Question

以下是我要清理的一段html示例：

<figure class="floatRight" style="margin-left: 30px">
<a class="zoomFunction alignLeft" href="https://www.thieme-connect.de/media/synthesis/EFirst/lookinside/ss-2015-c0259-st_10-1055_s-0034-1378861-1.jpg"><img src="https://www.thieme-connect.de/media/synthesis/EFirst/lookinside/thumbnails/ss-2015-c0259-st_10-1055_s-0034-1378861-1.jpg"/></a>
<figcaption></figcaption>
</figure>
<p>
<a name="N65743"></a>
</p><h3>Abstract</h3>
<p>2-<span class="i">tert</span>-Butyl-5-iodoindolizine underwent Sonogashira reaction with acetylenes in the presence of dichlorobis(triphenylphosphine)palladium, copper(I) iodide, and triethylamine in acetonitrile to give to the corresponding 5-ethynylindolizines in high yields; 5-iodo-2-phenylindolizine and 5-bromo-2-<span class="i">tert</span>-butylindolizine did not undergo the reaction. Several structures were characterized by X-ray. The 5-ethynylindolizines did not undergo cyclization to give cycl[3.2.2]azines.</p>
<div class="articleKeywords">
<a name="N65760"></a>
<h3>Key words</h3>
5-iodoindolizines - 
        Sonogashira reaction - 
        5-ethynylindolizine - 
        X-ray
      </div>
<a name="N67312"></a>
<h3>Supporting Information</h3>
<ul class="linkList">Supporting information for this article is available online at http://dx.doi.org/10.1055/s-0034-1378861.<li>
<a class="gotolink" href="https://www.thieme-connect.de/media/synthesis/EFirst/supmat/sup_ss-2015-c0259-st_10-1055_s-0034-1378861.pdf">Supporting Information</a>
</li>
</ul>

我基本上会做的是这样的事情：

from bs4 import BeautifulSoup


with open("test.xml", 'r') as file:
    soup = BeautifulSoup(file.read(), "lxml")

abstract = soup

[tag.extract() for tag in abstract("a", attrs={"name": True})]
[tag.extract() for tag in abstract("h3")]
[tag.extract() for tag in abstract("ul", attrs={"class": "linkList"})]
[tag.extract() for tag in abstract("a", attrs={"class": "gotolink"})]

print(abstract)

我希望多个extract（）行清除每个匹配的标记。但是，只有第一个有效！我可以摆脱＆＃34; a＆＃34;标签，但不是＆＃34; h3＆＃34;标签。如果我评论第一个提取行（＆＃34; a＆＃34;标记的那个），我可以摆脱＆＃34; h3＆＃34;标签，但不是其他标签。

有点奇怪。你知道我为什么会这样做吗？

我使用从pip

新安装的BeautifulSoup4 4.4.0

Answer 1

诀窍是在每次提取后创建一个新的Beautiful Soup对象，并在这个新对象上执行下一个提取。

这可能看起来有点难看，但它确实有效：

<强> clean.py

from bs4 import BeautifulSoup


with open("test.xml", 'r') as file:
    soup = BeautifulSoup(file.read(), "lxml")

abstract = soup

[tag.extract() for tag in abstract("a", attrs={"name": True})]
abstract = BeautifulSoup(str(abstract))
[tag.extract() for tag in abstract("h3")]
abstract = BeautifulSoup(str(abstract))
[tag.extract() for tag in abstract("ul", attrs={"class": "linkList"})]
abstract = BeautifulSoup(str(abstract))
[tag.extract() for tag in abstract("a", attrs={"class": "gotolink"})]

print(abstract)

输出

清洁前

(bs4extract)macbook:bs4extract joeyoung$ cat test.xml 
<figure class="floatRight" style="margin-left: 30px">
<a class="zoomFunction alignLeft" href="https://www.thieme-connect.de/media/synthesis/EFirst/lookinside/ss-2015-c0259-st_10-1055_s-0034-1378861-1.jpg"><img src="https://www.thieme-connect.de/media/synthesis/EFirst/lookinside/thumbnails/ss-2015-c0259-st_10-1055_s-0034-1378861-1.jpg"/></a>
<figcaption></figcaption>
</figure>
<p>
<a name="N65743"></a>
</p><h3>Abstract</h3>
<p>2-<span class="i">tert</span>-Butyl-5-iodoindolizine underwent Sonogashira reaction with acetylenes in the presence of dichlorobis(triphenylphosphine)palladium, copper(I) iodide, and triethylamine in acetonitrile to give to the corresponding 5-ethynylindolizines in high yields; 5-iodo-2-phenylindolizine and 5-bromo-2-<span class="i">tert</span>-butylindolizine did not undergo the reaction. Several structures were characterized by X-ray. The 5-ethynylindolizines did not undergo cyclization to give cycl[3.2.2]azines.</p>
<div class="articleKeywords">
<a name="N65760"></a>
<h3>Key words</h3>
5-iodoindolizines - 
        Sonogashira reaction - 
        5-ethynylindolizine - 
        X-ray
      </div>
<a name="N67312"></a>
<h3>Supporting Information</h3>
<ul class="linkList">Supporting information for this article is available online at http://dx.doi.org/10.1055/s-0034-1378861.<li>
<a class="gotolink" href="https://www.thieme-connect.de/media/synthesis/EFirst/supmat/sup_ss-2015-c0259-st_10-1055_s-0034-1378861.pdf">Supporting Information</a>
</li>
</ul>

清洁后

(bs4extract)macbook:bs4extract joeyoung$ python clean.py 
<html><body><figure class="floatRight" style="margin-left: 30px">
<a class="zoomFunction alignLeft" href="https://www.thieme-connect.de/media/synthesis/EFirst/lookinside/ss-2015-c0259-st_10-1055_s-0034-1378861-1.jpg"><img src="https://www.thieme-connect.de/media/synthesis/EFirst/lookinside/thumbnails/ss-2015-c0259-st_10-1055_s-0034-1378861-1.jpg"/></a>
<figcaption></figcaption>
</figure>
<p>
</p>
<p>2-<span class="i">tert</span>-Butyl-5-iodoindolizine underwent Sonogashira reaction with acetylenes in the presence of dichlorobis(triphenylphosphine)palladium, copper(I) iodide, and triethylamine in acetonitrile to give to the corresponding 5-ethynylindolizines in high yields; 5-iodo-2-phenylindolizine and 5-bromo-2-<span class="i">tert</span>-butylindolizine did not undergo the reaction. Several structures were characterized by X-ray. The 5-ethynylindolizines did not undergo cyclization to give cycl[3.2.2]azines.</p>
<div class="articleKeywords">

5-iodoindolizines - 
        Sonogashira reaction - 
        5-ethynylindolizine - 
        X-ray
      </div>
</body></html>

Answer 2

对不起，伙计们，这个bug实际上来自BeautifulSoup。降级到4.3.2-3时，完全相同的代码可以正常工作。我会举报。对不起，在发布之前我没有检查过。

使用BeautifulSoup进行多次提取并不起作用

2 个答案: