sports_scrapy/hg3535/spiders/roll_bangqiu.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
import datetime

import jsonpath

__author__ = 'ethereal'
__mtime__ = '2019/4/16'

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

import scrapy
from scrapy.http import Request
from ..items import Roll_Bangqiu


class BqrangqiuSpider(scrapy.Spider):
    name = 'roll_bangqiu'
    to_day = datetime.datetime.now()
    allowed_domains = ['hg3535z.com']
    custom_settings = {
        "ITEM_PIPELINES": {
            # "hg3535.pipeline.Roll_Banqiupipeline": 300,
            "hg3535.pipeline.roll_bangqiu.Roll_Banqiupipeline": 350,
        },
        # 'LOG_LEVEL': 'DEBUG',
        # 'LOG_FILE': "../hg3535/log/roll_bangqiu_{}_{}_{}.log".format(to_day.year, to_day.month, to_day.day)
    }

    # start_urls = ['https://hg3535z.com/odds2/d/getodds?sid=4&pt=4&ubt=am&pn=0&sb=2&dc=null&pid=0']  # 滚球菜单 棒球滚球列url
    start_urls = [
        'https://odata.jiushan6688.com/odds6i/d/getodds/zh-cn/sid/4/pt/4/ubt/am/pn/0/sb/2/dc/null/pid/0']  # 滚球菜单 棒球滚球列url

    # start_urls = ['http://hg3535z.com/odds2/d/getodds?sid=3&pt=4&ubt=am&pn=0&sb=2&dc=null&pid=0'] # 滚球菜单 网球测试滚球列url
    # start_urls = ['https://www.baidu.com']

