19.1. HTMLParser — Simple HTML and XHTML parser


The HTMLParser module has been renamed to html.parser in Python 3. The 2to3 tool will automatically adapt imports when converting your sources to Python 3.

New in version 2.2.

Source code: Lib/HTMLParser.py

This module defines a class HTMLParser which serves as the basis for parsing text files formatted in HTML (HyperText Mark-up Language) and XHTML. Unlike the parser in htmllib, this parser is not based on the SGML parser in sgmllib.

class HTMLParser.HTMLParser

An HTMLParser instance is fed HTML data and calls handler methods when start tags, end tags, text, comments, and other markup elements are encountered. The user should subclass HTMLParser and override its methods to implement the desired behavior.

The HTMLParser class is instantiated without arguments.

Unlike the parser in htmllib, this parser does not check that end tags match start tags or call the end-tag handler for elements which are closed implicitly by closing an outer element.

An exception is defined as well:

exception HTMLParser.HTMLParseError

HTMLParser is able to handle broken markup, but in some cases it might raise this exception when it encounters an error while parsing. This exception provides three attributes: msg is a brief message explaining the error, lineno is the number of the line on which the broken construct was detected, and offset is the number of characters into the line at which the construct starts.

19.1.1. Example HTML Parser Application

As a basic example, below is a simple HTML parser that uses the HTMLParser class to print out start tags, end tags and data as they are encountered:

from HTMLParser import HTMLParser

# create a subclass and override the handler methods
class MyHTMLParser(HTMLParser):
    def handle_starttag(self, tag, attrs):
        print "Encountered a start tag:", tag
    def handle_endtag(self, tag):
        print "Encountered an end tag :", tag
    def handle_data(self, data):
        print "Encountered some data  :", data

# instantiate the parser and fed it some HTML
parser = MyHTMLParser()
            '<body><h1>Parse me!</h1></body></html>')

The output will then be:

Encountered a start tag: html
Encountered a start tag: head
Encountered a start tag: title
Encountered some data  : Test
Encountered an end tag : title
Encountered an end tag : head
Encountered a start tag: body
Encountered a start tag: h1
Encountered some data  : Parse me!
Encountered an end tag : h1
Encountered an end tag : body
Encountered an end tag : html

19.1.2. HTMLParser Methods

HTMLParser instances have the following methods:


Feed some text to the parser. It is processed insofar as it consists of complete elements; incomplete data is buffered until more data is fed or close() is called. data can be either unicode or str, but passing unicode is advised.


Force processing of all buffered data as if it were followed by an end-of-file mark. This method may be redefined by a derived class to define additional processing at the end of the input, but the redefined version should always call the HTMLParser base class method close().


Reset the instance. Loses all unprocessed data. This is called implicitly at instantiation time.


Return current line number and offset.


Return the text of the most recently opened start tag. This should not normally be needed for structured processing, but may be useful in dealing with HTML “as deployed” or for re-generating input with minimal changes (whitespace between attributes can be preserved, etc.).

The following methods are called when data or markup elements are encountered and they are meant to be overridden in a subclass. The base class implementations do nothing (except for handle_startendtag()):

HTMLParser.handle_starttag(tag, attrs)

This method is called to handle the start of a tag (e.g. <div id="main">).

The tag argument is the name of the tag converted to lower case. The attrs argument is a list of (name, value) pairs containing the attributes found inside the tag’s <> brackets. The name will be translated to lower case, and quotes in the value have been removed, and character and entity references have been replaced.

For instance, for the tag <A HREF="http://www.cwi.nl/">, this method would be called as handle_starttag('a', [('href', 'http://www.cwi.nl/')]).

Changed in version 2.6: All entity references from htmlentitydefs are now replaced in the attribute values.


This method is called to handle the end tag of an element (e.g. </div>).

The tag argument is the name of the tag converted to lower case.

HTMLParser.handle_startendtag(tag, attrs)

Similar to handle_starttag(), but called when the parser encounters an XHTML-style empty tag (<img ... />). This method may be overridden by subclasses which require this particular lexical information; the default implementation simply calls handle_starttag() and handle_endtag().


This method is called to process arbitrary data (e.g. text nodes and the content of <script>...</script> and <style>...</style>).


This method is called to process a named character reference of the form &name; (e.g. &gt;), where name is a general entity reference (e.g. 'gt').


