++ed by:

1 non-PAUSE user.

Jeff Pinyan


YAPE::HTML - Yet Another Parser/Extractor for HTML


  use YAPE::HTML;
  use strict;
  my $content = "<html>...</html>";
  my $parser = YAPE::HTML->new($content);
  my ($extor,@fonts,@urls,@headings,@comments);
  # here is the tokenizing part
  while (my $chunk = $parser->next) {
    if ($chunk->type eq 'tag' and $chunk->tag eq 'font') {
      if (my $face = $chunk->get_attr('face')) {
        push @fonts, $face;
  # here we catch any errors
  unless ($parser->done) {
    die sprintf "bad HTML: %s (%s)",
      $parser->error, $parser->chunk;
  # here is the extracting part
  # <A> tags with HREF attributes
  # <IMG> tags with SRC attributes
  $extor = $parser->extract(a => ['href'], img => ['src']);
  while (my $chunk = $extor->()) {
    push @urls, $chunk->get_attr(
      $chunk->tag eq 'a' ? 'href' : 'src'
  # <H1>, <H2>, ..., <H6> tags
  $extor = $parser->extract(qr/^h[1-6]$/ => []);
  while (my $chunk = $extor->()) {
    push @headings, $chunk;
  # all comments
  $extor = $parser->extract(-COMMENT => []);
  while (my $chunk = $extor->()) {
    push @comments, $chunk;


The YAPE hierarchy of modules is an attempt at a unified means of parsing and extracting content. It attempts to maintain a generic interface, to promote simplicity and reusability. The API is powerful, yet simple. The modules do tokenization (which can be intercepted) and build trees, so that extraction of specific nodes is doable.


This module is yet another parser and tree-builder for HTML documents. It is designed to make extraction and modification of HTML documents simplistic. The API allows for easy custom additions to the document being parsed, and allows very specific tag, text, and comment extraction.


In addition to the base class, YAPE::HTML, there is the auxiliary class YAPE::HTML::Element (common to all YAPE base classes) that holds the individual nodes' classes. There is documentation for the node classes in that module's documentation.

HTML elements and their attributes are stored internally as lowercase strings. For clarification, that means that the tag <A HREF="FooBar.html"> is stored as

    TAG => 'a',
    ATTR => {
      href => 'FooBar.html',

This means that tags will be output in lowercase. There will be a feature in a future version to switch output case to capital letters.


  • YAPE::HTML::EMPTY(@tags)

    Adds to the internal hash of tags which never contain any out-of-tag content. This hash is %YAPE::HTML::EMPTY, and contains the following tag names: area, base, br, hr, img, input, link, meta, and param. Deletion from this hashmust be done manually. Adding to this hash automatically adds to the %OPEN hash, described next.

  • YAPE::HTML::OPEN(@tags)

    Adds to the internal hash of tags which do not require a closing tag. This hash is %YAPE::HTML::OPEN, and contains the following tag names: area, base, br, dd, dt, hr, img, input, li, link, meta, p, and param. Deletion from this hash must be done manually.

    There is a subtle difference between "empty" and "open" tags. For example, the <AREA> tag contains a few attributes, but there is no text associated with it (nor any other tags), and therefore, is "empty"; the <LI>, on the other hand,

    It is strongly suggested that for ease in parsing, any tags that you do not explicitly close have a / at the end of the tag:

      Here's my cat: <img src="cat.jpg" />

Methods for YAPE::HTML

  • use YAPE::HTML;

  • use YAPE::HTML qw( MyExt::Mod );

    If supplied no arguments, the module is loaded normally, and the node classes are given the proper inheritence (from YAPE::HTML::Element). If you supply a module (or list of modules), import will automatically include them (if needed) and set up their node classes with the proper inheritence -- that is, it will append YAPE::HTML to @MyExt::Mod::ISA, and YAPE::HTML::xxx to each node class's @ISA (where xxx is the name of the specific node class).

    It also copies the %OPEN and %EMPTY hashes, as well as the OPEN() and EMPTY() functions, into the MyExt::Mod namespace. This process is designed to save you from having to place @ISA assignments all over the place.

    It also copies the %SSI hash. This hash is not suggested to be altered, and therefore it does not have any public interface (you have to fiddle with it yourself). It exists to ensure an SSI is valid.

      package MyExt::Mod;
      use YAPE::HTML 'MyExt::Mod';
      # @MyExt::Mod::ISA = 'YAPE::HTML'
      # @MyExt::Mod::text::ISA = 'YAPE::HTML::text'
      # ...
      # being rather strict with the tags
      %OPEN = ();
      %EMPTY = ();
  • my $p = YAPE::HTML->new($HTML, $strict);

    Creates a YAPE::HTML object, using the contents of the $HTML string as its HTML to parse. The optional second argument determines whether this parser instance will demand strict comment parsing and require all tags to be closed with a closing tag or a / at the end of the tag (<HR />). Any true value (except for the special string -NO_STRICT) will turn strict parsing on. This is off by default. (This could be considered a bug.)

