3 # Copyright 1996-2007, Gisle Aas.
4 # Copyright 1999-2000, Michael A. Chase.
6 # This library is free software; you can redistribute it and/or
7 # modify it under the same terms as Perl itself.
10 use vars qw($VERSION @ISA);
12 $VERSION = '3.56'; # $Date: 2007/01/12 09:18:31 $
14 require HTML::Entities;
17 XSLoader::load('HTML::Parser', $VERSION);
22 my $self = bless {}, $class;
23 return $self->init(@_);
33 my $api_version = delete $arg{api_version} || (@_ ? 3 : 2);
34 if ($api_version >= 4) {
36 Carp::croak("API version $api_version not supported " .
37 "by HTML::Parser $VERSION");
40 if ($api_version < 3) {
41 # Set up method callbacks compatible with HTML-Parser-2.xx
42 $self->handler(text => "text", "self,text,is_cdata");
43 $self->handler(end => "end", "self,tagname,text");
44 $self->handler(process => "process", "self,token0,text");
45 $self->handler(start => "start",
46 "self,tagname,attr,attrseq,text");
48 $self->handler(comment =>
50 my($self, $tokens) = @_;
56 $self->handler(declaration =>
59 $self->declaration(substr($_[0], 2, -1));
63 if (my $h = delete $arg{handlers}) {
64 $h = {@$h} if ref($h) eq "ARRAY";
65 while (my($event, $cb) = each %$h) {
66 $self->handler($event => @$cb);
70 # In the end we try to assume plain attribute or handler
71 while (my($option, $val) = each %arg) {
72 if ($option =~ /^(\w+)_h$/) {
73 $self->handler($1 => @$val);
75 elsif ($option =~ /^(text|start|end|process|declaration|comment)$/) {
77 Carp::croak("Bad constructor option '$option'");
90 my($self, $file) = @_;
92 if (!ref($file) && ref(\$file) ne "GLOB") {
93 # Assume $file is a filename
95 open(F, $file) || return undef;
96 binmode(F); # should we? good for byte counts
101 while (read($file, $chunk, 512)) {
102 $self->parse($chunk) || last;
104 close($file) if $opened;
109 sub netscape_buggy_comment # legacy
113 Carp::carp("netscape_buggy_comment() is deprecated. " .
114 "Please use the strict_comment() method instead");
115 my $old = !$self->strict_comment;
116 $self->strict_comment(!shift) if @_;
120 # set up method stubs
125 *declaration = \&text;
135 HTML::Parser - HTML parser class
141 # Create parser object
142 $p = HTML::Parser->new( api_version => 3,
143 start_h => [\&start, "tagname, attr"],
144 end_h => [\&end, "tagname"],
145 marked_sections => 1,
148 # Parse document text chunk by chunk
152 $p->eof; # signal end of document
154 # Parse directly from file
155 $p->parse_file("foo.html");
157 open(my $fh, "<:utf8", "foo.html") || die;
162 Objects of the C<HTML::Parser> class will recognize markup and
163 separate it from plain text (alias data content) in HTML
164 documents. As different kinds of markup and text are recognized, the
165 corresponding event handlers are invoked.
167 C<HTML::Parser> is not a generic SGML parser. We have tried to
168 make it able to deal with the HTML that is actually "out there", and
169 it normally parses as closely as possible to the way the popular web
170 browsers do it instead of strictly following one of the many HTML
171 specifications from W3C. Where there is disagreement, there is often
172 an option that you can enable to get the official behaviour.
174 The document to be parsed may be supplied in arbitrary chunks. This
175 makes on-the-fly parsing as documents are received from the network
178 If event driven parsing does not feel right for your application, you
179 might want to use C<HTML::PullParser>. This is an C<HTML::Parser>
180 subclass that allows a more conventional program structure.
185 The following method is used to construct a new C<HTML::Parser> object:
189 =item $p = HTML::Parser->new( %options_and_handlers )
191 This class method creates a new C<HTML::Parser> object and
192 returns it. Key/value argument pairs may be provided to assign event
193 handlers or initialize parser options. The handlers and parser
194 options can also be set or modified later by the method calls described below.
196 If a top level key is in the form "<event>_h" (e.g., "text_h") then it
197 assigns a handler to that event, otherwise it initializes a parser
198 option. The event handler specification value must be an array
199 reference. Multiple handlers may also be assigned with the 'handlers
200 => [%handlers]' option. See examples below.
