从Ajax驱动的弹出框中包含的工具提示中刮取文本

时间:2019-04-30 15:51:05

标签: python selenium web-scraping beautifulsoup screen-scraping

我知道之前也曾问过类似的问题,但似乎没有一个问题适合这种特殊情况。我在几个站点上都遇到了这个问题,因此对于这个问题,我随机选择了the first page of SO's own tags list.

如果您查看首页上的第一个条目,则会看到以下内容:

enter image description here

显示今天和本周标签说明的开头,问题总数以及提出的问题数。该信息很容易选择:

from selenium.webdriver import Chrome
driver = Chrome()
driver.get('https://stackoverflow.com/tags')

例如,以JavaScript标记为重点

dat = driver.find_elements_by_xpath("//*[contains(text(), 'week')]/ancestor::div[5]/div/div[1]/span/parent::*")
for i in dat:
    print(i.text)

输出:

    javascript× 1801272
JavaScript (not to be confused with Java) is a high-level, dynamic, multi-paradigm, object-oriented, prototype-based, weakly-typed language used for both client-side and server-side scripting. Its pri…
703 asked today, 4757 this week

现在,它变得更加复杂了(至少对我而言):如果将鼠标悬停在JavaScript标记上,则会显示以下弹出框:

enter image description here

此框具有完整的标签说明,以及(四舍五入)问题和观察者的数量。如果将鼠标悬停在“ 1.2m watchers”元素上,则会看到以下工具提示:

enter image description here

这是此特定框的通话网址:

https://stackoverflow.com/tags/javascript/popup?_=1556571234452

该目标项目(以及问题总数)包含在以下html的title的{​​{1}}中:

span

对于第一页中提到的所有标签,我不知道如何将所有这些信息放在一起,以得到看起来像这样的输出(或数据框):

<div class="-container">
<div class="-arrow js-source-arrow"></div>
<div class="mb12">
        <span class="fc-orange-400 fw-bold mr8">
            <svg aria-hidden="true" class="svg-icon va-text-top iconFire" width="18" height="18" viewBox="0 0 18 18"><path d="M7.48.01c.87 2.4.44 3.74-.57 4.77-1.06 1.16-2.76 2.02-3.93 3.7C1.4 10.76 1.13 15.72 6.8 17c-2.38-1.28-2.9-5-.32-7.3-.66 2.24.57 3.67 2.1 3.16 1.5-.52 2.5.58 2.46 1.84-.02.86-.33 1.6-1.22 2A6.17 6.17 0 0 0 15 10.56c0-3.14-2.74-3.56-1.36-6.2-1.64.14-2.2 1.24-2.04 3.03.1 1.2-1.11 2-2.02 1.47-.73-.45-.72-1.31-.07-1.96 1.36-1.36 1.9-4.52-2.03-6.88L7.45 0l.03.01z"/></svg>
            <span title="1195903">1.2m</span> watchers
        </span>

        <span class="mr8"><span title="1801277">1.8m</span> questions</span>

        <a class="float-right fc-orange-400" href="/feeds/tag/javascript" title="Add this tag to your RSS reader"><svg aria-hidden="true" class="svg-icon iconRss" width="18" height="18" viewBox="0 0 18 18"><path d="M1 3c0-1.1.9-2 2-2h12a2 2 0 0 1 2 2v12a2 2 0 0 1-2 2H3a2 2 0 0 1-2-2V3zm14.5 12C15.5 8.1 9.9 2.5 3 2.5V5a10 10 0 0 1 10 10h2.5zm-5 0A7.5 7.5 0 0 0 3 7.5V10a5 5 0 0 1 5 5h2.5zm-5 0A2.5 2.5 0 0 0 3 12.5V15h2.5z"/></svg></a>
</div>
        <div>JavaScript (not to be confused with Java) is a high-level, dynamic, multi-paradigm, object-oriented, prototype-based, weakly-typed language used for both client-side and server-side scripting. Its primary use is in rendering and manipulating of web pages. Use this tag for questions regarding ECMAScript and its various dialects/implementations (excluding ActionScript and Google-Apps-Script). <a href="/questions/tagged/javascript">View tag</a></div></div>

为避免可能的评论,请让我补充:我知道SO具有用于此类搜索的API,但是(i)如我提到的,我随机选择了SO的标签页面,我想解决此问题尽可能地通用; (ii)如果我理解正确,this cannot be done with the SO API; (iii)即使可以,我仍然想学习如何使用抓取技术来做到这一点。

1 个答案:

答案 0 :(得分:1)

以下内容构造了检索该信息所需的最小URL,然后从这些URL中提取所需的信息,并将其插入作为列表row插入到最终列表results中的变量中。 。最终列表将最终转换为数据框。

您可以使用的构造循环所有页面

https://stackoverflow.com/tags?page={}

由于本周未对每个标签报告相同的时间段,因此不确定您想要的本周号等信息。如果您可以说明要如何处理,我将更新答案。看起来单位可以是日,周或月(其中2个)。

我认为在每周/每月等时间段内提出的问题是动态加载的,因此您不一定总是有两个度量值。为此,我添加了if语句来处理此问题。通过测试len的{​​{1}}直到== 2,您可以继续发出请求,直到获得该信息为止。

frequencies


结合使用原始代码和Selenium,以确保加载了动态内容:

import requests
from bs4 import BeautifulSoup as bs
import urllib.parse
import pandas as pd

url = 'https://stackoverflow.com/tags/{}/popup'
page_url = 'https://stackoverflow.com/tags?page={}'
results = []

with requests.Session() as s:
    r = s.get('https://stackoverflow.com/tags')
    soup = bs(r.content, 'lxml')
    num_pages = int(soup.select('.page-numbers')[-2].text)

    for page in range(1, 3): # for page in range(1, num_pages):
        frequency1 = []
        frequency2 = []
        if page > 1:
            r = s.get(page_url.format(page))
            soup = bs(r.content, 'lxml')

        tags = [(item.text, urllib.parse.quote(item.text)) for item in soup.select('.post-tag')]

        for item in soup.select('.stats-row'):
            frequencies = item.select('a')
            frequency1.append(frequencies[0].text)
            if len(frequencies) == 2:
                frequency2.append(frequencies[1].text)
            else:
                frequency2.append('Not loaded') 
        i = 0
        for tag in tags:
            r = s.get(url.format(tag[1]))
            soup = bs(r.content, 'lxml')
            description = soup.select_one('div:not([class])').text
            stats = [item['title'] for item in soup.select('[title]')]
            total_watchers = stats[0]
            total_questions = stats[1]
            row = [tag[0], description, total_watchers, total_questions, frequency1[i],  frequency2[i]]
            results.append(row)
            i+=1
df = pd.DataFrame(results, columns = ['Tag', 'Description', 'Total Watchers', 'Total Questions', 'Frequency1', 'Frequency2'])