1package HTML::Parser; 2 3use strict; 4 5our $VERSION = '3.76'; 6 7require HTML::Entities; 8 9require XSLoader; 10XSLoader::load('HTML::Parser', $VERSION); 11 12sub new 13{ 14 my $class = shift; 15 my $self = bless {}, $class; 16 return $self->init(@_); 17} 18 19 20sub init 21{ 22 my $self = shift; 23 $self->_alloc_pstate; 24 25 my %arg = @_; 26 my $api_version = delete $arg{api_version} || (@_ ? 3 : 2); 27 if ($api_version >= 4) { 28 require Carp; 29 Carp::croak("API version $api_version not supported " . 30 "by HTML::Parser $VERSION"); 31 } 32 33 if ($api_version < 3) { 34 # Set up method callbacks compatible with HTML-Parser-2.xx 35 $self->handler(text => "text", "self,text,is_cdata"); 36 $self->handler(end => "end", "self,tagname,text"); 37 $self->handler(process => "process", "self,token0,text"); 38 $self->handler(start => "start", 39 "self,tagname,attr,attrseq,text"); 40 41 $self->handler(comment => 42 sub { 43 my($self, $tokens) = @_; 44 for (@$tokens) { 45 $self->comment($_); 46 } 47 }, "self,tokens"); 48 49 $self->handler(declaration => 50 sub { 51 my $self = shift; 52 $self->declaration(substr($_[0], 2, -1)); 53 }, "self,text"); 54 } 55 56 if (my $h = delete $arg{handlers}) { 57 $h = {@$h} if ref($h) eq "ARRAY"; 58 while (my($event, $cb) = each %$h) { 59 $self->handler($event => @$cb); 60 } 61 } 62 63 # In the end we try to assume plain attribute or handler 64 while (my($option, $val) = each %arg) { 65 if ($option =~ /^(\w+)_h$/) { 66 $self->handler($1 => @$val); 67 } 68 elsif ($option =~ /^(text|start|end|process|declaration|comment)$/) { 69 require Carp; 70 Carp::croak("Bad constructor option '$option'"); 71 } 72 else { 73 $self->$option($val); 74 } 75 } 76 77 return $self; 78} 79 80 81sub parse_file 82{ 83 my($self, $file) = @_; 84 my $opened; 85 if (!ref($file) && ref(\$file) ne "GLOB") { 86 # Assume $file is a filename 87 local(*F); 88 open(F, "<", $file) || return undef; 89 binmode(F); # should we? good for byte counts 90 $opened++; 91 $file = *F; 92 } 93 my $chunk = ''; 94 while (read($file, $chunk, 512)) { 95 $self->parse($chunk) || last; 96 } 97 close($file) if $opened; 98 $self->eof; 99} 100 101 102sub netscape_buggy_comment # legacy 103{ 104 my $self = shift; 105 require Carp; 106 Carp::carp("netscape_buggy_comment() is deprecated. " . 107 "Please use the strict_comment() method instead"); 108 my $old = !$self->strict_comment; 109 $self->strict_comment(!shift) if @_; 110 return $old; 111} 112 113# set up method stubs 114sub text { } 115*start = \&text; 116*end = \&text; 117*comment = \&text; 118*declaration = \&text; 119*process = \&text; 120 1211; 122 123__END__ 124 125 126=head1 NAME 127 128HTML::Parser - HTML parser class 129 130=head1 SYNOPSIS 131 132 use strict; 133 use warnings; 134 use HTML::Parser (); 135 136 # Create parser object 137 my $p = HTML::Parser->new( 138 api_version => 3, 139 start_h => [\&start, "tagname, attr"], 140 end_h => [\&end, "tagname"], 141 marked_sections => 1, 142 ); 143 144 # Parse document text chunk by chunk 145 $p->parse($chunk1); 146 $p->parse($chunk2); 147 # ... 148 # signal end of document 149 $p->eof; 150 151 # Parse directly from file 152 $p->parse_file("foo.html"); 153 # or 154 open(my $fh, "<:utf8", "foo.html") || die; 155 $p->parse_file($fh); 156 157=head1 DESCRIPTION 158 159Objects of the C<HTML::Parser> class will recognize markup and 160separate it from plain text (alias data content) in HTML 161documents. As different kinds of markup and text are recognized, the 162corresponding event handlers are invoked. 163 164C<HTML::Parser> is not a generic SGML parser. We have tried to 165make it able to deal with the HTML that is actually "out there", and 166it normally parses as closely as possible to the way the popular web 167browsers do it instead of strictly following one of the many HTML 168specifications from W3C. Where there is disagreement, there is often 169an option that you can enable to get the official behaviour. 170 171The document to be parsed may be supplied in arbitrary chunks. This 172makes on-the-fly parsing as documents are received from the network 173possible. 174 175If event driven parsing does not feel right for your application, you 176might want to use C<HTML::PullParser>. This is an C<HTML::Parser> 177subclass that allows a more conventional program structure. 178 179 180=head1 METHODS 181 182The following method is used to construct a new C<HTML::Parser> object: 183 184=over 185 186=item $p = HTML::Parser->new( %options_and_handlers ) 187 188This class method creates a new C<HTML::Parser> object and 189returns it. Key/value argument pairs may be provided to assign event 190handlers or initialize parser options. The handlers and parser 191options can also be set or modified later by the method calls described below. 192 193If a top level key is in the form "<event>_h" (e.g., "text_h") then it 194assigns a handler to that event, otherwise it initializes a parser 195option. The event handler specification value must be an array 196reference. Multiple handlers may also be assigned with the 'handlers 197=> [%handlers]' option. See examples below. 198 199If new() is called without any arguments, it will create a parser that 200uses callback methods compatible with version 2 of C<HTML::Parser>. 201See the section on "version 2 compatibility" below for details. 