    def parse(self, response):
        datas = json.loads(response.text)
        # urls = "https://www.baidu.com"
        # yield  Request(url=urls,callback=self.parse_each)
        # datas ={"i": ["true", 4, 3110520, 2, 1, "sh", 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, "14 / 04", "", "", "false", "true", "\u68d2\u7403", 0], "eg": {"c": {"k": 29400, "n": "\u7f8e\u56fd\u804c\u4e1a\u68d2\u7403"}, "es": [{"dbg": "false", "egid": 0, "g": "", "i": ["\u4e9a\u7279\u5170\u5927\u52c7\u58eb\u961f", "\u7ebd\u7ea6\u5927\u90fd\u4f1a\u961f", "9", "True", "14 / 04", "", "9", "False", "0", "0", "2", "1", "", "", "", "", 0, "", "", 16, "True"], "ibs": "true", "ibsc": "false", "k": 3110520, "o": {"ah": {"s": 1, "v": ["h3164784133", "-1.5", "h3164784134", "+1.5", "o3164784133", "1.33", "o3164784134", "0.55"], "n": "\u6eda\u7403 \u8ba9\u7403"}, "ou": {"s": 2, "v": ["h3164784135", "6.5", "h3164784136", "6.5", "o3164784135", "0.70", "o3164784136", "1.04"], "n": "\u603b\u5f97\u5206:\u6eda\u7403 \u5927 / \u5c0f"}, "oe": {"s": 6, "v": ["o3164784137", "0.51", "o3164784138", "1.58"], "n": "\u603b\u5f97\u5206:\u6eda\u7403 \u5355 / \u53cc"}, "ml": {"s": 5, "v": ["o3164784139", "0.00", "o3164784140", "0.00"], "n": "\u6eda\u7403 \u72ec\u8d62\u76d8"}}, "p-o": [{"n": "\u51c0\u80dc\u7403\u6570", "st": "ha", "mgk": 9, "s": 7, "k": 301, "o": [["\u4e9a\u7279\u5170\u5927\u52c7\u58eb\u961f-\u8d62\u5f971\u5206", "o3164784186", "3.15", "0"], ["\u4e9a\u7279\u5170\u5927\u52c7\u58eb\u961f-\u8d62\u5f972\u5206", "o3164784187", "5.5", "3"], ["\u4e9a\u7279\u5170\u5927\u52c7\u58eb\u961f-\u8d62\u5f973\u5206", "o3164784188", "9.5", "6"], ["\u4e9a\u7279\u5170\u5927\u52c7\u58eb\u961f-\u8d62\u5f974\u5206\u6216\u66f4\u591a", "o3164784189", "7.5", "9"], ["\u7ebd\u7ea6\u5927\u90fd\u4f1a\u961f-\u8d62\u5f971\u5206", "o3164784190", "5.0", "2"], ["\u7ebd\u7ea6\u5927\u90fd\u4f1a\u961f-\u8d62\u5f972\u5206", "o3164784191", "7.5", "5"], ["\u7ebd\u7ea6\u5927\u90fd\u4f1a\u961f-\u8d62\u5f973\u5206", "o3164784192", "11.5", "8"], ["\u7ebd\u7ea6\u5927\u90fd\u4f1a\u961f-\u8d62\u5f974\u5206\u6216\u66f4\u591a", "o3164784193", "10.0", "11"]]}, {"n": "\u7b2c6\u5c40\u83b7\u80dc\u8005", "st": "ha", "mgk": 0, "s": 3000, "k": 4000, "o": [["\u548c\u5c40", "o3164784213", "1.50"], ["\u4e9a\u7279\u5170\u5927\u52c7\u58eb\u961f", "o3164784214", "5.5"], ["\u7ebd\u7ea6\u5927\u90fd\u4f1a\u961f", "o3164784215", "3.80"]]}, {"n": "\u52a0\u65f6", "st": "ha", "mgk": 312, "s": 8, "k": 13520, "o": [["\u662f", "o3164784228", "6.0"], ["\u4e0d\u662f", "o3164784229", "1.09"]]}], "pci": {"ctid": 0}, "sb": {"fb": "false", "sb": "false", "tb": "false", "ot": 0, "ps": [{"adv": -1, "a": 1.0, "e": 3110520, "h": 2.0, "p": "ft"}, {"adv": -1, "a": 0.0, "e": 3110520, "h": 0.0, "p": "ot"}, {"adv": -1, "a": 0.0, "e": 3110520, "h": 0.0, "p": "i1"}, {"adv": -1, "a": 1.0, "e": 3110520, "h": 1.0, "p": "i2"}, {"adv": -1, "a": 0.0, "e": 3110520, "h": 1.0, "p": "i3"}, {"adv": -1, "a": 0.0, "e": 3110520, "h": 0.0, "p": "i4"}, {"adv": -1, "a": 0.0, "e": 3110520, "h": 0.0, "p": "i5"}, {"adv": -1, "a": 0.0, "e": 3110520, "h": 0.0, "p": "i6"}, {"adv": -1, "a": 0.0, "e": 3110520, "h": 0.0, "p": "i7"}, {"adv": -1, "a": 0.0, "e": 3110520, "h": 0.0, "p": "i8"}, {"adv": -1, "a": 0.0, "e": 3110520, "h": 0.0, "p": "i9"}], "a": 1, "h": 2, "k": 3110520, "iwd": "false", "ls": -1, "s": 1, "ts": 0, "tt": 0, "ct": "00:00", "cp": "i5"}, "egn": ""}, {"dbg": "false", "egid": 0, "g": "", "i": ["ATL Braves - Team Runs - Over", "ATL Braves - Team Runs - Under", "9", "True", "14 / 04", "", "9", "False", "0", "0", "2", "0", "", "", "", "", 0, "", "", 16, "True"], "ibs": "true", "ibsc": "false", "k": 3110522, "o": {"ou": {"s": 3, "v": ["h3164784242", "3.5", "h3164784243", "3.5", "o3164784242", "1.12", "o3164784243", "0.63"], "n": "\u7403\u961f\u5f97\u5206:\u4e9a\u7279\u5170\u5927\u52c7\u58eb\u961f-\u6eda\u7403 \u5927 / \u5c0f"}}, "pci": {"ctid": 208, "ctn": "\u7403\u961f\u5f97\u5206"}, "egn": ""}, {"dbg": "false", "egid": 0, "g": "", "i": ["NY Mets - Team Runs - Over", "NY Mets - Team Runs - Under", "9", "True", "14 / 04", "", "9", "False", "0", "0", "1", "0", "", "", "", "", 0, "", "", 16, "True"], "ibs": "true", "ibsc": "false", "k": 3110524, "o": {"ou": {"s": 4, "v": ["h3164784261", "2.5", "h3164784263", "2.5", "o3164784261", "0.55", "o3164784263", "1.23"], "n": "\u7403\u961f\u5f97\u5206:\u7ebd\u7ea6\u5927\u90fd\u4f1a\u961f-\u6eda\u7403 \u5927 / \u5c0f"}}, "pci": {"ctid": 209, "ctn": "\u7403\u961f\u5f97\u5206"}, "egn": ""}]}, "ot": 2, "sc": {"3110520": {"a": 1, "h": 2}, "3110522": {"a": 0, "h": 2}, "3110524": {"a": 0, "h": 1}}, "v": 40576}
        # item = Today_all()
        ids = jsonpath.jsonpath(datas, '$..i-ot[0]..egs..es..i[16]')  # ids新列表
        # print(ids)
        if ids:
            # ids = set(ids)
            for i in ids:
                # urls = 'https://hg3535z.com/odds2/d/getamodds?eid={}&iip=true&ubt=am&isp=false'.format(i)
                urls = 'https://odata.jiushan6688.com/odds6i/d/getamodds/zh-cn/eid/{}/iip/true/ubt/am/isp/false'.format(
                    i)
                # print(urls)
                yield Request(url=urls, callback=self.parse_each, dont_filter=True)

