/[suikacvs]/markup/html/whatpm/Whatpm/HTML.pm
Suika

Contents of /markup/html/whatpm/Whatpm/HTML.pm

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1.109 - (hide annotations) (download)
Tue Mar 11 14:02:08 2008 UTC (18 years, 4 months ago) by wakaba
Branch: MAIN
Changes since 1.108: +19 -17 lines
++ whatpm/Whatpm/ChangeLog	11 Mar 2008 14:02:04 -0000
	* HTML.pm.src: Error type revised.

2008-03-11  Wakaba  <wakaba@suika.fam.cx>

1 wakaba 1.2 package Whatpm::HTML;
2 wakaba 1.1 use strict;
3 wakaba 1.109 our $VERSION=do{my @r=(q$Revision: 1.110 $=~/\d+/g);sprintf "%d."."%02d" x $#r,@r};
4 wakaba 1.65 use Error qw(:try);
5 wakaba 1.1
6 wakaba 1.18 ## ISSUE:
7     ## var doc = implementation.createDocument (null, null, null);
8     ## doc.write ('');
9     ## alert (doc.compatMode);
10 wakaba 1.1
11 wakaba 1.71 ## TODO: Control charcters and noncharacters are not allowed (HTML5 revision 1263)
12     ## TODO: 1252 parse error (revision 1264)
13     ## TODO: 8859-11 = 874 (revision 1271)
14 wakaba 1.31
15 wakaba 1.1 my $permitted_slash_tag_name = {
16     base => 1,
17     link => 1,
18     meta => 1,
19     hr => 1,
20     br => 1,
21 wakaba 1.71 img => 1,
22 wakaba 1.1 embed => 1,
23     param => 1,
24     area => 1,
25     col => 1,
26     input => 1,
27     };
28    
29 wakaba 1.4 my $c1_entity_char = {
30 wakaba 1.9 0x80 => 0x20AC,
31     0x81 => 0xFFFD,
32     0x82 => 0x201A,
33     0x83 => 0x0192,
34     0x84 => 0x201E,
35     0x85 => 0x2026,
36     0x86 => 0x2020,
37     0x87 => 0x2021,
38     0x88 => 0x02C6,
39     0x89 => 0x2030,
40     0x8A => 0x0160,
41     0x8B => 0x2039,
42     0x8C => 0x0152,
43     0x8D => 0xFFFD,
44     0x8E => 0x017D,
45     0x8F => 0xFFFD,
46     0x90 => 0xFFFD,
47     0x91 => 0x2018,
48     0x92 => 0x2019,
49     0x93 => 0x201C,
50     0x94 => 0x201D,
51     0x95 => 0x2022,
52     0x96 => 0x2013,
53     0x97 => 0x2014,
54     0x98 => 0x02DC,
55     0x99 => 0x2122,
56     0x9A => 0x0161,
57     0x9B => 0x203A,
58     0x9C => 0x0153,
59     0x9D => 0xFFFD,
60     0x9E => 0x017E,
61     0x9F => 0x0178,
62 wakaba 1.4 }; # $c1_entity_char
63 wakaba 1.1
64     my $special_category = {
65     address => 1, area => 1, base => 1, basefont => 1, bgsound => 1,
66     blockquote => 1, body => 1, br => 1, center => 1, col => 1, colgroup => 1,
67     dd => 1, dir => 1, div => 1, dl => 1, dt => 1, embed => 1, fieldset => 1,
68     form => 1, frame => 1, frameset => 1, h1 => 1, h2 => 1, h3 => 1,
69     h4 => 1, h5 => 1, h6 => 1, head => 1, hr => 1, iframe => 1, image => 1,
70     img => 1, input => 1, isindex => 1, li => 1, link => 1, listing => 1,
71     menu => 1, meta => 1, noembed => 1, noframes => 1, noscript => 1,
72     ol => 1, optgroup => 1, option => 1, p => 1, param => 1, plaintext => 1,
73     pre => 1, script => 1, select => 1, spacer => 1, style => 1, tbody => 1,
74     textarea => 1, tfoot => 1, thead => 1, title => 1, tr => 1, ul => 1, wbr => 1,
75     };
76     my $scoping_category = {
77 wakaba 1.102 applet => 1, button => 1, caption => 1, html => 1, marquee => 1, object => 1,
78 wakaba 1.1 table => 1, td => 1, th => 1,
79     };
80     my $formatting_category = {
81     a => 1, b => 1, big => 1, em => 1, font => 1, i => 1, nobr => 1,
82     s => 1, small => 1, strile => 1, strong => 1, tt => 1, u => 1,
83     };
84     # $phrasing_category: all other elements
85    
86 wakaba 1.65 sub parse_byte_string ($$$$;$) {
87     my $self = ref $_[0] ? shift : shift->new;
88     my $charset = shift;
89     my $bytes_s = ref $_[0] ? $_[0] : \($_[0]);
90     my $s;
91    
92     if (defined $charset) {
93 wakaba 1.66 require Encode; ## TODO: decode(utf8) don't delete BOM
94 wakaba 1.65 $s = \ (Encode::decode ($charset, $$bytes_s));
95 wakaba 1.66 $self->{input_encoding} = lc $charset; ## TODO: normalize name
96 wakaba 1.65 $self->{confident} = 1;
97     } else {
98 wakaba 1.67 ## TODO: Implement HTML5 detection algorithm
99     require Whatpm::Charset::UniversalCharDet;
100     $charset = Whatpm::Charset::UniversalCharDet->detect_byte_string
101     (substr ($$bytes_s, 0, 1024));
102     $charset ||= 'windows-1252';
103 wakaba 1.66 $s = \ (Encode::decode ($charset, $$bytes_s));
104     $self->{input_encoding} = $charset;
105 wakaba 1.65 $self->{confident} = 0;
106     }
107    
108     $self->{change_encoding} = sub {
109     my $self = shift;
110     my $charset = lc shift;
111     ## TODO: if $charset is supported
112     ## TODO: normalize charset name
113    
114     ## "Change the encoding" algorithm:
115    
116     ## Step 1
117     if ($charset eq 'utf-16') { ## ISSUE: UTF-16BE -> UTF-8? UTF-16LE -> UTF-8?
118     $charset = 'utf-8';
119     }
120    
121     ## Step 2
122     if (defined $self->{input_encoding} and
123     $self->{input_encoding} eq $charset) {
124     $self->{confident} = 1;
125     return;
126     }
127    
128 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'charset label detected:'.$self->{input_encoding}.
129 wakaba 1.66 ':'.$charset, level => 'w');
130 wakaba 1.65
131     ## Step 3
132     # if (can) {
133     ## change the encoding on the fly.
134     #$self->{confident} = 1;
135     #return;
136     # }
137    
138     ## Step 4
139     throw Whatpm::HTML::RestartParser (charset => $charset);
140     }; # $self->{change_encoding}
141    
142     my @args = @_; shift @args; # $s
143     my $return;
144     try {
145     $return = $self->parse_char_string ($s, @args);
146     } catch Whatpm::HTML::RestartParser with {
147     my $charset = shift->{charset};
148     $s = \ (Encode::decode ($charset, $$bytes_s));
149 wakaba 1.66 $self->{input_encoding} = $charset; ## TODO: normalize
150 wakaba 1.65 $self->{confident} = 1;
151     $return = $self->parse_char_string ($s, @args);
152     };
153     return $return;
154     } # parse_byte_string
155    
156 wakaba 1.71 ## NOTE: HTML5 spec says that the encoding layer MUST NOT strip BOM
157     ## and the HTML layer MUST ignore it. However, we does strip BOM in
158     ## the encoding layer and the HTML layer does not ignore any U+FEFF,
159     ## because the core part of our HTML parser expects a string of character,
160     ## not a string of bytes or code units or anything which might contain a BOM.
161     ## Therefore, any parser interface that accepts a string of bytes,
162     ## such as |parse_byte_string| in this module, must ensure that it does
163     ## strip the BOM and never strip any ZWNBSP.
164    
165 wakaba 1.65 *parse_char_string = \&parse_string;
166    
167 wakaba 1.1 sub parse_string ($$$;$) {
168 wakaba 1.65 my $self = ref $_[0] ? shift : shift->new;
169     my $s = ref $_[0] ? $_[0] : \($_[0]);
170 wakaba 1.1 $self->{document} = $_[1];
171 wakaba 1.65 @{$self->{document}->child_nodes} = ();
172 wakaba 1.1
173 wakaba 1.3 ## NOTE: |set_inner_html| copies most of this method's code
174    
175 wakaba 1.65 $self->{confident} = 1 unless exists $self->{confident};
176 wakaba 1.66 $self->{document}->input_encoding ($self->{input_encoding})
177     if defined $self->{input_encoding};
178 wakaba 1.65
179 wakaba 1.1 my $i = 0;
180 wakaba 1.3 my $line = 1;
181     my $column = 0;
182 wakaba 1.76 $self->{set_next_char} = sub {
183 wakaba 1.1 my $self = shift;
184 wakaba 1.13
185 wakaba 1.76 pop @{$self->{prev_char}};
186     unshift @{$self->{prev_char}}, $self->{next_char};
187 wakaba 1.13
188 wakaba 1.76 $self->{next_char} = -1 and return if $i >= length $$s;
189     $self->{next_char} = ord substr $$s, $i++, 1;
190 wakaba 1.3 $column++;
191 wakaba 1.1
192 wakaba 1.76 if ($self->{next_char} == 0x000A) { # LF
193 wakaba 1.4 $line++;
194     $column = 0;
195 wakaba 1.76 } elsif ($self->{next_char} == 0x000D) { # CR
196 wakaba 1.15 $i++ if substr ($$s, $i, 1) eq "\x0A";
197 wakaba 1.76 $self->{next_char} = 0x000A; # LF # MUST
198 wakaba 1.3 $line++;
199 wakaba 1.4 $column = 0;
200 wakaba 1.76 } elsif ($self->{next_char} > 0x10FFFF) {
201     $self->{next_char} = 0xFFFD; # REPLACEMENT CHARACTER # MUST
202     } elsif ($self->{next_char} == 0x0000) { # NULL
203 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'NULL');
204 wakaba 1.76 $self->{next_char} = 0xFFFD; # REPLACEMENT CHARACTER # MUST
205 wakaba 1.1 }
206     };
207 wakaba 1.76 $self->{prev_char} = [-1, -1, -1];
208     $self->{next_char} = -1;
209 wakaba 1.1
210 wakaba 1.3 my $onerror = $_[2] || sub {
211     my (%opt) = @_;
212     warn "Parse error ($opt{type}) at line $opt{line} column $opt{column}\n";
213     };
214     $self->{parse_error} = sub {
215     $onerror->(@_, line => $line, column => $column);
216 wakaba 1.1 };
217    
218     $self->_initialize_tokenizer;
219     $self->_initialize_tree_constructor;
220     $self->_construct_tree;
221     $self->_terminate_tree_constructor;
222    
223     return $self->{document};
224     } # parse_string
225    
226     sub new ($) {
227     my $class = shift;
228     my $self = bless {}, $class;
229 wakaba 1.76 $self->{set_next_char} = sub {
230     $self->{next_char} = -1;
231 wakaba 1.1 };
232     $self->{parse_error} = sub {
233     #
234     };
235 wakaba 1.65 $self->{change_encoding} = sub {
236     # if ($_[0] is a supported encoding) {
237     # run "change the encoding" algorithm;
238     # throw Whatpm::HTML::RestartParser (charset => $new_encoding);
239     # }
240     };
241 wakaba 1.63 $self->{application_cache_selection} = sub {
242     #
243     };
244 wakaba 1.1 return $self;
245     } # new
246    
247 wakaba 1.41 sub CM_ENTITY () { 0b001 } # & markup in data
248     sub CM_LIMITED_MARKUP () { 0b010 } # < markup in data (limited)
249     sub CM_FULL_MARKUP () { 0b100 } # < markup in data (any)
250    
251     sub PLAINTEXT_CONTENT_MODEL () { 0 }
252     sub CDATA_CONTENT_MODEL () { CM_LIMITED_MARKUP }
253     sub RCDATA_CONTENT_MODEL () { CM_ENTITY | CM_LIMITED_MARKUP }
254     sub PCDATA_CONTENT_MODEL () { CM_ENTITY | CM_FULL_MARKUP }
255    
256 wakaba 1.59 sub DATA_STATE () { 0 }
257     sub ENTITY_DATA_STATE () { 1 }
258     sub TAG_OPEN_STATE () { 2 }
259     sub CLOSE_TAG_OPEN_STATE () { 3 }
260     sub TAG_NAME_STATE () { 4 }
261     sub BEFORE_ATTRIBUTE_NAME_STATE () { 5 }
262     sub ATTRIBUTE_NAME_STATE () { 6 }
263     sub AFTER_ATTRIBUTE_NAME_STATE () { 7 }
264     sub BEFORE_ATTRIBUTE_VALUE_STATE () { 8 }
265     sub ATTRIBUTE_VALUE_DOUBLE_QUOTED_STATE () { 9 }
266     sub ATTRIBUTE_VALUE_SINGLE_QUOTED_STATE () { 10 }
267     sub ATTRIBUTE_VALUE_UNQUOTED_STATE () { 11 }
268     sub ENTITY_IN_ATTRIBUTE_VALUE_STATE () { 12 }
269     sub MARKUP_DECLARATION_OPEN_STATE () { 13 }
270     sub COMMENT_START_STATE () { 14 }
271     sub COMMENT_START_DASH_STATE () { 15 }
272     sub COMMENT_STATE () { 16 }
273     sub COMMENT_END_STATE () { 17 }
274     sub COMMENT_END_DASH_STATE () { 18 }
275     sub BOGUS_COMMENT_STATE () { 19 }
276     sub DOCTYPE_STATE () { 20 }
277     sub BEFORE_DOCTYPE_NAME_STATE () { 21 }
278     sub DOCTYPE_NAME_STATE () { 22 }
279     sub AFTER_DOCTYPE_NAME_STATE () { 23 }
280     sub BEFORE_DOCTYPE_PUBLIC_IDENTIFIER_STATE () { 24 }
281     sub DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED_STATE () { 25 }
282     sub DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED_STATE () { 26 }
283     sub AFTER_DOCTYPE_PUBLIC_IDENTIFIER_STATE () { 27 }
284     sub BEFORE_DOCTYPE_SYSTEM_IDENTIFIER_STATE () { 28 }
285     sub DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED_STATE () { 29 }
286     sub DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED_STATE () { 30 }
287     sub AFTER_DOCTYPE_SYSTEM_IDENTIFIER_STATE () { 31 }
288     sub BOGUS_DOCTYPE_STATE () { 32 }
289 wakaba 1.72 sub AFTER_ATTRIBUTE_VALUE_QUOTED_STATE () { 33 }
290 wakaba 1.59
291 wakaba 1.57 sub DOCTYPE_TOKEN () { 1 }
292     sub COMMENT_TOKEN () { 2 }
293     sub START_TAG_TOKEN () { 3 }
294     sub END_TAG_TOKEN () { 4 }
295     sub END_OF_FILE_TOKEN () { 5 }
296     sub CHARACTER_TOKEN () { 6 }
297    
298 wakaba 1.56 sub AFTER_HTML_IMS () { 0b100 }
299     sub HEAD_IMS () { 0b1000 }
300     sub BODY_IMS () { 0b10000 }
301 wakaba 1.58 sub BODY_TABLE_IMS () { 0b100000 }
302 wakaba 1.56 sub TABLE_IMS () { 0b1000000 }
303 wakaba 1.58 sub ROW_IMS () { 0b10000000 }
304 wakaba 1.56 sub BODY_AFTER_IMS () { 0b100000000 }
305     sub FRAME_IMS () { 0b1000000000 }
306 wakaba 1.101 sub SELECT_IMS () { 0b10000000000 }
307 wakaba 1.56
308 wakaba 1.85 ## NOTE: "initial" and "before html" insertion modes have no constants.
309 wakaba 1.84
310     ## NOTE: "after after body" insertion mode.
311 wakaba 1.56 sub AFTER_HTML_BODY_IM () { AFTER_HTML_IMS | BODY_AFTER_IMS }
312 wakaba 1.84
313     ## NOTE: "after after frameset" insertion mode.
314 wakaba 1.56 sub AFTER_HTML_FRAMESET_IM () { AFTER_HTML_IMS | FRAME_IMS }
315 wakaba 1.84
316 wakaba 1.56 sub IN_HEAD_IM () { HEAD_IMS | 0b00 }
317     sub IN_HEAD_NOSCRIPT_IM () { HEAD_IMS | 0b01 }
318     sub AFTER_HEAD_IM () { HEAD_IMS | 0b10 }
319     sub BEFORE_HEAD_IM () { HEAD_IMS | 0b11 }
320     sub IN_BODY_IM () { BODY_IMS }
321 wakaba 1.58 sub IN_CELL_IM () { BODY_IMS | BODY_TABLE_IMS | 0b01 }
322     sub IN_CAPTION_IM () { BODY_IMS | BODY_TABLE_IMS | 0b10 }
323     sub IN_ROW_IM () { TABLE_IMS | ROW_IMS | 0b01 }
324     sub IN_TABLE_BODY_IM () { TABLE_IMS | ROW_IMS | 0b10 }
325 wakaba 1.56 sub IN_TABLE_IM () { TABLE_IMS }
326     sub AFTER_BODY_IM () { BODY_AFTER_IMS }
327     sub IN_FRAMESET_IM () { FRAME_IMS | 0b01 }
328     sub AFTER_FRAMESET_IM () { FRAME_IMS | 0b10 }
329 wakaba 1.101 sub IN_SELECT_IM () { SELECT_IMS | 0b01 }
330     sub IN_SELECT_IN_TABLE_IM () { SELECT_IMS | 0b10 }
331 wakaba 1.56 sub IN_COLUMN_GROUP_IM () { 0b10 }
332    
333 wakaba 1.1 ## Implementations MUST act as if state machine in the spec
334    
335     sub _initialize_tokenizer ($) {
336     my $self = shift;
337 wakaba 1.59 $self->{state} = DATA_STATE; # MUST
338 wakaba 1.41 $self->{content_model} = PCDATA_CONTENT_MODEL; # be
339 wakaba 1.1 undef $self->{current_token}; # start tag, end tag, comment, or DOCTYPE
340     undef $self->{current_attribute};
341     undef $self->{last_emitted_start_tag_name};
342     undef $self->{last_attribute_value_state};
343     $self->{char} = [];
344 wakaba 1.76 # $self->{next_char}
345 wakaba 1.1
346     if (@{$self->{char}}) {
347 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
348 wakaba 1.1 } else {
349 wakaba 1.76 $self->{set_next_char}->($self);
350 wakaba 1.1 }
351    
352     $self->{token} = [];
353 wakaba 1.18 # $self->{escape}
354 wakaba 1.1 } # _initialize_tokenizer
355    
356     ## A token has:
357 wakaba 1.57 ## ->{type} == DOCTYPE_TOKEN, START_TAG_TOKEN, END_TAG_TOKEN, COMMENT_TOKEN,
358     ## CHARACTER_TOKEN, or END_OF_FILE_TOKEN
359     ## ->{name} (DOCTYPE_TOKEN)
360     ## ->{tag_name} (START_TAG_TOKEN, END_TAG_TOKEN)
361     ## ->{public_identifier} (DOCTYPE_TOKEN)
362     ## ->{system_identifier} (DOCTYPE_TOKEN)
363 wakaba 1.75 ## ->{quirks} == 1 or 0 (DOCTYPE_TOKEN): "force-quirks" flag
364 wakaba 1.57 ## ->{attributes} isa HASH (START_TAG_TOKEN, END_TAG_TOKEN)
365 wakaba 1.68 ## ->{name}
366     ## ->{value}
367     ## ->{has_reference} == 1 or 0
368 wakaba 1.57 ## ->{data} (COMMENT_TOKEN, CHARACTER_TOKEN)
369 wakaba 1.1
370     ## Emitted token MUST immediately be handled by the tree construction state.
371    
372     ## Before each step, UA MAY check to see if either one of the scripts in
373     ## "list of scripts that will execute as soon as possible" or the first
374     ## script in the "list of scripts that will execute asynchronously",
375     ## has completed loading. If one has, then it MUST be executed
376     ## and removed from the list.
377    
378 wakaba 1.61 ## NOTE: HTML5 "Writing HTML documents" section, applied to
379     ## documents and not to user agents and conformance checkers,
380     ## contains some requirements that are not detected by the
381     ## parsing algorithm:
382     ## - Some requirements on character encoding declarations. ## TODO
383     ## - "Elements MUST NOT contain content that their content model disallows."
384     ## ... Some are parse error, some are not (will be reported by c.c.).
385     ## - Polytheistic slash SHOULD NOT be used. (Applied only to atheists.) ## TODO
386     ## - Text (in elements, attributes, and comments) SHOULD NOT contain
387     ## control characters other than space characters. ## TODO: (what is control character? C0, C1 and DEL? Unicode control character?)
388    
389     ## TODO: HTML5 poses authors two SHOULD-level requirements that cannot
390     ## be detected by the HTML5 parsing algorithm:
391     ## - Text,
392    
393 wakaba 1.1 sub _get_next_token ($) {
394     my $self = shift;
395     if (@{$self->{token}}) {
396     return shift @{$self->{token}};
397     }
398    
399     A: {
400 wakaba 1.59 if ($self->{state} == DATA_STATE) {
401 wakaba 1.76 if ($self->{next_char} == 0x0026) { # &
402 wakaba 1.72 if ($self->{content_model} & CM_ENTITY and # PCDATA | RCDATA
403     not $self->{escape}) {
404 wakaba 1.77
405 wakaba 1.59 $self->{state} = ENTITY_DATA_STATE;
406 wakaba 1.1
407     if (@{$self->{char}}) {
408 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
409 wakaba 1.1 } else {
410 wakaba 1.76 $self->{set_next_char}->($self);
411 wakaba 1.1 }
412    
413     redo A;
414     } else {
415 wakaba 1.77
416 wakaba 1.1 #
417     }
418 wakaba 1.76 } elsif ($self->{next_char} == 0x002D) { # -
419 wakaba 1.41 if ($self->{content_model} & CM_LIMITED_MARKUP) { # RCDATA | CDATA
420 wakaba 1.13 unless ($self->{escape}) {
421 wakaba 1.76 if ($self->{prev_char}->[0] == 0x002D and # -
422     $self->{prev_char}->[1] == 0x0021 and # !