202 If new() is called without any arguments, it will create a parser that
203 uses callback methods compatible with version 2 of C<HTML::Parser>.
204 See the section on "version 2 compatibility" below for details.
206 The special constructor option 'api_version => 2' can be used to
207 initialize version 2 callbacks while still setting other options and
208 handlers. The 'api_version => 3' option can be used if you don't want
209 to set any options and don't want to fall back to v2 compatible
214 $p = HTML::Parser->new(api_version => 3,
215 text_h => [ sub {...}, "dtext" ]);
217 This creates a new parser object with a text event handler subroutine
218 that receives the original text with general entities decoded.
220 $p = HTML::Parser->new(api_version => 3,
221 start_h => [ 'my_start', "self,tokens" ]);
223 This creates a new parser object with a start event handler method
224 that receives the $p and the tokens array.
226 $p = HTML::Parser->new(api_version => 3,
227 handlers => { text => [\@array, "event,text"],
228 comment => [\@array, "event,text"],
231 This creates a new parser object that stores the event type and the
232 original text in @array for text and comment events.
236 The following methods feed the HTML document
237 to the C<HTML::Parser> object:
241 =item $p->parse( $string )
243 Parse $string as the next chunk of the HTML document. The return
244 value is normally a reference to the parser object (i.e. $p).
245 Handlers invoked should not attempt to modify the $string in-place until
248 If an invoked event handler aborts parsing by calling $p->eof, then
249 $p->parse() will return a FALSE value.
251 =item $p->parse( $code_ref )
253 If a code reference is passed as the argument to be parsed, then the
254 chunks to be parsed are obtained by invoking this function repeatedly.
255 Parsing continues until the function returns an empty (or undefined)
256 result. When this happens $p->eof is automatically signaled.
258 Parsing will also abort if one of the event handlers calls $p->eof.
260 The effect of this is the same as:
263 my $chunk = &$code_ref();
264 if (!defined($chunk) || !length($chunk)) {
268 $p->parse($chunk) || return undef;
271 But it is more efficient as this loop runs internally in XS code.
273 =item $p->parse_file( $file )
275 Parse text directly from a file. The $file argument can be a
276 filename, an open file handle, or a reference to an open file
279 If $file contains a filename and the file can't be opened, then the
280 method returns an undefined value and $! tells why it failed.
281 Otherwise the return value is a reference to the parser object.
283 If a file handle is passed as the $file argument, then the file will
284 normally be read until EOF, but not closed.
286 If an invoked event handler aborts parsing by calling $p->eof,
287 then $p->parse_file() may not have read the entire file.
289 On systems with multi-byte line terminators, the values passed for the
290 offset and length argspecs may be too low if parse_file() is called on
291 a file handle that is not in binary mode.
293 If a filename is passed in, then parse_file() will open the file in
298 Signals the end of the HTML document. Calling the $p->eof method
299 outside a handler callback will flush any remaining buffered text
300 (which triggers the C<text> event if there is any remaining text).
302 Calling $p->eof inside a handler will terminate parsing at that point
303 and cause $p->parse to return a FALSE value. This also terminates
304 parsing by $p->parse_file().
306 After $p->eof has been called, the parse() and parse_file() methods
307 can be invoked to feed new documents with the parser object.
309 The return value from eof() is a reference to the parser object.
314 Most parser options are controlled by boolean attributes.
315 Each boolean attribute is enabled by calling the corresponding method
316 with a TRUE argument and disabled with a FALSE argument. The
317 attribute value is left unchanged if no argument is given. The return
318 value from each method is the old attribute value.
320 Methods that can be used to get and/or set parser options are:
324 =item $p->attr_encoded
326 =item $p->attr_encoded( $bool )
328 By default, the C<attr> and C<@attr> argspecs will have general
329 entities for attribute values decoded. Enabling this attribute leaves
332 =item $p->boolean_attribute_value( $val )
334 This method sets the value reported for boolean attributes inside HTML
335 start tags. By default, the name of the attribute is also used as its
336 value. This affects the values reported for C<tokens> and C<attr>
339 =item $p->case_sensitive
341 =item $p->case_sensitive( $bool )
343 By default, tagnames and attribute names are down-cased. Enabling this
344 attribute leaves them as found in the HTML source document.
