python爬虫爬取的数据有10%和显示的不一样

Question

我用的是splash 爬取，返回的数据10%-20%和页面不一样. 80%是正常数据
试过爬取第一、二页，重复爬取。不一样的数据位置会变化，一条数据有时会正常爬到，有时不正常。
是网站随机对这返回数据 运算了吗？怎么能拿到原始数据。

或者怎么才能根据获取的不正常数据，通过运算转回对的数据。
找了一天，没找到答案。。也找不出它转码的规律。

求大神帮忙~~~~~

代码如下：

# -*- coding: utf-8 -*-
import json
import datetime
import scrapy
from scrapy import Spider
from nbaOdds.items import NbaOddsItem
from scrapy_splash import SplashRequest
from scrapy.http.headers import Headers 


RENDER_HTML_URL = "http://127.0.0.1:8050/render.html" 

class NbaOddsAHSpider(Spider):
    name = "odds"
    allowed_domains = ["oddsportal.com"]

    start_urls = []

    start_url ="http://www.oddsportal.com/basketball/usa/nba-2014-2015/results/#/page/"

    number_pages = 3

    # start_urls.append(start_url)
    for x in range(1, number_pages):

        start_urls.append(start_url + str(x))

    #转换队名
    def convert_name(self, name):
        if name == "Atlanta Hawks":
            return "ATL"
        elif name == "Boston Celtics":
            return "BOS"
        elif name == "Brooklyn Nets":
            return "BRK"
        elif name == "Cleveland Cavaliers":
            return "CLE"     
        elif name == "Chicago Bulls":
            return "CHI"
        elif name == "Charlotte Hornets":
            return "CHO"
        elif name == "Dallas Mavericks":
            return "DAL"
        elif name == "Denver Nuggets":
            return "DEN"
        elif name == "Detroit Pistons":
            return "DET"
        elif name == "Golden State Warriors":
            return "GSW"

        elif name == "Houston Rockets":
            return "HOU"
        elif name == "Indiana Pacers":
            return "IND"
        elif name == "Los Angeles Clippers":
            return "LAC"
        elif name == "Los Angeles Lakers":
            return "LAL"
        elif name == "New York Knicks":
            return "NYK"
        elif name == "New Orleans Pelicans":
            return "NOP"     
        elif name == "Memphis Grizzlies":
            return "MEM"
        elif name == "Miami Heat":
            return "MIA"
        elif name == "Minnesota Timberwolves":
            return "MIN"
        elif name == "Milwaukee Bucks":
            return "MIL"

        elif name == "Oklahoma City Thunder":
            return "OKC"
        elif name == "Orlando Magic":
            return "ORL"
        elif name == "Philadelphia 76ers":
            return "PHI"
        elif name == "Portland Trail Blazers":
            return "POR"
        elif name == "Phoenix Suns":
            return "PHO"
        elif name == "Sacramento Kings":
            return "SAC"
        elif name == "San Antonio Spurs":
            return "SAS"     
        elif name == "Toronto Raptors":
            return "TOR"
        elif name == "Utah Jazz":
            return "UTA"
        elif name == "Washington Wizards":
            return "WAS"



    def start_requests(self):

        for url in self.start_urls:
            body = json.dumps({"url":url, "wait":5})
            headers = Headers({'Content-Type': 'application/json',
                             #'Content-Type':'text/javascript;charset=UTF-8',
                               #'Content-Type': 'text/javascript; charset=UTF-8',
                               #"Host":"www.oddsportal.com",
                               #"Referer":"http://www.oddsportal.com/basketball/usa/nba-2015-2016/results/",
                               "User-Agent":"Mozilla/5.0 (Windows; U; Windows NT 5.2) AppleWebKit/525.13 (KHTML, like Gecko) Chrome/0.2.149.27 Safari/525.13"
                               })

            yield SplashRequest(url, self.parse, args={
                'wait': 0.5, 'html': 1, 'timeout': 3600,
            }, headers=headers,
            )


    def parse(self, response):

        base_url = "http://www.oddsportal.com"
        headers = Headers({'Content-Type': 'application/json',
                       #'Content-Type': 'text/javascript;charset=UTF-8',
                               "User-Agent":"Mozilla/5.0 (Windows; U; Windows NT 5.2) AppleWebKit/525.13 (KHTML, like Gecko) Chrome/0.2.149.27 Safari/525.13"
                               })

        games = response.xpath("//tr[contains(@class, 'deactivate')]")
        for game in games:

            child_url = base_url + game.xpath("td[contains(@class,'table-participant')]/a/@href").extract()[0] + "#over-under;8"

            yield SplashRequest(child_url, self.parse_ah_page, args={
                'wait': 0.5, 'html': 1, 'timeout': 3600,
            }, headers=headers,
            )

    def parse_ah_page(self, response):
        item = NbaOddsItem()
        matchup = response.xpath("//*[@id='col-content']/h1/text()").extract()[0]
        matchup_arr = matchup.strip().split('-')

        item["home_team_name"] = self.convert_name(matchup_arr[0].strip())
        item["away_team_name"] = self.convert_name(matchup_arr[1].strip())

        date_n_time = response.xpath("//*[@id='col-content']/p[1]/text()").extract()[0]
        tmp = date_n_time.split(',')
        date_time = datetime.datetime.strptime(tmp[1].strip() + " " + tmp[2].strip(), "%d %b  %Y %H:%M")
        date_time += datetime.timedelta(0,0,0,0,0,-5)
        date_time = date_time.strftime('%I:%M %p, %b %d, %Y')

        item["date"] = date_time

        ##########################
        #获取 整个表
        # Asian handicap -7.5 95.2%1.273.80(1)Compare odds
        # Asian handicap -6.5 (0)Compare odds
        ah_odds = response.xpath("//*[contains(@class, 'table-header-light')]")
        ah_odds_list = []

        for ah_odd in ah_odds:
            #每次只获取一个 左边 分数
            #如 Asian handicap -7.5 
            odds_number_temp = ah_odd.xpath("strong/a/text()").extract()[0]
            #分割  就只剩 -7.5
            odds_number = float(odds_number_temp.split()[-1])