423     $self->{prev_char}->[2] == 0x003C) { # <
424 wakaba 1.77
425 wakaba 1.13 $self->{escape} = 1;
426 wakaba 1.77 } else {
427    
428 wakaba 1.13 }
429 wakaba 1.77 } else {
430    
431 wakaba 1.13 }
432     }
433    
434     #
435 wakaba 1.76 } elsif ($self->{next_char} == 0x003C) { # <
436 wakaba 1.41 if ($self->{content_model} & CM_FULL_MARKUP or # PCDATA
437     (($self->{content_model} & CM_LIMITED_MARKUP) and # CDATA | RCDATA
438 wakaba 1.13 not $self->{escape})) {
439 wakaba 1.77
440 wakaba 1.59 $self->{state} = TAG_OPEN_STATE;
441 wakaba 1.1
442     if (@{$self->{char}}) {
443 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
444 wakaba 1.1 } else {
445 wakaba 1.76 $self->{set_next_char}->($self);
446 wakaba 1.1 }
447    
448     redo A;
449     } else {
450 wakaba 1.77
451 wakaba 1.1 #
452     }
453 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
454 wakaba 1.13 if ($self->{escape} and
455 wakaba 1.41 ($self->{content_model} & CM_LIMITED_MARKUP)) { # RCDATA | CDATA
456 wakaba 1.76 if ($self->{prev_char}->[0] == 0x002D and # -
457     $self->{prev_char}->[1] == 0x002D) { # -
458 wakaba 1.77
459 wakaba 1.13 delete $self->{escape};
460 wakaba 1.77 } else {
461    
462 wakaba 1.13 }
463 wakaba 1.77 } else {
464    
465 wakaba 1.13 }
466    
467     #
468 wakaba 1.76 } elsif ($self->{next_char} == -1) {
469 wakaba 1.77
470 wakaba 1.57 return ({type => END_OF_FILE_TOKEN});
471 wakaba 1.1 last A; ## TODO: ok?
472 wakaba 1.77 } else {
473    
474 wakaba 1.1 }
475     # Anything else
476 wakaba 1.57 my $token = {type => CHARACTER_TOKEN,
477 wakaba 1.76 data => chr $self->{next_char}};
478 wakaba 1.1 ## Stay in the data state
479    
480     if (@{$self->{char}}) {
481 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
482 wakaba 1.1 } else {
483 wakaba 1.76 $self->{set_next_char}->($self);
484 wakaba 1.1 }
485    
486    
487     return ($token);
488    
489     redo A;
490 wakaba 1.59 } elsif ($self->{state} == ENTITY_DATA_STATE) {
491 wakaba 1.1 ## (cannot happen in CDATA state)
492    
493 wakaba 1.72 my $token = $self->_tokenize_attempt_to_consume_an_entity (0, -1);
494 wakaba 1.1
495 wakaba 1.59 $self->{state} = DATA_STATE;
496 wakaba 1.1 # next-input-character is already done
497    
498     unless (defined $token) {
499 wakaba 1.77
500 wakaba 1.57 return ({type => CHARACTER_TOKEN, data => '&'});
501 wakaba 1.1 } else {
502 wakaba 1.77
503 wakaba 1.1 return ($token);
504     }
505    
506     redo A;
507 wakaba 1.59 } elsif ($self->{state} == TAG_OPEN_STATE) {
508 wakaba 1.41 if ($self->{content_model} & CM_LIMITED_MARKUP) { # RCDATA | CDATA
509 wakaba 1.76 if ($self->{next_char} == 0x002F) { # /
510 wakaba 1.1
511 wakaba 1.77
512 wakaba 1.1 if (@{$self->{char}}) {
513 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
514 wakaba 1.1 } else {
515 wakaba 1.76 $self->{set_next_char}->($self);
516 wakaba 1.1 }
517    
518 wakaba 1.59 $self->{state} = CLOSE_TAG_OPEN_STATE;
519 wakaba 1.1 redo A;
520     } else {
521 wakaba 1.77
522 wakaba 1.1 ## reconsume
523 wakaba 1.59 $self->{state} = DATA_STATE;
524 wakaba 1.1
525 wakaba 1.57 return ({type => CHARACTER_TOKEN, data => '<'});
526 wakaba 1.1
527     redo A;
528     }
529 wakaba 1.41 } elsif ($self->{content_model} & CM_FULL_MARKUP) { # PCDATA
530 wakaba 1.76 if ($self->{next_char} == 0x0021) { # !
531 wakaba 1.77
532 wakaba 1.59 $self->{state} = MARKUP_DECLARATION_OPEN_STATE;
533 wakaba 1.1
534     if (@{$self->{char}}) {
535 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
536 wakaba 1.1 } else {
537 wakaba 1.76 $self->{set_next_char}->($self);
538 wakaba 1.1 }
539    
540     redo A;
541 wakaba 1.76 } elsif ($self->{next_char} == 0x002F) { # /
542 wakaba 1.77
543 wakaba 1.59 $self->{state} = CLOSE_TAG_OPEN_STATE;
544 wakaba 1.1
545     if (@{$self->{char}}) {
546 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
547 wakaba 1.1 } else {
548 wakaba 1.76 $self->{set_next_char}->($self);
549 wakaba 1.1 }
550    
551     redo A;
552 wakaba 1.76 } elsif (0x0041 <= $self->{next_char} and
553     $self->{next_char} <= 0x005A) { # A..Z
554 wakaba 1.77
555 wakaba 1.1 $self->{current_token}
556 wakaba 1.57 = {type => START_TAG_TOKEN,
557 wakaba 1.76 tag_name => chr ($self->{next_char} + 0x0020)};
558 wakaba 1.59 $self->{state} = TAG_NAME_STATE;
559 wakaba 1.1
560     if (@{$self->{char}}) {
561 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
562 wakaba 1.1 } else {
563 wakaba 1.76 $self->{set_next_char}->($self);
564 wakaba 1.1 }
565    
566     redo A;
567 wakaba 1.76 } elsif (0x0061 <= $self->{next_char} and
568     $self->{next_char} <= 0x007A) { # a..z
569 wakaba 1.77
570 wakaba 1.57 $self->{current_token} = {type => START_TAG_TOKEN,
571 wakaba 1.76 tag_name => chr ($self->{next_char})};
572 wakaba 1.59 $self->{state} = TAG_NAME_STATE;
573 wakaba 1.1
574     if (@{$self->{char}}) {
575 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
576 wakaba 1.1 } else {
577 wakaba 1.76 $self->{set_next_char}->($self);
578 wakaba 1.1 }
579    
580     redo A;
581 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
582 wakaba 1.77
583 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'empty start tag');
584 wakaba 1.59 $self->{state} = DATA_STATE;
585 wakaba 1.1
586     if (@{$self->{char}}) {
587 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
588 wakaba 1.1 } else {
589 wakaba 1.76 $self->{set_next_char}->($self);
590 wakaba 1.1 }
591    
592    
593 wakaba 1.57 return ({type => CHARACTER_TOKEN, data => '<>'});
594 wakaba 1.1
595     redo A;
596 wakaba 1.76 } elsif ($self->{next_char} == 0x003F) { # ?
597 wakaba 1.77
598 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'pio');
599 wakaba 1.59 $self->{state} = BOGUS_COMMENT_STATE;
600 wakaba 1.76 ## $self->{next_char} is intentionally left as is
601 wakaba 1.1 redo A;
602     } else {
603 wakaba 1.77
604 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'bare stago');
605 wakaba 1.59 $self->{state} = DATA_STATE;
606 wakaba 1.1 ## reconsume
607    
608 wakaba 1.57 return ({type => CHARACTER_TOKEN, data => '<'});
609 wakaba 1.1
610     redo A;
611     }
612     } else {
613 wakaba 1.41 die "$0: $self->{content_model} in tag open";
614 wakaba 1.1 }
615 wakaba 1.59 } elsif ($self->{state} == CLOSE_TAG_OPEN_STATE) {
616 wakaba 1.41 if ($self->{content_model} & CM_LIMITED_MARKUP) { # RCDATA | CDATA
617 wakaba 1.23 if (defined $self->{last_emitted_start_tag_name}) {
618 wakaba 1.30 ## NOTE: <http://krijnhoetmer.nl/irc-logs/whatwg/20070626#l-564>
619 wakaba 1.23 my @next_char;
620     TAGNAME: for (my $i = 0; $i < length $self->{last_emitted_start_tag_name}; $i++) {
621 wakaba 1.76 push @next_char, $self->{next_char};
622 wakaba 1.23 my $c = ord substr ($self->{last_emitted_start_tag_name}, $i, 1);
623     my $C = 0x0061 <= $c && $c <= 0x007A ? $c - 0x0020 : $c;
624 wakaba 1.76 if ($self->{next_char} == $c or $self->{next_char} == $C) {
625 wakaba 1.23
626 wakaba 1.77
627 wakaba 1.1 if (@{$self->{char}}) {
628 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
629 wakaba 1.1 } else {
630 wakaba 1.76 $self->{set_next_char}->($self);
631 wakaba 1.1 }
632    
633 wakaba 1.23 next TAGNAME;
634     } else {
635 wakaba 1.77
636 wakaba 1.76 $self->{next_char} = shift @next_char; # reconsume
637 wakaba 1.23 unshift @{$self->{char}}, (@next_char);
638 wakaba 1.59 $self->{state} = DATA_STATE;
639 wakaba 1.23
640 wakaba 1.57 return ({type => CHARACTER_TOKEN, data => '</'});
641 wakaba 1.23
642     redo A;
643     }
644     }
645 wakaba 1.76 push @next_char, $self->{next_char};
646 wakaba 1.23
647 wakaba 1.76 unless ($self->{next_char} == 0x0009 or # HT
648     $self->{next_char} == 0x000A or # LF
649     $self->{next_char} == 0x000B or # VT
650     $self->{next_char} == 0x000C or # FF
651     $self->{next_char} == 0x0020 or # SP
652     $self->{next_char} == 0x003E or # >
653     $self->{next_char} == 0x002F or # /
654     $self->{next_char} == -1) {
655 wakaba 1.77
656 wakaba 1.76 $self->{next_char} = shift @next_char; # reconsume
657 wakaba 1.1 unshift @{$self->{char}}, (@next_char);
658 wakaba 1.59 $self->{state} = DATA_STATE;
659 wakaba 1.57 return ({type => CHARACTER_TOKEN, data => '</'});
660 wakaba 1.1 redo A;
661 wakaba 1.23 } else {
662 wakaba 1.77
663 wakaba 1.76 $self->{next_char} = shift @next_char;
664 wakaba 1.23 unshift @{$self->{char}}, (@next_char);
665     # and consume...
666 wakaba 1.1 }
667 wakaba 1.23 } else {
668     ## No start tag token has ever been emitted
669 wakaba 1.77
670 wakaba 1.23 # next-input-character is already done
671 wakaba 1.59 $self->{state} = DATA_STATE;
672 wakaba 1.57 return ({type => CHARACTER_TOKEN, data => '</'});
673 wakaba 1.1 redo A;
674     }
675     }
676    
677 wakaba 1.76 if (0x0041 <= $self->{next_char} and
678     $self->{next_char} <= 0x005A) { # A..Z
679 wakaba 1.77
680 wakaba 1.57 $self->{current_token} = {type => END_TAG_TOKEN,
681 wakaba 1.76 tag_name => chr ($self->{next_char} + 0x0020)};
682 wakaba 1.59 $self->{state} = TAG_NAME_STATE;
683 wakaba 1.1
684     if (@{$self->{char}}) {
685 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
686 wakaba 1.1 } else {
687 wakaba 1.76 $self->{set_next_char}->($self);
688 wakaba 1.1 }
689    
690     redo A;
691 wakaba 1.76 } elsif (0x0061 <= $self->{next_char} and
692     $self->{next_char} <= 0x007A) { # a..z
693 wakaba 1.77
694 wakaba 1.57 $self->{current_token} = {type => END_TAG_TOKEN,
695 wakaba 1.76 tag_name => chr ($self->{next_char})};
696 wakaba 1.59 $self->{state} = TAG_NAME_STATE;
697 wakaba 1.1
698     if (@{$self->{char}}) {
699 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
700 wakaba 1.1 } else {
701 wakaba 1.76 $self->{set_next_char}->($self);
702 wakaba 1.1 }
703    
704     redo A;
705 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
706 wakaba 1.77
707 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'empty end tag');
708 wakaba 1.59 $self->{state} = DATA_STATE;
709 wakaba 1.1
710     if (@{$self->{char}}) {
711 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
712 wakaba 1.1 } else {
713 wakaba 1.76 $self->{set_next_char}->($self);
714 wakaba 1.1 }
715    
716     redo A;
717 wakaba 1.76 } elsif ($self->{next_char} == -1) {
718 wakaba 1.77
719 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'bare etago');
720 wakaba 1.59 $self->{state} = DATA_STATE;
721 wakaba 1.1 # reconsume
722    
723 wakaba 1.57 return ({type => CHARACTER_TOKEN, data => '</'});
724 wakaba 1.1
725     redo A;
726     } else {
727 wakaba 1.77
728 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'bogus end tag');
729 wakaba 1.59 $self->{state} = BOGUS_COMMENT_STATE;
730 wakaba 1.76 ## $self->{next_char} is intentionally left as is
731 wakaba 1.1 redo A;
732     }
733 wakaba 1.59 } elsif ($self->{state} == TAG_NAME_STATE) {
734 wakaba 1.76 if ($self->{next_char} == 0x0009 or # HT
735     $self->{next_char} == 0x000A or # LF
736     $self->{next_char} == 0x000B or # VT
737     $self->{next_char} == 0x000C or # FF
738     $self->{next_char} == 0x0020) { # SP
739 wakaba 1.77
740 wakaba 1.59 $self->{state} = BEFORE_ATTRIBUTE_NAME_STATE;
741 wakaba 1.1
742     if (@{$self->{char}}) {
743 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
744 wakaba 1.1 } else {
745 wakaba 1.76 $self->{set_next_char}->($self);
746 wakaba 1.1 }
747    
748     redo A;
749 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
750 wakaba 1.57 if ($self->{current_token}->{type} == START_TAG_TOKEN) {
751 wakaba 1.77
752 wakaba 1.28 $self->{current_token}->{first_start_tag}
753     = not defined $self->{last_emitted_start_tag_name};
754 wakaba 1.1 $self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name};
755 wakaba 1.57 } elsif ($self->{current_token}->{type} == END_TAG_TOKEN) {
756 wakaba 1.41 $self->{content_model} = PCDATA_CONTENT_MODEL; # MUST
757 wakaba 1.78 #if ($self->{current_token}->{attributes}) {
758     # ## NOTE: This should never be reached.
759     # !!! cp (36);
760     # !!! parse-error (type => 'end tag attribute');
761     #} else {
762 wakaba 1.77
763 wakaba 1.78 #}
764 wakaba 1.1 } else {
765     die "$0: $self->{current_token}->{type}: Unknown token type";
766     }
767 wakaba 1.59 $self->{state} = DATA_STATE;
768 wakaba 1.1
769     if (@{$self->{char}}) {
770 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
771 wakaba 1.1 } else {
772 wakaba 1.76 $self->{set_next_char}->($self);
773 wakaba 1.1 }
774    
775    
776     return ($self->{current_token}); # start tag or end tag
777    
778     redo A;
779 wakaba 1.76 } elsif (0x0041 <= $self->{next_char} and
780     $self->{next_char} <= 0x005A) { # A..Z
781 wakaba 1.77
782 wakaba 1.76 $self->{current_token}->{tag_name} .= chr ($self->{next_char} + 0x0020);
783 wakaba 1.1 # start tag or end tag
784     ## Stay in this state
785    
786     if (@{$self->{char}}) {
787 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
788 wakaba 1.1 } else {
789 wakaba 1.76 $self->{set_next_char}->($self);
790 wakaba 1.1 }
791    
792     redo A;
793 wakaba 1.76 } elsif ($self->{next_char} == -1) {
794 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed tag');
795 wakaba 1.57 if ($self->{current_token}->{type} == START_TAG_TOKEN) {
796 wakaba 1.77
797 wakaba 1.28 $self->{current_token}->{first_start_tag}
798     = not defined $self->{last_emitted_start_tag_name};
799 wakaba 1.1 $self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name};
800 wakaba 1.57 } elsif ($self->{current_token}->{type} == END_TAG_TOKEN) {
801 wakaba 1.41 $self->{content_model} = PCDATA_CONTENT_MODEL; # MUST
802 wakaba 1.78 #if ($self->{current_token}->{attributes}) {
803     # ## NOTE: This state should never be reached.
804     # !!! cp (40);
805     # !!! parse-error (type => 'end tag attribute');
806     #} else {
807 wakaba 1.77
808 wakaba 1.78 #}
809 wakaba 1.1 } else {
810     die "$0: $self->{current_token}->{type}: Unknown token type";
811     }
812 wakaba 1.59 $self->{state} = DATA_STATE;
813 wakaba 1.1 # reconsume
814    
815     return ($self->{current_token}); # start tag or end tag
816    
817     redo A;
818 wakaba 1.76 } elsif ($self->{next_char} == 0x002F) { # /
819 wakaba 1.1
820     if (@{$self->{char}}) {
821 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
822 wakaba 1.1 } else {
823 wakaba 1.76 $self->{set_next_char}->($self);
824 wakaba 1.1 }
825    
826 wakaba 1.76 if ($self->{next_char} == 0x003E and # >
827 wakaba 1.57 $self->{current_token}->{type} == START_TAG_TOKEN and
828 wakaba 1.1 $permitted_slash_tag_name->{$self->{current_token}->{tag_name}}) {
829     # permitted slash
830 wakaba 1.77
831 wakaba 1.1 #
832     } else {
833 wakaba 1.77
834 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'nestc');
835 wakaba 1.1 }
836 wakaba 1.59 $self->{state} = BEFORE_ATTRIBUTE_NAME_STATE;
837 wakaba 1.1 # next-input-character is already done
838     redo A;
839     } else {
840 wakaba 1.77
841 wakaba 1.76 $self->{current_token}->{tag_name} .= chr $self->{next_char};
842 wakaba 1.1 # start tag or end tag
843     ## Stay in the state
844    
845     if (@{$self->{char}}) {
846 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
847 wakaba 1.1 } else {
848 wakaba 1.76 $self->{set_next_char}->($self);
849 wakaba 1.1 }
850    
851     redo A;
852     }
853 wakaba 1.59 } elsif ($self->{state} == BEFORE_ATTRIBUTE_NAME_STATE) {
854 wakaba 1.76 if ($self->{next_char} == 0x0009 or # HT
855     $self->{next_char} == 0x000A or # LF
856     $self->{next_char} == 0x000B or # VT
857     $self->{next_char} == 0x000C or # FF
858     $self->{next_char} == 0x0020) { # SP
859 wakaba 1.77
860 wakaba 1.1 ## Stay in the state
861    
862     if (@{$self->{char}}) {
863 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
864 wakaba 1.1 } else {
865 wakaba 1.76 $self->{set_next_char}->($self);
866 wakaba 1.1 }
867    
868     redo A;
869 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
870 wakaba 1.57 if ($self->{current_token}->{type} == START_TAG_TOKEN) {
871 wakaba 1.77
872 wakaba 1.28 $self->{current_token}->{first_start_tag}
873     = not defined $self->{last_emitted_start_tag_name};
874 wakaba 1.1 $self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name};
875 wakaba 1.57 } elsif ($self->{current_token}->{type} == END_TAG_TOKEN) {
876 wakaba 1.41 $self->{content_model} = PCDATA_CONTENT_MODEL; # MUST
877 wakaba 1.1 if ($self->{current_token}->{attributes}) {
878 wakaba 1.77
879 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'end tag attribute');
880 wakaba 1.77 } else {
881    
882 wakaba 1.1 }
883     } else {
884     die "$0: $self->{current_token}->{type}: Unknown token type";
885     }
886 wakaba 1.59 $self->{state} = DATA_STATE;
887 wakaba 1.1
888     if (@{$self->{char}}) {
889 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
890 wakaba 1.1 } else {
891 wakaba 1.76 $self->{set_next_char}->($self);
892 wakaba 1.1 }
893    
894    
895     return ($self->{current_token}); # start tag or end tag
896    
897     redo A;
898 wakaba 1.76 } elsif (0x0041 <= $self->{next_char} and
899     $self->{next_char} <= 0x005A) { # A..Z
900 wakaba 1.77
901 wakaba 1.76 $self->{current_attribute} = {name => chr ($self->{next_char} + 0x0020),
902 wakaba 1.1 value => ''};
903 wakaba 1.59 $self->{state} = ATTRIBUTE_NAME_STATE;
904 wakaba 1.1
905     if (@{$self->{char}}) {
906 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
907 wakaba 1.1 } else {
908 wakaba 1.76 $self->{set_next_char}->($self);
909 wakaba 1.1 }
910    
911     redo A;
912 wakaba 1.76 } elsif ($self->{next_char} == 0x002F) { # /
913 wakaba 1.1
914     if (@{$self->{char}}) {
915 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
916 wakaba 1.1 } else {
917 wakaba 1.76 $self->{set_next_char}->($self);
918 wakaba 1.1 }
919    
920 wakaba 1.76 if ($self->{next_char} == 0x003E and # >
921 wakaba 1.57 $self->{current_token}->{type} == START_TAG_TOKEN and
922 wakaba 1.1 $permitted_slash_tag_name->{$self->{current_token}->{tag_name}}) {
923     # permitted slash
924 wakaba 1.77
925 wakaba 1.1 #
926     } else {
927 wakaba 1.77
928 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'nestc');
929 wakaba 1.1 }
930     ## Stay in the state
931     # next-input-character is already done
932     redo A;
933 wakaba 1.76 } elsif ($self->{next_char} == -1) {
934 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed tag');
935 wakaba 1.57 if ($self->{current_token}->{type} == START_TAG_TOKEN) {
936 wakaba 1.77
937 wakaba 1.28 $self->{current_token}->{first_start_tag}
938     = not defined $self->{last_emitted_start_tag_name};
939 wakaba 1.1 $self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name};
940 wakaba 1.57 } elsif ($self->{current_token}->{type} == END_TAG_TOKEN) {
941 wakaba 1.41 $self->{content_model} = PCDATA_CONTENT_MODEL; # MUST
942 wakaba 1.1 if ($self->{current_token}->{attributes}) {
943 wakaba 1.77
944 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'end tag attribute');
945 wakaba 1.77 } else {
946    
947 wakaba 1.1 }
948     } else {
949     die "$0: $self->{current_token}->{type}: Unknown token type";
950     }
951 wakaba 1.59 $self->{state} = DATA_STATE;
952 wakaba 1.1 # reconsume
953    
954     return ($self->{current_token}); # start tag or end tag
955    
956     redo A;
957     } else {
958 wakaba 1.72 if ({
959     0x0022 => 1, # "
960     0x0027 => 1, # '
961     0x003D => 1, # =
962 wakaba 1.76 }->{$self->{next_char}}) {
963 wakaba 1.77
964 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'bad attribute name');
965 wakaba 1.77 } else {
966    
967 wakaba 1.72 }
968 wakaba 1.76 $self->{current_attribute} = {name => chr ($self->{next_char}),
969 wakaba 1.1 value => ''};
970 wakaba 1.59 $self->{state} = ATTRIBUTE_NAME_STATE;
971 wakaba 1.1
972     if (@{$self->{char}}) {
973 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
974 wakaba 1.1 } else {
975 wakaba 1.76 $self->{set_next_char}->($self);
976 wakaba 1.1 }
977    
978     redo A;
979     }
980 wakaba 1.59 } elsif ($self->{state} == ATTRIBUTE_NAME_STATE) {
981 wakaba 1.1 my $before_leave = sub {
982     if (exists $self->{current_token}->{attributes} # start tag or end tag
983     ->{$self->{current_attribute}->{name}}) { # MUST
984 wakaba 1.77
985 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'duplicate attribute:'.$self->{current_attribute}->{name});
986 wakaba 1.1 ## Discard $self->{current_attribute} # MUST
987     } else {
988 wakaba 1.77
989 wakaba 1.1 $self->{current_token}->{attributes}->{$self->{current_attribute}->{name}}
990     = $self->{current_attribute};
991     }
992     }; # $before_leave
993    
994 wakaba 1.76 if ($self->{next_char} == 0x0009 or # HT
995     $self->{next_char} == 0x000A or # LF
996     $self->{next_char} == 0x000B or # VT
997     $self->{next_char} == 0x000C or # FF
998     $self->{next_char} == 0x0020) { # SP
999 wakaba 1.77
1000 wakaba 1.1 $before_leave->();
1001 wakaba 1.59 $self->{state} = AFTER_ATTRIBUTE_NAME_STATE;
1002 wakaba 1.1
1003     if (@{$self->{char}}) {
1004 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1005 wakaba 1.1 } else {
1006 wakaba 1.76 $self->{set_next_char}->($self);
1007 wakaba 1.1 }
1008    
1009     redo A;
1010 wakaba 1.76 } elsif ($self->{next_char} == 0x003D) { # =
1011 wakaba 1.77
1012 wakaba 1.1 $before_leave->();
1013 wakaba 1.59 $self->{state} = BEFORE_ATTRIBUTE_VALUE_STATE;
1014 wakaba 1.1
1015     if (@{$self->{char}}) {
1016 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1017 wakaba 1.1 } else {
1018 wakaba 1.76 $self->{set_next_char}->($self);
1019 wakaba 1.1 }
1020    
1021     redo A;
1022 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
1023 wakaba 1.1 $before_leave->();
1024 wakaba 1.57 if ($self->{current_token}->{type} == START_TAG_TOKEN) {
1025 wakaba 1.77
1026 wakaba 1.28 $self->{current_token}->{first_start_tag}
1027     = not defined $self->{last_emitted_start_tag_name};
1028 wakaba 1.1 $self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name};
1029 wakaba 1.57 } elsif ($self->{current_token}->{type} == END_TAG_TOKEN) {
1030 wakaba 1.77
1031 wakaba 1.41 $self->{content_model} = PCDATA_CONTENT_MODEL; # MUST
1032 wakaba 1.1 if ($self->{current_token}->{attributes}) {
1033 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'end tag attribute');
1034 wakaba 1.1 }
1035     } else {
1036     die "$0: $self->{current_token}->{type}: Unknown token type";
1037     }
1038 wakaba 1.59 $self->{state} = DATA_STATE;
1039 wakaba 1.1
1040     if (@{$self->{char}}) {
1041 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1042 wakaba 1.1 } else {
1043 wakaba 1.76 $self->{set_next_char}->($self);
1044 wakaba 1.1 }
1045    
1046    
1047     return ($self->{current_token}); # start tag or end tag
1048    
1049     redo A;
1050 wakaba 1.76 } elsif (0x0041 <= $self->{next_char} and
1051     $self->{next_char} <= 0x005A) { # A..Z
1052 wakaba 1.77
1053 wakaba 1.76 $self->{current_attribute}->{name} .= chr ($self->{next_char} + 0x0020);
1054 wakaba 1.1 ## Stay in the state
1055    
1056     if (@{$self->{char}}) {
1057 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1058 wakaba 1.1 } else {
1059 wakaba 1.76 $self->{set_next_char}->($self);
1060 wakaba 1.1 }
1061    
1062     redo A;
1063 wakaba 1.76 } elsif ($self->{next_char} == 0x002F) { # /
1064 wakaba 1.1 $before_leave->();
1065    
1066     if (@{$self->{char}}) {
1067 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1068 wakaba 1.1 } else {
1069 wakaba 1.76 $self->{set_next_char}->($self);
1070 wakaba 1.1 }
1071    
1072 wakaba 1.76 if ($self->{next_char} == 0x003E and # >
1073 wakaba 1.57 $self->{current_token}->{type} == START_TAG_TOKEN and
1074 wakaba 1.1 $permitted_slash_tag_name->{$self->{current_token}->{tag_name}}) {
1075     # permitted slash
1076 wakaba 1.77
1077 wakaba 1.1 #
1078     } else {
1079 wakaba 1.77
1080 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'nestc');
1081 wakaba 1.1 }
1082 wakaba 1.59 $self->{state} = BEFORE_ATTRIBUTE_NAME_STATE;
1083 wakaba 1.1 # next-input-character is already done
1084     redo A;
1085 wakaba 1.76 } elsif ($self->{next_char} == -1) {
1086 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed tag');
1087 wakaba 1.1 $before_leave->();
1088 wakaba 1.57 if ($self->{current_token}->{type} == START_TAG_TOKEN) {
1089 wakaba 1.77
1090 wakaba 1.28 $self->{current_token}->{first_start_tag}
1091     = not defined $self->{last_emitted_start_tag_name};
1092 wakaba 1.1 $self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name};
1093 wakaba 1.57 } elsif ($self->{current_token}->{type} == END_TAG_TOKEN) {
1094 wakaba 1.41 $self->{content_model} = PCDATA_CONTENT_MODEL; # MUST
1095 wakaba 1.1 if ($self->{current_token}->{attributes}) {
1096 wakaba 1.77
1097 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'end tag attribute');
1098 wakaba 1.77 } else {
1099 wakaba 1.78 ## NOTE: This state should never be reached.