346 =item $p->closing_plaintext
348 =item $p->closing_plaintext( $bool )
350 By default, "plaintext" element can never be closed. Everything up to
351 the end of the document is parsed in CDATA mode. This historical
352 behaviour is what at least MSIE does. Enabling this attribute makes
353 closing "</plaintext>" tag effective and the parsing process will resume
354 after seeing this tag. This emulates gecko-based browsers.
356 =item $p->empty_element_tags
358 =item $p->empty_element_tags( $bool )
360 By default, empty element tags are not recognized as such and the "/"
361 before ">" is just treated like a normal name character (unless
362 C<strict_names> is enabled). Enabling this attribute make
363 C<HTML::Parser> recognize these tags.
365 Empty element tags look like start tags, but end with the character
366 sequence "/>" instead of ">". When recognized by C<HTML::Parser> they
367 cause an artificial end event in addition to the start event. The
368 C<text> for the artificial end event will be empty and the C<tokenpos>
369 array will be undefined even though the the token array will have one
370 element containing the tag name.
372 =item $p->marked_sections
374 =item $p->marked_sections( $bool )
376 By default, section markings like <![CDATA[...]]> are treated like
377 ordinary text. When this attribute is enabled section markings are
380 There are currently no events associated with the marked section
381 markup, but the text can be returned as C<skipped_text>.
383 =item $p->strict_comment
385 =item $p->strict_comment( $bool )
387 By default, comments are terminated by the first occurrence of "-->".
388 This is the behaviour of most popular browsers (like Mozilla, Opera and
389 MSIE), but it is not correct according to the official HTML
390 standard. Officially, you need an even number of "--" tokens before
391 the closing ">" is recognized and there may not be anything but
392 whitespace between an even and an odd "--".
394 The official behaviour is enabled by enabling this attribute.
396 Enabling of 'strict_comment' also disables recognizing these forms as
405 =item $p->strict_end( $bool )
407 By default, attributes and other junk are allowed to be present on end tags in a
408 manner that emulates MSIE's behaviour.
410 The official behaviour is enabled with this attribute. If enabled,
411 only whitespace is allowed between the tagname and the final ">".
413 =item $p->strict_names
415 =item $p->strict_names( $bool )
417 By default, almost anything is allowed in tag and attribute names.
418 This is the behaviour of most popular browsers and allows us to parse
419 some broken tags with invalid attribute values like:
421 <IMG SRC=newprevlstGr.gif ALT=[PREV LIST] BORDER=0>
423 By default, "LIST]" is parsed as a boolean attribute, not as
424 part of the ALT value as was clearly intended. This is also what
427 The official behaviour is enabled by enabling this attribute. If
428 enabled, it will cause the tag above to be reported as text
429 since "LIST]" is not a legal attribute name.
431 =item $p->unbroken_text
433 =item $p->unbroken_text( $bool )
435 By default, blocks of text are given to the text handler as soon as
436 possible (but the parser takes care always to break text at a
437 boundary between whitespace and non-whitespace so single words and
438 entities can always be decoded safely). This might create breaks that
439 make it hard to do transformations on the text. When this attribute is
440 enabled, blocks of text are always reported in one piece. This will
441 delay the text event until the following (non-text) event has been
442 recognized by the parser.
444 Note that the C<offset> argspec will give you the offset of the first
445 segment of text and C<length> is the combined length of the segments.
446 Since there might be ignored tags in between, these numbers can't be
447 used to directly index in the original document file.
451 =item $p->utf8_mode( $bool )
453 Enable this option when parsing raw undecoded UTF-8. This tells the
454 parser that the entities expanded for strings reported by C<attr>,
455 C<@attr> and C<dtext> should be expanded as decoded UTF-8 so they end
456 up compatible with the surrounding text.
458 If C<utf8_mode> is enabled then it is an error to pass strings
459 containing characters with code above 255 to the parse() method, and
460 the parse() method will croak if you try.
462 Example: The Unicode character "\x{2665}" is "\xE2\x99\xA5" when UTF-8
463 encoded. The character can also be represented by the entity
464 "♥" or "♥". If we feed the parser:
466 $p->parse("\xE2\x99\xA5♥");
468 then C<dtext> will be reported as "\xE2\x99\xA5\x{2665}" without
469 C<utf8_mode> enabled, but as "\xE2\x99\xA5\xE2\x99\xA5" when enabled.
470 The later string is what you want.
472 This option is only available with perl-5.8 or better.
476 =item $p->xml_mode( $bool )
478 Enabling this attribute changes the parser to allow some XML
479 constructs. This enables the behaviour controlled by individually by
480 the C<case_sensitive>, C<empty_element_tags>, C<strict_names> and
481 C<xml_pic> attributes and also suppresses special treatment of
482 elements that are parsed as CDATA for HTML.