202 203The special constructor option 'api_version => 2' can be used to 204initialize version 2 callbacks while still setting other options and 205handlers. The 'api_version => 3' option can be used if you don't want 206to set any options and don't want to fall back to v2 compatible 207mode. 208 209Examples: 210 211 $p = HTML::Parser->new( 212 api_version => 3, 213 text_h => [ sub {...}, "dtext" ] 214 ); 215 216This creates a new parser object with a text event handler subroutine 217that receives the original text with general entities decoded. 218 219 $p = HTML::Parser->new( 220 api_version => 3, 221 start_h => [ 'my_start', "self,tokens" ] 222 ); 223 224This creates a new parser object with a start event handler method 225that receives the $p and the tokens array. 226 227 $p = HTML::Parser->new( 228 api_version => 3, 229 handlers => { 230 text => [\@array, "event,text"], 231 comment => [\@array, "event,text"], 232 } 233 ); 234 235This creates a new parser object that stores the event type and the 236original text in @array for text and comment events. 237 238=back 239 240The following methods feed the HTML document 241to the C<HTML::Parser> object: 242 243=over 244 245=item $p->parse( $string ) 246 247Parse $string as the next chunk of the HTML document. Handlers invoked should 248not attempt to modify the $string in-place until $p->parse returns. 249 250If an invoked event handler aborts parsing by calling $p->eof, then $p->parse() 251will return a FALSE value. Otherwise the return value is a reference to the 252parser object ($p). 253 254=item $p->parse( $code_ref ) 255 256If a code reference is passed as the argument to be parsed, then the 257chunks to be parsed are obtained by invoking this function repeatedly. 258Parsing continues until the function returns an empty (or undefined) 259result. When this happens $p->eof is automatically signaled. 260 261Parsing will also abort if one of the event handlers calls $p->eof. 262 263The effect of this is the same as: 264 265 while (1) { 266 my $chunk = &$code_ref(); 267 if (!defined($chunk) || !length($chunk)) { 268 $p->eof; 269 return $p; 270 } 271 $p->parse($chunk) || return undef; 272 } 273 274But it is more efficient as this loop runs internally in XS code. 275 276=item $p->parse_file( $file ) 277 278Parse text directly from a file. The $file argument can be a 279filename, an open file handle, or a reference to an open file 280handle. 281 282If $file contains a filename and the file can't be opened, then the 283method returns an undefined value and $! tells why it failed. 284Otherwise the return value is a reference to the parser object. 285 286If a file handle is passed as the $file argument, then the file will 287normally be read until EOF, but not closed. 288 289If an invoked event handler aborts parsing by calling $p->eof, 290then $p->parse_file() may not have read the entire file. 291 292On systems with multi-byte line terminators, the values passed for the 293offset and length argspecs may be too low if parse_file() is called on 294a file handle that is not in binary mode. 295 296If a filename is passed in, then parse_file() will open the file in 297binary mode. 298 299=item $p->eof 300 301Signals the end of the HTML document. Calling the $p->eof method 302outside a handler callback will flush any remaining buffered text 303(which triggers the C<text> event if there is any remaining text). 304 305Calling $p->eof inside a handler will terminate parsing at that point 306and cause $p->parse to return a FALSE value. This also terminates 307parsing by $p->parse_file(). 308 309After $p->eof has been called, the parse() and parse_file() methods 310can be invoked to feed new documents with the parser object. 311 312The return value from eof() is a reference to the parser object. 313 314=back 315 316 317Most parser options are controlled by boolean attributes. 318Each boolean attribute is enabled by calling the corresponding method 319with a TRUE argument and disabled with a FALSE argument. The 320attribute value is left unchanged if no argument is given. The return 321value from each method is the old attribute value. 322 323Methods that can be used to get and/or set parser options are: 324 325=over 326 327=item $p->attr_encoded 328 329=item $p->attr_encoded( $bool ) 330 331By default, the C<attr> and C<@attr> argspecs will have general 332entities for attribute values decoded. Enabling this attribute leaves 333entities alone. 334 335=item $p->backquote 336 337=item $p->backquote( $bool ) 338 339By default, only ' and " are recognized as quote characters around 340attribute values. MSIE also recognizes backquotes for some reason. 341Enabling this attribute provides compatibility with this behaviour. 342 343=item $p->boolean_attribute_value( $val ) 344 345This method sets the value reported for boolean attributes inside HTML 346start tags. By default, the name of the attribute is also used as its 347value. This affects the values reported for C<tokens> and C<attr> 348argspecs. 349 350=item $p->case_sensitive 351 352=item $p->case_sensitive( $bool ) 353 354By default, tag names and attribute names are down-cased. Enabling this 355attribute leaves them as found in the HTML source document. 