1100 wakaba 1.77
1101 wakaba 1.1 }
1102     } else {
1103     die "$0: $self->{current_token}->{type}: Unknown token type";
1104     }
1105 wakaba 1.59 $self->{state} = DATA_STATE;
1106 wakaba 1.1 # reconsume
1107    
1108     return ($self->{current_token}); # start tag or end tag
1109    
1110     redo A;
1111     } else {
1112 wakaba 1.76 if ($self->{next_char} == 0x0022 or # "
1113     $self->{next_char} == 0x0027) { # '
1114 wakaba 1.77
1115 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'bad attribute name');
1116 wakaba 1.77 } else {
1117    
1118 wakaba 1.72 }
1119 wakaba 1.76 $self->{current_attribute}->{name} .= chr ($self->{next_char});
1120 wakaba 1.1 ## Stay in the state
1121    
1122     if (@{$self->{char}}) {
1123 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1124 wakaba 1.1 } else {
1125 wakaba 1.76 $self->{set_next_char}->($self);
1126 wakaba 1.1 }
1127    
1128     redo A;
1129     }
1130 wakaba 1.59 } elsif ($self->{state} == AFTER_ATTRIBUTE_NAME_STATE) {
1131 wakaba 1.76 if ($self->{next_char} == 0x0009 or # HT
1132     $self->{next_char} == 0x000A or # LF
1133     $self->{next_char} == 0x000B or # VT
1134     $self->{next_char} == 0x000C or # FF
1135     $self->{next_char} == 0x0020) { # SP
1136 wakaba 1.77
1137 wakaba 1.1 ## Stay in the state
1138    
1139     if (@{$self->{char}}) {
1140 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1141 wakaba 1.1 } else {
1142 wakaba 1.76 $self->{set_next_char}->($self);
1143 wakaba 1.1 }
1144    
1145     redo A;
1146 wakaba 1.76 } elsif ($self->{next_char} == 0x003D) { # =
1147 wakaba 1.77
1148 wakaba 1.59 $self->{state} = BEFORE_ATTRIBUTE_VALUE_STATE;
1149 wakaba 1.1
1150     if (@{$self->{char}}) {
1151 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1152 wakaba 1.1 } else {
1153 wakaba 1.76 $self->{set_next_char}->($self);
1154 wakaba 1.1 }
1155    
1156     redo A;
1157 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
1158 wakaba 1.57 if ($self->{current_token}->{type} == START_TAG_TOKEN) {
1159 wakaba 1.77
1160 wakaba 1.28 $self->{current_token}->{first_start_tag}
1161     = not defined $self->{last_emitted_start_tag_name};
1162 wakaba 1.1 $self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name};
1163 wakaba 1.57 } elsif ($self->{current_token}->{type} == END_TAG_TOKEN) {
1164 wakaba 1.41 $self->{content_model} = PCDATA_CONTENT_MODEL; # MUST
1165 wakaba 1.1 if ($self->{current_token}->{attributes}) {
1166 wakaba 1.77
1167 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'end tag attribute');
1168 wakaba 1.77 } else {
1169 wakaba 1.78 ## NOTE: This state should never be reached.
1170 wakaba 1.77
1171 wakaba 1.1 }
1172     } else {
1173     die "$0: $self->{current_token}->{type}: Unknown token type";
1174     }
1175 wakaba 1.59 $self->{state} = DATA_STATE;
1176 wakaba 1.1
1177     if (@{$self->{char}}) {
1178 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1179 wakaba 1.1 } else {
1180 wakaba 1.76 $self->{set_next_char}->($self);
1181 wakaba 1.1 }
1182    
1183    
1184     return ($self->{current_token}); # start tag or end tag
1185    
1186     redo A;
1187 wakaba 1.76 } elsif (0x0041 <= $self->{next_char} and
1188     $self->{next_char} <= 0x005A) { # A..Z
1189 wakaba 1.77
1190 wakaba 1.76 $self->{current_attribute} = {name => chr ($self->{next_char} + 0x0020),
1191 wakaba 1.1 value => ''};
1192 wakaba 1.59 $self->{state} = ATTRIBUTE_NAME_STATE;
1193 wakaba 1.1
1194     if (@{$self->{char}}) {
1195 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1196 wakaba 1.1 } else {
1197 wakaba 1.76 $self->{set_next_char}->($self);
1198 wakaba 1.1 }
1199    
1200     redo A;
1201 wakaba 1.76 } elsif ($self->{next_char} == 0x002F) { # /
1202 wakaba 1.1
1203     if (@{$self->{char}}) {
1204 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1205 wakaba 1.1 } else {
1206 wakaba 1.76 $self->{set_next_char}->($self);
1207 wakaba 1.1 }
1208    
1209 wakaba 1.76 if ($self->{next_char} == 0x003E and # >
1210 wakaba 1.57 $self->{current_token}->{type} == START_TAG_TOKEN and
1211 wakaba 1.1 $permitted_slash_tag_name->{$self->{current_token}->{tag_name}}) {
1212     # permitted slash
1213 wakaba 1.77
1214 wakaba 1.1 #
1215     } else {
1216 wakaba 1.77
1217 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'nestc');
1218 wakaba 1.33 ## TODO: Different error type for <aa / bb> than <aa/>
1219 wakaba 1.1 }
1220 wakaba 1.59 $self->{state} = BEFORE_ATTRIBUTE_NAME_STATE;
1221 wakaba 1.1 # next-input-character is already done
1222     redo A;
1223 wakaba 1.76 } elsif ($self->{next_char} == -1) {
1224 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed tag');
1225 wakaba 1.57 if ($self->{current_token}->{type} == START_TAG_TOKEN) {
1226 wakaba 1.77
1227 wakaba 1.28 $self->{current_token}->{first_start_tag}
1228     = not defined $self->{last_emitted_start_tag_name};
1229 wakaba 1.1 $self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name};
1230 wakaba 1.57 } elsif ($self->{current_token}->{type} == END_TAG_TOKEN) {
1231 wakaba 1.41 $self->{content_model} = PCDATA_CONTENT_MODEL; # MUST
1232 wakaba 1.1 if ($self->{current_token}->{attributes}) {
1233 wakaba 1.77
1234 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'end tag attribute');
1235 wakaba 1.77 } else {
1236 wakaba 1.78 ## NOTE: This state should never be reached.
1237 wakaba 1.77
1238 wakaba 1.1 }
1239     } else {
1240     die "$0: $self->{current_token}->{type}: Unknown token type";
1241     }
1242 wakaba 1.59 $self->{state} = DATA_STATE;
1243 wakaba 1.1 # reconsume
1244    
1245     return ($self->{current_token}); # start tag or end tag
1246    
1247     redo A;
1248     } else {
1249 wakaba 1.77
1250 wakaba 1.76 $self->{current_attribute} = {name => chr ($self->{next_char}),
1251 wakaba 1.1 value => ''};
1252 wakaba 1.59 $self->{state} = ATTRIBUTE_NAME_STATE;
1253 wakaba 1.1
1254     if (@{$self->{char}}) {
1255 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1256 wakaba 1.1 } else {
1257 wakaba 1.76 $self->{set_next_char}->($self);
1258 wakaba 1.1 }
1259    
1260     redo A;
1261     }
1262 wakaba 1.59 } elsif ($self->{state} == BEFORE_ATTRIBUTE_VALUE_STATE) {
1263 wakaba 1.76 if ($self->{next_char} == 0x0009 or # HT
1264     $self->{next_char} == 0x000A or # LF
1265     $self->{next_char} == 0x000B or # VT
1266     $self->{next_char} == 0x000C or # FF
1267     $self->{next_char} == 0x0020) { # SP
1268 wakaba 1.77
1269 wakaba 1.1 ## Stay in the state
1270    
1271     if (@{$self->{char}}) {
1272 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1273 wakaba 1.1 } else {
1274 wakaba 1.76 $self->{set_next_char}->($self);
1275 wakaba 1.1 }
1276    
1277     redo A;
1278 wakaba 1.76 } elsif ($self->{next_char} == 0x0022) { # "
1279 wakaba 1.77
1280 wakaba 1.59 $self->{state} = ATTRIBUTE_VALUE_DOUBLE_QUOTED_STATE;
1281 wakaba 1.1
1282     if (@{$self->{char}}) {
1283 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1284 wakaba 1.1 } else {
1285 wakaba 1.76 $self->{set_next_char}->($self);
1286 wakaba 1.1 }
1287    
1288     redo A;
1289 wakaba 1.76 } elsif ($self->{next_char} == 0x0026) { # &
1290 wakaba 1.77
1291 wakaba 1.59 $self->{state} = ATTRIBUTE_VALUE_UNQUOTED_STATE;
1292 wakaba 1.1 ## reconsume
1293     redo A;
1294 wakaba 1.76 } elsif ($self->{next_char} == 0x0027) { # '
1295 wakaba 1.77
1296 wakaba 1.59 $self->{state} = ATTRIBUTE_VALUE_SINGLE_QUOTED_STATE;
1297 wakaba 1.1
1298     if (@{$self->{char}}) {
1299 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1300 wakaba 1.1 } else {
1301 wakaba 1.76 $self->{set_next_char}->($self);
1302 wakaba 1.1 }
1303    
1304     redo A;
1305 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
1306 wakaba 1.57 if ($self->{current_token}->{type} == START_TAG_TOKEN) {
1307 wakaba 1.77
1308 wakaba 1.28 $self->{current_token}->{first_start_tag}
1309     = not defined $self->{last_emitted_start_tag_name};
1310 wakaba 1.1 $self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name};
1311 wakaba 1.57 } elsif ($self->{current_token}->{type} == END_TAG_TOKEN) {
1312 wakaba 1.41 $self->{content_model} = PCDATA_CONTENT_MODEL; # MUST
1313 wakaba 1.1 if ($self->{current_token}->{attributes}) {
1314 wakaba 1.77
1315 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'end tag attribute');
1316 wakaba 1.77 } else {
1317 wakaba 1.78 ## NOTE: This state should never be reached.
1318 wakaba 1.77
1319 wakaba 1.1 }
1320     } else {
1321     die "$0: $self->{current_token}->{type}: Unknown token type";
1322     }
1323 wakaba 1.59 $self->{state} = DATA_STATE;
1324 wakaba 1.1
1325     if (@{$self->{char}}) {
1326 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1327 wakaba 1.1 } else {
1328 wakaba 1.76 $self->{set_next_char}->($self);
1329 wakaba 1.1 }
1330    
1331    
1332     return ($self->{current_token}); # start tag or end tag
1333    
1334     redo A;
1335 wakaba 1.76 } elsif ($self->{next_char} == -1) {
1336 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed tag');
1337 wakaba 1.57 if ($self->{current_token}->{type} == START_TAG_TOKEN) {
1338 wakaba 1.77
1339 wakaba 1.28 $self->{current_token}->{first_start_tag}
1340     = not defined $self->{last_emitted_start_tag_name};
1341 wakaba 1.1 $self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name};
1342 wakaba 1.57 } elsif ($self->{current_token}->{type} == END_TAG_TOKEN) {
1343 wakaba 1.41 $self->{content_model} = PCDATA_CONTENT_MODEL; # MUST
1344 wakaba 1.1 if ($self->{current_token}->{attributes}) {
1345 wakaba 1.77
1346 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'end tag attribute');
1347 wakaba 1.77 } else {
1348 wakaba 1.78 ## NOTE: This state should never be reached.
1349 wakaba 1.77
1350 wakaba 1.1 }
1351     } else {
1352     die "$0: $self->{current_token}->{type}: Unknown token type";
1353     }
1354 wakaba 1.59 $self->{state} = DATA_STATE;
1355 wakaba 1.1 ## reconsume
1356    
1357     return ($self->{current_token}); # start tag or end tag
1358    
1359     redo A;
1360     } else {
1361 wakaba 1.76 if ($self->{next_char} == 0x003D) { # =
1362 wakaba 1.77
1363 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'bad attribute value');
1364 wakaba 1.77 } else {
1365    
1366 wakaba 1.72 }
1367 wakaba 1.76 $self->{current_attribute}->{value} .= chr ($self->{next_char});
1368 wakaba 1.59 $self->{state} = ATTRIBUTE_VALUE_UNQUOTED_STATE;
1369 wakaba 1.1
1370     if (@{$self->{char}}) {
1371 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1372 wakaba 1.1 } else {
1373 wakaba 1.76 $self->{set_next_char}->($self);
1374 wakaba 1.1 }
1375    
1376     redo A;
1377     }
1378 wakaba 1.59 } elsif ($self->{state} == ATTRIBUTE_VALUE_DOUBLE_QUOTED_STATE) {
1379 wakaba 1.76 if ($self->{next_char} == 0x0022) { # "
1380 wakaba 1.77
1381 wakaba 1.72 $self->{state} = AFTER_ATTRIBUTE_VALUE_QUOTED_STATE;
1382 wakaba 1.1
1383     if (@{$self->{char}}) {
1384 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1385 wakaba 1.1 } else {
1386 wakaba 1.76 $self->{set_next_char}->($self);
1387 wakaba 1.1 }
1388    
1389     redo A;
1390 wakaba 1.76 } elsif ($self->{next_char} == 0x0026) { # &
1391 wakaba 1.77
1392 wakaba 1.59 $self->{last_attribute_value_state} = $self->{state};
1393     $self->{state} = ENTITY_IN_ATTRIBUTE_VALUE_STATE;
1394 wakaba 1.1
1395     if (@{$self->{char}}) {
1396 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1397 wakaba 1.1 } else {
1398 wakaba 1.76 $self->{set_next_char}->($self);
1399 wakaba 1.1 }
1400    
1401     redo A;
1402 wakaba 1.76 } elsif ($self->{next_char} == -1) {
1403 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed attribute value');
1404 wakaba 1.57 if ($self->{current_token}->{type} == START_TAG_TOKEN) {
1405 wakaba 1.77
1406 wakaba 1.28 $self->{current_token}->{first_start_tag}
1407     = not defined $self->{last_emitted_start_tag_name};
1408 wakaba 1.1 $self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name};
1409 wakaba 1.57 } elsif ($self->{current_token}->{type} == END_TAG_TOKEN) {
1410 wakaba 1.41 $self->{content_model} = PCDATA_CONTENT_MODEL; # MUST
1411 wakaba 1.1 if ($self->{current_token}->{attributes}) {
1412 wakaba 1.77
1413 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'end tag attribute');
1414 wakaba 1.77 } else {
1415 wakaba 1.78 ## NOTE: This state should never be reached.
1416 wakaba 1.77
1417 wakaba 1.1 }
1418     } else {
1419     die "$0: $self->{current_token}->{type}: Unknown token type";
1420     }
1421 wakaba 1.59 $self->{state} = DATA_STATE;
1422 wakaba 1.1 ## reconsume
1423    
1424     return ($self->{current_token}); # start tag or end tag
1425    
1426     redo A;
1427     } else {
1428 wakaba 1.77
1429 wakaba 1.76 $self->{current_attribute}->{value} .= chr ($self->{next_char});
1430 wakaba 1.1 ## Stay in the state
1431    
1432     if (@{$self->{char}}) {
1433 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1434 wakaba 1.1 } else {
1435 wakaba 1.76 $self->{set_next_char}->($self);
1436 wakaba 1.1 }
1437    
1438     redo A;
1439     }
1440 wakaba 1.59 } elsif ($self->{state} == ATTRIBUTE_VALUE_SINGLE_QUOTED_STATE) {
1441 wakaba 1.76 if ($self->{next_char} == 0x0027) { # '
1442 wakaba 1.77
1443 wakaba 1.72 $self->{state} = AFTER_ATTRIBUTE_VALUE_QUOTED_STATE;
1444 wakaba 1.1
1445     if (@{$self->{char}}) {
1446 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1447 wakaba 1.1 } else {
1448 wakaba 1.76 $self->{set_next_char}->($self);
1449 wakaba 1.1 }
1450    
1451     redo A;
1452 wakaba 1.76 } elsif ($self->{next_char} == 0x0026) { # &
1453 wakaba 1.77
1454 wakaba 1.59 $self->{last_attribute_value_state} = $self->{state};
1455     $self->{state} = ENTITY_IN_ATTRIBUTE_VALUE_STATE;
1456 wakaba 1.1
1457     if (@{$self->{char}}) {
1458 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1459 wakaba 1.1 } else {
1460 wakaba 1.76 $self->{set_next_char}->($self);
1461 wakaba 1.1 }
1462    
1463     redo A;
1464 wakaba 1.76 } elsif ($self->{next_char} == -1) {
1465 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed attribute value');
1466 wakaba 1.57 if ($self->{current_token}->{type} == START_TAG_TOKEN) {
1467 wakaba 1.77
1468 wakaba 1.28 $self->{current_token}->{first_start_tag}
1469     = not defined $self->{last_emitted_start_tag_name};
1470 wakaba 1.1 $self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name};
1471 wakaba 1.57 } elsif ($self->{current_token}->{type} == END_TAG_TOKEN) {
1472 wakaba 1.41 $self->{content_model} = PCDATA_CONTENT_MODEL; # MUST
1473 wakaba 1.1 if ($self->{current_token}->{attributes}) {
1474 wakaba 1.77
1475 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'end tag attribute');
1476 wakaba 1.77 } else {
1477 wakaba 1.78 ## NOTE: This state should never be reached.