  • my $text = $p->chunk($len);

    Returns the next $len characters in the input string; $len defaults to 30 characters. This is useful for figuring out why a parsing error occurs.

  • my $done = $p->done;

    Returns true if the parser is done with the input string, and false otherwise.

  • my $errstr = $p->error;

    Returns the parser error message.

  • my $coderef = $p->extract(...);

    Returns a code reference that returns the next object that matches the criteria given in the arguments. This is a fundamental feature of the module, and you can extract that from "Extracting Sections".

  • my $node = $p->display(...);

    Returns a string representation of the entire content. It calls the parse method in case there is more data that has not yet been parsed. This calls the fullstring method on the root nodes. Check the YAPE::HTML::Element docs on the arguments to fullstring.

  • my $node = $p->next;

    Returns the next token, or undef if there is no valid token. There will be an error message (accessible with the error method) if there was a problem in the parsing.

  • my $node = $p->parse;

    Calls next until all the data has been parsed.

  • my $attr = $p->quote($string);

    Returns a quoted string, suitable for using as an attribute. It turns any embedded " characters into &quot;. This can also be called as a raw function:

      my $quoted = YAPE::HTML::quote($string);
  • my $root = $p->root;

    Returns an array reference holding the root of the tree structure -- for documents that contain multiple top-level tags, this will have more than one element.

  • my $state = $p->state;

    Returns the current state of the parser. It is one of the following values: close(TAG), comment, done, dtd, error, open(TAG), pi, ssi, text, text(script), or text(xmp). The open and close states contain the name of the element in parentheses (ex. open(img)). Tag names, as well as the names of attributes, are converted to lowercase. The state of text(script) refers to text found inside an <SCRIPT> element, and likewise for text(xmp).

  • my $HTMLnode = $p->top;

    Returns the first <HTML> node it finds in the tree structure.

Extracting Sections

YAPE::HTML allows comprehensive extraction of tags, text, comments, DTDs, PIs, and SSIs, using a simple, yet rich, syntax:

  my $extor = $parser->extract(
    TYPE => [ REQS ],

TYPE can be either the name of a tag ("table"), a regular expression that matches tags (qr/^t[drh]$/), or a special string to match all tags (-TAG), all text (-TEXT), all comments (-COMMENT), all DTDs (-DTD), all PIs (-PI), and all SSIs (-SSI).

REQS varies from element to element:

  • -TAG, -DTD, -PI, -SSI

    A list of attributes that the tag/DTD/PI/SSI must have.


    A list of strings and regexes that the content of the text/comment must have or match.

Here are some example uses:

  • all tags starting with "h"

      my $extor = $parser->extract(qr/^h/ => []);
  • all tags with an "align" attribute

      my $extor = $parser->extract(-TAG => ['align']);
  • all text containing the word "japhy"

      my $extor = $parser->extract(-TEXT => [qr/\bjaphy\b/i]);
  • tags involving links

      my $extor = $parser->extract(
        a => ['href'],
        area => ['href'],
        base => ['href'],
        body => ['background'],
        img => ['src'],
        # ...


This is a list of special features of YAPE::HTML.

  • On-the-fly cleaning of HTML

    If you aren't enforcing strict HTML syntax, then in the act of parsing HTML, if a tag that should be closed is not closed, it will be flagged for closing. That means that input like:


    will appear as:


    upon request for output. In addition, tags that are left dangling open at the end of an HTML document get closed. That means:


    will appear as:

  • Syntax-checking

    If strict checking is off, the only error you'll receive from mismatched HTML tags is a closing tag out-of-place.

    On the other hand, if you do enforce strict HTML syntax, you'll be informed of tags that do not get closed as well (that should be closed).


This is a listing of things to add to future versions of this module.


  • HTML entity translation (via HTML::Entities no doubt)

    Add a flag to the fullstring method of objects, -EXPAND, which will display &...; HTML escapes as the character representing them.

  • Toggle case of output (lower/upper case)

    Add a flag to the fullstring method of objects, -UPPER, which will display tag and attribute names in uppercase.

  • Super-strict syntax checking

    DTD-like strictness in regards to nesting of elements -- <LI> is not allowed to be outside an <OL> or <UL> element.


  • Make it faster, of course

    There's probably some inherent slowness to this method, but it works. And it supports the robust extract method.

  • Combine CLOSED and IMPLICIT

    Make three constants, CLOSED_NO, CLOSED_YES, and CLOSED_IMPL.


Following is a list of known or reported bugs.


  • Inheritence fixed again (fixed in 1.11)

  • Inheritence was fouled up (fixed in 1.10)


  • The above features aren't in here yet. ;)

  • Strict syntax-checking is not on by default.

  • This documentation might be incomplete.

  • DTD, PI, and SSI support is incomplete.

  • Probably need some more test cases.

  • SSI conditional tags don't contain content.


Visit YAPE's web site at http://www.pobox.com/~japhy/YAPE/.


The YAPE::HTML::Element documentation, for information on the node classes.


  Jeff "japhy" Pinyan

1 POD Error

The following errors were encountered while parsing the POD:

Around line 725:

You forgot a '=back' before '=head1'