486 =item $p->xml_pic( $bool )
488 By default, I<processing instructions> are terminated by ">". When
489 this attribute is enabled, processing instructions are terminated by
494 As markup and text is recognized, handlers are invoked. The following
495 method is used to set up handlers for different events:
499 =item $p->handler( event => \&subroutine, $argspec )
501 =item $p->handler( event => $method_name, $argspec )
503 =item $p->handler( event => \@accum, $argspec )
505 =item $p->handler( event => "" );
507 =item $p->handler( event => undef );
509 =item $p->handler( event );
511 This method assigns a subroutine, method, or array to handle an event.
513 Event is one of C<text>, C<start>, C<end>, C<declaration>, C<comment>,
514 C<process>, C<start_document>, C<end_document> or C<default>.
516 The C<\&subroutine> is a reference to a subroutine which is called to handle
519 The C<$method_name> is the name of a method of $p which is called to handle
522 The C<@accum> is an array that will hold the event information as
525 If the second argument is "", the event is ignored.
526 If it is undef, the default handler is invoked for the event.
528 The C<$argspec> is a string that describes the information to be reported
529 for the event. Any requested information that does not apply to a
530 specific event is passed as C<undef>. If argspec is omitted, then it
533 The return value from $p->handler is the old callback routine or a
534 reference to the accumulator array.
536 Any return values from handler callback routines/methods are always
537 ignored. A handler callback can request parsing to be aborted by
538 invoking the $p->eof method. A handler callback is not allowed to
539 invoke the $p->parse() or $p->parse_file() method. An exception will
540 be raised if it tries.
544 $p->handler(start => "start", 'self, attr, attrseq, text' );
546 This causes the "start" method of object $p to be called for 'start' events.
547 The callback signature is $p->start(\%attr, \@attr_seq, $text).
549 $p->handler(start => \&start, 'attr, attrseq, text' );
551 This causes subroutine start() to be called for 'start' events.
552 The callback signature is start(\%attr, \@attr_seq, $text).
554 $p->handler(start => \@accum, '"S", attr, attrseq, text' );
556 This causes 'start' event information to be saved in @accum.
557 The array elements will be ['S', \%attr, \@attr_seq, $text].
559 $p->handler(start => "");
561 This causes 'start' events to be ignored. It also suppresses
562 invocations of any default handler for start events. It is in most
563 cases equivalent to $p->handler(start => sub {}), but is more
564 efficient. It is different from the empty-sub-handler in that
565 C<skipped_text> is not reset by it.
567 $p->handler(start => undef);
569 This causes no handler to be associated with start events.
570 If there is a default handler it will be invoked.
574 Filters based on tags can be set up to limit the number of events
575 reported. The main bottleneck during parsing is often the huge number
576 of callbacks made from the parser. Applying filters can improve
577 performance significantly.
579 The following methods control filters:
583 =item $p->ignore_elements( @tags )
585 Both the C<start> event and the C<end> event as well as any events that
586 would be reported in between are suppressed. The ignored elements can
587 contain nested occurrences of itself. Example:
589 $p->ignore_elements(qw(script style));
591 The C<script> and C<style> tags will always nest properly since their
592 content is parsed in CDATA mode. For most other tags
593 C<ignore_elements> must be used with caution since HTML is often not
596 =item $p->ignore_tags( @tags )
598 Any C<start> and C<end> events involving any of the tags given are
599 suppressed. To reset the filter (i.e. don't suppress any C<start> and
600 C<end> events), call C<ignore_tags> without an argument.
602 =item $p->report_tags( @tags )
604 Any C<start> and C<end> events involving any of the tags I<not> given
605 are suppressed. To reset the filter (i.e. report all C<start> and
606 C<end> events), call C<report_tags> without an argument.
610 Internally, the system has two filter lists, one for C<report_tags>
611 and one for C<ignore_tags>, and both filters are applied. This
612 effectively gives C<ignore_tags> precedence over C<report_tags>.
616 $p->ignore_tags(qw(style));
617 $p->report_tags(qw(script style));
619 results in only C<script> events being reported.
