| 1 | package HTML::PullParser; |
| 2 | |
| 3 | # $Id: PullParser.pm,v 2.6 2001/04/02 23:26:18 gisle Exp $ |
| 4 | |
| 5 | require HTML::Parser; |
| 6 | @ISA=qw(HTML::Parser); |
| 7 | $VERSION = sprintf("%d.%02d", q$Revision: 2.6 $ =~ /(\d+)\.(\d+)/); |
| 8 | |
| 9 | use strict; |
| 10 | use Carp (); |
| 11 | |
| 12 | sub new |
| 13 | { |
| 14 | my($class, %cnf) = @_; |
| 15 | |
| 16 | # Construct argspecs for the various events |
| 17 | my %argspec; |
| 18 | for (qw(start end text declaration comment process default)) { |
| 19 | my $tmp = delete $cnf{$_}; |
| 20 | next unless defined $tmp; |
| 21 | $argspec{$_} = $tmp; |
| 22 | } |
| 23 | Carp::croak("Info not collected for any events") |
| 24 | unless %argspec; |
| 25 | |
| 26 | my $file = delete $cnf{file}; |
| 27 | my $doc = delete $cnf{doc}; |
| 28 | Carp::croak("Can't parse from both 'doc' and 'file' at the same time") |
| 29 | if defined($file) && defined($doc); |
| 30 | Carp::croak("No 'doc' or 'file' given to parse from") |
| 31 | unless defined($file) || defined($doc); |
| 32 | |
| 33 | # Create object |
| 34 | $cnf{api_version} = 3; |
| 35 | my $self = $class->SUPER::new(%cnf); |
| 36 | |
| 37 | my $accum = $self->{pullparser_accum} = []; |
| 38 | while (my($event, $argspec) = each %argspec) { |
| 39 | $self->SUPER::handler($event => $accum, $argspec); |
| 40 | } |
| 41 | |
| 42 | if (defined $doc) { |
| 43 | $self->{pullparser_str_ref} = ref($doc) ? $doc : \$doc; |
| 44 | $self->{pullparser_str_pos} = 0; |
| 45 | } |
| 46 | else { |
| 47 | if (!ref($file) && ref(\$file) ne "GLOB") { |
| 48 | require IO::File; |
| 49 | $file = IO::File->new($file, "r") || return; |
| 50 | } |
| 51 | |
| 52 | $self->{pullparser_file} = $file; |
| 53 | } |
| 54 | $self; |
| 55 | } |
| 56 | |
| 57 | |
| 58 | sub handler |
| 59 | { |
| 60 | Carp::croak("Can't set handlers for HTML::PullParser"); |
| 61 | } |
| 62 | |
| 63 | |
| 64 | sub get_token |
| 65 | { |
| 66 | my $self = shift; |
| 67 | while (!@{$self->{pullparser_accum}} && !$self->{pullparser_eof}) { |
| 68 | if (my $f = $self->{pullparser_file}) { |
| 69 | # must try to parse more from the file |
| 70 | my $buf; |
| 71 | if (read($f, $buf, 512)) { |
| 72 | $self->parse($buf); |
| 73 | } else { |
| 74 | $self->eof; |
| 75 | $self->{pullparser_eof}++; |
| 76 | delete $self->{pullparser_file}; |
| 77 | } |
| 78 | } |
| 79 | elsif (my $sref = $self->{pullparser_str_ref}) { |
| 80 | # must try to parse more from the scalar |
| 81 | my $pos = $self->{pullparser_str_pos}; |
| 82 | my $chunk = substr($$sref, $pos, 512); |
| 83 | $self->parse($chunk); |
| 84 | $pos += length($chunk); |
| 85 | if ($pos < length($$sref)) { |
| 86 | $self->{pullparser_str_pos} = $pos; |
| 87 | } |
| 88 | else { |
| 89 | $self->eof; |
| 90 | $self->{pullparser_eof}++; |
| 91 | delete $self->{pullparser_str_ref}; |
| 92 | delete $self->{pullparser_str_pos}; |
| 93 | } |
| 94 | } |
| 95 | else { |
| 96 | die; |
| 97 | } |
| 98 | } |
| 99 | shift @{$self->{pullparser_accum}}; |
| 100 | } |
| 101 | |
| 102 | |
| 103 | sub unget_token |
| 104 | { |
| 105 | my $self = shift; |
| 106 | unshift @{$self->{pullparser_accum}}, @_; |
| 107 | $self; |
| 108 | } |
| 109 | |
| 110 | 1; |
| 111 | |
| 112 | |
| 113 | __END__ |
| 114 | |
| 115 | =head1 NAME |
| 116 | |
| 117 | HTML::PullParser - Alternative HTML::Parser interface |
| 118 | |
| 119 | =head1 SYNOPSIS |
| 120 | |
| 121 | use HTML::PullParser; |
| 122 | |
| 123 | $p = HTML::PullParser->new(file => "index.html", |
| 124 | start => 'event, tagname, @attr', |
| 125 | end => 'event, tagname', |
| 126 | ignore_elements => [qw(script style)], |
