How to prettify HTML so tag attributes will remain

2020-08-17 17:48发布

问题:

I got this little piece of code:

text = """<html><head></head><body>
    <h1 style="
    text-align: center;
">Main site</h1>
    <div>
        <p style="
    color: blue;
    text-align: center;
">text1
        </p>
        <p style="
    color: blueviolet;
    text-align: center;
">text2
        </p>
    </div>
    <div>
        <p style="text-align:center">
            <img src="./foo/test.jpg" alt="Testing static images" style="
">
        </p>
    </div>
</body></html>
"""

import sys
import re
import bs4


def prettify(soup, indent_width=4):
    r = re.compile(r'^(\s*)', re.MULTILINE)
    return r.sub(r'\1' * indent_width, soup.prettify())

soup = bs4.BeautifulSoup(text, "html.parser")
print(prettify(soup))

The output of the above snippet right now is:

<html>
    <head>
    </head>
    <body>
        <h1 style="
                text-align: center;
">
            Main site
        </h1>
        <div>
            <p style="
                color: blue;
                text-align: center;
">
                text1
            </p>
            <p style="
                color: blueviolet;
                text-align: center;
">
                text2
            </p>
        </div>
        <div>
            <p style="text-align:center">
                <img alt="Testing static images" src="./foo/test.jpg" style="
"/>
            </p>
        </div>
    </body>
</html>

I'd like to figure out how to format the output so it becomes this instead:

<html>
    <head>
    </head>
    <body>
        <h1 style="text-align: center;">
            Main site
        </h1>
        <div>
            <p style="color: blue;text-align: center;">
                text1
            </p>
            <p style="color: blueviolet;text-align: center;">
                text2
            </p>
        </div>
        <div>
            <p style="text-align:center">
                <img alt="Testing static images" src="./foo/test.jpg" style=""/>
            </p>
        </div>
    </body>
</html>

Said otherwise, I'd like to keep html statements such as <tag attrib1=value1 attrib2=value2 ... attribn=valuen> in one single line if possible. When I say "if possible" I mean without screwing up the value of the attributes themselves (value1, value2, ..., valuen).

Is this possible to achieve with beautifulsoup4? As far I've read in the docs it seems you can use a custom formatter but I don't know how I could have a custom formatter so it can accomplish the described requirements.

EDIT:

@alecxe solution is quite simple, unfortunately fails in some more complex cases like the below one, ie:

test1 = """
<div id="dialer-capmaign-console" class="fill-vertically" style="flex: 1 1 auto;">
    <div id="sessionsGrid" data-columns="[
        { field: 'dialerSession.startTime', format:'{0:G}', title:'Start time', width:122 },
        { field: 'dialerSession.endTime', format:'{0:G}', title:'End time', width:122, attributes: {class:'tooltip-column'}},
        { field: 'conversationStartTime', template: cty.ui.gct.duration_dialerSession_conversationStartTime_endTime, title:'Duration', width:80},
        { field: 'dialerSession.caller.lastName',template: cty.ui.gct.person_dialerSession_caller_link, title:'Caller', width:160 },
        { field: 'noteType',template:cty.ui.gct.nameDescription_noteType, title:'Note type', width:150, attributes: {class:'tooltip-column'}},
        { field: 'note', title:'Note'}
        ]">
</div>
</div>
"""

from bs4 import BeautifulSoup
import re


def prettify(soup, indent_width=4, single_lines=True):
    if single_lines:
        for tag in soup():
            for attr in tag.attrs:
                print(tag.attrs[attr], tag.attrs[attr].__class__)
                tag.attrs[attr] = " ".join(
                    tag.attrs[attr].replace("\n", " ").split())

    r = re.compile(r'^(\s*)', re.MULTILINE)
    return r.sub(r'\1' * indent_width, soup.prettify())


def html_beautify(text):
    soup = BeautifulSoup(text, "html.parser")
    return prettify(soup)

print(html_beautify(test1))