    def parse_each(self, response):
        datas = json.loads(response.text)
        # 倒三角形  sa向下，sh向上
        triangle = datas['i'][5]
        try:
            # new_datas = json.loads(response.text)['eg']
            new_datas = datas['eg']
        except:
            new_datas = ""
        try:
            pt = response.meta['pt']
        except:
            pt = 0
        if new_datas:
            # 联赛id
            league_id = new_datas["c"]["k"]
            # 联赛名
            league_name = new_datas["c"]["n"]
            new_data = new_datas["es"]
            for result in new_data:
                if result['pci']["ctid"] == 0:
                    item = Roll_Bangqiu()
                    # 比赛id
                    game_id = str(result['k'])
                    # 球队1
                    team_home = result['i'][0]
                    # 球队2
                    team_guest = result['i'][1]
                    # 数量(97>)
                    number = result['i'][2]
                    # 比赛状态
                    zhuangtai = result['i'][3]
                    # 日期
                    data_game = result['i'][4]
                    # 开赛时间
                    time_game = result['i'][5]
                    # 队1分数
                    score_home1 = result['i'][10]
                    # 队2分数
                    score_guest1 = result['i'][11]
                    # 第几盘
                    # jijie = result['i'][12]
                    # 球队得分
                    qiudui = result['pci'].get('ctn', "")
                    # 是否出局 不一定存在
                    ot = result['sb'].get("ot", "")

                    ps_list = result['sb']['ps']
                    # 右菱形bool值
                    fb = result['sb']['fb']
                    # 左菱形bool值
                    tb = result['sb']['tb']
                    # 上菱形bool值
                    sb = result['sb']['tb']
                    # 局数或者节或者分
                    cp = result['sb']['cp']
                    # 第一个队伍
                    a = result['sb']['a']
                    # 第二个队伍
                    h = result['sb']['h']
                    d = {}
                    d['ps'] = ps_list
                    d['cp'] = cp
                    d['ot'] = ot
                    d['triangle'] = triangle
                    d['fb'] = fb
                    d['tb'] = tb
                    d['sb'] = sb
                    # --------------------------------------------------让球分割线-----------------------------------------------------------
                    concedes_dict = {}
                    concedes_dict_rule = {}
                    try:
                        concedes = result['o']["ah"]["v"]
                        concedes_dict_rule['coh'] = concedes[1]
                        concedes_dict['coh'] = concedes[5]
                        concedes_dict_rule['cog'] = concedes[3]
                        concedes_dict['cog'] = concedes[7]
                    except:
                        concedes_dict_rule['coh'] = ""
                        concedes_dict['coh'] = ""
                        concedes_dict_rule['cog'] = ""
                        concedes_dict['cog'] = ""
                    # 让球 前5局-------------------------------------------------------------------------------------------------------------
                    try:
                        concedes = result['o']["ahf5in"]["v"]
                        concedes_dict_rule['coh_5'] = concedes[1]
                        concedes_dict['coh_5'] = concedes[5]
                        concedes_dict_rule['cog_5'] = concedes[3]
                        concedes_dict['cog_5'] = concedes[7]
                    except:
                        concedes_dict_rule['coh_5'] = ""
                        concedes_dict['coh_5'] = ""
                        concedes_dict_rule['cog_5'] = ""
                        concedes_dict['cog_5'] = ""
                    # ---------------------------------------------------独赢----------------------------------------------------------------
                    capots_dict = {}
                    try:
                        capots = result['o']["ml"]["v"]
                        capots_dict['ch'] = capots[1]
                        capots_dict['cg'] = capots[3]

                    except:
                        capots_dict['ch'] = ""
                        capots_dict['cg'] = ""
                    # ---------------------------------------------------总得分:大/小--------------------------------------------------------

                    total_size_dict = {}
                    total_size_dict_rule = {}
                    try:
                        total_sizes = result['o']["ou"]["v"]
                        total_size_dict_rule['tnb'] = total_sizes[1]
                        total_size_dict['tnb'] = total_sizes[5]
                        total_size_dict_rule['tns'] = total_sizes[3]
                        total_size_dict['tns'] = total_sizes[7]
                    except:
                        total_size_dict_rule['tnb'] = ""
                        total_size_dict['tnb'] = ""
                        total_size_dict_rule['tns'] = ""
                        total_size_dict['tns'] = ""
                    # 总得分: 大/小 前5局-----------------------------------------------------------------------------------------------------
                    try:
                        total_sizes = result['o']["ouf5in"]["v"]
                        total_size_dict_rule['tnb_5'] = total_sizes[1]
                        total_size_dict['tnb_5'] = total_sizes[5]
                        total_size_dict_rule['tns_5'] = total_sizes[3]
                        total_size_dict['tns_5'] = total_sizes[7]
                    except:
                        total_size_dict_rule['tnb_5'] = ""
                        total_size_dict['tnb_5'] = ""
                        total_size_dict_rule['tns_5'] = ""
                        total_size_dict['tns_5'] = ""