1478 wakaba 1.77
1479 wakaba 1.1 }
1480     } else {
1481     die "$0: $self->{current_token}->{type}: Unknown token type";
1482     }
1483 wakaba 1.59 $self->{state} = DATA_STATE;
1484 wakaba 1.1 ## reconsume
1485    
1486     return ($self->{current_token}); # start tag or end tag
1487    
1488     redo A;
1489     } else {
1490 wakaba 1.77
1491 wakaba 1.76 $self->{current_attribute}->{value} .= chr ($self->{next_char});
1492 wakaba 1.1 ## Stay in the state
1493    
1494     if (@{$self->{char}}) {
1495 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1496 wakaba 1.1 } else {
1497 wakaba 1.76 $self->{set_next_char}->($self);
1498 wakaba 1.1 }
1499    
1500     redo A;
1501     }
1502 wakaba 1.59 } elsif ($self->{state} == ATTRIBUTE_VALUE_UNQUOTED_STATE) {
1503 wakaba 1.76 if ($self->{next_char} == 0x0009 or # HT
1504     $self->{next_char} == 0x000A or # LF
1505     $self->{next_char} == 0x000B or # HT
1506     $self->{next_char} == 0x000C or # FF
1507     $self->{next_char} == 0x0020) { # SP
1508 wakaba 1.77
1509 wakaba 1.59 $self->{state} = BEFORE_ATTRIBUTE_NAME_STATE;
1510 wakaba 1.1
1511     if (@{$self->{char}}) {
1512 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1513 wakaba 1.1 } else {
1514 wakaba 1.76 $self->{set_next_char}->($self);
1515 wakaba 1.1 }
1516    
1517     redo A;
1518 wakaba 1.76 } elsif ($self->{next_char} == 0x0026) { # &
1519 wakaba 1.77
1520 wakaba 1.59 $self->{last_attribute_value_state} = $self->{state};
1521     $self->{state} = ENTITY_IN_ATTRIBUTE_VALUE_STATE;
1522 wakaba 1.1
1523     if (@{$self->{char}}) {
1524 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1525 wakaba 1.1 } else {
1526 wakaba 1.76 $self->{set_next_char}->($self);
1527 wakaba 1.1 }
1528    
1529     redo A;
1530 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
1531 wakaba 1.57 if ($self->{current_token}->{type} == START_TAG_TOKEN) {
1532 wakaba 1.77
1533 wakaba 1.28 $self->{current_token}->{first_start_tag}
1534     = not defined $self->{last_emitted_start_tag_name};
1535 wakaba 1.1 $self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name};
1536 wakaba 1.57 } elsif ($self->{current_token}->{type} == END_TAG_TOKEN) {
1537 wakaba 1.41 $self->{content_model} = PCDATA_CONTENT_MODEL; # MUST
1538 wakaba 1.1 if ($self->{current_token}->{attributes}) {
1539 wakaba 1.77
1540 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'end tag attribute');
1541 wakaba 1.77 } else {
1542 wakaba 1.78 ## NOTE: This state should never be reached.
1543 wakaba 1.77
1544 wakaba 1.1 }
1545     } else {
1546     die "$0: $self->{current_token}->{type}: Unknown token type";
1547     }
1548 wakaba 1.59 $self->{state} = DATA_STATE;
1549 wakaba 1.1
1550     if (@{$self->{char}}) {
1551 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1552 wakaba 1.1 } else {
1553 wakaba 1.76 $self->{set_next_char}->($self);
1554 wakaba 1.1 }
1555    
1556    
1557     return ($self->{current_token}); # start tag or end tag
1558    
1559     redo A;
1560 wakaba 1.76 } elsif ($self->{next_char} == -1) {
1561 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed tag');
1562 wakaba 1.57 if ($self->{current_token}->{type} == START_TAG_TOKEN) {
1563 wakaba 1.77
1564 wakaba 1.28 $self->{current_token}->{first_start_tag}
1565     = not defined $self->{last_emitted_start_tag_name};
1566 wakaba 1.1 $self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name};
1567 wakaba 1.57 } elsif ($self->{current_token}->{type} == END_TAG_TOKEN) {
1568 wakaba 1.41 $self->{content_model} = PCDATA_CONTENT_MODEL; # MUST
1569 wakaba 1.1 if ($self->{current_token}->{attributes}) {
1570 wakaba 1.77
1571 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'end tag attribute');
1572 wakaba 1.77 } else {
1573 wakaba 1.78 ## NOTE: This state should never be reached.
1574 wakaba 1.77
1575 wakaba 1.1 }
1576     } else {
1577     die "$0: $self->{current_token}->{type}: Unknown token type";
1578     }
1579 wakaba 1.59 $self->{state} = DATA_STATE;
1580 wakaba 1.1 ## reconsume
1581    
1582     return ($self->{current_token}); # start tag or end tag
1583    
1584     redo A;
1585     } else {
1586 wakaba 1.72 if ({
1587     0x0022 => 1, # "
1588     0x0027 => 1, # '
1589     0x003D => 1, # =
1590 wakaba 1.76 }->{$self->{next_char}}) {
1591 wakaba 1.77
1592 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'bad attribute value');
1593 wakaba 1.77 } else {
1594    
1595 wakaba 1.72 }
1596 wakaba 1.76 $self->{current_attribute}->{value} .= chr ($self->{next_char});
1597 wakaba 1.1 ## Stay in the state
1598    
1599     if (@{$self->{char}}) {
1600 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1601 wakaba 1.1 } else {
1602 wakaba 1.76 $self->{set_next_char}->($self);
1603 wakaba 1.1 }
1604    
1605     redo A;
1606     }
1607 wakaba 1.59 } elsif ($self->{state} == ENTITY_IN_ATTRIBUTE_VALUE_STATE) {
1608 wakaba 1.72 my $token = $self->_tokenize_attempt_to_consume_an_entity
1609     (1,
1610     $self->{last_attribute_value_state}
1611     == ATTRIBUTE_VALUE_DOUBLE_QUOTED_STATE ? 0x0022 : # "
1612     $self->{last_attribute_value_state}
1613     == ATTRIBUTE_VALUE_SINGLE_QUOTED_STATE ? 0x0027 : # '
1614     -1);
1615 wakaba 1.1
1616     unless (defined $token) {
1617 wakaba 1.77
1618 wakaba 1.1 $self->{current_attribute}->{value} .= '&';
1619     } else {
1620 wakaba 1.77
1621 wakaba 1.1 $self->{current_attribute}->{value} .= $token->{data};
1622 wakaba 1.68 $self->{current_attribute}->{has_reference} = $token->{has_reference};
1623 wakaba 1.1 ## ISSUE: spec says "append the returned character token to the current attribute's value"
1624     }
1625    
1626     $self->{state} = $self->{last_attribute_value_state};
1627     # next-input-character is already done
1628     redo A;
1629 wakaba 1.72 } elsif ($self->{state} == AFTER_ATTRIBUTE_VALUE_QUOTED_STATE) {
1630 wakaba 1.76 if ($self->{next_char} == 0x0009 or # HT
1631     $self->{next_char} == 0x000A or # LF
1632     $self->{next_char} == 0x000B or # VT
1633     $self->{next_char} == 0x000C or # FF
1634     $self->{next_char} == 0x0020) { # SP
1635 wakaba 1.77
1636 wakaba 1.72 $self->{state} = BEFORE_ATTRIBUTE_NAME_STATE;
1637    
1638     if (@{$self->{char}}) {
1639 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1640 wakaba 1.72 } else {
1641 wakaba 1.76 $self->{set_next_char}->($self);
1642 wakaba 1.72 }
1643    
1644     redo A;
1645 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
1646 wakaba 1.72 if ($self->{current_token}->{type} == START_TAG_TOKEN) {
1647 wakaba 1.77
1648 wakaba 1.72 $self->{current_token}->{first_start_tag}
1649     = not defined $self->{last_emitted_start_tag_name};
1650     $self->{last_emitted_start_tag_name} = $self->{current_token}->{tag_name};
1651     } elsif ($self->{current_token}->{type} == END_TAG_TOKEN) {
1652     $self->{content_model} = PCDATA_CONTENT_MODEL; # MUST
1653     if ($self->{current_token}->{attributes}) {
1654 wakaba 1.77
1655 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'end tag attribute');
1656 wakaba 1.77 } else {
1657 wakaba 1.78 ## NOTE: This state should never be reached.
1658 wakaba 1.77
1659 wakaba 1.72 }
1660     } else {
1661     die "$0: $self->{current_token}->{type}: Unknown token type";
1662     }
1663     $self->{state} = DATA_STATE;
1664    
1665     if (@{$self->{char}}) {
1666 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1667 wakaba 1.72 } else {
1668 wakaba 1.76 $self->{set_next_char}->($self);
1669 wakaba 1.72 }
1670    
1671    
1672     return ($self->{current_token}); # start tag or end tag
1673    
1674     redo A;
1675 wakaba 1.76 } elsif ($self->{next_char} == 0x002F) { # /
1676 wakaba 1.72
1677     if (@{$self->{char}}) {
1678 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1679 wakaba 1.72 } else {
1680 wakaba 1.76 $self->{set_next_char}->($self);
1681 wakaba 1.72 }
1682    
1683 wakaba 1.76 if ($self->{next_char} == 0x003E and # >
1684 wakaba 1.72 $self->{current_token}->{type} == START_TAG_TOKEN and
1685     $permitted_slash_tag_name->{$self->{current_token}->{tag_name}}) {
1686     # permitted slash
1687 wakaba 1.77
1688 wakaba 1.72 #
1689     } else {
1690 wakaba 1.77
1691 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'nestc');
1692 wakaba 1.72 }
1693     $self->{state} = BEFORE_ATTRIBUTE_NAME_STATE;
1694     # next-input-character is already done
1695     redo A;
1696     } else {
1697 wakaba 1.77
1698 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'no space between attributes');
1699 wakaba 1.72 $self->{state} = BEFORE_ATTRIBUTE_NAME_STATE;
1700     ## reconsume
1701     redo A;
1702     }
1703 wakaba 1.59 } elsif ($self->{state} == BOGUS_COMMENT_STATE) {
1704 wakaba 1.1 ## (only happen if PCDATA state)
1705    
1706 wakaba 1.57 my $token = {type => COMMENT_TOKEN, data => ''};
1707 wakaba 1.1
1708     BC: {
1709 wakaba 1.76 if ($self->{next_char} == 0x003E) { # >
1710 wakaba 1.77
1711 wakaba 1.59 $self->{state} = DATA_STATE;
1712 wakaba 1.1
1713     if (@{$self->{char}}) {
1714 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1715 wakaba 1.1 } else {
1716 wakaba 1.76 $self->{set_next_char}->($self);
1717 wakaba 1.1 }
1718    
1719    
1720     return ($token);
1721    
1722     redo A;
1723 wakaba 1.76 } elsif ($self->{next_char} == -1) {
1724 wakaba 1.77
1725 wakaba 1.59 $self->{state} = DATA_STATE;
1726 wakaba 1.1 ## reconsume
1727    
1728     return ($token);
1729    
1730     redo A;
1731     } else {
1732 wakaba 1.77
1733 wakaba 1.76 $token->{data} .= chr ($self->{next_char});
1734 wakaba 1.1
1735     if (@{$self->{char}}) {
1736 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1737 wakaba 1.1 } else {
1738 wakaba 1.76 $self->{set_next_char}->($self);
1739 wakaba 1.1 }
1740    
1741     redo BC;
1742     }
1743     } # BC
1744 wakaba 1.77
1745     die "$0: _get_next_token: unexpected case [BC]";
1746 wakaba 1.59 } elsif ($self->{state} == MARKUP_DECLARATION_OPEN_STATE) {
1747 wakaba 1.1 ## (only happen if PCDATA state)
1748    
1749     my @next_char;
1750 wakaba 1.76 push @next_char, $self->{next_char};
1751 wakaba 1.1
1752 wakaba 1.76 if ($self->{next_char} == 0x002D) { # -
1753 wakaba 1.1
1754     if (@{$self->{char}}) {
1755 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1756 wakaba 1.1 } else {
1757 wakaba 1.76 $self->{set_next_char}->($self);
1758 wakaba 1.1 }
1759    
1760 wakaba 1.76 push @next_char, $self->{next_char};
1761     if ($self->{next_char} == 0x002D) { # -
1762 wakaba 1.77
1763 wakaba 1.57 $self->{current_token} = {type => COMMENT_TOKEN, data => ''};
1764 wakaba 1.59 $self->{state} = COMMENT_START_STATE;
1765 wakaba 1.1
1766     if (@{$self->{char}}) {
1767 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1768 wakaba 1.1 } else {
1769 wakaba 1.76 $self->{set_next_char}->($self);
1770 wakaba 1.1 }
1771    
1772     redo A;
1773 wakaba 1.77 } else {
1774    
1775 wakaba 1.1 }
1776 wakaba 1.76 } elsif ($self->{next_char} == 0x0044 or # D
1777     $self->{next_char} == 0x0064) { # d
1778 wakaba 1.1
1779     if (@{$self->{char}}) {
1780 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1781 wakaba 1.1 } else {
1782 wakaba 1.76 $self->{set_next_char}->($self);
1783 wakaba 1.1 }
1784    
1785 wakaba 1.76 push @next_char, $self->{next_char};
1786     if ($self->{next_char} == 0x004F or # O
1787     $self->{next_char} == 0x006F) { # o
1788 wakaba 1.1
1789     if (@{$self->{char}}) {
1790 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1791 wakaba 1.1 } else {
1792 wakaba 1.76 $self->{set_next_char}->($self);
1793 wakaba 1.1 }
1794    
1795 wakaba 1.76 push @next_char, $self->{next_char};
1796     if ($self->{next_char} == 0x0043 or # C
1797     $self->{next_char} == 0x0063) { # c
1798 wakaba 1.1
1799     if (@{$self->{char}}) {
1800 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1801 wakaba 1.1 } else {
1802 wakaba 1.76 $self->{set_next_char}->($self);
1803 wakaba 1.1 }
1804    
1805 wakaba 1.76 push @next_char, $self->{next_char};
1806     if ($self->{next_char} == 0x0054 or # T
1807     $self->{next_char} == 0x0074) { # t
1808 wakaba 1.1
1809     if (@{$self->{char}}) {
1810 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1811 wakaba 1.1 } else {
1812 wakaba 1.76 $self->{set_next_char}->($self);
1813 wakaba 1.1 }
1814    
1815 wakaba 1.76 push @next_char, $self->{next_char};
1816     if ($self->{next_char} == 0x0059 or # Y
1817     $self->{next_char} == 0x0079) { # y
1818 wakaba 1.1
1819     if (@{$self->{char}}) {
1820 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1821 wakaba 1.1 } else {
1822 wakaba 1.76 $self->{set_next_char}->($self);
1823 wakaba 1.1 }
1824    
1825 wakaba 1.76 push @next_char, $self->{next_char};
1826     if ($self->{next_char} == 0x0050 or # P
1827     $self->{next_char} == 0x0070) { # p
1828 wakaba 1.1
1829     if (@{$self->{char}}) {
1830 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1831 wakaba 1.1 } else {
1832 wakaba 1.76 $self->{set_next_char}->($self);
1833 wakaba 1.1 }
1834    
1835 wakaba 1.76 push @next_char, $self->{next_char};
1836     if ($self->{next_char} == 0x0045 or # E
1837     $self->{next_char} == 0x0065) { # e
1838 wakaba 1.77
1839     ## TODO: What a stupid code this is!
1840 wakaba 1.59 $self->{state} = DOCTYPE_STATE;
1841 wakaba 1.1
1842     if (@{$self->{char}}) {
1843 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1844 wakaba 1.1 } else {
1845 wakaba 1.76 $self->{set_next_char}->($self);
1846 wakaba 1.1 }
1847    
1848     redo A;
1849 wakaba 1.77 } else {
1850    
1851 wakaba 1.1 }
1852 wakaba 1.77 } else {
1853    
1854 wakaba 1.1 }
1855 wakaba 1.77 } else {
1856    
1857 wakaba 1.1 }
1858 wakaba 1.77 } else {
1859    
1860 wakaba 1.1 }
1861 wakaba 1.77 } else {
1862    
1863 wakaba 1.1 }
1864 wakaba 1.77 } else {
1865    
1866 wakaba 1.1 }
1867 wakaba 1.77 } else {
1868    
1869 wakaba 1.1 }
1870    
1871 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'bogus comment');
1872 wakaba 1.76 $self->{next_char} = shift @next_char;
1873 wakaba 1.1 unshift @{$self->{char}}, (@next_char);
1874 wakaba 1.59 $self->{state} = BOGUS_COMMENT_STATE;
1875 wakaba 1.1 redo A;
1876    
1877     ## ISSUE: typos in spec: chacacters, is is a parse error
1878     ## ISSUE: spec is somewhat unclear on "is the first character that will be in the comment"; what is "that will be in the comment" is what the algorithm defines, isn't it?