356 357=item $p->closing_plaintext 358 359=item $p->closing_plaintext( $bool ) 360 361By default, C<plaintext> element can never be closed. Everything up to 362the end of the document is parsed in CDATA mode. This historical 363behaviour is what at least MSIE does. Enabling this attribute makes 364closing C< </plaintext> > tag effective and the parsing process will resume 365after seeing this tag. This emulates early gecko-based browsers. 366 367=item $p->empty_element_tags 368 369=item $p->empty_element_tags( $bool ) 370 371By default, empty element tags are not recognized as such and the "/" 372before ">" is just treated like a normal name character (unless 373C<strict_names> is enabled). Enabling this attribute make 374C<HTML::Parser> recognize these tags. 375 376Empty element tags look like start tags, but end with the character 377sequence "/>" instead of ">". When recognized by C<HTML::Parser> they 378cause an artificial end event in addition to the start event. The 379C<text> for the artificial end event will be empty and the C<tokenpos> 380array will be undefined even though the token array will have one 381element containing the tag name. 382 383=item $p->marked_sections 384 385=item $p->marked_sections( $bool ) 386 387By default, section markings like <![CDATA[...]]> are treated like 388ordinary text. When this attribute is enabled section markings are 389honoured. 390 391There are currently no events associated with the marked section 392markup, but the text can be returned as C<skipped_text>. 393 394=item $p->strict_comment 395 396=item $p->strict_comment( $bool ) 397 398By default, comments are terminated by the first occurrence of "-->". 399This is the behaviour of most popular browsers (like Mozilla, Opera and 400MSIE), but it is not correct according to the official HTML 401standard. Officially, you need an even number of "--" tokens before 402the closing ">" is recognized and there may not be anything but 403whitespace between an even and an odd "--". 404 405The official behaviour is enabled by enabling this attribute. 406 407Enabling of 'strict_comment' also disables recognizing these forms as 408comments: 409 410 </ comment> 411 <! comment> 412 413 414=item $p->strict_end 415 416=item $p->strict_end( $bool ) 417 418By default, attributes and other junk are allowed to be present on end tags in a 419manner that emulates MSIE's behaviour. 420 421The official behaviour is enabled with this attribute. If enabled, 422only whitespace is allowed between the tagname and the final ">". 423 424=item $p->strict_names 425 426=item $p->strict_names( $bool ) 427 428By default, almost anything is allowed in tag and attribute names. 429This is the behaviour of most popular browsers and allows us to parse 430some broken tags with invalid attribute values like: 431 432 <IMG SRC=newprevlstGr.gif ALT=[PREV LIST] BORDER=0> 433 434By default, "LIST]" is parsed as a boolean attribute, not as 435part of the ALT value as was clearly intended. This is also what 436Mozilla sees. 437 438The official behaviour is enabled by enabling this attribute. If 439enabled, it will cause the tag above to be reported as text 440since "LIST]" is not a legal attribute name. 441 442=item $p->unbroken_text 443 444=item $p->unbroken_text( $bool ) 445 446By default, blocks of text are given to the text handler as soon as 447possible (but the parser takes care always to break text at a 448boundary between whitespace and non-whitespace so single words and 449entities can always be decoded safely). This might create breaks that 450make it hard to do transformations on the text. When this attribute is 451enabled, blocks of text are always reported in one piece. This will 452delay the text event until the following (non-text) event has been 453recognized by the parser. 454 455Note that the C<offset> argspec will give you the offset of the first 456segment of text and C<length> is the combined length of the segments. 457Since there might be ignored tags in between, these numbers can't be 458used to directly index in the original document file. 459 460=item $p->utf8_mode 461 462=item $p->utf8_mode( $bool ) 463 464Enable this option when parsing raw undecoded UTF-8. This tells the 465parser that the entities expanded for strings reported by C<attr>, 466C<@attr> and C<dtext> should be expanded as decoded UTF-8 so they end 467up compatible with the surrounding text. 468 469If C<utf8_mode> is enabled then it is an error to pass strings 470containing characters with code above 255 to the parse() method, and 471the parse() method will croak if you try. 472 473Example: The Unicode character "\x{2665}" is "\xE2\x99\xA5" when UTF-8 474encoded. The character can also be represented by the entity 475"♥" or "♥". If we feed the parser: 476 477 $p->parse("\xE2\x99\xA5♥"); 478 479then C<dtext> will be reported as "\xE2\x99\xA5\x{2665}" without 480C<utf8_mode> enabled, but as "\xE2\x99\xA5\xE2\x99\xA5" when enabled. 481The later string is what you want. 482 483This option is only available with perl-5.8 or better. 484 485=item $p->xml_mode 486 487=item $p->xml_mode( $bool ) 488 489Enabling this attribute changes the parser to allow some XML 490constructs. This enables the behaviour controlled by individually by 491the C<case_sensitive>, C<empty_element_tags>, C<strict_names> and 492C<xml_pic> attributes and also suppresses special treatment of 493elements that are parsed as CDATA for HTML. 494 495=item $p->xml_pic 496 497=item $p->xml_pic( $bool ) 498 499By default, I<processing instructions> are terminated by ">". When 500this attribute is enabled, processing instructions are terminated by 501"?