This method is called to process decimal and hexadecimal numeric character references of the form &#NNN; and &#xNNN;. For example, the decimal equivalent for &gt; is &#62;, whereas the hexadecimal is &#x3E;; in this case the method will receive '62' or 'x3E'.


This method is called when a comment is encountered (e.g. <!--comment-->).

For example, the comment <!-- comment --> will cause this method to be called with the argument ' comment '.

The content of Internet Explorer conditional comments (condcoms) will also be sent to this method, so, for <!--[if IE 9]>IE9-specific content<![endif]-->, this method will receive '[if IE 9]>IE-specific content<![endif]'.


This method is called to handle an HTML doctype declaration (e.g. <!DOCTYPE html>).

The decl parameter will be the entire contents of the declaration inside the <!...> markup (e.g. 'DOCTYPE html').


This method is called when a processing instruction is encountered. The data parameter will contain the entire processing instruction. For example, for the processing instruction <?proc color='red'>, this method would be called as handle_pi("proc color='red'").


The HTMLParser class uses the SGML syntactic rules for processing instructions. An XHTML processing instruction using the trailing '?' will cause the '?' to be included in data.


This method is called when an unrecognized declaration is read by the parser.

The data parameter will be the entire contents of the declaration inside the <![...]> markup. It is sometimes useful to be overridden by a derived class.

19.1.3. Examples

The following class implements a parser that will be used to illustrate more examples:

from HTMLParser import HTMLParser
from htmlentitydefs import name2codepoint

class MyHTMLParser(HTMLParser):
    def handle_starttag(self, tag, attrs):
        print "Start tag:", tag
        for attr in attrs:
            print "     attr:", attr
    def handle_endtag(self, tag):
        print "End tag  :", tag
    def handle_data(self, data):
        print "Data     :", data
    def handle_comment(self, data):
        print "Comment  :", data
    def handle_entityref(self, name):
        c = unichr(name2codepoint[name])
        print "Named ent:", c
    def handle_charref(self, name):
        if name.startswith('x'):
            c = unichr(int(name[1:], 16))
            c = unichr(int(name))
        print "Num ent  :", c
    def handle_decl(self, data):
        print "Decl     :", data

parser = MyHTMLParser()

Parsing a doctype:

>>> parser.feed('<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" '
...             '"http://www.w3.org/TR/html4/strict.dtd">')
Decl     : DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd"

Parsing an element with a few attributes and a title:

>>> parser.feed('<img src="python-logo.png" alt="The Python logo">')
Start tag: img
     attr: ('src', 'python-logo.png')
     attr: ('alt', 'The Python logo')
>>> parser.feed('<h1>Python</h1>')
Start tag: h1
Data     : Python
End tag  : h1

The content of script and style elements is returned as is, without further parsing:

>>> parser.feed('<style type="text/css">#python { color: green }</style>')
Start tag: style
     attr: ('type', 'text/css')
Data     : #python { color: green }
End tag  : style
>>> parser.feed('<script type="text/javascript">'
...             'alert("<strong>hello!</strong>");</script>')
Start tag: script
     attr: ('type', 'text/javascript')
Data     : alert("<strong>hello!</strong>");
End tag  : script

Parsing comments:

>>> parser.feed('<!-- a comment -->'
...             '<!--[if IE 9]>IE-specific content<![endif]-->')
Comment  :  a comment
Comment  : [if IE 9]>IE-specific content<![endif]

Parsing named and numeric character references and converting them to the correct char (note: these 3 references are all equivalent to '>'):

>>> parser.feed('&gt;&#62;&#x3E;')
Named ent: >
Num ent  : >
Num ent  : >

Feeding incomplete chunks to feed() works, but handle_data() might be called more than once:

>>> for chunk in ['<sp', 'an>buff', 'ered ', 'text</s', 'pan>']:
...     parser.feed(chunk)
Start tag: span
Data     : buff
Data     : ered
Data     : text
End tag  : span

Parsing invalid HTML (e.g. unquoted attributes) also works:

>>> parser.feed('<p><a class=link href=#main>tag soup</p ></a>')
Start tag: p
Start tag: a
     attr: ('class', 'link')
     attr: ('href', '#main')
Data     : tag soup
End tag  : p
End tag  : a