我知道之前有人问过类似的问题，但似乎没有一个适用于这种特殊情况。我在几个网站上遇到过它，所以对于这个问题，我随机选择了 the first page of SO's own tags list.
如果您查看第一页的第一个条目，您会看到:
其中显示标签描述的开头、问题总数以及今天和本周提出的问题数。这些信息很容易被选中:
from selenium.webdriver import Chrome
driver = Chrome()
driver.get('https://stackoverflow.com/tags')
例如，关注 JavaScript 标签:
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
现在它变得更复杂了(至少对我而言):如果你将鼠标悬停在 JavaScript 标签上，你会看到这个弹出框:
该框包含完整的标签描述，以及问题和观察者的(四舍五入)数量。如果将鼠标悬停在“1.2m watchers”元素上，您会看到此工具提示:
这是调用此特定框的 url:
https://stackoverflow.com/tags/javascript/popup?_=1556571234452
该目标项目(以及问题总数)包含在此 html 中 span 的 title 中:
<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>
我想不通的是如何将所有这些信息放在一起以获得看起来像这样的输出(或数据框)，对于第一页中提到的所有标签:
Tag: JavaScript
Total questions: 1801277 #or whatever it is at the time this is performed
Watchers: 1195902 #same
.
.
etc.
为了抢占可能的评论，请让我补充:我知道 SO 有一个用于此类搜索的 API，但是 (i) 正如我提到的，我随机选择了 SO 的标签页，我想解决这个问题尽可能通用； (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。
请您参考如下方法:
以下构造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，并插入到变量中，这些变量作为列表 row 插入到最终列表 结果。最终列表在最后转换为数据框。
你可以循环所有页面的结构
https://stackoverflow.com/tags?page={}
不确定您想要什么关于这周的数字等，因为没有为每个标签报告相同的时间段。如果您可以说明您希望如何处理，我会更新答案。看起来单位可以是天、周或月(其中 2 个)。
我认为在周/月等时间段内提出的问题是动态加载的，因此您并不总是存在两个测量值。为此，我添加了一个 if 语句来处理这个问题。您可以继续发出请求，直到通过测试 frequencies 的 len 直到 == 2 获得该信息。
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'])
使用原始代码结合Selenium确保加载动态内容:
import requests
from bs4 import BeautifulSoup as bs
import urllib.parse
import pandas as pd
from selenium import webdriver
from selenium.webdriver.common.by import By
from selenium.webdriver.support.ui import WebDriverWait
from selenium.webdriver.support import expected_conditions as EC
url = 'https://stackoverflow.com/tags/{}/popup'
page_url = 'https://stackoverflow.com/tags?page={}'
results = []
d = webdriver.Chrome()
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 + 1):
if page > 1:
r = d.get(page_url.format(page))
WebDriverWait(d,10).until(EC.presence_of_all_elements_located((By.CSS_SELECTOR, '.stats-row a')))
soup = bs(d.page_source, 'lxml')
tags = [(item.text, urllib.parse.quote(item.text)) for item in soup.select('.post-tag')]
how_many = [item.text for item in soup.select('.stats-row a')]
frequency1 = how_many[0::2]
frequency2 = how_many[1::2]
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'])
d.quit()
print(df.head())
