爬取京东生鲜的商品数据和评论数据

20次阅读

共计 8513 个字符,预计需要花费 22 分钟才能阅读完成。

很久以前,爬了京东的生鲜页面,鞋子想把代码发出来共享

首先简明,爬取 商品数据 采用 selenium 操作 chrome 模拟浏览器动态渲染页面 +ajax 加载评论

具体的看下面的说明

所需内容:
商品小分类名称(苹果,橙子等)
商品名称(烟台红富士苹果 5kg 一级铂金大果 单果 230-320g 新鲜水果)
商品总评论数量
商品好评率
评论星级
评论长度
评论点赞数量
评论回复数量
评论文本内容
评论者等级
评论发表距抓取的天数(days)
抓取部分带有追评的评论:追评文本内容、追评与初评相距时间

以上是这次任务的需求

这个页面大部分的信息都是动态渲染出来的,所以要用 selenium


可以看到,要找到评论不是去常见的 HXR 而是 JS,peoduct 开头的就是评论信息

Request URL: https://sclub.jd.com/comment/…
Request Method: GET
Status Code: 200
Remote Address: 117.148.129.129:443
Referrer Policy: no-referrer-when-downgrade

在这个 url 中,查询字符串中的大部分参数不是必须的

def make_url(baseurl, page=0, score=0, productId='3756271'):
    data1 = {
        'callback': 'fetchJSON_comment98vv7490',
        'productId': productId,
        'score': score,
        'sortType': '6',
        'page': page,
        'pageSize': '10',
        'isShadowSku': '0',  #
        'fold': '1',  #
    }
    url = baseurl + urlencode(data1)
    return url

具体的可以在代码中体现。

下面我要贴代码了,坐稳扶好,不想复制的话,可以去我的 github 上下载。

# https://www.jd.com/allSort.aspx
import requests
from pyquery import PyQuery as pq
from prettyprinter import cpprint
import json
from urllib.parse import urlencode
from selenium import webdriver
from selenium.webdriver.support.wait import WebDriverWait
from selenium.webdriver.common.by import By
from selenium.webdriver.support import expected_conditions as EC
import time
import csv
import datetime
import sys


def get_ajax(url):
    headers = {
        'referer': 'https://item.jd.com/3756271.html',  # referer: https://item.jd.com/3756271.html
        'user-agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.110 Safari/537.36',
    }
    response = requests.get(url, headers=headers)
    return json.loads(response.text[26:-2])


def make_url(baseurl, page=0, score=0, productId='3756271'):
    data1 = {
        'callback': 'fetchJSON_comment98vv7490',
        'productId': productId,
        'score': score,
        'sortType': '6',
        'page': page,
        'pageSize': '10',
        'isShadowSku': '0',  #
        'fold': '1',  #
    }
    url = baseurl + urlencode(data1)
    return url


def parse_json(rjson, url=None):
    for comment in rjson.get('comments'):
        item = {}
        item['url'] = url
        item['评论星级'] = comment.get('score')
        item['评论长度'] = len(comment.get('content'))
        item['评论点赞数量'] = comment.get('usefulVoteCount')
        item['评论回复数量'] = comment.get('replyCount')
        item['评论文本内容'] = comment.get('content')
        item['评论者等级'] = comment.get('userLevelId')
        try:
            date1 = time.strptime(comment.get('creationTime'), "%Y-%m-%d %H:%M:%S")
            date2 = time.localtime(time.time())
            date1 = datetime.datetime(date1[0], date1[1], date1[2])
            date2 = datetime.datetime(date2[0], date2[1], date2[2])
            item['评论发表距抓取的天数(days)'] = str((date2 - date1).days)
        except Exception as error:
            print('error is >>>', error)
            item['评论发表距抓取的天数(days)'] = ''if comment.get('afterUserComment', {}).get('hAfterUserComment', {}).get('content','') == '此用户未填写评价内容':
            item['追评文本内容'] = ''
        else:
            item['追评文本内容'] = comment.get('afterUserComment', {}).get('hAfterUserComment', {}).get('content', '')
        try:
            date1 = time.strptime(comment.get('afterUserComment', {}).get('created', ''),"%Y-%m-%d %H:%M:%S")
            date2 = time.localtime(time.time())
            date1 = datetime.datetime(date1[0], date1[1], date1[2])
            date2 = datetime.datetime(date2[0], date2[1], date2[2])
            item['追评与初评相距时间'] = str((date2 - date1).days)
        except Exception:
            item['追评与初评相距时间'] = ''if item[' 追评文本内容 '] =='':
            item['追评与初评相距时间'] = ''
        yield item