>" instead. 502 503=back 504 505As markup and text is recognized, handlers are invoked. The following 506method is used to set up handlers for different events: 507 508=over 509 510=item $p->handler( event => \&subroutine, $argspec ) 511 512=item $p->handler( event => $method_name, $argspec ) 513 514=item $p->handler( event => \@accum, $argspec ) 515 516=item $p->handler( event => "" ); 517 518=item $p->handler( event => undef ); 519 520=item $p->handler( event ); 521 522This method assigns a subroutine, method, or array to handle an event. 523 524Event is one of C<text>, C<start>, C<end>, C<declaration>, C<comment>, 525C<process>, C<start_document>, C<end_document> or C<default>. 526 527The C<\&subroutine> is a reference to a subroutine which is called to handle 528the event. 529 530The C<$method_name> is the name of a method of $p which is called to handle 531the event. 532 533The C<@accum> is an array that will hold the event information as 534sub-arrays. 535 536If the second argument is "", the event is ignored. 537If it is undef, the default handler is invoked for the event. 538 539The C<$argspec> is a string that describes the information to be reported 540for the event. Any requested information that does not apply to a 541specific event is passed as C<undef>. If argspec is omitted, then it 542is left unchanged. 543 544The return value from $p->handler is the old callback routine or a 545reference to the accumulator array. 546 547Any return values from handler callback routines/methods are always 548ignored. A handler callback can request parsing to be aborted by 549invoking the $p->eof method. A handler callback is not allowed to 550invoke the $p->parse() or $p->parse_file() method. An exception will 551be raised if it tries. 552 553Examples: 554 555 $p->handler(start => "start", 'self, attr, attrseq, text' ); 556 557This causes the "start" method of object C<$p> to be called for 'start' events. 558The callback signature is C<< $p->start(\%attr, \@attr_seq, $text) >>. 559 560 $p->handler(start => \&start, 'attr, attrseq, text' ); 561 562This causes subroutine start() to be called for 'start' events. 563The callback signature is start(\%attr, \@attr_seq, $text). 564 565 $p->handler(start => \@accum, '"S", attr, attrseq, text' ); 566 567This causes 'start' event information to be saved in @accum. 568The array elements will be ['S', \%attr, \@attr_seq, $text]. 569 570 $p->handler(start => ""); 571 572This causes 'start' events to be ignored. It also suppresses 573invocations of any default handler for start events. It is in most 574cases equivalent to $p->handler(start => sub {}), but is more 575efficient. It is different from the empty-sub-handler in that 576C<skipped_text> is not reset by it. 577 578 $p->handler(start => undef); 579 580This causes no handler to be associated with start events. 581If there is a default handler it will be invoked. 582 583=back 584 585Filters based on tags can be set up to limit the number of events 586reported. The main bottleneck during parsing is often the huge number 587of callbacks made from the parser. Applying filters can improve 588performance significantly. 589 590The following methods control filters: 591 592=over 593 594=item $p->ignore_elements( @tags ) 595 596Both the C<start> event and the C<end> event as well as any events that 597would be reported in between are suppressed. The ignored elements can 598contain nested occurrences of itself. Example: 599 600 $p->ignore_elements(qw(script style)); 601 602The C<script> and C<style> tags will always nest properly since their 603content is parsed in CDATA mode. For most other tags 604C<ignore_elements> must be used with caution since HTML is often not 605I<well formed>. 606 607=item $p->ignore_tags( @tags ) 608 609Any C<start> and C<end> events involving any of the tags given are 610suppressed. To reset the filter (i.e. don't suppress any C<start> and 611C<end> events), call C<ignore_tags> without an argument. 612 613=item $p->report_tags( @tags ) 614 615Any C<start> and C<end> events involving any of the tags I<not> given 616are suppressed. To reset the filter (i.e. report all C<start> and 617C<end> events), call C<report_tags> without an argument. 618 619=back 620 621Internally, the system has two filter lists, one for C<report_tags> 622and one for C<ignore_tags>, and both filters are applied. This 623effectively gives C<ignore_tags> precedence over C<report_tags>. 624 625Examples: 626 627 $p->ignore_tags(qw(style)); 628 $p->report_tags(qw(script style)); 629 630results in only C<script> events being reported. 631 632=head2 Argspec 633 634Argspec is a string containing a comma-separated list that describes 635the information reported by the event. The following argspec 636identifier names can be used: 637 638=over 639 640=item C<attr> 641 642Attr causes a reference to a hash of attribute name/value pairs to be 643passed. 644 645Boolean attributes' values are either the value set by 646$p->boolean_attribute_value, or the attribute name if no value has been 647set by $p->boolean_attribute_value. 