                    # ------------------------------------------------------总得分:单/双-----------------------------------------------------
                    odd_evens_dict = {}
                    odd_evens_dict_rule = {}
                    try:
                        odd_evens = result['o']["oe"]["v"]
                        # 上半场 主队让球条件
                        odd_evens_dict['tss'] = odd_evens[1]
                        odd_evens_dict_rule['tss'] = "单"
                        odd_evens_dict['tsd'] = odd_evens[3]
                        odd_evens_dict_rule['tsd'] = "双"

                    except:
                        odd_evens_dict['tss'] = ""
                        odd_evens_dict['tsd'] = ""
                        odd_evens_dict_rule['tss'] = ""
                        odd_evens_dict_rule['tsd'] = ""

                    team_source_dict = {}
                    try:
                        team_sources = result['sb']['ps']
                        for team_source in team_sources:
                            if team_source['p'] == 'i1':
                                team_source_dict['one_score'] = str(team_source['h']) + '-' + str(team_source['a'])
                            if team_source['p'] == 'i2':
                                team_source_dict['two_score'] = str(team_source['h']) + '-' + str(team_source['a'])
                            if team_source['p'] == 'i3':
                                team_source_dict['three_score'] = str(team_source['h']) + '-' + str(team_source['a'])
                            if team_source['p'] == 'i4':
                                team_source_dict['four_score'] = str(team_source['h']) + '-' + str(team_source['a'])
                            if team_source['p'] == 'i5':
                                team_source_dict['five_score'] = str(team_source['h']) + '-' + str(team_source['a'])
                            if team_source['p'] == 'i6':
                                team_source_dict['six_score'] = str(team_source['h']) + '-' + str(team_source['a'])
                            if team_source['p'] == 'i7':
                                team_source_dict['seven_score'] = str(team_source['h']) + '-' + str(team_source['a'])
                            if team_source['p'] == 'i8':
                                team_source_dict['eight_score'] = str(team_source['h']) + '-' + str(team_source['a'])
                            if team_source['p'] == 'i9':
                                team_source_dict['nine_score'] = str(team_source['h']) + '-' + str(team_source['a'])
                            # 加时
                            if team_source['p'] == 'ot':
                                team_source_dict['overtime_score'] = str(team_source['h']) + '-' + str(team_source['a'])
                            # 总局数比分
                            if team_source['p'] == 'ft':
                                team_source_dict['total_inning'] = int(team_source['h']) + int(team_source['a'])
                                team_source_dict['total_source'] = str(team_source['h']) + '-' + str(team_source['a'])
                    except:
                        team_source_dict['one_score'] = ''
                        team_source_dict['two_score'] = ''
                        team_source_dict['three_score'] = ''
                        team_source_dict['four_score'] = ''
                        team_source_dict['five_score'] = ''
                        team_source_dict['six_score'] = ''
                        team_source_dict['seven_score'] = ''
                        team_source_dict['eight_score'] = ''
                        team_source_dict['nine_score'] = ''
                        team_source_dict['overtime_score'] = ''
                        team_source_dict['total_inning'] = 0
                        team_source_dict['total_source'] = ''

                    # 联赛id
                    item['league_id'] = league_id
                    # 联赛名
                    item['league_name'] = league_name
                    # 比赛id
                    item['game_id'] = game_id
                    # 球队1
                    item['team_home'] = team_home
                    # 球队2
                    item['team_guest'] = team_guest
                    # 数量(97>)
                    item['number'] = number
                    # 比赛状态
                    item['zhuangtai'] = zhuangtai
                    # 日期
                    item['data_game'] = data_game
                    # 开赛时间
                    item['time_game'] = time_game
                    # 队1分数
                    item['score_home'] = score_home1
                    # 队2分数
                    item['score_guest'] = score_guest1
                    # 第几节
                    item['jijie'] = cp
                    item['pt'] = pt
                    # 球队得分
                    item['qiudui'] = "{}:{}".format(a, h)
                    # 让球
                    item['concedes_dict'] = concedes_dict
                    item['concedes_dict_rule'] = concedes_dict_rule
                    # 独赢
                    item['capots_dict'] = capots_dict
                    # 总得分：大/小
                    item['total_size_dict'] = total_size_dict
                    item['total_size_dict_rule'] = total_size_dict_rule
                    # 总得分：单/双
                    item['odd_evens_dict'] = odd_evens_dict
                    item['odd_evens_dict_rule'] = odd_evens_dict_rule
                    item['match_score_dict'] = d
                    item['team_source_dict'] = team_source_dict
                    yield item