=head1 NAME Whatpm::HTML - An HTML Parser and Serializer =head1 SYNOPSIS use Whatpm::HTML; my $s = q<...>; # $doc = an empty DOM |Document| object my $on_error = sub { my %error = @_; warn $error{type}, "\n"; }; Whatpm::HTML->parse_string ($s => $doc, $onerror); ## Now, |$doc| is the DOM representation of |$s|. =head1 DESCRIPTION The C module contains HTML parser and serializer. The HTML parser can be used to construct the DOM tree representation from an HTML document. The parsing and tree construction are done as described in the Web Application 1.0 specification. The HTML serializer can be used to obtain the HTML document representation of a DOM tree (or a tree fragment thereof). The serialization is performed as described in the Web Applications 1.0 specification for C DOM attribute. This module is part of Whatpm - Perl Modules for Web Hypertext Application Technologies. =head1 METHODS =over 4 =item [I<$doc> =] Whatpm::HTML->parse_string (I<$s>, I<$doc>[, I<$onerror>]); Parse a string I<$s> as an HTML document. The first argument, I<$s>, MUST be a string. It is parsed as a sequence of characters representing an HTML document. The second argument, I<$doc>, MUST be an empty read-write DOM C object. The HTML DOM tree is constructed onto this C object. The third argument, I<$onerror>, MUST be a reference to the error handler code. Whenever a parse error is detected, this code is invoked with an argument that contains a useless string that might describe what is wrong. The code MAY throw an exception, so that whole the parsing process aborts. Otherwise, the parser will continue to process the input. The code MUST NOT modify I<$s> or I<$doc>. If it does, then the result is undefined. This argument is optional; if missing, any parse error makes that string being Ced. B: To be a conforming user agent, the code MUST either abort the processing by throwing an exception at the first invocation or MUST continue the processing until the parser stops. The method returns the DOM C object (i.e. the second argument). Note that the C module provides a non-conforming implementation of DOM that only implements a subset that is necessary for the purpose of C's parsing and serializing. With this module, creating a new HTML C object from a string containing HTML document might be coded as: use Whatpm::HTML; use Whatpm::NanoDOM; my $doc = Whatpm::HTML->parse_string ($s => Whatpm::NanoDOM::Document->new, $onerror); =back =head1 LOW-LEVEL INTERFACE @@ TBW =head2 Application Cache Selection Algorithm Hook Once a parser I<$p> is instantiated by method C, a C reference can be set to C<< I<$p>->{application_cache_selection} >>. That C will be called back when the application cache selection algorithm MUST be run per HTML5. By default, C<< I<$p>->{application_cache_selection} >> is set to an empty subroutine. The subroutine will be invoked with an argument I, which is set to the manifest URI when the algorithm MUST be invoked with a manifest URI, or is set to C when the algorithm MUST be invoked without no manifest URI. =head1 ERROR REPORTS @@ TBW The list of the error types is available in Whatpm Error Types . =head1 TO DO Tokenizer should emit a sequence of character tokens as one token to improve performance. A method that accepts a byte stream as an input. Charset detection algorithm. Documentation for the setter of inner_html. And there are many "TODO"s and "ISSUE"s in the source code. =head1 SEE ALSO Whatpm . Whatpm Error Types . HTML5 . L. L. L. =head1 AUTHOR Wakaba . =head1 LICENSE Copyright 2007 Wakaba This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself. =cut # $Date: 2007/11/11 04:59:36 $