"""
@author: luojunhui
"""

from pandas import DataFrame

from applications import bot
from applications.const import ColdStartTaskConst
from applications.db import DatabaseConnector
from config import long_articles_config

from coldStartTasks.publish.basic import filter_by_status
from coldStartTasks.publish.basic import filter_by_sensitive_words
from coldStartTasks.publish.basic import filter_by_title_length
from coldStartTasks.publish.basic import update_published_articles_status
from coldStartTasks.publish.basic import get_article_from_meta_table
from coldStartTasks.publish.basic import update_article_status_after_publishing
from coldStartTasks.publish.basic import create_crawler_plan
from coldStartTasks.publish.basic import insert_into_article_crawler_plan
from coldStartTasks.publish.basic import bind_to_generate_plan

const = ColdStartTaskConst()


def filter_articles_before_create_plan(article_df: DataFrame) -> DataFrame:
    """
    filter articles before create plan
    """
    total_length = article_df.shape[0]

    # filter by status
    filter_df = filter_by_status(article_df)
    filter_length0 = filter_df.shape[0]

    # filter by sensitive words
    filter_df = filter_by_sensitive_words(filter_df)
    filter_length1 = filter_df.shape[0]

    # filter by title length
    filter_df = filter_by_title_length(filter_df)
    filter_length2 = filter_df.shape[0]

    bot(
        title="文章联想任务，开始创建抓取计划",
        detail={
            "文章总数": total_length,
            "发布状态过滤": "过滤： {}, 剩余： {}".format(
                total_length - filter_length0, filter_length0
            ),
            "敏感词过滤": "过滤： {}, 剩余： {}".format(
                filter_length0 - filter_length1, filter_length1
            ),
            "标题长度过滤": "过滤： {}, 剩余： {}".format(
                filter_length1 - filter_length2, filter_length2
            ),
        },
        mention=False,
    )

    return filter_df


class ArticleAssociationPublish(object):
    """
    publish i2i articles
    """

    def __init__(self):
        self.db_client = DatabaseConnector(db_config=long_articles_config)
        self.db_client.connect()

    def deal(self):
        """
        class entrance
        """
        # update published articles
        update_published_articles_status(db_client=self.db_client)

        # get data from meta table
        article_dataframe = get_article_from_meta_table(
            db_client=self.db_client, category="article_association", platform="weixin"
        )

        # fileter articles
        filter_dataframe = filter_articles_before_create_plan(article_dataframe)

        # create crawler plan
        url_list = filter_dataframe["link"].values.tolist()
        if url_list:
            crawler_plan_id, crawler_plan_name, create_timestamp = create_crawler_plan(
                url_list=url_list, plan_tag="article_association", platform="weixin"
            )

            # insert crawler plan
            insert_into_article_crawler_plan(
                db_client=self.db_client,
                crawler_plan_id=crawler_plan_id,
                crawler_plan_name=crawler_plan_name,
                create_timestamp=create_timestamp,
            )

            # bind to generate plan
            bind_to_generate_plan(
                category="article_association",
                crawler_plan_id=crawler_plan_id,
                crawler_plan_name=crawler_plan_name,
                platform="weixin",
            )

            # update status
            article_id_list = filter_dataframe["article_id"].values.tolist()
            update_article_status_after_publishing(
                db_client=self.db_client, article_id_list=article_id_list
            )

            bot(
                title="文章联想任务，创建抓取计划成功",
                detail={
                    "抓取计划id": crawler_plan_id,
                    "抓取计划名称": crawler_plan_name,
                    "抓取条数": len(url_list),
                    "冷启动类型": "article_association",
                },
                mention=False,
            )