| 127 | ) || die "Can't open: $!"; |
| 128 | while (my $token = $p->get_token) { |
| 129 | #...do something with $token |
| 130 | } |
| 131 | |
| 132 | =head1 DESCRIPTION |
| 133 | |
| 134 | The HTML::PullParser is an alternative interface to the HTML::Parser class. |
| 135 | It basically turns the HTML::Parser inside out. You associate a file |
| 136 | (or any IO::Handle object or string) with the parser at construction time and |
| 137 | then repeatedly call $parser->get_token to obtain the tags and text |
| 138 | found in the parsed document. |
| 139 | |
| 140 | The following methods are provided: |
| 141 | |
| 142 | =over 4 |
| 143 | |
| 144 | =item $p = HTML::PullParser->new( file => $file, %options ) |
| 145 | |
| 146 | =item $p = HTML::PullParser->new( doc => \$doc, %options ) |
| 147 | |
| 148 | A C<HTML::PullParser> can be made to parse from either a file or a |
| 149 | literal document based on whether the C<file> or C<doc> option is |
| 150 | passed to the parser's constructor. |
| 151 | |
| 152 | The C<file> passed in can either be a file name or a file handle |
| 153 | object. If a file name is passed, and it can't be opened for reading, |
| 154 | then the constructor will return an undefined value and $! will tell |
| 155 | you why it failed. Otherwise the argument is taken to be some object |
| 156 | that the C<HTML::PullParser> can read() from when it needs more data. |
| 157 | The stream will be read() until EOF, but not closed. |
| 158 | |
| 159 | A C<doc> can be passed plain or as a reference |
| 160 | to a scalar. If a reference is passed then the value of this scalar |
| 161 | should not be changed before all tokens have been extracted. |
| 162 | |
| 163 | Next the information to be returned for the different token types must |
| 164 | be set up. This is done by simply assosiating an argspec (as defined |
| 165 | in L<HTML::Parser>) with the events you have an interrest in. For |
| 166 | instance, if you want C<start> tokens to be reported as the string |
| 167 | C<'S'> followed by the tagname and the attributes you might pass an |
| 168 | C<start>-option like this: |
| 169 | |
| 170 | $p = HTML::Parser-New( doc => $doc_to_parse, |
| 171 | start => '"S", tagname, @attr', |
| 172 | end => '"E", tagname', |
| 173 | ); |
| 174 | |
| 175 | At last other C<HTML::Parser> options, like C<ignore_tags>, and |
| 176 | C<unbroken_text>, can be passed in. Note that you should not use the |
| 177 | I<event>_h options to set up parser handlers. |
| 178 | |
| 179 | =item $token = $p->get_token |
| 180 | |
| 181 | This method will return the next I<token> found in the HTML document, |
| 182 | or C<undef> at the end of the document. The token is usually returned |
| 183 | as an array reference. The content of this array match the argspec |
| 184 | set up during C<HTML::PullParser> construction. |
| 185 | |
| 186 | =item $p->unget_token($token,...) |
| 187 | |
| 188 | If you find out you have read too many tokens you can push them back, |
| 189 | so that they are returned again the next time $p->get_token is called. |
| 190 | |
| 191 | =head1 EXAMPLES |
| 192 | |
| 193 | The 'eg/hform' script shows how we might parse the form section of |
| 194 | HTML::Documents using HTML::PullParser. |
| 195 | |
| 196 | =head1 SEE ALSO |
| 197 | |
| 198 | L<HTML::Parser>, L<HTML::TokeParser> |
| 199 | |
| 200 | =head1 COPYRIGHT |
| 201 | |
| 202 | Copyright 1998-2001 Gisle Aas. |
| 203 | |
| 204 | This library is free software; you can redistribute it and/or |
| 205 | modify it under the same terms as Perl itself. |
| 206 | |
| 207 | =cut |