648 649This passes undef except for C<start> events. 650 651Unless C<xml_mode> or C<case_sensitive> is enabled, the attribute 652names are forced to lower case. 653 654General entities are decoded in the attribute values and 655one layer of matching quotes enclosing the attribute values is removed. 656 657The Unicode character set is assumed for entity decoding. 658 659=item C<@attr> 660 661Basically the same as C<attr>, but keys and values are passed as 662individual arguments and the original sequence of the attributes is 663kept. The parameters passed will be the same as the @attr calculated 664here: 665 666 @attr = map { $_ => $attr->{$_} } @$attrseq; 667 668assuming $attr and $attrseq here are the hash and array passed as the 669result of C<attr> and C<attrseq> argspecs. 670 671This passes no values for events besides C<start>. 672 673=item C<attrseq> 674 675Attrseq causes a reference to an array of attribute names to be 676passed. This can be useful if you want to walk the C<attr> hash in 677the original sequence. 678 679This passes undef except for C<start> events. 680 681Unless C<xml_mode> or C<case_sensitive> is enabled, the attribute 682names are forced to lower case. 683 684=item C<column> 685 686Column causes the column number of the start of the event to be passed. 687The first column on a line is 0. 688 689=item C<dtext> 690 691Dtext causes the decoded text to be passed. General entities are 692automatically decoded unless the event was inside a CDATA section or 693was between literal start and end tags (C<script>, C<style>, 694C<xmp>, C<iframe>, C<title>, C<textarea> and C<plaintext>). 695 696The Unicode character set is assumed for entity decoding. With Perl 697version 5.6 or earlier only the Latin-1 range is supported, and 698entities for characters outside the range 0..255 are left unchanged. 699 700This passes undef except for C<text> events. 701 702=item C<event> 703 704Event causes the event name to be passed. 705 706The event name is one of C<text>, C<start>, C<end>, C<declaration>, 707C<comment>, C<process>, C<start_document> or C<end_document>. 708 709=item C<is_cdata> 710 711Is_cdata causes a TRUE value to be passed if the event is inside a CDATA 712section or between literal start and end tags (C<script>, 713C<style>, C<xmp>, C<iframe>, C<title>, C<textarea> and C<plaintext>). 714 715if the flag is FALSE for a text event, then you should normally 716either use C<dtext> or decode the entities yourself before the text is 717processed further. 718 719=item C<length> 720 721Length causes the number of bytes of the source text of the event to 722be passed. 723 724=item C<line> 725 726Line causes the line number of the start of the event to be passed. 727The first line in the document is 1. Line counting doesn't start 728until at least one handler requests this value to be reported. 729 730=item C<offset> 731 732Offset causes the byte position in the HTML document of the start of 733the event to be passed. The first byte in the document has offset 0. 734 735=item C<offset_end> 736 737Offset_end causes the byte position in the HTML document of the end of 738the event to be passed. This is the same as C<offset> + C<length>. 739 740=item C<self> 741 742Self causes the current object to be passed to the handler. If the 743handler is a method, this must be the first element in the argspec. 744 745An alternative to passing self as an argspec is to register closures 746that capture $self by themselves as handlers. Unfortunately this 747creates circular references which prevent the HTML::Parser object 748from being garbage collected. Using the C<self> argspec avoids this 749problem. 750 751=item C<skipped_text> 752 753Skipped_text returns the concatenated text of all the events that have 754been skipped since the last time an event was reported. Events might 755be skipped because no handler is registered for them or because some 756filter applies. Skipped text also includes marked section markup, 757since there are no events that can catch it. 758 759If an C<"">-handler is registered for an event, then the text for this 760event is not included in C<skipped_text>. Skipped text both before 761and after the C<"">-event is included in the next reported 762C<skipped_text>. 763 764=item C<tag> 765 766Same as C<tagname>, but prefixed with "/" if it belongs to an C<end> 767event and "!" for a declaration. The C<tag> does not have any prefix 768for C<start> events, and is in this case identical to C<tagname>. 769 770=item C<tagname> 771 772This is the element name (or I<generic identifier> in SGML jargon) for 773start and end tags. Since HTML is case insensitive, this name is 774forced to lower case to ease string matching. 775 776Since XML is case sensitive, the tagname case is not changed when 777C<xml_mode> is enabled. The same happens if the C<case_sensitive> attribute 778is set. 779 780The declaration type of declaration elements is also passed as a tagname, 781even if that is a bit strange. 782In fact, in the current implementation tagname is 783identical to C<token0> except that the name may be forced to lower case. 784 785=item C<token0> 786 787Token0 causes the original text of the first token string to be 788passed. This should always be the same as $tokens->[0]. 