1879 wakaba 1.59 } elsif ($self->{state} == COMMENT_START_STATE) {
1880 wakaba 1.76 if ($self->{next_char} == 0x002D) { # -
1881 wakaba 1.77
1882 wakaba 1.59 $self->{state} = COMMENT_START_DASH_STATE;
1883 wakaba 1.23
1884     if (@{$self->{char}}) {
1885 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1886 wakaba 1.23 } else {
1887 wakaba 1.76 $self->{set_next_char}->($self);
1888 wakaba 1.23 }
1889    
1890     redo A;
1891 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
1892 wakaba 1.77
1893 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'bogus comment');
1894 wakaba 1.59 $self->{state} = DATA_STATE;
1895 wakaba 1.23
1896     if (@{$self->{char}}) {
1897 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1898 wakaba 1.23 } else {
1899 wakaba 1.76 $self->{set_next_char}->($self);
1900 wakaba 1.23 }
1901    
1902    
1903     return ($self->{current_token}); # comment
1904    
1905     redo A;
1906 wakaba 1.76 } elsif ($self->{next_char} == -1) {
1907 wakaba 1.77
1908 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed comment');
1909 wakaba 1.59 $self->{state} = DATA_STATE;
1910 wakaba 1.23 ## reconsume
1911    
1912     return ($self->{current_token}); # comment
1913    
1914 wakaba 1.77 redo A;
1915     } else {
1916    
1917 wakaba 1.23 $self->{current_token}->{data} # comment
1918 wakaba 1.76 .= chr ($self->{next_char});
1919 wakaba 1.59 $self->{state} = COMMENT_STATE;
1920 wakaba 1.23
1921     if (@{$self->{char}}) {
1922 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1923 wakaba 1.23 } else {
1924 wakaba 1.76 $self->{set_next_char}->($self);
1925 wakaba 1.23 }
1926    
1927     redo A;
1928     }
1929 wakaba 1.59 } elsif ($self->{state} == COMMENT_START_DASH_STATE) {
1930 wakaba 1.76 if ($self->{next_char} == 0x002D) { # -
1931 wakaba 1.77
1932 wakaba 1.59 $self->{state} = COMMENT_END_STATE;
1933 wakaba 1.23
1934     if (@{$self->{char}}) {
1935 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1936 wakaba 1.23 } else {
1937 wakaba 1.76 $self->{set_next_char}->($self);
1938 wakaba 1.23 }
1939    
1940     redo A;
1941 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
1942 wakaba 1.77
1943 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'bogus comment');
1944 wakaba 1.59 $self->{state} = DATA_STATE;
1945 wakaba 1.23
1946     if (@{$self->{char}}) {
1947 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1948 wakaba 1.23 } else {
1949 wakaba 1.76 $self->{set_next_char}->($self);
1950 wakaba 1.23 }
1951    
1952    
1953     return ($self->{current_token}); # comment
1954    
1955     redo A;
1956 wakaba 1.76 } elsif ($self->{next_char} == -1) {
1957 wakaba 1.77
1958 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed comment');
1959 wakaba 1.59 $self->{state} = DATA_STATE;
1960 wakaba 1.23 ## reconsume
1961    
1962     return ($self->{current_token}); # comment
1963    
1964     redo A;
1965     } else {
1966 wakaba 1.77
1967 wakaba 1.23 $self->{current_token}->{data} # comment
1968 wakaba 1.76 .= '-' . chr ($self->{next_char});
1969 wakaba 1.59 $self->{state} = COMMENT_STATE;
1970 wakaba 1.23
1971     if (@{$self->{char}}) {
1972 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1973 wakaba 1.23 } else {
1974 wakaba 1.76 $self->{set_next_char}->($self);
1975 wakaba 1.23 }
1976    
1977     redo A;
1978     }
1979 wakaba 1.59 } elsif ($self->{state} == COMMENT_STATE) {
1980 wakaba 1.76 if ($self->{next_char} == 0x002D) { # -
1981 wakaba 1.77
1982 wakaba 1.59 $self->{state} = COMMENT_END_DASH_STATE;
1983 wakaba 1.1
1984     if (@{$self->{char}}) {
1985 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
1986 wakaba 1.1 } else {
1987 wakaba 1.76 $self->{set_next_char}->($self);
1988 wakaba 1.1 }
1989    
1990     redo A;
1991 wakaba 1.76 } elsif ($self->{next_char} == -1) {
1992 wakaba 1.77
1993 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed comment');
1994 wakaba 1.59 $self->{state} = DATA_STATE;
1995 wakaba 1.1 ## reconsume
1996    
1997     return ($self->{current_token}); # comment
1998    
1999     redo A;
2000     } else {
2001 wakaba 1.77
2002 wakaba 1.76 $self->{current_token}->{data} .= chr ($self->{next_char}); # comment
2003 wakaba 1.1 ## Stay in the state
2004    
2005     if (@{$self->{char}}) {
2006 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2007 wakaba 1.1 } else {
2008 wakaba 1.76 $self->{set_next_char}->($self);
2009 wakaba 1.1 }
2010    
2011     redo A;
2012     }
2013 wakaba 1.59 } elsif ($self->{state} == COMMENT_END_DASH_STATE) {
2014 wakaba 1.76 if ($self->{next_char} == 0x002D) { # -
2015 wakaba 1.77
2016 wakaba 1.59 $self->{state} = COMMENT_END_STATE;
2017 wakaba 1.1
2018     if (@{$self->{char}}) {
2019 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2020 wakaba 1.1 } else {
2021 wakaba 1.76 $self->{set_next_char}->($self);
2022 wakaba 1.1 }
2023    
2024     redo A;
2025 wakaba 1.76 } elsif ($self->{next_char} == -1) {
2026 wakaba 1.77
2027 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed comment');
2028 wakaba 1.59 $self->{state} = DATA_STATE;
2029 wakaba 1.1 ## reconsume
2030    
2031     return ($self->{current_token}); # comment
2032    
2033     redo A;
2034     } else {
2035 wakaba 1.77
2036 wakaba 1.76 $self->{current_token}->{data} .= '-' . chr ($self->{next_char}); # comment
2037 wakaba 1.59 $self->{state} = COMMENT_STATE;
2038 wakaba 1.1
2039     if (@{$self->{char}}) {
2040 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2041 wakaba 1.1 } else {
2042 wakaba 1.76 $self->{set_next_char}->($self);
2043 wakaba 1.1 }
2044    
2045     redo A;
2046     }
2047 wakaba 1.59 } elsif ($self->{state} == COMMENT_END_STATE) {
2048 wakaba 1.76 if ($self->{next_char} == 0x003E) { # >
2049 wakaba 1.77
2050 wakaba 1.59 $self->{state} = DATA_STATE;
2051 wakaba 1.1
2052     if (@{$self->{char}}) {
2053 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2054 wakaba 1.1 } else {
2055 wakaba 1.76 $self->{set_next_char}->($self);
2056 wakaba 1.1 }
2057    
2058    
2059     return ($self->{current_token}); # comment
2060    
2061     redo A;
2062 wakaba 1.76 } elsif ($self->{next_char} == 0x002D) { # -
2063 wakaba 1.77
2064 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'dash in comment');
2065 wakaba 1.1 $self->{current_token}->{data} .= '-'; # comment
2066     ## Stay in the state
2067    
2068     if (@{$self->{char}}) {
2069 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2070 wakaba 1.1 } else {
2071 wakaba 1.76 $self->{set_next_char}->($self);
2072 wakaba 1.1 }
2073    
2074     redo A;
2075 wakaba 1.76 } elsif ($self->{next_char} == -1) {
2076 wakaba 1.77
2077 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed comment');
2078 wakaba 1.59 $self->{state} = DATA_STATE;
2079 wakaba 1.1 ## reconsume
2080    
2081     return ($self->{current_token}); # comment
2082    
2083     redo A;
2084     } else {
2085 wakaba 1.77
2086 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'dash in comment');
2087 wakaba 1.76 $self->{current_token}->{data} .= '--' . chr ($self->{next_char}); # comment
2088 wakaba 1.59 $self->{state} = COMMENT_STATE;
2089 wakaba 1.1
2090     if (@{$self->{char}}) {
2091 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2092 wakaba 1.1 } else {
2093 wakaba 1.76 $self->{set_next_char}->($self);
2094 wakaba 1.1 }
2095    
2096     redo A;
2097     }
2098 wakaba 1.59 } elsif ($self->{state} == DOCTYPE_STATE) {
2099 wakaba 1.76 if ($self->{next_char} == 0x0009 or # HT
2100     $self->{next_char} == 0x000A or # LF
2101     $self->{next_char} == 0x000B or # VT
2102     $self->{next_char} == 0x000C or # FF
2103     $self->{next_char} == 0x0020) { # SP
2104 wakaba 1.77
2105 wakaba 1.59 $self->{state} = BEFORE_DOCTYPE_NAME_STATE;
2106 wakaba 1.1
2107     if (@{$self->{char}}) {
2108 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2109 wakaba 1.1 } else {
2110 wakaba 1.76 $self->{set_next_char}->($self);
2111 wakaba 1.1 }
2112    
2113     redo A;
2114     } else {
2115 wakaba 1.77
2116 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'no space before DOCTYPE name');
2117 wakaba 1.59 $self->{state} = BEFORE_DOCTYPE_NAME_STATE;
2118 wakaba 1.1 ## reconsume
2119     redo A;
2120     }
2121 wakaba 1.59 } elsif ($self->{state} == BEFORE_DOCTYPE_NAME_STATE) {
2122 wakaba 1.76 if ($self->{next_char} == 0x0009 or # HT
2123     $self->{next_char} == 0x000A or # LF
2124     $self->{next_char} == 0x000B or # VT
2125     $self->{next_char} == 0x000C or # FF
2126     $self->{next_char} == 0x0020) { # SP
2127 wakaba 1.77
2128 wakaba 1.1 ## Stay in the state
2129    
2130     if (@{$self->{char}}) {
2131 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2132 wakaba 1.1 } else {
2133 wakaba 1.76 $self->{set_next_char}->($self);
2134 wakaba 1.1 }
2135    
2136     redo A;
2137 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
2138 wakaba 1.77
2139 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'no DOCTYPE name');
2140 wakaba 1.59 $self->{state} = DATA_STATE;
2141 wakaba 1.18
2142     if (@{$self->{char}}) {
2143 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2144 wakaba 1.18 } else {
2145 wakaba 1.76 $self->{set_next_char}->($self);
2146 wakaba 1.18 }
2147    
2148    
2149 wakaba 1.75 return ({type => DOCTYPE_TOKEN, quirks => 1});
2150 wakaba 1.18
2151     redo A;
2152 wakaba 1.77 } elsif ($self->{next_char} == -1) {
2153    
2154 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'no DOCTYPE name');
2155 wakaba 1.59 $self->{state} = DATA_STATE;
2156 wakaba 1.18 ## reconsume
2157    
2158 wakaba 1.75 return ({type => DOCTYPE_TOKEN, quirks => 1});
2159 wakaba 1.18
2160     redo A;
2161     } else {
2162 wakaba 1.77
2163 wakaba 1.18 $self->{current_token}
2164 wakaba 1.57 = {type => DOCTYPE_TOKEN,
2165 wakaba 1.76 name => chr ($self->{next_char}),
2166 wakaba 1.75 #quirks => 0,
2167     };
2168 wakaba 1.4 ## ISSUE: "Set the token's name name to the" in the spec
2169 wakaba 1.59 $self->{state} = DOCTYPE_NAME_STATE;
2170 wakaba 1.1
2171     if (@{$self->{char}}) {
2172 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2173 wakaba 1.1 } else {
2174 wakaba 1.76 $self->{set_next_char}->($self);
2175 wakaba 1.1 }
2176    
2177     redo A;
2178 wakaba 1.18 }
2179 wakaba 1.59 } elsif ($self->{state} == DOCTYPE_NAME_STATE) {
2180 wakaba 1.18 ## ISSUE: Redundant "First," in the spec.
2181 wakaba 1.76 if ($self->{next_char} == 0x0009 or # HT
2182     $self->{next_char} == 0x000A or # LF
2183     $self->{next_char} == 0x000B or # VT
2184     $self->{next_char} == 0x000C or # FF
2185     $self->{next_char} == 0x0020) { # SP
2186 wakaba 1.77
2187 wakaba 1.59 $self->{state} = AFTER_DOCTYPE_NAME_STATE;
2188 wakaba 1.18
2189     if (@{$self->{char}}) {
2190 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2191 wakaba 1.18 } else {
2192 wakaba 1.76 $self->{set_next_char}->($self);
2193 wakaba 1.18 }
2194    
2195     redo A;
2196 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
2197 wakaba 1.77
2198 wakaba 1.59 $self->{state} = DATA_STATE;
2199 wakaba 1.1
2200     if (@{$self->{char}}) {
2201 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2202 wakaba 1.1 } else {
2203 wakaba 1.76 $self->{set_next_char}->($self);
2204 wakaba 1.1 }
2205    
2206    
2207 wakaba 1.18 return ($self->{current_token}); # DOCTYPE
2208 wakaba 1.1
2209     redo A;
2210 wakaba 1.76 } elsif ($self->{next_char} == -1) {
2211 wakaba 1.77
2212 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed DOCTYPE');
2213 wakaba 1.59 $self->{state} = DATA_STATE;
2214 wakaba 1.1 ## reconsume
2215    
2216 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2217 wakaba 1.18 return ($self->{current_token}); # DOCTYPE
2218 wakaba 1.1
2219     redo A;
2220     } else {
2221 wakaba 1.77
2222 wakaba 1.18 $self->{current_token}->{name}
2223 wakaba 1.76 .= chr ($self->{next_char}); # DOCTYPE
2224 wakaba 1.18 ## Stay in the state
2225 wakaba 1.1
2226     if (@{$self->{char}}) {
2227 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2228 wakaba 1.1 } else {
2229 wakaba 1.76 $self->{set_next_char}->($self);
2230 wakaba 1.1 }
2231    
2232     redo A;
2233     }
2234 wakaba 1.59 } elsif ($self->{state} == AFTER_DOCTYPE_NAME_STATE) {
2235 wakaba 1.76 if ($self->{next_char} == 0x0009 or # HT
2236     $self->{next_char} == 0x000A or # LF
2237     $self->{next_char} == 0x000B or # VT
2238     $self->{next_char} == 0x000C or # FF
2239     $self->{next_char} == 0x0020) { # SP
2240 wakaba 1.77
2241 wakaba 1.18 ## Stay in the state
2242 wakaba 1.1
2243     if (@{$self->{char}}) {
2244 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2245 wakaba 1.1 } else {
2246 wakaba 1.76 $self->{set_next_char}->($self);
2247 wakaba 1.1 }
2248    
2249     redo A;
2250 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
2251 wakaba 1.77
2252 wakaba 1.59 $self->{state} = DATA_STATE;
2253 wakaba 1.1
2254     if (@{$self->{char}}) {
2255 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2256 wakaba 1.1 } else {
2257 wakaba 1.76 $self->{set_next_char}->($self);
2258 wakaba 1.1 }
2259    
2260    
2261     return ($self->{current_token}); # DOCTYPE
2262    
2263     redo A;
2264 wakaba 1.76 } elsif ($self->{next_char} == -1) {
2265 wakaba 1.77
2266 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed DOCTYPE');
2267 wakaba 1.59 $self->{state} = DATA_STATE;
2268 wakaba 1.18 ## reconsume
2269    
2270 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2271 wakaba 1.18 return ($self->{current_token}); # DOCTYPE
2272    
2273     redo A;
2274 wakaba 1.76 } elsif ($self->{next_char} == 0x0050 or # P
2275     $self->{next_char} == 0x0070) { # p
2276 wakaba 1.18
2277     if (@{$self->{char}}) {
2278 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2279 wakaba 1.18 } else {
2280 wakaba 1.76 $self->{set_next_char}->($self);
2281 wakaba 1.18 }
2282    
2283 wakaba 1.76 if ($self->{next_char} == 0x0055 or # U
2284     $self->{next_char} == 0x0075) { # u
2285 wakaba 1.18
2286     if (@{$self->{char}}) {
2287 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2288 wakaba 1.18 } else {
2289 wakaba 1.76 $self->{set_next_char}->($self);
2290 wakaba 1.18 }
2291    
2292 wakaba 1.76 if ($self->{next_char} == 0x0042 or # B
2293     $self->{next_char} == 0x0062) { # b
2294 wakaba 1.18
2295     if (@{$self->{char}}) {
2296 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2297 wakaba 1.18 } else {
2298 wakaba 1.76 $self->{set_next_char}->($self);
2299 wakaba 1.18 }
2300    
2301 wakaba 1.76 if ($self->{next_char} == 0x004C or # L
2302     $self->{next_char} == 0x006C) { # l
2303 wakaba 1.18
2304     if (@{$self->{char}}) {
2305 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2306 wakaba 1.18 } else {
2307 wakaba 1.76 $self->{set_next_char}->($self);
2308 wakaba 1.18 }
2309    
2310 wakaba 1.76 if ($self->{next_char} == 0x0049 or # I
2311     $self->{next_char} == 0x0069) { # i
2312 wakaba 1.18
2313     if (@{$self->{char}}) {
2314 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2315 wakaba 1.18 } else {
2316 wakaba 1.76 $self->{set_next_char}->($self);
2317 wakaba 1.18 }
2318    
2319 wakaba 1.76 if ($self->{next_char} == 0x0043 or # C
2320     $self->{next_char} == 0x0063) { # c
2321 wakaba 1.77
2322 wakaba 1.59 $self->{state} = BEFORE_DOCTYPE_PUBLIC_IDENTIFIER_STATE;
2323 wakaba 1.18
2324     if (@{$self->{char}}) {
2325 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2326 wakaba 1.18 } else {
2327 wakaba 1.76 $self->{set_next_char}->($self);
2328 wakaba 1.18 }
2329    
2330     redo A;
2331 wakaba 1.77 } else {
2332    
2333 wakaba 1.18 }
2334 wakaba 1.77 } else {
2335    
2336 wakaba 1.18 }
2337 wakaba 1.77 } else {
2338    
2339 wakaba 1.18 }
2340 wakaba 1.77 } else {
2341    
2342 wakaba 1.18 }
2343 wakaba 1.77 } else {
2344    
2345 wakaba 1.18 }
2346    
2347     #
2348 wakaba 1.76 } elsif ($self->{next_char} == 0x0053 or # S
2349     $self->{next_char} == 0x0073) { # s
2350 wakaba 1.18
2351     if (@{$self->{char}}) {
2352 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2353 wakaba 1.18 } else {
2354 wakaba 1.76 $self->{set_next_char}->($self);
2355 wakaba 1.18 }
2356    
2357 wakaba 1.76 if ($self->{next_char} == 0x0059 or # Y
2358     $self->{next_char} == 0x0079) { # y
2359 wakaba 1.18
2360     if (@{$self->{char}}) {
2361 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2362 wakaba 1.18 } else {
2363 wakaba 1.76 $self->{set_next_char}->($self);
2364 wakaba 1.18 }
2365    
2366 wakaba 1.76 if ($self->{next_char} == 0x0053 or # S
2367     $self->{next_char} == 0x0073) { # s
2368 wakaba 1.18
2369     if (@{$self->{char}}) {
2370 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2371 wakaba 1.18 } else {
2372 wakaba 1.76 $self->{set_next_char}->($self);
2373 wakaba 1.18 }
2374    
2375 wakaba 1.76 if ($self->{next_char} == 0x0054 or # T
2376     $self->{next_char} == 0x0074) { # t
2377 wakaba 1.18
2378     if (@{$self->{char}}) {
2379 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2380 wakaba 1.18 } else {
2381 wakaba 1.76 $self->{set_next_char}->($self);
2382 wakaba 1.18 }
2383    
2384 wakaba 1.76 if ($self->{next_char} == 0x0045 or # E
2385     $self->{next_char} == 0x0065) { # e
2386 wakaba 1.18
2387     if (@{$self->{char}}) {
2388 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2389 wakaba 1.18 } else {
2390 wakaba 1.76 $self->{set_next_char}->($self);
2391 wakaba 1.18 }
2392    
2393 wakaba 1.76 if ($self->{next_char} == 0x004D or # M
2394     $self->{next_char} == 0x006D) { # m
2395 wakaba 1.77
2396 wakaba 1.59 $self->{state} = BEFORE_DOCTYPE_SYSTEM_IDENTIFIER_STATE;
2397 wakaba 1.18
2398     if (@{$self->{char}}) {
2399 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2400 wakaba 1.18 } else {
2401 wakaba 1.76 $self->{set_next_char}->($self);
2402 wakaba 1.18 }
2403    
2404     redo A;
2405 wakaba 1.77 } else {
2406    
2407 wakaba 1.18 }
2408 wakaba 1.77 } else {
2409    
2410 wakaba 1.18 }
2411 wakaba 1.77 } else {
2412    
2413 wakaba 1.18 }
2414 wakaba 1.77 } else {
2415    
2416 wakaba 1.18 }
2417 wakaba 1.77 } else {
2418    
2419 wakaba 1.18 }
2420    
2421     #
2422     } else {
2423    
2424 wakaba 1.77
2425 wakaba 1.18 if (@{$self->{char}}) {
2426 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2427 wakaba 1.18 } else {
2428 wakaba 1.76 $self->{set_next_char}->($self);
2429 wakaba 1.18 }
2430    
2431     #
2432     }
2433    
2434 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'string after DOCTYPE name');
2435 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2436 wakaba 1.73
2437 wakaba 1.59 $self->{state} = BOGUS_DOCTYPE_STATE;
2438 wakaba 1.18 # next-input-character is already done
2439     redo A;
2440 wakaba 1.59 } elsif ($self->{state} == BEFORE_DOCTYPE_PUBLIC_IDENTIFIER_STATE) {
2441 wakaba 1.18 if ({
2442     0x0009 => 1, 0x000A => 1, 0x000B => 1, 0x000C => 1, 0x0020 => 1,
2443     #0x000D => 1, # HT, LF, VT, FF, SP, CR
2444 wakaba 1.76 }->{$self->{next_char}}) {
2445 wakaba 1.77
2446 wakaba 1.1 ## Stay in the state
2447    
2448     if (@{$self->{char}}) {
2449 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2450 wakaba 1.1 } else {
2451 wakaba 1.76 $self->{set_next_char}->($self);
2452 wakaba 1.1 }
2453    
2454     redo A;
2455 wakaba 1.76 } elsif ($self->{next_char} eq 0x0022) { # "
2456 wakaba 1.77
2457 wakaba 1.18 $self->{current_token}->{public_identifier} = ''; # DOCTYPE
2458 wakaba 1.59 $self->{state} = DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED_STATE;
2459 wakaba 1.18
2460     if (@{$self->{char}}) {
2461 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2462 wakaba 1.18 } else {
2463 wakaba 1.76 $self->{set_next_char}->($self);
2464 wakaba 1.18 }
2465    
2466     redo A;
2467 wakaba 1.76 } elsif ($self->{next_char} eq 0x0027) { # '
2468 wakaba 1.77
2469 wakaba 1.18 $self->{current_token}->{public_identifier} = ''; # DOCTYPE
2470 wakaba 1.59 $self->{state} = DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED_STATE;
2471 wakaba 1.18
2472     if (@{$self->{char}}) {
2473 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2474 wakaba 1.18 } else {
2475 wakaba 1.76 $self->{set_next_char}->($self);
2476 wakaba 1.18 }
2477    
2478     redo A;
2479 wakaba 1.76 } elsif ($self->{next_char} eq 0x003E) { # >
2480 wakaba 1.77
2481 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'no PUBLIC literal');
2482 wakaba 1.18
2483 wakaba 1.59 $self->{state} = DATA_STATE;
2484 wakaba 1.18
2485     if (@{$self->{char}}) {
2486 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2487 wakaba 1.18 } else {
2488 wakaba 1.76 $self->{set_next_char}->($self);
2489 wakaba 1.18 }
2490    
2491    
2492 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2493 wakaba 1.18 return ($self->{current_token}); # DOCTYPE
2494    
2495     redo A;
2496 wakaba 1.76 } elsif ($self->{next_char} == -1) {
2497 wakaba 1.77
2498 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed DOCTYPE');
2499 wakaba 1.18
2500 wakaba 1.59 $self->{state} = DATA_STATE;
2501 wakaba 1.1 ## reconsume
2502    
2503 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2504 wakaba 1.18 return ($self->{current_token}); # DOCTYPE
2505 wakaba 1.1
2506     redo A;
2507     } else {
2508 wakaba 1.77
2509 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'string after PUBLIC');
2510 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2511 wakaba 1.73
2512 wakaba 1.59 $self->{state} = BOGUS_DOCTYPE_STATE;
2513 wakaba 1.18
2514     if (@{$self->{char}}) {
2515 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2516 wakaba 1.18 } else {
2517 wakaba 1.76 $self->{set_next_char}->($self);
2518 wakaba 1.18 }
2519    
2520     redo A;
2521     }
2522 wakaba 1.59 } elsif ($self->{state} == DOCTYPE_PUBLIC_IDENTIFIER_DOUBLE_QUOTED_STATE) {
2523 wakaba 1.76 if ($self->{next_char} == 0x0022) { # "
2524 wakaba 1.77
2525 wakaba 1.59 $self->{state} = AFTER_DOCTYPE_PUBLIC_IDENTIFIER_STATE;
2526 wakaba 1.18
2527     if (@{$self->{char}}) {
2528 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2529 wakaba 1.18 } else {
2530 wakaba 1.76 $self->{set_next_char}->($self);
2531 wakaba 1.18 }
2532    
2533     redo A;
2534 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
2535 wakaba 1.77
2536 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed PUBLIC literal');
2537 wakaba 1.70
2538     $self->{state} = DATA_STATE;
2539    
2540     if (@{$self->{char}}) {
2541 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2542 wakaba 1.70 } else {
2543 wakaba 1.76 $self->{set_next_char}->($self);
2544 wakaba 1.70 }
2545    
2546    
2547 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2548 wakaba 1.70 return ($self->{current_token}); # DOCTYPE
2549    
2550     redo A;
2551 wakaba 1.76 } elsif ($self->{next_char} == -1) {
2552 wakaba 1.77
2553 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed PUBLIC literal');
2554 wakaba 1.18
2555 wakaba 1.59 $self->{state} = DATA_STATE;
2556 wakaba 1.18 ## reconsume
2557    
2558 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2559 wakaba 1.18 return ($self->{current_token}); # DOCTYPE
2560    
2561     redo A;
2562     } else {
2563 wakaba 1.77
2564 wakaba 1.18 $self->{current_token}->{public_identifier} # DOCTYPE
2565 wakaba 1.76 .= chr $self->{next_char};
2566 wakaba 1.18 ## Stay in the state
2567    
2568     if (@{$self->{char}}) {
2569 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2570 wakaba 1.18 } else {
2571 wakaba 1.76 $self->{set_next_char}->($self);
2572 wakaba 1.18 }
2573    
2574     redo A;
2575     }
2576 wakaba 1.59 } elsif ($self->{state} == DOCTYPE_PUBLIC_IDENTIFIER_SINGLE_QUOTED_STATE) {
