BeautifulSoup Parser

BeautifulSoup is a Python package that parses broken HTML, just like lxml supports it based on the parser of libxml2. BeautifulSoup uses a different parsing approach. It is not a real HTML parser but uses regular expressions to dive through tag soup. It is therefore more forgiving in some cases and less good in others. It is not uncommon that lxml/libxml2 parses and fixes broken HTML better, but BeautifulSoup has superiour support for encoding detection. It very much depends on the input which parser works better.

To prevent users from having to choose their parser library in advance, lxml can interface to the parsing capabilities of BeautifulSoup through the lxml.html.soupparser module. It provides three main functions: fromstring() and parse() to parse a string or file using BeautifulSoup into an lxml.html document, and convert_tree() to convert an existing BeautifulSoup tree into a list of top-level Elements.

Contents

Parsing with the soupparser

The functions fromstring() and parse() behave as known from ElementTree. The first returns a root Element, the latter returns an ElementTree.

There is also a legacy module called lxml.html.ElementSoup, which mimics the interface provided by ElementTree's own ElementSoup module. Note that the soupparser module was added in lxml 2.0.3. Previous versions of lxml 2.0.x only have the ElementSoup module.

Here is a document full of tag soup, similar to, but not quite like, HTML:

>>> tag_soup = '<meta><head><title>Hello</head><body onload=crash()>Hi all<p>'

all you need to do is pass it to the fromstring() function:

>>> from lxml.html.soupparser import fromstring
>>> root = fromstring(tag_soup)

To see what we have here, you can serialise it:

>>> from lxml.etree import tostring
>>> print(tostring(root, pretty_print=True).strip())
<html>
  <meta/>
  <head>
    <title>Hello</title>
  </head>
  <body onload="crash()">Hi all<p/></body>
</html>

Not quite what you'd expect from an HTML page, but, well, it was broken already, right? BeautifulSoup did its best, and so now it's a tree.

To control which Element implementation is used, you can pass a makeelement factory function to parse() and fromstring(). By default, this is based on the HTML parser defined in lxml.html.

For a quick comparison, libxml2 2.6.32 parses the same tag soup as follows. The main difference is that libxml2 tries harder to adhere to the structure of an HTML document and moves misplaced tags where they (likely) belong. Note, however, that the result can vary between parser versions.

<html>
  <head>
    <meta/>
    <title>Hello</title>
  </head>
  <body>
    <p>Hi all</p>
    <p/>
  </body>
</html>

Entity handling

By default, the BeautifulSoup parser also replaces the entities it finds by their character equivalent.

>>> tag_soup = '<body>&copy;&euro;&#45;&#245;&#445;<p>'
>>> body = fromstring(tag_soup).find('.//body')
>>> body.text
u'\xa9\u20ac-\xf5\u01bd'

If you want them back on the way out, you can just serialise with the default encoding, which is 'US-ASCII'.

>>> tostring(body)
'<body>&#169;&#8364;-&#245;&#445;<p/></body>'

>>> tostring(body, method="html")
'<body>&#169;&#8364;-&#245;&#445;<p></p></body>'

Any other encoding will output the respective byte sequences.

>>> tostring(body, encoding="utf-8")
'<body>\xc2\xa9\xe2\x82\xac-\xc3\xb5\xc6\xbd<p/></body>'

>>> tostring(body, method="html", encoding="utf-8")
'<body>\xc2\xa9\xe2\x82\xac-\xc3\xb5\xc6\xbd<p></p></body>'

>>> tostring(body, encoding='unicode')
u'<body>\xa9\u20ac-\xf5\u01bd<p/></body>'

>>> tostring(body, method="html", encoding='unicode')
u'<body>\xa9\u20ac-\xf5\u01bd<p></p></body>'

Using soupparser as a fallback

The downside of using this parser is that it is much slower than the HTML parser of lxml. So if performance matters, you might want to consider using soupparser only as a fallback for certain cases.

One common problem of lxml's parser is that it might not get the encoding right in cases where the document contains a <meta> tag at the wrong place. In this case, you can exploit the fact that lxml serialises much faster than most other HTML libraries for Python. Just serialise the document to unicode and if that gives you an exception, re-parse it with BeautifulSoup to see if that works better.

>>> tag_soup = '''\
... <meta http-equiv="Content-Type"
...       content="text/html;charset=utf-8" />
... <html>
...   <head>
...     <title>Hello W\xc3\xb6rld!</title>
...   </head>
...   <body>Hi all</body>
... </html>'''

>>> import lxml.html
>>> import lxml.html.soupparser

>>> root = lxml.html.fromstring(tag_soup)
>>> try:
...     ignore = tostring(root, encoding='unicode')
... except UnicodeDecodeError:
...     root = lxml.html.soupparser.fromstring(tag_soup)

Using only the encoding detection

If you prefer a 'real' (and fast) HTML parser instead of the regular expression based one in BeautifulSoup, you can still benefit from BeautifulSoup's support for encoding detection in the UnicodeDammit class.

>>> from BeautifulSoup import UnicodeDammit

>>> def decode_html(html_string):
...     converted = UnicodeDammit(html_string, isHTML=True)
...     if not converted.unicode:
...         raise UnicodeDecodeError(
...             "Failed to detect encoding, tried [%s]",
...             ', '.join(converted.triedEncodings))
...     # print converted.originalEncoding
...     return converted.unicode

>>> root = lxml.html.fromstring(decode_html(tag_soup))