789 790For C<declaration> events, this is the declaration type. 791 792For C<start> and C<end> events, this is the tag name. 793 794For C<process> and non-strict C<comment> events, this is everything 795inside the tag. 796 797This passes undef if there are no tokens in the event. 798 799=item C<tokenpos> 800 801Tokenpos causes a reference to an array of token positions to be 802passed. For each string that appears in C<tokens>, this array 803contains two numbers. The first number is the offset of the start of 804the token in the original C<text> and the second number is the length 805of the token. 806 807Boolean attributes in a C<start> event will have (0,0) for the 808attribute value offset and length. 809 810This passes undef if there are no tokens in the event (e.g., C<text>) 811and for artificial C<end> events triggered by empty element tags. 812 813If you are using these offsets and lengths to modify C<text>, you 814should either work from right to left, or be very careful to calculate 815the changes to the offsets. 816 817=item C<tokens> 818 819Tokens causes a reference to an array of token strings to be passed. 820The strings are exactly as they were found in the original text, 821no decoding or case changes are applied. 822 823For C<declaration> events, the array contains each word, comment, and 824delimited string starting with the declaration type. 825 826For C<comment> events, this contains each sub-comment. If 827$p->strict_comments is disabled, there will be only one sub-comment. 828 829For C<start> events, this contains the original tag name followed by 830the attribute name/value pairs. The values of boolean attributes will 831be either the value set by $p->boolean_attribute_value, or the 832attribute name if no value has been set by 833$p->boolean_attribute_value. 834 835For C<end> events, this contains the original tag name (always one token). 836 837For C<process> events, this contains the process instructions (always one 838token). 839 840This passes C<undef> for C<text> events. 841 842=item C<text> 843 844Text causes the source text (including markup element delimiters) to be 845passed. 846 847=item C<undef> 848 849Pass an undefined value. Useful as padding where the same handler 850routine is registered for multiple events. 851 852=item C<'...'> 853 854A literal string of 0 to 255 characters enclosed 855in single (') or double (") quotes is passed as entered. 856 857=back 858 859The whole argspec string can be wrapped up in C<'@{...}'> to signal 860that the resulting event array should be flattened. This only makes a 861difference if an array reference is used as the handler target. 862Consider this example: 863 864 $p->handler(text => [], 'text'); 865 $p->handler(text => [], '@{text}']); 866 867With two text events; C<"foo">, C<"bar">; then the first example will end 868up with [["foo"], ["bar"]] and the second with ["foo", "bar"] in 869the handler target array. 870 871 872=head2 Events 873 874Handlers for the following events can be registered: 875 876=over 877 878=item C<comment> 879 880This event is triggered when a markup comment is recognized. 881 882Example: 883 884 <!-- This is a comment -- -- So is this --> 885 886=item C<declaration> 887 888This event is triggered when a I<markup declaration> is recognized. 889 890For typical HTML documents, the only declaration you are 891likely to find is <!DOCTYPE ...>. 892 893Example: 894 895 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" 896 "http://www.w3.org/TR/html4/strict.dtd"> 897 898DTDs inside <!DOCTYPE ...> will confuse HTML::Parser. 899 900=item C<default> 901 902This event is triggered for events that do not have a specific 903handler. You can set up a handler for this event to catch stuff you 904did not want to catch explicitly. 905 906=item C<end> 907 908This event is triggered when an end tag is recognized. 909 910Example: 911 912 </A> 913 914=item C<end_document> 915 916This event is triggered when $p->eof is called and after any remaining 917text is flushed. There is no document text associated with this event. 918 919=item C<process> 920 921This event is triggered when a processing instructions markup is 922recognized. 923 924The format and content of processing instructions are system and 925application dependent. 926 927Examples: 928 929 <? HTML processing instructions > 930 <? XML processing instructions ?> 931 932=item C<start> 933 934This event is triggered when a start tag is recognized. 935 936Example: 937 938 <A HREF="http://www.perl.com/"> 939 940=item C<start_document> 941 942This event is triggered before any other events for a new document. A 943handler for it can be used to initialize stuff. There is no document 944text associated with this event. 945 946=item C<text> 947 948This event is triggered when plain text (characters) is recognized. 949The text may contain multiple lines. A sequence of text may be broken 950between several text events unless $p->unbroken_text is enabled. 951 952The parser will make sure that it does not break a word or a sequence 953of whitespace between two text events. 954 955=back 956 957=head2 Unicode 958 959C<HTML::Parser> can parse Unicode strings when running under 960perl-5.8 or better. If Unicode is passed to $p->parse() then chunks 961of Unicode will be reported to the handlers. The offset and length 962argspecs will also report their position in terms of characters. 