            #每次只获取一个 右边的开这个odds的有几个 Compare odds 
            # 如  (1)
            odds_cnt_temp = ah_odd.xpath("span[contains(@class,'odds-cnt')]/text()").extract()[0]
            #print odds_cnt_temp
            # 剩下1，去除() 去除第一个和最后一个字符
            odds_cnt = int(odds_cnt_temp[1:-1])

            odds_over = 0
            odds_under = 0

            if odds_cnt > 0:
                #获取第一个odds，
                #如 1.27
                #   3.80
                odds = ah_odd.xpath("span[contains(@class,'chunk-odd')]/a/text()").extract()

                #如果 odds 有数据，0为客odds 1为主odds
                if len(odds) > 1:
                    # 判断是否有错误数据 XX/XX
                    #---------------下面这段是有时返回和页面不同的数据---------
                    if "/" in odds[1]:
                        # 分子
                        molecule = float(odds[0].split("/")[0])
                        # 分母
                        Denominator = float(odds[0].split("/")[1])
                        odds_under = molecule/Denominator+1

                        molecule = float(odds[1].split("/")[0])
                        # 分母
                        Denominator = float(odds[1].split("/")[1])
                        odds_over = molecule/Denominator+1
                    #---------------上面这段是有时返回和页面不同的数据---------
                    else:
                        odds_over = float(odds[1])
                        odds_under = float(odds[0])

            ah_odds_list.append((odds_number, odds_cnt, odds_over, odds_under))

            print (odds_number, odds_cnt, odds_over, odds_under)

        ah_odds_list = sorted(ah_odds_list, reverse=True, key=lambda a: a[1])

        ah_odds_list = ah_odds_list[0]

        item['over_1'] = ah_odds_list[2]
        item['part_1_ou'] = ah_odds_list[0]
        item['under_1'] = ah_odds_list[3]


        yield item

第一段'over_1'和'under_1'错误的数据，第二段为正常

2017-10-25 21:53:07 [scrapy] DEBUG: Crawled (200) <GET http://www.oddsportal.com/basketball/usa/nba-2014-2015/milwaukee-bucks-chicago-bulls-rPsOwGX8/#over-under;8 via http://127.0.0.1:8050/render.html> (referer: None)
2017-10-25 21:53:07 [stdout] INFO: (47.0, 1, -125.0, 104.0)
2017-10-25 21:53:07 [stdout] INFO: (47.5, 4, -109.0, -110.0)
2017-10-25 21:53:07 [stdout] INFO: (48.0, 5, -105.0, -117.0)
2017-10-25 21:53:07 [stdout] INFO: (48.5, 0, 0, 0)
2017-10-25 21:53:07 [stdout] INFO: ('insert:', 12)
2017-10-25 21:53:07 [scrapy] DEBUG: Scraped from <200 http://www.oddsportal.com/basketball/usa/nba-2014-2015/milwaukee-bucks-chicago-bulls-rPsOwGX8/#over-under;8>
{'away_team_name': 'CHI',
 'date': '05:30 PM, Apr 25, 2015',
 'home_team_name': 'MIL',
 'over_1': -105.0,
 'part_1_ou': 48.0,
 'under_1': -117.0}
2017-10-25 21:53:11 [scrapy] DEBUG: Crawled (200) <GET http://www.oddsportal.com/basketball/usa/nba-2014-2015/new-orleans-pelicans-golden-state-warriors-Ei2haTbg/#over-under;8 via http://127.0.0.1:8050/render.html> (referer: None)
2017-10-25 21:53:11 [stdout] INFO: (51.5, 1, 1.77, 2.1)
2017-10-25 21:53:11 [stdout] INFO: (52.0, 5, 1.87, 1.97)
2017-10-25 21:53:11 [stdout] INFO: (52.5, 4, 1.9, 1.91)
2017-10-25 21:53:11 [stdout] INFO: (53.0, 0, 0, 0)
2017-10-25 21:53:11 [stdout] INFO: ('insert:', 13)
2017-10-25 21:53:11 [scrapy] DEBUG: Scraped from <200 http://www.oddsportal.com/basketball/usa/nba-2014-2015/new-orleans-pelicans-golden-state-warriors-Ei2haTbg/#over-under;8>
{'away_team_name': 'GSW',
 'date': '08:00 PM, Apr 25, 2015',
 'home_team_name': 'NOP',
 'over_1': 1.87,
 'part_1_ou': 52.0,
 'under_1': 1.97}

爬虫获取的错误数据，和对应网页的数据.大概有10%+会出现错误数据，其他正常
另一种错误数据格式如 "20/25" ,对应正常数据 "1.8" ,
我在代码中转换用, 20/25+1 转换了

错误的数据我也找不到规律转回正确的。。。