sports_scrapy/hg3535/spiders/guanjun.py

# -*- coding: utf-8 -*-
# import copy
import datetime
import json

import scrapy
from scrapy.http import Request

from .. items import Guanjun
# from scrapy_redis.spiders import RedisSpider


class LqguanjunSpider(scrapy.Spider):
    name = 'guanjun'
    allowed_domains = ['hg3535z.com']
    to_day = datetime.datetime.now()
    # redis_key = "guanjun: start_urls"
    custom_settings = {
        "ITEM_PIPELINES": {
            'hg3535.pipeline.guanjun.Guanjunpipeline': 300,
        },
        # 'LOG_LEVEL': 'DEBUG',
        # 'LOG_FILE': "./log/guanjun_{}_{}_{}.log".format(to_day.year, to_day.month, to_day.day)
    }

    def start_requests(self):
        for z in range(1, 5):
            for x in range(0, 4):
                # for y in range()
                url = 'https://odata.jiushan6688.com/odds6i/d/getodds/zh-cn/sid/{}/pt/1/ubt/or/pn/{}/sb/2/dc/null/pid/0'.format(z, x)
                # url = 'https://odata.jiushan6688.com/odds6i/d/getodds/zh-cn/sid/1/pt/1/ubt/or/pn/1/sb/2/dc/null/pid/0'
                # redis_key = "guanjun: start_urls"
                # https://odata.jiushan6688.com/odds6i/d/getodds/zh-cn/sid/1/pt/1/ubt/or/pn/0/sb/2/dc/null/pid/0
                yield Request(url=url, callback=self.parse, dont_filter=True)

    def parse(self, response):
        try:
            datas = json.loads(response.text)
        except:
            datas = ""
        try:
            results = datas['n-ot']['egs']
        except:
            results = ""
        try:
            ball = datas['n-ot']['s']['k']
        except:
            ball = ""
        if results:
            for result in results:
                # 联赛id
                league_id = result['c']['k']
                # 联赛名
                league_name = result['c']['n']
                new_results = result['es']
                for new_result in new_results:
                    date_game = new_result['i'][4]
                    time_game = new_result['i'][5]
                    new_league_name = new_result['egn'].replace('\t', '')
                    data = new_result['n-o']
                    item = Guanjun()
                    item['league_name'] = league_name
                    item['new_league_name'] = new_league_name
                    item['time_game'] = time_game
                    item['data_game'] = date_game
                    item['ball'] = ball
                    item['champion_data'] = data
                    item['league_id'] = league_id
                    item['set_id'] = new_result['k']
                    yield item
                    # for i in new_result['n-o']:
                    #     new_champion = i['mn']
                        # for y in i['o']:
                        #     item = Guanjun()
                        #     game_id = str(new_result['i'][16])
                        #     tema_home = y[0]
                        #     champion_team = float(y[2]) - 1
                        #
                        #     item['tema_home'] = tema_home
                        #     item['league_id'] = league_id
                        #     item['game_id'] = game_id
                        #     item['data_game'] = data_game
                        #     item['time_game'] = time_game
                        #     item['new_league_name'] = new_league_name
                        #     item['champion_team'] = champion_team
                        #     item['new_champion'] = new_champion
                        #     # item['pt'] = pt
                        #     item['ball'] = ball
                        #     yield item