963 964It is safe to parse raw undecoded UTF-8 if you either avoid decoding 965entities and make sure to not use I<argspecs> that do, or enable the 966C<utf8_mode> for the parser. Parsing of undecoded UTF-8 might be 967useful when parsing from a file where you need the reported offsets 968and lengths to match the byte offsets in the file. 969 970If a filename is passed to $p->parse_file() then the file will be read 971in binary mode. This will be fine if the file contains only ASCII or 972Latin-1 characters. If the file contains UTF-8 encoded text then care 973must be taken when decoding entities as described in the previous 974paragraph, but better is to open the file with the UTF-8 layer so that 975it is decoded properly: 976 977 open(my $fh, "<:utf8", "index.html") || die "...: $!"; 978 $p->parse_file($fh); 979 980If the file contains text encoded in a charset besides ASCII, Latin-1 981or UTF-8 then decoding will always be needed. 982 983=head1 VERSION 2 COMPATIBILITY 984 985When an C<HTML::Parser> object is constructed with no arguments, a set 986of handlers is automatically provided that is compatible with the old 987HTML::Parser version 2 callback methods. 988 989This is equivalent to the following method calls: 990 991 $p->handler(start => "start", "self, tagname, attr, attrseq, text"); 992 $p->handler(end => "end", "self, tagname, text"); 993 $p->handler(text => "text", "self, text, is_cdata"); 994 $p->handler(process => "process", "self, token0, text"); 995 $p->handler( 996 comment => sub { 997 my($self, $tokens) = @_; 998 for (@$tokens) {$self->comment($_);} 999 }, 1000 "self, tokens" 1001 ); 1002 $p->handler( 1003 declaration => sub { 1004 my $self = shift; 1005 $self->declaration(substr($_[0], 2, -1)); 1006 }, 1007 "self, text" 1008 ); 1009 1010Setting up these handlers can also be requested with the "api_version => 10112" constructor option. 1012 1013=head1 SUBCLASSING 1014 1015The C<HTML::Parser> class is able to be subclassed. Parser objects are plain 1016hashes and C<HTML::Parser> reserves only hash keys that start with 1017"_hparser". The parser state can be set up by invoking the init() 1018method, which takes the same arguments as new(). 1019 1020=head1 EXAMPLES 1021 1022The first simple example shows how you might strip out comments from 1023an HTML document. We achieve this by setting up a comment handler that 1024does nothing and a default handler that will print out anything else: 1025 1026 use HTML::Parser; 1027 HTML::Parser->new( 1028 default_h => [sub { print shift }, 'text'], 1029 comment_h => [""], 1030 )->parse_file(shift || die) || die $!; 1031 1032An alternative implementation is: 1033 1034 use HTML::Parser; 1035 HTML::Parser->new( 1036 end_document_h => [sub { print shift }, 'skipped_text'], 1037 comment_h => [""], 1038 )->parse_file(shift || die) || die $!; 1039 1040This will in most cases be much more efficient since only a single 1041callback will be made. 1042 1043The next example prints out the text that is inside the <title> 1044element of an HTML document. Here we start by setting up a start 1045handler. When it sees the title start tag it enables a text handler 1046that prints any text found and an end handler that will terminate 1047parsing as soon as the title end tag is seen: 1048 1049 use HTML::Parser (); 1050 1051 sub start_handler { 1052 return if shift ne "title"; 1053 my $self = shift; 1054 $self->handler(text => sub { print shift }, "dtext"); 1055 $self->handler( 1056 end => sub { 1057 shift->eof if shift eq "title"; 1058 }, 1059 "tagname,self" 1060 ); 1061 } 1062 1063 my $p = HTML::Parser->new(api_version => 3); 1064 $p->handler(start => \&start_handler, "tagname,self"); 1065 $p->parse_file(shift || die) || die $!; 1066 print "\n"; 1067 1068More examples are found in the F<eg/> directory of the C<HTML-Parser> 1069distribution: the program C<hrefsub> shows how you can edit all links 1070found in a document; the program C<htextsub> shows how to edit the text only; the 1071program C<hstrip> shows how you can strip out certain tags/elements 1072and/or attributes; and the program C<htext> show how to obtain the 1073plain text, but not any script/style content. 1074 1075You can browse the F<eg/> directory online from the I<[Browse]> link on 1076the http://search.cpan.org/~gaas/HTML-Parser/ page. 1077 1078=head1 BUGS 1079 1080The <style> and <script> sections do not end with the first "</", but 1081need the complete corresponding end tag. The standard behaviour is 1082not really practical. 1083 1084When the I<strict_comment> option is enabled, we still recognize 1085comments where there is something other than whitespace between even 1086and odd "--" markers. 1087 1088Once $p->boolean_attribute_value has been set, there is no way to 1089restore the default behaviour. 1090 1091There is currently no way to get both quote characters 1092into the same literal argspec. 1093 1094Empty tags, e.g. "<>" and "</>", are not recognized. SGML allows them 1095to repeat the previous start tag or close the previous start tag 1096respectively. 1097 1098NET tags, e.g. "code/.../" are not recognized. This is SGML 1099shorthand for "<code>...</code>". 1100 1101Incomplete start or end tags, e.g. "<tt<b>...</b</tt>" are not 1102recognized. 