623 Argspec is a string containing a comma-separated list that describes
624 the information reported by the event. The following argspec
625 identifier names can be used:
631 Attr causes a reference to a hash of attribute name/value pairs to be
634 Boolean attributes' values are either the value set by
635 $p->boolean_attribute_value, or the attribute name if no value has been
636 set by $p->boolean_attribute_value.
638 This passes undef except for C<start> events.
640 Unless C<xml_mode> or C<case_sensitive> is enabled, the attribute
641 names are forced to lower case.
643 General entities are decoded in the attribute values and
644 one layer of matching quotes enclosing the attribute values is removed.
646 The Unicode character set is assumed for entity decoding. With Perl
647 version 5.6 or earlier only the Latin-1 range is supported, and
648 entities for characters outside the range 0..255 are left unchanged.
652 Basically the same as C<attr>, but keys and values are passed as
653 individual arguments and the original sequence of the attributes is
654 kept. The parameters passed will be the same as the @attr calculated
657 @attr = map { $_ => $attr->{$_} } @$attrseq;
659 assuming $attr and $attrseq here are the hash and array passed as the
660 result of C<attr> and C<attrseq> argspecs.
662 This passes no values for events besides C<start>.
666 Attrseq causes a reference to an array of attribute names to be
667 passed. This can be useful if you want to walk the C<attr> hash in
668 the original sequence.
670 This passes undef except for C<start> events.
672 Unless C<xml_mode> or C<case_sensitive> is enabled, the attribute
673 names are forced to lower case.
677 Column causes the column number of the start of the event to be passed.
678 The first column on a line is 0.
682 Dtext causes the decoded text to be passed. General entities are
683 automatically decoded unless the event was inside a CDATA section or
684 was between literal start and end tags (C<script>, C<style>,
685 C<xmp>, and C<plaintext>).
687 The Unicode character set is assumed for entity decoding. With Perl
688 version 5.6 or earlier only the Latin-1 range is supported, and
689 entities for characters outside the range 0..255 are left unchanged.
691 This passes undef except for C<text> events.
695 Event causes the event name to be passed.
697 The event name is one of C<text>, C<start>, C<end>, C<declaration>,
698 C<comment>, C<process>, C<start_document> or C<end_document>.
702 Is_cdata causes a TRUE value to be passed if the event is inside a CDATA
703 section or between literal start and end tags (C<script>,
704 C<style>, C<xmp>, and C<plaintext>).
706 if the flag is FALSE for a text event, then you should normally
707 either use C<dtext> or decode the entities yourself before the text is
712 Length causes the number of bytes of the source text of the event to
717 Line causes the line number of the start of the event to be passed.
718 The first line in the document is 1. Line counting doesn't start
719 until at least one handler requests this value to be reported.
723 Offset causes the byte position in the HTML document of the start of
724 the event to be passed. The first byte in the document has offset 0.
728 Offset_end causes the byte position in the HTML document of the end of
729 the event to be passed. This is the same as C<offset> + C<length>.
733 Self causes the current object to be passed to the handler. If the
734 handler is a method, this must be the first element in the argspec.
736 An alternative to passing self as an argspec is to register closures
737 that capture $self by themselves as handlers. Unfortunately this
738 creates circular references which prevent the HTML::Parser object
739 from being garbage collected. Using the C<self> argspec avoids this
742 =item C<skipped_text>
744 Skipped_text returns the concatenated text of all the events that have
745 been skipped since the last time an event was reported. Events might
746 be skipped because no handler is registered for them or because some
747 filter applies. Skipped text also includes marked section markup,
748 since there are no events that can catch it.
750 If an C<"">-handler is registered for an event, then the text for this
751 event is not included in C<skipped_text>. Skipped text both before
752 and after the C<"">-event is included in the next reported
757 Same as C<tagname>, but prefixed with "/" if it belongs to an C<end>
758 event and "!" for a declaration. The C<tag> does not have any prefix
759 for C<start> events, and is in this case identical to C<tagname>.
763 This is the element name (or I<generic identifier> in SGML jargon) for
764 start and end tags. Since HTML is case insensitive, this name is
765 forced to lower case to ease string matching.
767 Since XML is case sensitive, the tagname case is not changed when
768 C<xml_mode> is enabled. The same happens if the C<case_sensitive> attribute
771 The declaration type of declaration elements is also passed as a tagname,
772 even if that is a bit strange.
773 In fact, in the current implementation tagname is
774 identical to C<token0> except that the name may be forced to lower case.
778 Token0 causes the original text of the first token string to be
779 passed. This should always be the same as $tokens->[0].
