11

I'm trying to 'defrontpagify' the html of a MS FrontPage generated website, and I'm writing a BeautifulSoup script to do it.

However, I've gotten stuck on the part where I try to strip a particular attribute (or list attributes) from every tag in the document that contains them. The code snippet:

REMOVE_ATTRIBUTES = ['lang','language','onmouseover','onmouseout','script','style','font',
                        'dir','face','size','color','style','class','width','height','hspace',
                        'border','valign','align','background','bgcolor','text','link','vlink',
                        'alink','cellpadding','cellspacing']

# remove all attributes in REMOVE_ATTRIBUTES from all tags, 
# but preserve the tag and its content. 
for attribute in REMOVE_ATTRIBUTES:
    for tag in soup.findAll(attribute=True):
        del(tag[attribute])

It runs without error, but doesn't actually strip any of the attributes. When I run it without the outer loop, just hard coding a single attribute (soup.findAll('style'=True), it works.

Anyone see know the problem here?

PS - I don't much like the nested loops either. If anyone knows a more functional, map/filter-ish style, I'd love to see it.

bgibson
  • 17,379
  • 8
  • 29
  • 45
  • 2
    For me, it works if `soup.findAll(attribute=True)` is changed to simply `soup.findAll()`. – mzjn Jan 29 '12 at 09:36
  • Nice catch, that does indeed work. Pretty obvious in hindsight, don't need to check the attribute value twice. Only problem is it checks all the attributes of every tag in the doc, and takes twice as long to run, but 5s vs 2.5s for ~15 pages isn't a big deal here. – bgibson Jan 30 '12 at 07:28

5 Answers5

10

The line

for tag in soup.findAll(attribute=True):

does not find any tags. There might be a way to use findAll; I'm not sure. However, this works:

import BeautifulSoup
REMOVE_ATTRIBUTES = [
    'lang','language','onmouseover','onmouseout','script','style','font',
    'dir','face','size','color','style','class','width','height','hspace',
    'border','valign','align','background','bgcolor','text','link','vlink',
    'alink','cellpadding','cellspacing']

doc = '''<html><head><title>Page title</title></head><body><p id="firstpara" align="center">This is <i>paragraph</i> <a onmouseout="">one</a>.<p id="secondpara" align="blah">This is <i>paragraph</i> <b>two</b>.</html>'''
soup = BeautifulSoup.BeautifulSoup(doc)
for tag in soup.recursiveChildGenerator():
    try:
        tag.attrs = [(key,value) for key,value in tag.attrs
                     if key not in REMOVE_ATTRIBUTES]
    except AttributeError: 
        # 'NavigableString' object has no attribute 'attrs'
        pass
print(soup.prettify())

Note this this code will only work in Python 3. If you need it to work in Python 2, see Nóra's answer below.

Pikamander2
  • 7,332
  • 3
  • 48
  • 69
unutbu
  • 842,883
  • 184
  • 1,785
  • 1,677
  • Good enough, thanks! As for findAll, I'm sure I'm just referencing the `attribute` variable wrong somehow, since hardcoding the attribute name in its place does work. Will dig into that more on the next pass, after I get the whole script working. – bgibson Jan 29 '12 at 00:30
6

Here's a Python 2 version of unutbu's answer:

REMOVE_ATTRIBUTES = ['lang','language','onmouseover']

doc = '''<html><head><title>Page title</title></head><body><p id="firstpara" align="center">This is <i>paragraph</i> <a onmouseout="">one</a>.<p id="secondpara" align="blah">This is <i>paragraph</i> <b>two</b>.</html>'''

soup = BeautifulSoup.BeautifulSoup(doc)

for tag in soup.recursiveChildGenerator():
    if hasattr(tag, 'attrs'):
        tag.attrs = {key:value for key,value in tag.attrs.iteritems()
                    if key not in REMOVE_ATTRIBUTES}
Pikamander2
  • 7,332
  • 3
  • 48
  • 69
Nóra
  • 61
  • 1
  • 1
6

Just ftr: the problem here is that if you pass HTML attributes as keyword arguments, the keyword is the name of the attribute. So your code is searching for tags with an attribute of name attribute, as the variable does not get expanded.

This is why

  1. hard-coding your attribute name worked[0]
  2. the code does not fail. The search just doesn't match any tags

To fix the problem, pass the attribute you are looking for as a dict:

for attribute in REMOVE_ATTRIBUTES:
    for tag in soup.find_all(attrs={attribute: True}):
        del tag[attribute]

Hth someone in the future, dtk

[0]: Although it needs to be find_all(style=True) in your example, without the quotes, because SyntaxError: keyword can't be an expression

dtk
  • 2,197
  • 2
  • 26
  • 19
3

I use this one:

if "align" in div.attrs:
    del div.attrs["align"]

or

if "align" in div.attrs:
    div.attrs.pop("align")

Thanks to https://stackoverflow.com/a/22497855/1907997

oxidworks
  • 1,563
  • 1
  • 14
  • 37
2

I use this method to remove a list of attributes, very compact :

attributes_to_del = ["style", "border", "rowspan", "colspan", "width", "height", 
                     "align", "valign", "color", "bgcolor", "cellspacing", 
                     "cellpadding", "onclick", "alt", "title"]
for attr_del in attributes_to_del: 
    [s.attrs.pop(attr_del) for s in soup.find_all() if attr_del in s.attrs]
Cyril
  • 157
  • 2
  • 2