2577 wakaba 1.76 if ($self->{next_char} == 0x0027) { # '
2578 wakaba 1.77
2579 wakaba 1.59 $self->{state} = AFTER_DOCTYPE_PUBLIC_IDENTIFIER_STATE;
2580 wakaba 1.18
2581     if (@{$self->{char}}) {
2582 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2583 wakaba 1.18 } else {
2584 wakaba 1.76 $self->{set_next_char}->($self);
2585 wakaba 1.18 }
2586    
2587     redo A;
2588 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
2589 wakaba 1.77
2590 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed PUBLIC literal');
2591 wakaba 1.70
2592     $self->{state} = DATA_STATE;
2593    
2594     if (@{$self->{char}}) {
2595 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2596 wakaba 1.70 } else {
2597 wakaba 1.76 $self->{set_next_char}->($self);
2598 wakaba 1.70 }
2599    
2600    
2601 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2602 wakaba 1.70 return ($self->{current_token}); # DOCTYPE
2603    
2604     redo A;
2605 wakaba 1.76 } elsif ($self->{next_char} == -1) {
2606 wakaba 1.77
2607 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed PUBLIC literal');
2608 wakaba 1.18
2609 wakaba 1.59 $self->{state} = DATA_STATE;
2610 wakaba 1.18 ## reconsume
2611    
2612 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2613 wakaba 1.18 return ($self->{current_token}); # DOCTYPE
2614    
2615     redo A;
2616     } else {
2617 wakaba 1.77
2618 wakaba 1.18 $self->{current_token}->{public_identifier} # DOCTYPE
2619 wakaba 1.76 .= chr $self->{next_char};
2620 wakaba 1.18 ## Stay in the state
2621    
2622     if (@{$self->{char}}) {
2623 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2624 wakaba 1.18 } else {
2625 wakaba 1.76 $self->{set_next_char}->($self);
2626 wakaba 1.18 }
2627    
2628     redo A;
2629     }
2630 wakaba 1.59 } elsif ($self->{state} == AFTER_DOCTYPE_PUBLIC_IDENTIFIER_STATE) {
2631 wakaba 1.18 if ({
2632     0x0009 => 1, 0x000A => 1, 0x000B => 1, 0x000C => 1, 0x0020 => 1,
2633     #0x000D => 1, # HT, LF, VT, FF, SP, CR
2634 wakaba 1.76 }->{$self->{next_char}}) {
2635 wakaba 1.77
2636 wakaba 1.1 ## Stay in the state
2637    
2638     if (@{$self->{char}}) {
2639 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2640 wakaba 1.1 } else {
2641 wakaba 1.76 $self->{set_next_char}->($self);
2642 wakaba 1.1 }
2643    
2644     redo A;
2645 wakaba 1.76 } elsif ($self->{next_char} == 0x0022) { # "
2646 wakaba 1.77
2647 wakaba 1.18 $self->{current_token}->{system_identifier} = ''; # DOCTYPE
2648 wakaba 1.59 $self->{state} = DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED_STATE;
2649 wakaba 1.18
2650     if (@{$self->{char}}) {
2651 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2652 wakaba 1.18 } else {
2653 wakaba 1.76 $self->{set_next_char}->($self);
2654 wakaba 1.18 }
2655    
2656     redo A;
2657 wakaba 1.76 } elsif ($self->{next_char} == 0x0027) { # '
2658 wakaba 1.77
2659 wakaba 1.18 $self->{current_token}->{system_identifier} = ''; # DOCTYPE
2660 wakaba 1.59 $self->{state} = DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED_STATE;
2661 wakaba 1.18
2662     if (@{$self->{char}}) {
2663 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2664 wakaba 1.18 } else {
2665 wakaba 1.76 $self->{set_next_char}->($self);
2666 wakaba 1.18 }
2667    
2668     redo A;
2669 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
2670 wakaba 1.77
2671 wakaba 1.59 $self->{state} = DATA_STATE;
2672 wakaba 1.18
2673     if (@{$self->{char}}) {
2674 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2675 wakaba 1.18 } else {
2676 wakaba 1.76 $self->{set_next_char}->($self);
2677 wakaba 1.18 }
2678    
2679    
2680     return ($self->{current_token}); # DOCTYPE
2681    
2682     redo A;
2683 wakaba 1.76 } elsif ($self->{next_char} == -1) {
2684 wakaba 1.77
2685 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed DOCTYPE');
2686 wakaba 1.18
2687 wakaba 1.59 $self->{state} = DATA_STATE;
2688 wakaba 1.26 ## reconsume
2689 wakaba 1.18
2690 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2691 wakaba 1.18 return ($self->{current_token}); # DOCTYPE
2692    
2693     redo A;
2694     } else {
2695 wakaba 1.77
2696 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'string after PUBLIC literal');
2697 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2698 wakaba 1.73
2699 wakaba 1.59 $self->{state} = BOGUS_DOCTYPE_STATE;
2700 wakaba 1.18
2701     if (@{$self->{char}}) {
2702 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2703 wakaba 1.18 } else {
2704 wakaba 1.76 $self->{set_next_char}->($self);
2705 wakaba 1.18 }
2706    
2707     redo A;
2708 wakaba 1.1 }
2709 wakaba 1.59 } elsif ($self->{state} == BEFORE_DOCTYPE_SYSTEM_IDENTIFIER_STATE) {
2710 wakaba 1.18 if ({
2711     0x0009 => 1, 0x000A => 1, 0x000B => 1, 0x000C => 1, 0x0020 => 1,
2712     #0x000D => 1, # HT, LF, VT, FF, SP, CR
2713 wakaba 1.76 }->{$self->{next_char}}) {
2714 wakaba 1.77
2715 wakaba 1.1 ## Stay in the state
2716    
2717     if (@{$self->{char}}) {
2718 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2719 wakaba 1.1 } else {
2720 wakaba 1.76 $self->{set_next_char}->($self);
2721 wakaba 1.1 }
2722    
2723     redo A;
2724 wakaba 1.76 } elsif ($self->{next_char} == 0x0022) { # "
2725 wakaba 1.77
2726 wakaba 1.18 $self->{current_token}->{system_identifier} = ''; # DOCTYPE
2727 wakaba 1.59 $self->{state} = DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED_STATE;
2728 wakaba 1.18
2729     if (@{$self->{char}}) {
2730 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2731 wakaba 1.18 } else {
2732 wakaba 1.76 $self->{set_next_char}->($self);
2733 wakaba 1.18 }
2734    
2735     redo A;
2736 wakaba 1.76 } elsif ($self->{next_char} == 0x0027) { # '
2737 wakaba 1.77
2738 wakaba 1.18 $self->{current_token}->{system_identifier} = ''; # DOCTYPE
2739 wakaba 1.59 $self->{state} = DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED_STATE;
2740 wakaba 1.18
2741     if (@{$self->{char}}) {
2742 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2743 wakaba 1.18 } else {
2744 wakaba 1.76 $self->{set_next_char}->($self);
2745 wakaba 1.18 }
2746    
2747     redo A;
2748 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
2749 wakaba 1.77
2750 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'no SYSTEM literal');
2751 wakaba 1.59 $self->{state} = DATA_STATE;
2752 wakaba 1.1
2753     if (@{$self->{char}}) {
2754 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2755 wakaba 1.1 } else {
2756 wakaba 1.76 $self->{set_next_char}->($self);
2757 wakaba 1.1 }
2758    
2759    
2760 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2761 wakaba 1.1 return ($self->{current_token}); # DOCTYPE
2762    
2763     redo A;
2764 wakaba 1.76 } elsif ($self->{next_char} == -1) {
2765 wakaba 1.77
2766 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed DOCTYPE');
2767 wakaba 1.18
2768 wakaba 1.59 $self->{state} = DATA_STATE;
2769 wakaba 1.26 ## reconsume
2770 wakaba 1.18
2771 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2772 wakaba 1.18 return ($self->{current_token}); # DOCTYPE
2773    
2774     redo A;
2775     } else {
2776 wakaba 1.77
2777 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'string after SYSTEM');
2778 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2779 wakaba 1.73
2780 wakaba 1.59 $self->{state} = BOGUS_DOCTYPE_STATE;
2781 wakaba 1.18
2782     if (@{$self->{char}}) {
2783 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2784 wakaba 1.18 } else {
2785 wakaba 1.76 $self->{set_next_char}->($self);
2786 wakaba 1.18 }
2787    
2788     redo A;
2789     }
2790 wakaba 1.59 } elsif ($self->{state} == DOCTYPE_SYSTEM_IDENTIFIER_DOUBLE_QUOTED_STATE) {
2791 wakaba 1.76 if ($self->{next_char} == 0x0022) { # "
2792 wakaba 1.77
2793 wakaba 1.59 $self->{state} = AFTER_DOCTYPE_SYSTEM_IDENTIFIER_STATE;
2794 wakaba 1.18
2795     if (@{$self->{char}}) {
2796 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2797 wakaba 1.18 } else {
2798 wakaba 1.76 $self->{set_next_char}->($self);
2799 wakaba 1.18 }
2800    
2801     redo A;
2802 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
2803 wakaba 1.77
2804 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed PUBLIC literal');
2805 wakaba 1.70
2806     $self->{state} = DATA_STATE;
2807    
2808     if (@{$self->{char}}) {
2809 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2810 wakaba 1.70 } else {
2811 wakaba 1.76 $self->{set_next_char}->($self);
2812 wakaba 1.70 }
2813    
2814    
2815 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2816 wakaba 1.70 return ($self->{current_token}); # DOCTYPE
2817    
2818     redo A;
2819 wakaba 1.76 } elsif ($self->{next_char} == -1) {
2820 wakaba 1.77
2821 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed SYSTEM literal');
2822 wakaba 1.18
2823 wakaba 1.59 $self->{state} = DATA_STATE;
2824 wakaba 1.1 ## reconsume
2825    
2826 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2827 wakaba 1.1 return ($self->{current_token}); # DOCTYPE
2828    
2829     redo A;
2830     } else {
2831 wakaba 1.77
2832 wakaba 1.18 $self->{current_token}->{system_identifier} # DOCTYPE
2833 wakaba 1.76 .= chr $self->{next_char};
2834 wakaba 1.18 ## Stay in the state
2835    
2836     if (@{$self->{char}}) {
2837 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2838 wakaba 1.18 } else {
2839 wakaba 1.76 $self->{set_next_char}->($self);
2840 wakaba 1.18 }
2841    
2842     redo A;
2843     }
2844 wakaba 1.59 } elsif ($self->{state} == DOCTYPE_SYSTEM_IDENTIFIER_SINGLE_QUOTED_STATE) {
2845 wakaba 1.76 if ($self->{next_char} == 0x0027) { # '
2846 wakaba 1.77
2847 wakaba 1.59 $self->{state} = AFTER_DOCTYPE_SYSTEM_IDENTIFIER_STATE;
2848 wakaba 1.18
2849     if (@{$self->{char}}) {
2850 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2851 wakaba 1.18 } else {
2852 wakaba 1.76 $self->{set_next_char}->($self);
2853 wakaba 1.18 }
2854    
2855     redo A;
2856 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
2857 wakaba 1.77
2858 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed PUBLIC literal');
2859 wakaba 1.70
2860     $self->{state} = DATA_STATE;
2861    
2862     if (@{$self->{char}}) {
2863 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2864 wakaba 1.70 } else {
2865 wakaba 1.76 $self->{set_next_char}->($self);
2866 wakaba 1.70 }
2867    
2868    
2869 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2870 wakaba 1.70 return ($self->{current_token}); # DOCTYPE
2871    
2872     redo A;
2873 wakaba 1.76 } elsif ($self->{next_char} == -1) {
2874 wakaba 1.77
2875 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed SYSTEM literal');
2876 wakaba 1.18
2877 wakaba 1.59 $self->{state} = DATA_STATE;
2878 wakaba 1.18 ## reconsume
2879    
2880 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2881 wakaba 1.18 return ($self->{current_token}); # DOCTYPE
2882    
2883     redo A;
2884     } else {
2885 wakaba 1.77
2886 wakaba 1.18 $self->{current_token}->{system_identifier} # DOCTYPE
2887 wakaba 1.76 .= chr $self->{next_char};
2888 wakaba 1.18 ## Stay in the state
2889    
2890     if (@{$self->{char}}) {
2891 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2892 wakaba 1.18 } else {
2893 wakaba 1.76 $self->{set_next_char}->($self);
2894 wakaba 1.18 }
2895    
2896     redo A;
2897     }
2898 wakaba 1.59 } elsif ($self->{state} == AFTER_DOCTYPE_SYSTEM_IDENTIFIER_STATE) {
2899 wakaba 1.18 if ({
2900     0x0009 => 1, 0x000A => 1, 0x000B => 1, 0x000C => 1, 0x0020 => 1,
2901     #0x000D => 1, # HT, LF, VT, FF, SP, CR
2902 wakaba 1.76 }->{$self->{next_char}}) {
2903 wakaba 1.77
2904 wakaba 1.18 ## Stay in the state
2905    
2906     if (@{$self->{char}}) {
2907 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2908 wakaba 1.18 } else {
2909 wakaba 1.76 $self->{set_next_char}->($self);
2910 wakaba 1.18 }
2911    
2912     redo A;
2913 wakaba 1.76 } elsif ($self->{next_char} == 0x003E) { # >
2914 wakaba 1.77
2915 wakaba 1.59 $self->{state} = DATA_STATE;
2916 wakaba 1.18
2917     if (@{$self->{char}}) {
2918 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2919 wakaba 1.18 } else {
2920 wakaba 1.76 $self->{set_next_char}->($self);
2921 wakaba 1.18 }
2922    
2923    
2924     return ($self->{current_token}); # DOCTYPE
2925    
2926     redo A;
2927 wakaba 1.76 } elsif ($self->{next_char} == -1) {
2928 wakaba 1.77
2929 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed DOCTYPE');
2930 wakaba 1.18
2931 wakaba 1.59 $self->{state} = DATA_STATE;
2932 wakaba 1.26 ## reconsume
2933 wakaba 1.18
2934 wakaba 1.75 $self->{current_token}->{quirks} = 1;
2935 wakaba 1.18 return ($self->{current_token}); # DOCTYPE
2936    
2937     redo A;
2938     } else {
2939 wakaba 1.77
2940 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'string after SYSTEM literal');
2941 wakaba 1.75 #$self->{current_token}->{quirks} = 1;
2942 wakaba 1.73
2943 wakaba 1.59 $self->{state} = BOGUS_DOCTYPE_STATE;
2944 wakaba 1.1
2945     if (@{$self->{char}}) {
2946 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2947 wakaba 1.1 } else {
2948 wakaba 1.76 $self->{set_next_char}->($self);
2949 wakaba 1.1 }
2950    
2951     redo A;
2952     }
2953 wakaba 1.59 } elsif ($self->{state} == BOGUS_DOCTYPE_STATE) {
2954 wakaba 1.76 if ($self->{next_char} == 0x003E) { # >
2955 wakaba 1.77
2956 wakaba 1.59 $self->{state} = DATA_STATE;
2957 wakaba 1.1
2958     if (@{$self->{char}}) {
2959 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2960 wakaba 1.1 } else {
2961 wakaba 1.76 $self->{set_next_char}->($self);
2962 wakaba 1.1 }
2963    
2964    
2965     return ($self->{current_token}); # DOCTYPE
2966    
2967     redo A;
2968 wakaba 1.76 } elsif ($self->{next_char} == -1) {
2969 wakaba 1.77
2970 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unclosed DOCTYPE');
2971 wakaba 1.59 $self->{state} = DATA_STATE;
2972 wakaba 1.1 ## reconsume
2973    
2974     return ($self->{current_token}); # DOCTYPE
2975    
2976     redo A;
2977     } else {
2978 wakaba 1.77
2979 wakaba 1.1 ## Stay in the state
2980    
2981     if (@{$self->{char}}) {
2982 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
2983 wakaba 1.1 } else {
2984 wakaba 1.76 $self->{set_next_char}->($self);
2985 wakaba 1.1 }
2986    
2987     redo A;
2988     }
2989     } else {
2990     die "$0: $self->{state}: Unknown state";
2991     }
2992     } # A
2993    
2994     die "$0: _get_next_token: unexpected case";
2995     } # _get_next_token
2996    
2997 wakaba 1.72 sub _tokenize_attempt_to_consume_an_entity ($$$) {
2998     my ($self, $in_attr, $additional) = @_;
2999 wakaba 1.20
3000     if ({
3001     0x0009 => 1, 0x000A => 1, 0x000B => 1, 0x000C => 1, # HT, LF, VT, FF,
3002     0x0020 => 1, 0x003C => 1, 0x0026 => 1, -1 => 1, # SP, <, & # 0x000D # CR
3003 wakaba 1.72 $additional => 1,
3004 wakaba 1.76 }->{$self->{next_char}}) {
3005 wakaba 1.78
3006 wakaba 1.20 ## Don't consume
3007     ## No error
3008     return undef;
3009 wakaba 1.76 } elsif ($self->{next_char} == 0x0023) { # #
3010 wakaba 1.1
3011     if (@{$self->{char}}) {
3012 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
3013 wakaba 1.1 } else {
3014 wakaba 1.76 $self->{set_next_char}->($self);
3015 wakaba 1.1 }
3016    
3017 wakaba 1.76 if ($self->{next_char} == 0x0078 or # x
3018     $self->{next_char} == 0x0058) { # X
3019 wakaba 1.26 my $code;
3020 wakaba 1.1 X: {
3021 wakaba 1.76 my $x_char = $self->{next_char};
3022 wakaba 1.1
3023     if (@{$self->{char}}) {
3024 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
3025 wakaba 1.1 } else {
3026 wakaba 1.76 $self->{set_next_char}->($self);
3027 wakaba 1.1 }
3028    
3029 wakaba 1.76 if (0x0030 <= $self->{next_char} and
3030     $self->{next_char} <= 0x0039) { # 0..9
3031 wakaba 1.78
3032 wakaba 1.26 $code ||= 0;
3033     $code *= 0x10;
3034 wakaba 1.76 $code += $self->{next_char} - 0x0030;
3035 wakaba 1.1 redo X;
3036 wakaba 1.76 } elsif (0x0061 <= $self->{next_char} and
3037     $self->{next_char} <= 0x0066) { # a..f
3038 wakaba 1.78
3039 wakaba 1.26 $code ||= 0;
3040     $code *= 0x10;
3041 wakaba 1.76 $code += $self->{next_char} - 0x0060 + 9;
3042 wakaba 1.1 redo X;
3043 wakaba 1.76 } elsif (0x0041 <= $self->{next_char} and
3044     $self->{next_char} <= 0x0046) { # A..F
3045 wakaba 1.78
3046 wakaba 1.26 $code ||= 0;
3047     $code *= 0x10;
3048 wakaba 1.76 $code += $self->{next_char} - 0x0040 + 9;
3049 wakaba 1.1 redo X;
3050 wakaba 1.26 } elsif (not defined $code) { # no hexadecimal digit
3051 wakaba 1.78
3052 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'bare hcro');
3053 wakaba 1.76 unshift @{$self->{char}}, ($x_char, $self->{next_char});
3054     $self->{next_char} = 0x0023; # #
3055 wakaba 1.1 return undef;
3056 wakaba 1.76 } elsif ($self->{next_char} == 0x003B) { # ;
3057 wakaba 1.1
3058 wakaba 1.78
3059 wakaba 1.1 if (@{$self->{char}}) {
3060 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
3061 wakaba 1.1 } else {
3062 wakaba 1.76 $self->{set_next_char}->($self);
3063 wakaba 1.1 }
3064    
3065     } else {
3066 wakaba 1.78
3067 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'no refc');
3068 wakaba 1.1 }
3069    
3070 wakaba 1.26 if ($code == 0 or (0xD800 <= $code and $code <= 0xDFFF)) {
3071 wakaba 1.78
3072 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => sprintf 'invalid character reference:U+%04X', $code);
3073 wakaba 1.26 $code = 0xFFFD;
3074     } elsif ($code > 0x10FFFF) {
3075 wakaba 1.78
3076 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => sprintf 'invalid character reference:U-%08X', $code);
3077 wakaba 1.26 $code = 0xFFFD;
3078     } elsif ($code == 0x000D) {
3079 wakaba 1.78
3080 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'CR character reference');
3081 wakaba 1.26 $code = 0x000A;
3082     } elsif (0x80 <= $code and $code <= 0x9F) {
3083 wakaba 1.78
3084 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => sprintf 'C1 character reference:U+%04X', $code);
3085 wakaba 1.26 $code = $c1_entity_char->{$code};
3086 wakaba 1.1 }
3087    
3088 wakaba 1.68 return {type => CHARACTER_TOKEN, data => chr $code,
3089     has_reference => 1};
3090 wakaba 1.1 } # X
3091 wakaba 1.76 } elsif (0x0030 <= $self->{next_char} and
3092     $self->{next_char} <= 0x0039) { # 0..9
3093     my $code = $self->{next_char} - 0x0030;
3094 wakaba 1.1
3095     if (@{$self->{char}}) {
3096 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
3097 wakaba 1.1 } else {
3098 wakaba 1.76 $self->{set_next_char}->($self);
3099 wakaba 1.1 }
3100    
3101    
3102 wakaba 1.76 while (0x0030 <= $self->{next_char} and
3103     $self->{next_char} <= 0x0039) { # 0..9
3104 wakaba 1.78
3105 wakaba 1.1 $code *= 10;
3106 wakaba 1.76 $code += $self->{next_char} - 0x0030;
3107 wakaba 1.1
3108    
3109     if (@{$self->{char}}) {
3110 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
3111 wakaba 1.1 } else {
3112 wakaba 1.76 $self->{set_next_char}->($self);
3113 wakaba 1.1 }
3114    
3115     }
3116    
3117 wakaba 1.76 if ($self->{next_char} == 0x003B) { # ;
3118 wakaba 1.1
3119 wakaba 1.78
3120 wakaba 1.1 if (@{$self->{char}}) {
3121 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
3122 wakaba 1.1 } else {
3123 wakaba 1.76 $self->{set_next_char}->($self);
3124 wakaba 1.1 }
3125    
3126     } else {
3127 wakaba 1.78
3128 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'no refc');
3129 wakaba 1.1 }
3130    
3131 wakaba 1.26 if ($code == 0 or (0xD800 <= $code and $code <= 0xDFFF)) {
3132 wakaba 1.78
3133 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => sprintf 'invalid character reference:U+%04X', $code);
3134 wakaba 1.26 $code = 0xFFFD;
3135     } elsif ($code > 0x10FFFF) {
3136 wakaba 1.78
3137 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => sprintf 'invalid character reference:U-%08X', $code);
3138 wakaba 1.26 $code = 0xFFFD;
3139     } elsif ($code == 0x000D) {
3140 wakaba 1.78
3141 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'CR character reference');
3142 wakaba 1.26 $code = 0x000A;
3143 wakaba 1.4 } elsif (0x80 <= $code and $code <= 0x9F) {
3144 wakaba 1.78
3145 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => sprintf 'C1 character reference:U+%04X', $code);
3146 wakaba 1.4 $code = $c1_entity_char->{$code};
3147 wakaba 1.1 }
3148    
3149 wakaba 1.68 return {type => CHARACTER_TOKEN, data => chr $code, has_reference => 1};
3150 wakaba 1.1 } else {
3151 wakaba 1.78
3152 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'bare nero');
3153 wakaba 1.76 unshift @{$self->{char}}, ($self->{next_char});
3154     $self->{next_char} = 0x0023; # #
3155 wakaba 1.1 return undef;
3156     }
3157 wakaba 1.76 } elsif ((0x0041 <= $self->{next_char} and
3158     $self->{next_char} <= 0x005A) or
3159     (0x0061 <= $self->{next_char} and
3160     $self->{next_char} <= 0x007A)) {
3161     my $entity_name = chr $self->{next_char};
3162 wakaba 1.1
3163     if (@{$self->{char}}) {
3164 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
3165 wakaba 1.1 } else {
3166 wakaba 1.76 $self->{set_next_char}->($self);
3167 wakaba 1.1 }
3168    
3169    
3170     my $value = $entity_name;
3171 wakaba 1.37 my $match = 0;
3172 wakaba 1.16 require Whatpm::_NamedEntityList;
3173     our $EntityChar;
3174 wakaba 1.1
3175     while (length $entity_name < 10 and
3176     ## NOTE: Some number greater than the maximum length of entity name
3177 wakaba 1.76 ((0x0041 <= $self->{next_char} and # a
3178     $self->{next_char} <= 0x005A) or # x
3179     (0x0061 <= $self->{next_char} and # a
3180     $self->{next_char} <= 0x007A) or # z
3181     (0x0030 <= $self->{next_char} and # 0
3182     $self->{next_char} <= 0x0039) or # 9
3183     $self->{next_char} == 0x003B)) { # ;
3184     $entity_name .= chr $self->{next_char};
3185 wakaba 1.16 if (defined $EntityChar->{$entity_name}) {
3186 wakaba 1.76 if ($self->{next_char} == 0x003B) { # ;
3187 wakaba 1.78
3188 wakaba 1.26 $value = $EntityChar->{$entity_name};
3189 wakaba 1.16 $match = 1;
3190    
3191     if (@{$self->{char}}) {
3192 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
3193 wakaba 1.16 } else {
3194 wakaba 1.76 $self->{set_next_char}->($self);
3195 wakaba 1.16 }
3196    
3197     last;
3198 wakaba 1.37 } else {
3199 wakaba 1.78
3200 wakaba 1.26 $value = $EntityChar->{$entity_name};
3201     $match = -1;
3202 wakaba 1.37
3203     if (@{$self->{char}}) {
3204 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
3205 wakaba 1.37 } else {
3206 wakaba 1.76 $self->{set_next_char}->($self);
3207 wakaba 1.37 }
3208    
3209 wakaba 1.16 }
3210 wakaba 1.1 } else {
3211 wakaba 1.78
3212 wakaba 1.76 $value .= chr $self->{next_char};
3213 wakaba 1.37 $match *= 2;
3214    
3215 wakaba 1.1 if (@{$self->{char}}) {
3216 wakaba 1.76 $self->{next_char} = shift @{$self->{char}};
3217 wakaba 1.1 } else {
3218 wakaba 1.76 $self->{set_next_char}->($self);
3219 wakaba 1.1 }
3220    
3221 wakaba 1.37 }
3222 wakaba 1.1 }
3223    
3224 wakaba 1.16 if ($match > 0) {
3225 wakaba 1.78
3226 wakaba 1.68 return {type => CHARACTER_TOKEN, data => $value, has_reference => 1};
3227 wakaba 1.16 } elsif ($match < 0) {
3228 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'no refc');
3229 wakaba 1.37 if ($in_attr and $match < -1) {
3230 wakaba 1.78
3231 wakaba 1.57 return {type => CHARACTER_TOKEN, data => '&'.$entity_name};
3232 wakaba 1.37 } else {
3233 wakaba 1.78
3234 wakaba 1.68 return {type => CHARACTER_TOKEN, data => $value, has_reference => 1};
3235 wakaba 1.37 }
3236 wakaba 1.1 } else {
3237 wakaba 1.78
3238 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'bare ero');
3239 wakaba 1.68 ## NOTE: "No characters are consumed" in the spec.