781 For C<declaration> events, this is the declaration type.
783 For C<start> and C<end> events, this is the tag name.
785 For C<process> and non-strict C<comment> events, this is everything
788 This passes undef if there are no tokens in the event.
792 Tokenpos causes a reference to an array of token positions to be
793 passed. For each string that appears in C<tokens>, this array
794 contains two numbers. The first number is the offset of the start of
795 the token in the original C<text> and the second number is the length
798 Boolean attributes in a C<start> event will have (0,0) for the
799 attribute value offset and length.
801 This passes undef if there are no tokens in the event (e.g., C<text>)
802 and for artificial C<end> events triggered by empty element tags.
804 If you are using these offsets and lengths to modify C<text>, you
805 should either work from right to left, or be very careful to calculate
806 the changes to the offsets.
810 Tokens causes a reference to an array of token strings to be passed.
811 The strings are exactly as they were found in the original text,
812 no decoding or case changes are applied.
814 For C<declaration> events, the array contains each word, comment, and
815 delimited string starting with the declaration type.
817 For C<comment> events, this contains each sub-comment. If
818 $p->strict_comments is disabled, there will be only one sub-comment.
820 For C<start> events, this contains the original tag name followed by
821 the attribute name/value pairs. The values of boolean attributes will
822 be either the value set by $p->boolean_attribute_value, or the
823 attribute name if no value has been set by
824 $p->boolean_attribute_value.
826 For C<end> events, this contains the original tag name (always one token).
828 For C<process> events, this contains the process instructions (always one
831 This passes C<undef> for C<text> events.
835 Text causes the source text (including markup element delimiters) to be
840 Pass an undefined value. Useful as padding where the same handler
841 routine is registered for multiple events.
845 A literal string of 0 to 255 characters enclosed
846 in single (') or double (") quotes is passed as entered.
850 The whole argspec string can be wrapped up in C<'@{...}'> to signal
851 that the resulting event array should be flattened. This only makes a
852 difference if an array reference is used as the handler target.
853 Consider this example:
855 $p->handler(text => [], 'text');
856 $p->handler(text => [], '@{text}']);
858 With two text events; C<"foo">, C<"bar">; then the first example will end
859 up with [["foo"], ["bar"]] and the second with ["foo", "bar"] in
860 the handler target array.
865 Handlers for the following events can be registered:
871 This event is triggered when a markup comment is recognized.
875 <!-- This is a comment -- -- So is this -->
879 This event is triggered when a I<markup declaration> is recognized.
881 For typical HTML documents, the only declaration you are
882 likely to find is <!DOCTYPE ...>.
886 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
887 "http://www.w3.org/TR/html40/strict.dtd">
889 DTDs inside <!DOCTYPE ...> will confuse HTML::Parser.
893 This event is triggered for events that do not have a specific
894 handler. You can set up a handler for this event to catch stuff you
895 did not want to catch explicitly.
899 This event is triggered when an end tag is recognized.
905 =item C<end_document>
907 This event is triggered when $p->eof is called and after any remaining
908 text is flushed. There is no document text associated with this event.
912 This event is triggered when a processing instructions markup is
915 The format and content of processing instructions are system and
916 application dependent.
920 <? HTML processing instructions >
921 <? XML processing instructions ?>
925 This event is triggered when a start tag is recognized.
929 <A HREF="http://www.perl.com/">
931 =item C<start_document>
933 This event is triggered before any other events for a new document. A
934 handler for it can be used to initialize stuff. There is no document
935 text associated with this event.
939 This event is triggered when plain text (characters) is recognized.
940 The text may contain multiple lines. A sequence of text may be broken
941 between several text events unless $p->unbroken_text is enabled.
943 The parser will make sure that it does not break a word or a sequence
944 of whitespace between two text events.
950 The C<HTML::Parser> can parse Unicode strings when running under
951 perl-5.8 or better. If Unicode is passed to $p->parse() then chunks
952 of Unicode will be reported to the handlers. The offset and length
953 argspecs will also report their position in terms of characters.
955 It is safe to parse raw undecoded UTF-8 if you either avoid decoding
956 entities and make sure to not use I<argspecs> that do, or enable the
957 C<utf8_mode> for the parser. Parsing of undecoded UTF-8 might be
958 useful when parsing from a file where you need the reported offsets
959 and lengths to match the byte offsets in the file.