TRACEBACK:

dialer-capmaign-console <class 'str'>
['fill-vertically'] <class 'list'>
Traceback (most recent call last):
  File "d:\mcve\x.py", line 35, in <module>
    print(html_beautify(test1))
  File "d:\mcve\x.py", line 33, in html_beautify
    return prettify(soup)
  File "d:\mcve\x.py", line 25, in prettify
    tag.attrs[attr].replace("\n", " ").split())
AttributeError: 'list' object has no attribute 'replace'

回答1:

BeautifulSoup tried to preserve the newlines and multiple spaces you had in the attribute values in the input HTML.

One workaround here would be to iterate over the element attributes and clean them up prior to prettifying - removing the newlines and replacing multiple consecutive spaces with a single space:

for tag in soup():
    for attr in tag.attrs:
        tag.attrs[attr] = " ".join(tag.attrs[attr].replace("\n", " ").split())

print(soup.prettify())

Prints:

<html>
 <head>
 </head>
 <body>
  <h1 style="text-align: center;">
   Main site
  </h1>
  <div>
   <p style="color: blue; text-align: center;">
    text1
   </p>
   <p style="color: blueviolet; text-align: center;">
    text2
   </p>
  </div>
  <div>
   <p style="text-align:center">
    <img alt="Testing static images" src="./foo/test.jpg" style=""/>
   </p>
  </div>
 </body>
</html>

Update (to address the multi-valued attributes like class):

You just need to add a slight modification adding special handling for the case when an attribute is of a list type:

for tag in soup():
    tag.attrs = {
        attr: [" ".join(attr_value.replace("\n", " ").split()) for attr_value in value] 
              if isinstance(value, list)
              else " ".join(value.replace("\n", " ").split())
        for attr, value in tag.attrs.items()
    }


回答2:

While BeautifulSoup is more commonly used, HTML Tidy may be a better choice if you're working with quirks and have more specific requirements.

After installing the library for Python (pip install pytidylib) try the following code:

from tidylib import Tidy
tidy = Tidy()
# assign string to text
config = {
    "doctype": "omit",
    # "show-body-only": True
}
print tidy.tidy_document(text, options=config)[0]

tidy.tidy_document returns a tuple with the HTML and any errors that may have occurred. This code will output

<html>
  <head>
    <title></title>
  </head>
  <body>
    <h1 style="text-align: center;">
      Main site
    </h1>
    <div>
      <p style="color: blue; text-align: center;">
        text1
      </p>
      <p style="color: blueviolet; text-align: center;">
        text2
      </p>
    </div>
    <div>
      <p style="text-align:center">
        <img src="./foo/test.jpg" alt="Testing static images" style="">
      </p>
    </div>
  </body>
</html>

By uncommenting the "show-body-only": True for the second sample.

<div id="dialer-capmaign-console" class="fill-vertically" style="flex: 1 1 auto;">
  <div id="sessionsGrid" data-columns="[ { field: 'dialerSession.startTime', format:'{0:G}', title:'Start time', width:122 }, { field: 'dialerSession.endTime', format:'{0:G}', title:'End time', width:122, attributes: {class:'tooltip-column'}}, { field: 'conversationStartTime', template: cty.ui.gct.duration_dialerSession_conversationStartTime_endTime, title:'Duration', width:80}, { field: 'dialerSession.caller.lastName',template: cty.ui.gct.person_dialerSession_caller_link, title:'Caller', width:160 }, { field: 'noteType',template:cty.ui.gct.nameDescription_noteType, title:'Note type', width:150, attributes: {class:'tooltip-column'}}, { field: 'note', title:'Note'} ]"></div>
</div>

See more configuration for further options and customization. There are wrapping options specific to attributes which may help. As you can see, empty elements will only take up one line, and html-tidy will automatically try to add things like DOCTYPE, head and title tags.