3240 wakaba 1.57 return {type => CHARACTER_TOKEN, data => '&'.$value};
3241 wakaba 1.1 }
3242     } else {
3243 wakaba 1.78
3244 wakaba 1.1 ## no characters are consumed
3245 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'bare ero');
3246 wakaba 1.1 return undef;
3247     }
3248     } # _tokenize_attempt_to_consume_an_entity
3249    
3250     sub _initialize_tree_constructor ($) {
3251     my $self = shift;
3252     ## NOTE: $self->{document} MUST be specified before this method is called
3253     $self->{document}->strict_error_checking (0);
3254     ## TODO: Turn mutation events off # MUST
3255     ## TODO: Turn loose Document option (manakai extension) on
3256 wakaba 1.18 $self->{document}->manakai_is_html (1); # MUST
3257 wakaba 1.1 } # _initialize_tree_constructor
3258    
3259     sub _terminate_tree_constructor ($) {
3260     my $self = shift;
3261     $self->{document}->strict_error_checking (1);
3262     ## TODO: Turn mutation events on
3263     } # _terminate_tree_constructor
3264    
3265     ## ISSUE: Should append_child (for example) in script executed in tree construction stage fire mutation events?
3266    
3267 wakaba 1.3 { # tree construction stage
3268     my $token;
3269    
3270 wakaba 1.1 sub _construct_tree ($) {
3271     my ($self) = @_;
3272    
3273     ## When an interactive UA render the $self->{document} available
3274     ## to the user, or when it begin accepting user input, are
3275     ## not defined.
3276    
3277     ## Append a character: collect it and all subsequent consecutive
3278     ## characters and insert one Text node whose data is concatenation
3279     ## of all those characters. # MUST
3280    
3281     $token = $self->_get_next_token;
3282    
3283 wakaba 1.3 undef $self->{form_element};
3284     undef $self->{head_element};
3285     $self->{open_elements} = [];
3286     undef $self->{inner_html_node};
3287    
3288 wakaba 1.84 ## NOTE: The "initial" insertion mode.
3289 wakaba 1.3 $self->_tree_construction_initial; # MUST
3290 wakaba 1.84
3291 wakaba 1.85 ## NOTE: The "before html" insertion mode.
3292 wakaba 1.3 $self->_tree_construction_root_element;
3293 wakaba 1.84 $self->{insertion_mode} = BEFORE_HEAD_IM;
3294    
3295     ## NOTE: The "before head" insertion mode and so on.
3296 wakaba 1.3 $self->_tree_construction_main;
3297     } # _construct_tree
3298    
3299     sub _tree_construction_initial ($) {
3300     my $self = shift;
3301 wakaba 1.84
3302     ## NOTE: "initial" insertion mode
3303    
3304 wakaba 1.18 INITIAL: {
3305 wakaba 1.57 if ($token->{type} == DOCTYPE_TOKEN) {
3306 wakaba 1.18 ## NOTE: Conformance checkers MAY, instead of reporting "not HTML5"
3307     ## error, switch to a conformance checking mode for another
3308     ## language.
3309     my $doctype_name = $token->{name};
3310     $doctype_name = '' unless defined $doctype_name;
3311     $doctype_name =~ tr/a-z/A-Z/;
3312     if (not defined $token->{name} or # <!DOCTYPE>
3313     defined $token->{public_identifier} or
3314     defined $token->{system_identifier}) {
3315 wakaba 1.79
3316 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not HTML5');
3317 wakaba 1.18 } elsif ($doctype_name ne 'HTML') {
3318 wakaba 1.79
3319 wakaba 1.18 ## ISSUE: ASCII case-insensitive? (in fact it does not matter)
3320 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not HTML5');
3321 wakaba 1.79 } else {
3322    
3323 wakaba 1.18 }
3324    
3325     my $doctype = $self->{document}->create_document_type_definition
3326     ($token->{name}); ## ISSUE: If name is missing (e.g. <!DOCTYPE>)?
3327     $doctype->public_id ($token->{public_identifier})
3328     if defined $token->{public_identifier};
3329     $doctype->system_id ($token->{system_identifier})
3330     if defined $token->{system_identifier};
3331     ## NOTE: Other DocumentType attributes are null or empty lists.
3332     ## ISSUE: internalSubset = null??
3333     $self->{document}->append_child ($doctype);
3334    
3335 wakaba 1.75 if ($token->{quirks} or $doctype_name ne 'HTML') {
3336 wakaba 1.79
3337 wakaba 1.18 $self->{document}->manakai_compat_mode ('quirks');
3338     } elsif (defined $token->{public_identifier}) {
3339     my $pubid = $token->{public_identifier};
3340     $pubid =~ tr/a-z/A-z/;
3341     if ({
3342     "+//SILMARIL//DTD HTML PRO V0R11 19970101//EN" => 1,
3343     "-//ADVASOFT LTD//DTD HTML 3.0 ASWEDIT + EXTENSIONS//EN" => 1,
3344     "-//AS//DTD HTML 3.0 ASWEDIT + EXTENSIONS//EN" => 1,
3345     "-//IETF//DTD HTML 2.0 LEVEL 1//EN" => 1,
3346     "-//IETF//DTD HTML 2.0 LEVEL 2//EN" => 1,
3347     "-//IETF//DTD HTML 2.0 STRICT LEVEL 1//EN" => 1,
3348     "-//IETF//DTD HTML 2.0 STRICT LEVEL 2//EN" => 1,
3349     "-//IETF//DTD HTML 2.0 STRICT//EN" => 1,
3350     "-//IETF//DTD HTML 2.0//EN" => 1,
3351     "-//IETF//DTD HTML 2.1E//EN" => 1,
3352     "-//IETF//DTD HTML 3.0//EN" => 1,
3353     "-//IETF//DTD HTML 3.0//EN//" => 1,
3354     "-//IETF//DTD HTML 3.2 FINAL//EN" => 1,
3355     "-//IETF//DTD HTML 3.2//EN" => 1,
3356     "-//IETF//DTD HTML 3//EN" => 1,
3357     "-//IETF//DTD HTML LEVEL 0//EN" => 1,
3358     "-//IETF//DTD HTML LEVEL 0//EN//2.0" => 1,
3359     "-//IETF//DTD HTML LEVEL 1//EN" => 1,
3360     "-//IETF//DTD HTML LEVEL 1//EN//2.0" => 1,
3361     "-//IETF//DTD HTML LEVEL 2//EN" => 1,
3362     "-//IETF//DTD HTML LEVEL 2//EN//2.0" => 1,
3363     "-//IETF//DTD HTML LEVEL 3//EN" => 1,
3364     "-//IETF//DTD HTML LEVEL 3//EN//3.0" => 1,
3365     "-//IETF//DTD HTML STRICT LEVEL 0//EN" => 1,
3366     "-//IETF//DTD HTML STRICT LEVEL 0//EN//2.0" => 1,
3367     "-//IETF//DTD HTML STRICT LEVEL 1//EN" => 1,
3368     "-//IETF//DTD HTML STRICT LEVEL 1//EN//2.0" => 1,
3369     "-//IETF//DTD HTML STRICT LEVEL 2//EN" => 1,
3370     "-//IETF//DTD HTML STRICT LEVEL 2//EN//2.0" => 1,
3371     "-//IETF//DTD HTML STRICT LEVEL 3//EN" => 1,
3372     "-//IETF//DTD HTML STRICT LEVEL 3//EN//3.0" => 1,
3373     "-//IETF//DTD HTML STRICT//EN" => 1,
3374     "-//IETF//DTD HTML STRICT//EN//2.0" => 1,
3375     "-//IETF//DTD HTML STRICT//EN//3.0" => 1,
3376     "-//IETF//DTD HTML//EN" => 1,
3377     "-//IETF//DTD HTML//EN//2.0" => 1,
3378     "-//IETF//DTD HTML//EN//3.0" => 1,
3379     "-//METRIUS//DTD METRIUS PRESENTATIONAL//EN" => 1,
3380     "-//MICROSOFT//DTD INTERNET EXPLORER 2.0 HTML STRICT//EN" => 1,
3381     "-//MICROSOFT//DTD INTERNET EXPLORER 2.0 HTML//EN" => 1,
3382     "-//MICROSOFT//DTD INTERNET EXPLORER 2.0 TABLES//EN" => 1,
3383     "-//MICROSOFT//DTD INTERNET EXPLORER 3.0 HTML STRICT//EN" => 1,
3384     "-//MICROSOFT//DTD INTERNET EXPLORER 3.0 HTML//EN" => 1,
3385     "-//MICROSOFT//DTD INTERNET EXPLORER 3.0 TABLES//EN" => 1,
3386     "-//NETSCAPE COMM. CORP.//DTD HTML//EN" => 1,
3387     "-//NETSCAPE COMM. CORP.//DTD STRICT HTML//EN" => 1,
3388     "-//O'REILLY AND ASSOCIATES//DTD HTML 2.0//EN" => 1,
3389     "-//O'REILLY AND ASSOCIATES//DTD HTML EXTENDED 1.0//EN" => 1,
3390 wakaba 1.72 "-//O'REILLY AND ASSOCIATES//DTD HTML EXTENDED RELAXED 1.0//EN" => 1,
3391     "-//SOFTQUAD SOFTWARE//DTD HOTMETAL PRO 6.0::19990601::EXTENSIONS TO HTML 4.0//EN" => 1,
3392     "-//SOFTQUAD//DTD HOTMETAL PRO 4.0::19971010::EXTENSIONS TO HTML 4.0//EN" => 1,
3393 wakaba 1.18 "-//SPYGLASS//DTD HTML 2.0 EXTENDED//EN" => 1,
3394     "-//SQ//DTD HTML 2.0 HOTMETAL + EXTENSIONS//EN" => 1,
3395     "-//SUN MICROSYSTEMS CORP.//DTD HOTJAVA HTML//EN" => 1,
3396     "-//SUN MICROSYSTEMS CORP.//DTD HOTJAVA STRICT HTML//EN" => 1,
3397     "-//W3C//DTD HTML 3 1995-03-24//EN" => 1,
3398     "-//W3C//DTD HTML 3.2 DRAFT//EN" => 1,
3399     "-//W3C//DTD HTML 3.2 FINAL//EN" => 1,
3400     "-//W3C//DTD HTML 3.2//EN" => 1,
3401     "-//W3C//DTD HTML 3.2S DRAFT//EN" => 1,
3402     "-//W3C//DTD HTML 4.0 FRAMESET//EN" => 1,
3403     "-//W3C//DTD HTML 4.0 TRANSITIONAL//EN" => 1,
3404     "-//W3C//DTD HTML EXPERIMETNAL 19960712//EN" => 1,
3405     "-//W3C//DTD HTML EXPERIMENTAL 970421//EN" => 1,
3406     "-//W3C//DTD W3 HTML//EN" => 1,
3407     "-//W3O//DTD W3 HTML 3.0//EN" => 1,
3408     "-//W3O//DTD W3 HTML 3.0//EN//" => 1,
3409     "-//W3O//DTD W3 HTML STRICT 3.0//EN//" => 1,
3410     "-//WEBTECHS//DTD MOZILLA HTML 2.0//EN" => 1,
3411     "-//WEBTECHS//DTD MOZILLA HTML//EN" => 1,
3412     "-/W3C/DTD HTML 4.0 TRANSITIONAL/EN" => 1,
3413     "HTML" => 1,
3414     }->{$pubid}) {
3415 wakaba 1.79
3416 wakaba 1.18 $self->{document}->manakai_compat_mode ('quirks');
3417     } elsif ($pubid eq "-//W3C//DTD HTML 4.01 FRAMESET//EN" or
3418     $pubid eq "-//W3C//DTD HTML 4.01 TRANSITIONAL//EN") {
3419     if (defined $token->{system_identifier}) {
3420 wakaba 1.79
3421 wakaba 1.18 $self->{document}->manakai_compat_mode ('quirks');
3422     } else {
3423 wakaba 1.79
3424 wakaba 1.18 $self->{document}->manakai_compat_mode ('limited quirks');
3425 wakaba 1.3 }
3426 wakaba 1.80 } elsif ($pubid eq "-//W3C//DTD XHTML 1.0 FRAMESET//EN" or
3427     $pubid eq "-//W3C//DTD XHTML 1.0 TRANSITIONAL//EN") {
3428 wakaba 1.79
3429 wakaba 1.18 $self->{document}->manakai_compat_mode ('limited quirks');
3430 wakaba 1.79 } else {
3431    
3432 wakaba 1.18 }
3433 wakaba 1.79 } else {
3434    
3435 wakaba 1.18 }
3436     if (defined $token->{system_identifier}) {
3437     my $sysid = $token->{system_identifier};
3438     $sysid =~ tr/A-Z/a-z/;
3439     if ($sysid eq "http://www.ibm.com/data/dtd/v11/ibmxhtml1-transitional.dtd") {
3440 wakaba 1.80 ## TODO: Check the spec: PUBLIC "(limited quirks)" "(quirks)"
3441 wakaba 1.18 $self->{document}->manakai_compat_mode ('quirks');
3442 wakaba 1.79
3443     } else {
3444    
3445 wakaba 1.18 }
3446 wakaba 1.79 } else {
3447    
3448 wakaba 1.18 }
3449    
3450 wakaba 1.85 ## Go to the "before html" insertion mode.
3451 wakaba 1.18 $token = $self->_get_next_token;
3452     return;
3453     } elsif ({
3454 wakaba 1.57 START_TAG_TOKEN, 1,
3455     END_TAG_TOKEN, 1,
3456     END_OF_FILE_TOKEN, 1,
3457 wakaba 1.18 }->{$token->{type}}) {
3458 wakaba 1.79
3459 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'no DOCTYPE');
3460 wakaba 1.18 $self->{document}->manakai_compat_mode ('quirks');
3461 wakaba 1.85 ## Go to the "before html" insertion mode.
3462 wakaba 1.18 ## reprocess
3463     return;
3464 wakaba 1.57 } elsif ($token->{type} == CHARACTER_TOKEN) {
3465 wakaba 1.18 if ($token->{data} =~ s/^([\x09\x0A\x0B\x0C\x20]+)//) { # \x0D
3466     ## Ignore the token
3467 wakaba 1.26
3468 wakaba 1.18 unless (length $token->{data}) {
3469 wakaba 1.79
3470 wakaba 1.84 ## Stay in the insertion mode.
3471 wakaba 1.18 $token = $self->_get_next_token;
3472     redo INITIAL;
3473 wakaba 1.79 } else {
3474    
3475 wakaba 1.3 }
3476 wakaba 1.79 } else {
3477    
3478 wakaba 1.3 }
3479 wakaba 1.18
3480 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'no DOCTYPE');
3481 wakaba 1.18 $self->{document}->manakai_compat_mode ('quirks');
3482 wakaba 1.85 ## Go to the "before html" insertion mode.
3483 wakaba 1.18 ## reprocess
3484     return;
3485 wakaba 1.57 } elsif ($token->{type} == COMMENT_TOKEN) {
3486 wakaba 1.79
3487 wakaba 1.18 my $comment = $self->{document}->create_comment ($token->{data});
3488     $self->{document}->append_child ($comment);
3489    
3490 wakaba 1.84 ## Stay in the insertion mode.
3491 wakaba 1.18 $token = $self->_get_next_token;
3492     redo INITIAL;
3493     } else {
3494 wakaba 1.57 die "$0: $token->{type}: Unknown token type";
3495 wakaba 1.18 }
3496     } # INITIAL
3497 wakaba 1.79
3498     die "$0: _tree_construction_initial: This should be never reached";
3499 wakaba 1.3 } # _tree_construction_initial
3500    
3501     sub _tree_construction_root_element ($) {
3502     my $self = shift;
3503 wakaba 1.84
3504 wakaba 1.85 ## NOTE: "before html" insertion mode.
3505 wakaba 1.3
3506     B: {
3507 wakaba 1.57 if ($token->{type} == DOCTYPE_TOKEN) {
3508 wakaba 1.79
3509 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in html:#DOCTYPE');
3510 wakaba 1.3 ## Ignore the token
3511 wakaba 1.84 ## Stay in the insertion mode.
3512 wakaba 1.3 $token = $self->_get_next_token;
3513     redo B;
3514 wakaba 1.57 } elsif ($token->{type} == COMMENT_TOKEN) {
3515 wakaba 1.79
3516 wakaba 1.3 my $comment = $self->{document}->create_comment ($token->{data});
3517     $self->{document}->append_child ($comment);
3518 wakaba 1.84 ## Stay in the insertion mode.
3519 wakaba 1.3 $token = $self->_get_next_token;
3520     redo B;
3521 wakaba 1.57 } elsif ($token->{type} == CHARACTER_TOKEN) {
3522 wakaba 1.26 if ($token->{data} =~ s/^([\x09\x0A\x0B\x0C\x20]+)//) { # \x0D
3523     ## Ignore the token.
3524    
3525 wakaba 1.3 unless (length $token->{data}) {
3526 wakaba 1.79
3527 wakaba 1.84 ## Stay in the insertion mode.
3528 wakaba 1.3 $token = $self->_get_next_token;
3529     redo B;
3530 wakaba 1.79 } else {
3531    
3532 wakaba 1.3 }
3533 wakaba 1.79 } else {
3534    
3535 wakaba 1.3 }
3536 wakaba 1.63
3537     $self->{application_cache_selection}->(undef);
3538    
3539     #
3540     } elsif ($token->{type} == START_TAG_TOKEN) {
3541 wakaba 1.84 if ($token->{tag_name} eq 'html') {
3542     my $root_element;
3543 wakaba 1.79
3544 wakaba 1.84 $root_element = $self->{document}->create_element_ns
3545     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
3546    
3547     for my $attr_name (keys %{ $token->{attributes}}) {
3548     $root_element->set_attribute_ns (undef, [undef, $attr_name],
3549     $token->{attributes} ->{$attr_name}->{value});
3550     }
3551    
3552     $self->{document}->append_child ($root_element);
3553     push @{$self->{open_elements}}, [$root_element, 'html'];
3554    
3555     if ($token->{attributes}->{manifest}) {
3556    
3557     $self->{application_cache_selection}
3558     ->($token->{attributes}->{manifest}->{value});
3559     ## ISSUE: No relative reference resolution?
3560     } else {
3561    
3562     $self->{application_cache_selection}->(undef);
3563     }
3564    
3565     $token = $self->_get_next_token;
3566     return; ## Go to the "before head" insertion mode.
3567 wakaba 1.63 } else {
3568 wakaba 1.79
3569 wakaba 1.84 #
3570 wakaba 1.63 }
3571 wakaba 1.3 } elsif ({
3572 wakaba 1.57 END_TAG_TOKEN, 1,
3573     END_OF_FILE_TOKEN, 1,
3574 wakaba 1.3 }->{$token->{type}}) {
3575 wakaba 1.79
3576 wakaba 1.3 #
3577     } else {
3578 wakaba 1.57 die "$0: $token->{type}: Unknown token type";
3579 wakaba 1.3 }
3580 wakaba 1.63
3581 wakaba 1.84 my $root_element;
3582 wakaba 1.3 $root_element = $self->{document}->create_element_ns
3583     (q<http://www.w3.org/1999/xhtml>, [undef, 'html']);
3584    
3585 wakaba 1.84 $self->{document}->append_child ($root_element);
3586     push @{$self->{open_elements}}, [$root_element, 'html'];
3587    
3588     $self->{application_cache_selection}->(undef);
3589    
3590     ## NOTE: Reprocess the token.
3591     return; ## Go to the "before head" insertion mode.