961 If a filename is passed to $p->parse_file() then the file will be read
962 in binary mode. This will be fine if the file contains only ASCII or
963 Latin-1 characters. If the file contains UTF-8 encoded text then care
964 must be taken when decoding entities as described in the previous
965 paragraph, but better is to open the file with the UTF-8 layer so that
966 it is decoded properly:
968 open(my $fh, "<:utf8", "index.html") || die "...: $!";
971 If the file contains text encoded in a charset besides ASCII, Latin-1
972 or UTF-8 then decoding will always be needed.
974 =head1 VERSION 2 COMPATIBILITY
976 When an C<HTML::Parser> object is constructed with no arguments, a set
977 of handlers is automatically provided that is compatible with the old
978 HTML::Parser version 2 callback methods.
980 This is equivalent to the following method calls:
982 $p->handler(start => "start", "self, tagname, attr, attrseq, text");
983 $p->handler(end => "end", "self, tagname, text");
984 $p->handler(text => "text", "self, text, is_cdata");
985 $p->handler(process => "process", "self, token0, text");
986 $p->handler(comment =>
988 my($self, $tokens) = @_;
989 for (@$tokens) {$self->comment($_);}},
991 $p->handler(declaration =>
994 $self->declaration(substr($_[0], 2, -1));},
997 Setting up these handlers can also be requested with the "api_version =>
998 2" constructor option.
1002 The C<HTML::Parser> class is subclassable. Parser objects are plain
1003 hashes and C<HTML::Parser> reserves only hash keys that start with
1004 "_hparser". The parser state can be set up by invoking the init()
1005 method, which takes the same arguments as new().
1009 The first simple example shows how you might strip out comments from
1010 an HTML document. We achieve this by setting up a comment handler that
1011 does nothing and a default handler that will print out anything else:
1014 HTML::Parser->new(default_h => [sub { print shift }, 'text'],
1016 )->parse_file(shift || die) || die $!;
1018 An alternative implementation is:
1021 HTML::Parser->new(end_document_h => [sub { print shift },
1024 )->parse_file(shift || die) || die $!;
1026 This will in most cases be much more efficient since only a single
1027 callback will be made.
1029 The next example prints out the text that is inside the <title>
1030 element of an HTML document. Here we start by setting up a start
1031 handler. When it sees the title start tag it enables a text handler
1032 that prints any text found and an end handler that will terminate
1033 parsing as soon as the title end tag is seen:
1035 use HTML::Parser ();
1039 return if shift ne "title";
1041 $self->handler(text => sub { print shift }, "dtext");
1042 $self->handler(end => sub { shift->eof if shift eq "title"; },
1046 my $p = HTML::Parser->new(api_version => 3);
1047 $p->handler( start => \&start_handler, "tagname,self");
1048 $p->parse_file(shift || die) || die $!;
1051 On a Debian box, more examples can be found in the
1052 /usr/share/doc/libhtml-parser-perl/examples directory.
1053 The program C<hrefsub> shows how you can edit all links
1054 found in a document and C<htextsub> how to edit the text only; the
1055 program C<hstrip> shows how you can strip out certain tags/elements
1056 and/or attributes; and the program C<htext> show how to obtain the
1057 plain text, but not any script/style content.
1059 You can browse the F<eg/> directory online from the I<[Browse]> link on
1060 the http://search.cpan.org/~gaas/HTML-Parser/ page.
1064 The <style> and <script> sections do not end with the first "</", but
1065 need the complete corresponding end tag. The standard behaviour is
1066 not really practical.
1068 When the I<strict_comment> option is enabled, we still recognize
1069 comments where there is something other than whitespace between even
1070 and odd "--" markers.
1072 Once $p->boolean_attribute_value has been set, there is no way to
1073 restore the default behaviour.
1075 There is currently no way to get both quote characters
1076 into the same literal argspec.
1078 Empty tags, e.g. "<>" and "</>", are not recognized. SGML allows them
1079 to repeat the previous start tag or close the previous start tag
1082 NET tags, e.g. "code/.../" are not recognized. This is SGML
1083 shorthand for "<code>...</code>".
1085 Unclosed start or end tags, e.g. "<tt<b>...</b</tt>" are not
1090 The following messages may be produced by HTML::Parser. The notation
1091 in this listing is the same as used in L<perldiag>:
1095 =item Not a reference to a hash
1097 (F) The object blessed into or subclassed from HTML::Parser is not a
1098 hash as required by the HTML::Parser methods.
