sports_scrapy/hg3535/spiders/jieshu.py

import datetime
import json
# import re
import logging

import redis
import scrapy

from ..items import Hgjieshu
from ..settings import R_HOST, R_PASSWORD, R_POST, R_DB


class HgjieshuSpider(scrapy.Spider):
    name = 'jieshu'
    to_day = datetime.datetime.now()
    allowed_domains = ['hg3535z.com']
    custom_settings = {
        "ITEM_PIPELINES": {
            'hg3535.pipeline.jieshu.Jieshuqiupipeline': 300,
        },
        # 'LOG_LEVEL': 'DEBUG',
        # 'LOG_FILE': "../hg3535/log/saiguo{}_{}_{}.log".format(to_day.year, to_day.month, to_day.day)
    }
    rls = redis.Redis(host=R_HOST, port=R_POST, db=R_DB, password=R_PASSWORD)

    def start_requests(self):
        match_ids = self.rls.smembers("hg3535.gunqiu.ids")
        if match_ids:
            for match_id in match_ids:
                match_id = match_id.decode()
                url = 'https://odata.jiushan6688.com/odds6i/d/getamodds/zh-cn/eid/{}/iip/true/ubt/am/isp/false'.format(match_id)
                yield scrapy.Request(url=url, callback=self.parse, dont_filter=True)

    def parse(self, response):
        logger = logging.getLogger(__name__)
        try:
            data = json.loads(response.text)
            status = data['i'][0]
            # if not status:
            ball = data['i'][31]
            match_id = data['i'][2]
            item = Hgjieshu()
            item['ball'] = ball
            item['match_id'] = match_id
            item['status'] = status
            yield item
            # else:
            #     return
        except Exception as e:
            logger.warning(e)
            return