = f"""
            insert into account_avg_info_v3
            (gh_id, position, update_time, account_name, fans, read_avg, like_avg, status, account_type, 
            account_mode, account_source, account_status, business_type, read_rate_avg, read_avg_ci_upper)
            values
            (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s);
        """
        try:
            await self.pool.async_save(
                query=insert_sql,
                db_name="piaoquan_crawler",
                params=(
                    gh_id,
                    index,
                    dt,
                    account["account_name"],
                    fans,
                    read_avg,
                    self.DEFAULT_LIKE,
                    self.USING_STATUS,
                    account["account_type"],
                    account["mode_type"],
                    account["account_source"],
                    account["status"],
                    business_type,
                    read_rate_avg,
                    read_avg_ci_upper,
                ),
            )
        except Exception as e:
            print(e)
            update_sql = f"""
                update account_avg_info_v3
                set fans = %s, read_avg = %s, read_rate_avg = %s, read_avg_ci_upper = %s
                where gh_id = %s and position = %s and update_time = %s
            """
            try:
                await self.pool.async_save(
                    query=update_sql,
                    db_name="piaoquan_crawler",
                    params=(
                        fans,
                        read_avg,
                        read_rate_avg,
                        read_avg_ci_upper,
                        account["gh_id"],
                        index,
                        dt,
                    ),
                )
            except Exception as e:
                print(e)

        # 修改前一天的状态为 0
        update_status_sql = f"""
            UPDATE account_avg_info_v3
            SET status = %s
            WHERE update_time != %s AND gh_id = %s AND position = %s;
        """
        await self.pool.async_save(
            query=update_status_sql,
            db_name="piaoquan_crawler",
            params=(self.NOT_USING_STATUS, dt, gh_id, index),
        )

    async def deal(self, end_date: str | None):
        if not end_date:
            end_date = datetime.now().strftime("%Y-%m-%d")

        dt = (datetime.strptime(end_date, "%Y-%m-%d") - timedelta(days=1)).strftime(
            "%Y-%m-%d"
        )
        account_with_read_rate_avg = await self.get_accounts_read_avg(dt)
        accounts = await self.get_publishing_accounts()

        for account in tqdm(accounts, desc="计算单个账号的阅读均值"):
            if account["gh_id"] in self.FORBIDDEN_GH_IDS:
                continue

            try:
                await self.cal_read_avg_detail(
                    account=account,
                    dt=dt,
                    account_with_read_rate_avg=account_with_read_rate_avg,
                )
            except Exception as e:
                print(f"计算账号 {account['account_name']} 阅读均值失败 : {e}")
                print(traceback.format_exc())


class AccountPositonOpenRateAvg(AccountPositionReadRateAvg):
    pass