1100 =item Bad signature in parser state object at %p
1102 (F) The _hparser_xs_state element does not refer to a valid state structure.
1103 Something must have changed the internal value
1104 stored in this hash element, or the memory has been overwritten.
1106 =item _hparser_xs_state element is not a reference
1108 (F) The _hparser_xs_state element has been destroyed.
1110 =item Can't find '_hparser_xs_state' element in HTML::Parser hash
1112 (F) The _hparser_xs_state element is missing from the parser hash.
1113 It was either deleted, or not created when the object was created.
1115 =item API version %s not supported by HTML::Parser %s
1117 (F) The constructor option 'api_version' with an argument greater than
1118 or equal to 4 is reserved for future extensions.
1120 =item Bad constructor option '%s'
1122 (F) An unknown constructor option key was passed to the new() or
1125 =item Parse loop not allowed
1127 (F) A handler invoked the parse() or parse_file() method.
1128 This is not permitted.
1130 =item marked sections not supported
1132 (F) The $p->marked_sections() method was invoked in a HTML::Parser
1133 module that was compiled without support for marked sections.
1135 =item Unknown boolean attribute (%d)
1137 (F) Something is wrong with the internal logic that set up aliases for
1140 =item Only code or array references allowed as handler
1142 (F) The second argument for $p->handler must be either a subroutine
1143 reference, then name of a subroutine or method, or a reference to an
1146 =item No handler for %s events
1148 (F) The first argument to $p->handler must be a valid event name; i.e. one
1149 of "start", "end", "text", "process", "declaration" or "comment".
1151 =item Unrecognized identifier %s in argspec
1153 (F) The identifier is not a known argspec name.
1154 Use one of the names mentioned in the argspec section above.
1156 =item Literal string is longer than 255 chars in argspec
1158 (F) The current implementation limits the length of literals in
1159 an argspec to 255 characters. Make the literal shorter.
1161 =item Backslash reserved for literal string in argspec
1163 (F) The backslash character "\" is not allowed in argspec literals.
1164 It is reserved to permit quoting inside a literal in a later version.
1166 =item Unterminated literal string in argspec
1168 (F) The terminating quote character for a literal was not found.
1170 =item Bad argspec (%s)
1172 (F) Only identifier names, literals, spaces and commas
1173 are allowed in argspecs.
1175 =item Missing comma separator in argspec
1177 (F) Identifiers in an argspec must be separated with ",".
1179 =item Parsing of undecoded UTF-8 will give garbage when decoding entities
1181 (W) The first chunk parsed appears to contain undecoded UTF-8 and one
1182 or more argspecs that decode entities are used for the callback
1185 The result of decoding will be a mix of encoded and decoded characters
1186 for any entities that expand to characters with code above 127. This
1187 is not a good thing.
1189 The solution is to use the Encode::encode_utf8() on the data before
1190 feeding it to the $p->parse(). For $p->parse_file() pass a file that
1191 has been opened in ":utf8" mode.
1193 The parser can process raw undecoded UTF-8 sanely if the C<utf8_mode>
1194 is enabled or if the "attr", "@attr" or "dtext" argspecs is avoided.
1196 =item Parsing string decoded with wrong endianess
1198 (W) The first character in the document is U+FFFE. This is not a
1199 legal Unicode character but a byte swapped BOM. The result of parsing
1200 will likely be garbage.
1202 =item Parsing of undecoded UTF-32
1204 (W) The parser found the Unicode UTF-32 BOM signature at the start
1205 of the document. The result of parsing will likely be garbage.
1207 =item Parsing of undecoded UTF-16
1209 (W) The parser found the Unicode UTF-16 BOM signature at the start of
1210 the document. The result of parsing will likely be garbage.
1216 L<HTML::Entities>, L<HTML::PullParser>, L<HTML::TokeParser>, L<HTML::HeadParser>,
1217 L<HTML::LinkExtor>, L<HTML::Form>
1219 L<HTML::TreeBuilder> (part of the I<HTML-Tree> distribution)
1221 http://www.w3.org/TR/html4
1223 More information about marked sections and processing instructions may
1224 be found at C<http://www.sgml.u-net.com/book/sgml-8.htm>.
1228 Copyright 1996-2007 Gisle Aas. All rights reserved.
1229 Copyright 1999-2000 Michael A. Chase. All rights reserved.
1231 This library is free software; you can redistribute it and/or
1232 modify it under the same terms as Perl itself.