def save_csv_merinfo(item):
    with open(FILENAME_MER, 'a', encoding=ENCODING, newline='') as f:
        writer = csv.DictWriter(f, fieldnames=fieldnames_merinfo)
        # writer.writeheader()
        writer.writerow(item)


def save_csv_cominfo(item):
    with open(FILENAME_COM, 'a', encoding=ENCODING, newline='') as f:
        writer = csv.DictWriter(f, fieldnames=fieldnames_cominfo)
        # writer.writeheader()
        writer.writerow(item)


def get_page(url):
    browser.get(url)
    submit = wait.until(EC.presence_of_element_located((By.XPATH, '//div[contains(@class,"tab-main")]/ul/li[5]')))
    time.sleep(2)
    for i in range(30):
        browser.execute_script("window.scrollBy(0,50)")
        time.sleep(0.1)
    submit.click()
    time.sleep(3)
    return browser.page_source


def parse_page(html, url):
    page_item = {}
    doc = pq(html, parser='html')
    page_item['url'] = url
    page_item['商品小分类名称'] = doc('#crumb-wrap > div > div.crumb.fl.clearfix > div:nth-child(5) > a').text()
    page_item['商品名称'] = doc('div.itemInfo-wrap div.sku-name').text()
    page_item['商品总评论数量'] = doc('#detail > div.tab-main.large > ul > li.current > s').text().replace('(', '').replace(')','')
    page_item['商品好评率'] = doc('#comment > div.mc > div.comment-info.J-comment-info > div.comment-percent > div').text()
    ##comment > div.mc > div.comment-info.J-comment-info > div.comment-percent > div
    return page_item


def csv_create():
    with open(FILENAME_MER, 'w', encoding=ENCODING, newline='') as f:
        writer = csv.writer(f)
        writer.writerow(fieldnames_merinfo)

    with open(FILENAME_COM, 'w', encoding=ENCODING, newline='') as f:
        writer = csv.writer(f)
        writer.writerow(fieldnames_cominfo)


def crawl_all_page_url():
    global ALL_PAGE_URL
    browser = webdriver.Chrome()
    wait = WebDriverWait(browser, 20)

    browser.get('https://www.jd.com/allSort.aspx')
    wait.until(EC.presence_of_element_located((By.XPATH, '/html/body/div[5]/div[2]/div[1]/div[2]/div[2]/div[9]/div[2]/div[3]')))

    CASE = []
    for i in range(10):  # 水果
        initcase = '/html/body/div[5]/div[2]/div[1]/div[2]/div[2]/div[9]/div[2]/div[3]/dl[2]/dd/a[{}]'.format(i + 1)
        CASE.append(initcase)
    for i in range(4):  # 猪肉羊肉
        initcase = '/html/body/div[5]/div[2]/div[1]/div[2]/div[2]/div[9]/div[2]/div[3]/dl[3]/dd/a[{}]'.format(i + 1)
        CASE.append(initcase)
    for i in range(8):  # 海鲜水产
        initcase = '/html/body/div[5]/div[2]/div[1]/div[2]/div[2]/div[9]/div[2]/div[3]/dl[4]/dd/a[{}]'.format(i + 1)
        CASE.append(initcase)
    for i in range(4):  # 禽肉蛋白
        initcase = '/html/body/div[5]/div[2]/div[1]/div[2]/div[2]/div[9]/div[2]/div[3]/dl[5]/dd/a[{}]'.format(i + 1)
        CASE.append(initcase)
    for i in range(6):  # 冷冻食品
        initcase = '/html/body/div[5]/div[2]/div[1]/div[2]/div[2]/div[9]/div[2]/div[3]/dl[6]/dd/a[{}]'.format(i + 1)
        CASE.append(initcase)
    # 规则只要更改 range 里面的值和 dl[]里面的值,可高度扩展

    for case in CASE:
        print('>>>>>>>>>')
        submit = wait.until(EC.element_to_be_clickable((By.XPATH, case)))
        submit.click()

        print(browser.current_url)

        handle = browser.current_window_handle
        handles = browser.window_handles
        for newhandle in handles:
            if newhandle != handle:
                browser.switch_to.window(newhandle)
        time.sleep(1.5)
        wait.until(EC.presence_of_element_located((By.XPATH, '//div[@id="plist"]/ul[contains(@class,"gl-warp")]')))
        doc = pq(browser.page_source, parser='html')
        for li in list(doc('div#plist ul.gl-warp li').items())[:10]:
            res = 'https:' + str(li('div div.p-commit-n strong a').attr('href')).replace('#comment', '')
            print(res)
            ALL_PAGE_URL.append(res)
        time.sleep(1.5)
        browser.close()
        browser.switch_to.window(handle)