1103 1104=head1 DIAGNOSTICS 1105 1106The following messages may be produced by HTML::Parser. The notation 1107in this listing is the same as used in L<perldiag>: 1108 1109=over 1110 1111=item Not a reference to a hash 1112 1113(F) The object blessed into or subclassed from HTML::Parser is not a 1114hash as required by the HTML::Parser methods. 1115 1116=item Bad signature in parser state object at %p 1117 1118(F) The _hparser_xs_state element does not refer to a valid state structure. 1119Something must have changed the internal value 1120stored in this hash element, or the memory has been overwritten. 1121 1122=item _hparser_xs_state element is not a reference 1123 1124(F) The _hparser_xs_state element has been destroyed. 1125 1126=item Can't find '_hparser_xs_state' element in HTML::Parser hash 1127 1128(F) The _hparser_xs_state element is missing from the parser hash. 1129It was either deleted, or not created when the object was created. 1130 1131=item API version %s not supported by HTML::Parser %s 1132 1133(F) The constructor option 'api_version' with an argument greater than 1134or equal to 4 is reserved for future extensions. 1135 1136=item Bad constructor option '%s' 1137 1138(F) An unknown constructor option key was passed to the new() or 1139init() methods. 1140 1141=item Parse loop not allowed 1142 1143(F) A handler invoked the parse() or parse_file() method. 1144This is not permitted. 1145 1146=item marked sections not supported 1147 1148(F) The $p->marked_sections() method was invoked in a HTML::Parser 1149module that was compiled without support for marked sections. 1150 1151=item Unknown boolean attribute (%d) 1152 1153(F) Something is wrong with the internal logic that set up aliases for 1154boolean attributes. 1155 1156=item Only code or array references allowed as handler 1157 1158(F) The second argument for $p->handler must be either a subroutine 1159reference, then name of a subroutine or method, or a reference to an 1160array. 1161 1162=item No handler for %s events 1163 1164(F) The first argument to $p->handler must be a valid event name; i.e. one 1165of "start", "end", "text", "process", "declaration" or "comment". 1166 1167=item Unrecognized identifier %s in argspec 1168 1169(F) The identifier is not a known argspec name. 1170Use one of the names mentioned in the argspec section above. 1171 1172=item Literal string is longer than 255 chars in argspec 1173 1174(F) The current implementation limits the length of literals in 1175an argspec to 255 characters. Make the literal shorter. 1176 1177=item Backslash reserved for literal string in argspec 1178 1179(F) The backslash character "\" is not allowed in argspec literals. 1180It is reserved to permit quoting inside a literal in a later version. 1181 1182=item Unterminated literal string in argspec 1183 1184(F) The terminating quote character for a literal was not found. 1185 1186=item Bad argspec (%s) 1187 1188(F) Only identifier names, literals, spaces and commas 1189are allowed in argspecs. 1190 1191=item Missing comma separator in argspec 1192 1193(F) Identifiers in an argspec must be separated with ",". 1194 1195=item Parsing of undecoded UTF-8 will give garbage when decoding entities 1196 1197(W) The first chunk parsed appears to contain undecoded UTF-8 and one 1198or more argspecs that decode entities are used for the callback 1199handlers. 1200 1201The result of decoding will be a mix of encoded and decoded characters 1202for any entities that expand to characters with code above 127. This 1203is not a good thing. 1204 1205The recommended solution is to apply Encode::decode_utf8() on the data before 1206feeding it to the $p->parse(). For $p->parse_file() pass a file that has been 1207opened in ":utf8" mode. 1208 1209The alternative solution is to enable the C<utf8_mode> and not decode before 1210passing strings to $p->parse(). The parser can process raw undecoded UTF-8 1211sanely if the C<utf8_mode> is enabled, or if the C<attr>, C<@attr> or C<dtext> 1212argspecs are avoided. 1213 1214=item Parsing string decoded with wrong endian selection 1215 1216(W) The first character in the document is U+FFFE. This is not a 1217legal Unicode character but a byte swapped C<BOM>. The result of parsing 1218will likely be garbage. 1219 1220=item Parsing of undecoded UTF-32 1221 1222(W) The parser found the Unicode UTF-32 C<BOM> signature at the start 1223of the document. The result of parsing will likely be garbage. 1224 1225=item Parsing of undecoded UTF-16 1226 1227(W) The parser found the Unicode UTF-16 C<BOM> signature at the start of 1228the document. The result of parsing will likely be garbage. 1229 1230=back 1231 1232=head1 SEE ALSO 1233 1234L<HTML::Entities>, L<HTML::PullParser>, L<HTML::TokeParser>, L<HTML::HeadParser>, 1235L<HTML::LinkExtor>, L<HTML::Form> 1236 1237L<HTML::TreeBuilder> (part of the I<HTML-Tree> distribution) 1238 1239L<http://www.w3.org/TR/html4/> 1240 1241More information about marked sections and processing instructions may 1242be found at L<http://www.is-thought.co.uk/book/sgml-8.htm>. 1243 1244=head1 COPYRIGHT 1245 1246 Copyright 1996-2016 Gisle Aas. All rights reserved. 1247 Copyright 1999-2000 Michael A. Chase. All rights reserved. 1248 1249This library is free software; you can redistribute it and/or 1250modify it under the same terms as Perl itself. 1251 1252=cut 1253