def load_all_page_url():
    global ALL_PAGE_URL
    with open(FILENAME_CACHE, 'r', encoding='utf-8') as f:
        reader = csv.reader(f)
        for item in reader:
            ALL_PAGE_URL.append(item[0])


if __name__ == '__main__':
    # 前期准备 >>>>>>>>>>
    browser = webdriver.Chrome()  # selenium 模拟浏览器
    wait = WebDriverWait(browser, 20)
    MAXINDEX = 7  # 最大请求评论页数, 为了控制评论数量在 500 条左右,应该设置为 35 左右,35 时略大于 500(网页评论非无限下拉)

    # 用户自定义配置区 ********************************
    TIMESLEEP = 2  # 睡眠间隔
    FILENAME_MER = 'merinfo_test.csv'  # 商品信息的文件名
    FILENAME_COM = 'cominfo_test.csv'  # 评论信息的文件名
    FILENAME_CACHE = 'cache.csv'
    ENCODING = 'UTF-8'  # 保存的 CSV 的编码
    # **********************************************

    # csv 文件的字段
    fieldnames_merinfo = ['url', '商品小分类名称', '商品名称', '商品总评论数量', '商品好评率']
    fieldnames_cominfo = ['url', '评论星级', '评论长度', '评论点赞数量', '评论回复数量', '评论文本内容', '评论者等级', '评论发表距抓取的天数(days)', '追评文本内容',
                          '追评与初评相距时间']
    # <<<<<<<<<<<<<<<<<

    start = time.time()

    # csv_create()  # 重置
    # 去重模块 >>>
    URLSET = []  # 已存在的 url 的集合
    with open(FILENAME_MER, 'r', encoding=ENCODING) as f:
        reader = csv.reader(f)
        for res in reader:
            URLSET.append(res[0])
    print('URLSET is', URLSET)
    # 爬取商品信息
    ALL_PAGE_URL = []  # 所有的网页链接
    load_all_page_url()  # 这两个函数要二选一,load_all_page_url 会从本地的 cache.csv 载入,速度更快,脱机工作,不占用网络
    # crawl_all_page_url() # 这两个函数要二选一,load_all_page_url 会从本地的 cache.csv 载入,速度更快,脱机工作,不占用网络
    for page_url in ALL_PAGE_URL:
        if page_url not in URLSET:
            URLSET.append(page_url)  # 动态去重
            try:
                html = get_page(page_url)  # 请求网页,selenium 动态渲染
                item_mer = parse_page(html, url=page_url)  # 解析网页,pyquery
                cpprint(item_mer)

                # 爬取评论信息,ajax
                Flag = 0  # 计数器
                ITEMS = []
                baseurl = 'https://sclub.jd.com/comment/productPageComments.action?'
                for score in [5, 3, 2, 1]:  # 0 全部评论,5 追评,3 好评,2 中评,1 差评
                    if score == 5:
                        MAXINDEX_TEMP = MAXINDEX
                    else:
                        MAXINDEX_TEMP = int(MAXINDEX / 7)  # 控制比例为 7:1:1:1
                    for index in range(MAXINDEX_TEMP):
                        time.sleep(TIMESLEEP)
                        url = make_url(baseurl, page=index, score=score,
                                       productId=''.join(list(filter(str.isdigit, page_url))))  # 构造 url
                        try:
                            json_ = get_ajax(url)  # 进行 ajax 请求
                            if len(json_.get('comments')) != 0:
                                for item in parse_json(json_, url=page_url):  # 解析 json
                                    cpprint(item)
                                    ITEMS.append(item)
                                    Flag += 1
                            else:
                                break
                        except Exception as error:
                            print('AJAX 请求发生错误{}>>>'.format(error))
                            print('url is {}'.format(url))
                            print(str(datetime.datetime.now()))
                            sys.exit(0)  # ajax 请求出错时退出程序,确保数据完整性
                # 一个网页的商品信息和评论信息都爬取完毕时,保存数据
                save_csv_merinfo(item_mer)  # 保存商品信息

                for item in ITEMS:  # 保存评论信息
                    try:
                        save_csv_cominfo(item)
                    except Exception as error:
                        print(error)
                print("保存了 {} 条评论".format(Flag))
            except Exception as error:
                print('网页请求发生错误{}>>>'.format(error))
        print('一个网页请求已经结束 >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>')
        # time.sleep(TIMESLEEP)

    end = time.time()
    print('总共用时 {} 秒'.format(end - start))

还是写了一些注释的,不懂得可以在评论中问。暂时这样吧!

正文完
 0