3592    
3593     ## ISSUE: There is an issue in the spec
3594 wakaba 1.3 } # B
3595 wakaba 1.79
3596     die "$0: _tree_construction_root_element: This should never be reached";
3597 wakaba 1.3 } # _tree_construction_root_element
3598    
3599     sub _reset_insertion_mode ($) {
3600     my $self = shift;
3601    
3602     ## Step 1
3603     my $last;
3604    
3605     ## Step 2
3606     my $i = -1;
3607     my $node = $self->{open_elements}->[$i];
3608    
3609     ## Step 3
3610     S3: {
3611 wakaba 1.29 if ($self->{open_elements}->[0]->[0] eq $node->[0]) {
3612     $last = 1;
3613     if (defined $self->{inner_html_node}) {
3614     if ($self->{inner_html_node}->[1] eq 'td' or
3615     $self->{inner_html_node}->[1] eq 'th') {
3616 wakaba 1.79
3617 wakaba 1.29 #
3618     } else {
3619 wakaba 1.79
3620 wakaba 1.29 $node = $self->{inner_html_node};
3621     }
3622 wakaba 1.3 }
3623     }
3624    
3625     ## Step 4..13
3626     my $new_mode = {
3627 wakaba 1.56 select => IN_SELECT_IM,
3628 wakaba 1.83 ## NOTE: |option| and |optgroup| do not set
3629     ## insertion mode to "in select" by themselves.
3630 wakaba 1.56 td => IN_CELL_IM,
3631     th => IN_CELL_IM,
3632     tr => IN_ROW_IM,
3633     tbody => IN_TABLE_BODY_IM,
3634     thead => IN_TABLE_BODY_IM,
3635     tfoot => IN_TABLE_BODY_IM,
3636     caption => IN_CAPTION_IM,
3637     colgroup => IN_COLUMN_GROUP_IM,
3638     table => IN_TABLE_IM,
3639     head => IN_BODY_IM, # not in head!
3640     body => IN_BODY_IM,
3641     frameset => IN_FRAMESET_IM,
3642 wakaba 1.3 }->{$node->[1]};
3643     $self->{insertion_mode} = $new_mode and return if defined $new_mode;
3644    
3645     ## Step 14
3646     if ($node->[1] eq 'html') {
3647     unless (defined $self->{head_element}) {
3648 wakaba 1.79
3649 wakaba 1.56 $self->{insertion_mode} = BEFORE_HEAD_IM;
3650 wakaba 1.3 } else {
3651 wakaba 1.81 ## ISSUE: Can this state be reached?
3652 wakaba 1.79
3653 wakaba 1.56 $self->{insertion_mode} = AFTER_HEAD_IM;
3654 wakaba 1.3 }
3655     return;
3656 wakaba 1.79 } else {
3657    
3658 wakaba 1.3 }
3659    
3660     ## Step 15
3661 wakaba 1.56 $self->{insertion_mode} = IN_BODY_IM and return if $last;
3662 wakaba 1.3
3663     ## Step 16
3664     $i--;
3665     $node = $self->{open_elements}->[$i];
3666    
3667     ## Step 17
3668     redo S3;
3669     } # S3
3670 wakaba 1.79
3671     die "$0: _reset_insertion_mode: This line should never be reached";
3672 wakaba 1.3 } # _reset_insertion_mode
3673    
3674     sub _tree_construction_main ($) {
3675     my $self = shift;
3676    
3677 wakaba 1.1 my $active_formatting_elements = [];
3678    
3679     my $reconstruct_active_formatting_elements = sub { # MUST
3680     my $insert = shift;
3681    
3682     ## Step 1
3683     return unless @$active_formatting_elements;
3684    
3685     ## Step 3
3686     my $i = -1;
3687     my $entry = $active_formatting_elements->[$i];
3688    
3689     ## Step 2
3690     return if $entry->[0] eq '#marker';
3691 wakaba 1.3 for (@{$self->{open_elements}}) {
3692 wakaba 1.1 if ($entry->[0] eq $_->[0]) {
3693 wakaba 1.79
3694 wakaba 1.1 return;
3695     }
3696     }
3697    
3698     S4: {
3699     ## Step 4
3700     last S4 if $active_formatting_elements->[0]->[0] eq $entry->[0];
3701    
3702     ## Step 5
3703     $i--;
3704     $entry = $active_formatting_elements->[$i];
3705    
3706     ## Step 6
3707     if ($entry->[0] eq '#marker') {
3708 wakaba 1.79
3709 wakaba 1.1 #
3710     } else {
3711     my $in_open_elements;
3712 wakaba 1.3 OE: for (@{$self->{open_elements}}) {
3713 wakaba 1.1 if ($entry->[0] eq $_->[0]) {
3714 wakaba 1.79
3715 wakaba 1.1 $in_open_elements = 1;
3716     last OE;
3717     }
3718     }
3719     if ($in_open_elements) {
3720 wakaba 1.79
3721 wakaba 1.1 #
3722     } else {
3723 wakaba 1.81 ## NOTE: <!DOCTYPE HTML><p><b><i><u></p> <p>X
3724 wakaba 1.79
3725 wakaba 1.1 redo S4;
3726     }
3727     }
3728    
3729     ## Step 7
3730     $i++;
3731     $entry = $active_formatting_elements->[$i];
3732     } # S4
3733    
3734     S7: {
3735     ## Step 8
3736     my $clone = [$entry->[0]->clone_node (0), $entry->[1]];
3737    
3738     ## Step 9
3739     $insert->($clone->[0]);
3740 wakaba 1.3 push @{$self->{open_elements}}, $clone;
3741 wakaba 1.1
3742     ## Step 10
3743 wakaba 1.3 $active_formatting_elements->[$i] = $self->{open_elements}->[-1];
3744 wakaba 1.1
3745     ## Step 11
3746     unless ($clone->[0] eq $active_formatting_elements->[-1]->[0]) {
3747 wakaba 1.79
3748 wakaba 1.1 ## Step 7'
3749     $i++;
3750     $entry = $active_formatting_elements->[$i];
3751    
3752     redo S7;
3753     }
3754 wakaba 1.79
3755    
3756 wakaba 1.1 } # S7
3757     }; # $reconstruct_active_formatting_elements
3758    
3759     my $clear_up_to_marker = sub {
3760     for (reverse 0..$#$active_formatting_elements) {
3761     if ($active_formatting_elements->[$_]->[0] eq '#marker') {
3762 wakaba 1.79
3763 wakaba 1.1 splice @$active_formatting_elements, $_;
3764     return;
3765     }
3766     }
3767 wakaba 1.79
3768    
3769 wakaba 1.1 }; # $clear_up_to_marker
3770    
3771 wakaba 1.96 my $insert;
3772    
3773     my $parse_rcdata = sub ($) {
3774     my ($content_model_flag) = @_;
3775 wakaba 1.25
3776     ## Step 1
3777     my $start_tag_name = $token->{tag_name};
3778     my $el;
3779    
3780     $el = $self->{document}->create_element_ns
3781     (q<http://www.w3.org/1999/xhtml>, [undef, $start_tag_name]);
3782 wakaba 1.1
3783 wakaba 1.6 for my $attr_name (keys %{ $token->{attributes}}) {
3784 wakaba 1.25 $el->set_attribute_ns (undef, [undef, $attr_name],
3785 wakaba 1.6 $token->{attributes} ->{$attr_name}->{value});
3786     }
3787    
3788 wakaba 1.25
3789     ## Step 2
3790 wakaba 1.96 $insert->($el);
3791 wakaba 1.25
3792     ## Step 3
3793 wakaba 1.41 $self->{content_model} = $content_model_flag; # CDATA or RCDATA
3794 wakaba 1.13 delete $self->{escape}; # MUST
3795 wakaba 1.25
3796     ## Step 4
3797 wakaba 1.1 my $text = '';
3798     $token = $self->_get_next_token;
3799 wakaba 1.57 while ($token->{type} == CHARACTER_TOKEN) { # or until stop tokenizing
3800 wakaba 1.79
3801 wakaba 1.1 $text .= $token->{data};
3802     $token = $self->_get_next_token;
3803 wakaba 1.25 }
3804    
3805     ## Step 5
3806 wakaba 1.1 if (length $text) {
3807 wakaba 1.79
3808 wakaba 1.25 my $text = $self->{document}->create_text_node ($text);
3809     $el->append_child ($text);
3810 wakaba 1.1 }
3811 wakaba 1.25
3812     ## Step 6
3813 wakaba 1.41 $self->{content_model} = PCDATA_CONTENT_MODEL;
3814 wakaba 1.25
3815     ## Step 7
3816 wakaba 1.79 if ($token->{type} == END_TAG_TOKEN and
3817     $token->{tag_name} eq $start_tag_name) {
3818    
3819 wakaba 1.1 ## Ignore the token
3820     } else {
3821 wakaba 1.96 ## NOTE: An end-of-file token.
3822     if ($content_model_flag == CDATA_CONTENT_MODEL) {
3823    
3824 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in CDATA:#'.$token->{type});
3825 wakaba 1.96 } elsif ($content_model_flag == RCDATA_CONTENT_MODEL) {
3826    
3827 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in RCDATA:#'.$token->{type});
3828 wakaba 1.96 } else {
3829     die "$0: $content_model_flag in parse_rcdata";
3830     }
3831 wakaba 1.1 }
3832     $token = $self->_get_next_token;
3833 wakaba 1.25 }; # $parse_rcdata
3834 wakaba 1.1
3835 wakaba 1.96 my $script_start_tag = sub () {
3836 wakaba 1.1 my $script_el;
3837    
3838     $script_el = $self->{document}->create_element_ns
3839     (q<http://www.w3.org/1999/xhtml>, [undef, 'script']);
3840    
3841     for my $attr_name (keys %{ $token->{attributes}}) {
3842     $script_el->set_attribute_ns (undef, [undef, $attr_name],
3843     $token->{attributes} ->{$attr_name}->{value});
3844     }
3845    
3846     ## TODO: mark as "parser-inserted"
3847    
3848 wakaba 1.41 $self->{content_model} = CDATA_CONTENT_MODEL;
3849 wakaba 1.13 delete $self->{escape}; # MUST
3850 wakaba 1.1
3851     my $text = '';
3852     $token = $self->_get_next_token;
3853 wakaba 1.57 while ($token->{type} == CHARACTER_TOKEN) {
3854 wakaba 1.79
3855 wakaba 1.1 $text .= $token->{data};
3856     $token = $self->_get_next_token;
3857     } # stop if non-character token or tokenizer stops tokenising
3858     if (length $text) {
3859 wakaba 1.79
3860 wakaba 1.1 $script_el->manakai_append_text ($text);
3861     }
3862    
3863 wakaba 1.41 $self->{content_model} = PCDATA_CONTENT_MODEL;
3864 wakaba 1.1
3865 wakaba 1.57 if ($token->{type} == END_TAG_TOKEN and
3866 wakaba 1.1 $token->{tag_name} eq 'script') {
3867 wakaba 1.79
3868 wakaba 1.1 ## Ignore the token
3869     } else {
3870 wakaba 1.79
3871 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in CDATA:#'.$token->{type});
3872 wakaba 1.1 ## ISSUE: And ignore?
3873     ## TODO: mark as "already executed"
3874     }
3875    
3876 wakaba 1.3 if (defined $self->{inner_html_node}) {
3877 wakaba 1.79
3878 wakaba 1.3 ## TODO: mark as "already executed"
3879     } else {
3880 wakaba 1.79
3881 wakaba 1.1 ## TODO: $old_insertion_point = current insertion point
3882     ## TODO: insertion point = just before the next input character
3883 wakaba 1.25
3884     $insert->($script_el);
3885 wakaba 1.1
3886     ## TODO: insertion point = $old_insertion_point (might be "undefined")
3887    
3888     ## TODO: if there is a script that will execute as soon as the parser resume, then...
3889     }
3890    
3891     $token = $self->_get_next_token;
3892     }; # $script_start_tag
3893    
3894 wakaba 1.102 ## NOTE: $open_tables->[-1]->[0] is the "current table" element node.
3895     ## NOTE: $open_tables->[-1]->[1] is the "tainted" flag.
3896     my $open_tables = [[$self->{open_elements}->[0]->[0]]];
3897    
3898 wakaba 1.1 my $formatting_end_tag = sub {
3899     my $tag_name = shift;
3900    
3901 wakaba 1.102 ## NOTE: The adoption agency algorithm (AAA).
3902    
3903 wakaba 1.1 FET: {
3904     ## Step 1
3905     my $formatting_element;
3906     my $formatting_element_i_in_active;
3907     AFE: for (reverse 0..$#$active_formatting_elements) {
3908     if ($active_formatting_elements->[$_]->[1] eq $tag_name) {
3909 wakaba 1.79
3910 wakaba 1.1 $formatting_element = $active_formatting_elements->[$_];
3911     $formatting_element_i_in_active = $_;
3912     last AFE;
3913     } elsif ($active_formatting_elements->[$_]->[0] eq '#marker') {
3914 wakaba 1.79
3915 wakaba 1.1 last AFE;
3916     }
3917     } # AFE
3918     unless (defined $formatting_element) {
3919 wakaba 1.79
3920 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$tag_name);
3921 wakaba 1.1 ## Ignore the token
3922     $token = $self->_get_next_token;
3923     return;
3924     }
3925     ## has an element in scope
3926     my $in_scope = 1;
3927     my $formatting_element_i_in_open;
3928 wakaba 1.3 INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
3929     my $node = $self->{open_elements}->[$_];
3930 wakaba 1.1 if ($node->[0] eq $formatting_element->[0]) {
3931     if ($in_scope) {
3932 wakaba 1.79
3933 wakaba 1.1 $formatting_element_i_in_open = $_;
3934     last INSCOPE;
3935     } else { # in open elements but not in scope
3936 wakaba 1.79
3937 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
3938 wakaba 1.1 ## Ignore the token
3939     $token = $self->_get_next_token;
3940     return;
3941     }
3942     } elsif ({
3943 wakaba 1.102 applet => 1, table => 1, caption => 1, td => 1, th => 1,
3944 wakaba 1.1 button => 1, marquee => 1, object => 1, html => 1,
3945     }->{$node->[1]}) {
3946 wakaba 1.79
3947 wakaba 1.1 $in_scope = 0;
3948     }
3949     } # INSCOPE
3950     unless (defined $formatting_element_i_in_open) {
3951 wakaba 1.79
3952 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
3953 wakaba 1.1 pop @$active_formatting_elements; # $formatting_element
3954     $token = $self->_get_next_token; ## TODO: ok?
3955     return;
3956     }
3957 wakaba 1.3 if (not $self->{open_elements}->[-1]->[0] eq $formatting_element->[0]) {
3958 wakaba 1.79
3959 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not closed:'.$self->{open_elements}->[-1]->[1]);
3960 wakaba 1.1 }
3961    
3962     ## Step 2
3963     my $furthest_block;
3964     my $furthest_block_i_in_open;
3965 wakaba 1.3 OE: for (reverse 0..$#{$self->{open_elements}}) {
3966     my $node = $self->{open_elements}->[$_];
3967 wakaba 1.1 if (not $formatting_category->{$node->[1]} and
3968     #not $phrasing_category->{$node->[1]} and
3969     ($special_category->{$node->[1]} or
3970 wakaba 1.102 $scoping_category->{$node->[1]})) { ## Scoping is redundant, maybe
3971 wakaba 1.79
3972 wakaba 1.1 $furthest_block = $node;
3973     $furthest_block_i_in_open = $_;
3974     } elsif ($node->[0] eq $formatting_element->[0]) {
3975 wakaba 1.79
3976 wakaba 1.1 last OE;
3977     }
3978     } # OE
3979    
3980     ## Step 3
3981     unless (defined $furthest_block) { # MUST
3982 wakaba 1.79
3983 wakaba 1.3 splice @{$self->{open_elements}}, $formatting_element_i_in_open;
3984 wakaba 1.1 splice @$active_formatting_elements, $formatting_element_i_in_active, 1;
3985     $token = $self->_get_next_token;
3986     return;
3987     }
3988    
3989     ## Step 4
3990 wakaba 1.3 my $common_ancestor_node = $self->{open_elements}->[$formatting_element_i_in_open - 1];
3991 wakaba 1.1
3992     ## Step 5
3993     my $furthest_block_parent = $furthest_block->[0]->parent_node;
3994     if (defined $furthest_block_parent) {
3995 wakaba 1.79
3996 wakaba 1.1 $furthest_block_parent->remove_child ($furthest_block->[0]);
3997     }
3998    
3999     ## Step 6
4000     my $bookmark_prev_el
4001     = $active_formatting_elements->[$formatting_element_i_in_active - 1]
4002     ->[0];
4003    
4004     ## Step 7
4005     my $node = $furthest_block;
4006     my $node_i_in_open = $furthest_block_i_in_open;
4007     my $last_node = $furthest_block;
4008     S7: {
4009     ## Step 1
4010     $node_i_in_open--;
4011 wakaba 1.3 $node = $self->{open_elements}->[$node_i_in_open];
4012 wakaba 1.1
4013     ## Step 2
4014     my $node_i_in_active;
4015     S7S2: {
4016     for (reverse 0..$#$active_formatting_elements) {
4017     if ($active_formatting_elements->[$_]->[0] eq $node->[0]) {
4018 wakaba 1.79
4019 wakaba 1.1 $node_i_in_active = $_;
4020     last S7S2;
4021     }
4022     }
4023 wakaba 1.3 splice @{$self->{open_elements}}, $node_i_in_open, 1;
4024 wakaba 1.1 redo S7;
4025     } # S7S2
4026    
4027     ## Step 3
4028     last S7 if $node->[0] eq $formatting_element->[0];
4029    
4030     ## Step 4
4031     if ($last_node->[0] eq $furthest_block->[0]) {
4032 wakaba 1.79
4033 wakaba 1.1 $bookmark_prev_el = $node->[0];
4034     }
4035    
4036     ## Step 5
4037     if ($node->[0]->has_child_nodes ()) {
4038 wakaba 1.79
4039 wakaba 1.1 my $clone = [$node->[0]->clone_node (0), $node->[1]];
4040     $active_formatting_elements->[$node_i_in_active] = $clone;
4041 wakaba 1.3 $self->{open_elements}->[$node_i_in_open] = $clone;
4042 wakaba 1.1 $node = $clone;
4043     }
4044    
4045     ## Step 6
4046     $node->[0]->append_child ($last_node->[0]);
4047    
4048     ## Step 7
4049     $last_node = $node;
4050    
4051     ## Step 8
4052     redo S7;
4053     } # S7
4054    
4055     ## Step 8
4056 wakaba 1.102 if ({
4057     table => 1, tbody => 1, tfoot => 1, thead => 1, tr => 1,
4058     }->{$common_ancestor_node->[1]}) {
4059     my $foster_parent_element;
4060     my $next_sibling;
4061     OE: for (reverse 0..$#{$self->{open_elements}}) {
4062     if ($self->{open_elements}->[$_]->[1] eq 'table') {
4063     my $parent = $self->{open_elements}->[$_]->[0]->parent_node;
4064     if (defined $parent and $parent->node_type == 1) {
4065    
4066     $foster_parent_element = $parent;
4067     $next_sibling = $self->{open_elements}->[$_]->[0];
4068     } else {
4069    
4070     $foster_parent_element
4071     = $self->{open_elements}->[$_ - 1]->[0];
4072     }
4073     last OE;
4074     }
4075     } # OE
4076     $foster_parent_element = $self->{open_elements}->[0]->[0]
4077     unless defined $foster_parent_element;
4078     $foster_parent_element->insert_before ($last_node->[0], $next_sibling);
4079     $open_tables->[-1]->[1] = 1; # tainted
4080     } else {
4081    
4082     $common_ancestor_node->[0]->append_child ($last_node->[0]);
4083     }
4084 wakaba 1.1
4085     ## Step 9
4086     my $clone = [$formatting_element->[0]->clone_node (0),
4087     $formatting_element->[1]];
4088    
4089     ## Step 10
4090     my @cn = @{$furthest_block->[0]->child_nodes};
4091     $clone->[0]->append_child ($_) for @cn;
4092    
4093     ## Step 11
4094     $furthest_block->[0]->append_child ($clone->[0]);
4095    
4096     ## Step 12
4097     my $i;
4098     AFE: for (reverse 0..$#$active_formatting_elements) {
4099     if ($active_formatting_elements->[$_]->[0] eq $formatting_element->[0]) {
4100 wakaba 1.79
4101 wakaba 1.1 splice @$active_formatting_elements, $_, 1;
4102     $i-- and last AFE if defined $i;
4103     } elsif ($active_formatting_elements->[$_]->[0] eq $bookmark_prev_el) {
4104 wakaba 1.79
4105 wakaba 1.1 $i = $_;
4106     }
4107     } # AFE
4108     splice @$active_formatting_elements, $i + 1, 0, $clone;
4109    
4110     ## Step 13
4111     undef $i;
4112 wakaba 1.3 OE: for (reverse 0..$#{$self->{open_elements}}) {
4113     if ($self->{open_elements}->[$_]->[0] eq $formatting_element->[0]) {
4114 wakaba 1.79
4115 wakaba 1.3 splice @{$self->{open_elements}}, $_, 1;
4116 wakaba 1.1 $i-- and last OE if defined $i;
4117 wakaba 1.3 } elsif ($self->{open_elements}->[$_]->[0] eq $furthest_block->[0]) {
4118 wakaba 1.79
4119 wakaba 1.1 $i = $_;
4120     }
4121     } # OE
4122 wakaba 1.3 splice @{$self->{open_elements}}, $i + 1, 1, $clone;
4123 wakaba 1.1
4124     ## Step 14
4125     redo FET;
4126     } # FET
4127     }; # $formatting_end_tag
4128    
4129 wakaba 1.96 $insert = my $insert_to_current = sub {
4130 wakaba 1.25 $self->{open_elements}->[-1]->[0]->append_child ($_[0]);
4131 wakaba 1.1 }; # $insert_to_current
4132    
4133     my $insert_to_foster = sub {
4134 wakaba 1.95 my $child = shift;
4135     if ({
4136     table => 1, tbody => 1, tfoot => 1, thead => 1, tr => 1,
4137     }->{$self->{open_elements}->[-1]->[1]}) {
4138     # MUST
4139     my $foster_parent_element;
4140     my $next_sibling;
4141 wakaba 1.3 OE: for (reverse 0..$#{$self->{open_elements}}) {
4142     if ($self->{open_elements}->[$_]->[1] eq 'table') {
4143     my $parent = $self->{open_elements}->[$_]->[0]->parent_node;
4144 wakaba 1.1 if (defined $parent and $parent->node_type == 1) {
4145 wakaba 1.79
4146 wakaba 1.1 $foster_parent_element = $parent;
4147 wakaba 1.3 $next_sibling = $self->{open_elements}->[$_]->[0];
4148 wakaba 1.1 } else {
4149 wakaba 1.79
4150 wakaba 1.1 $foster_parent_element
4151 wakaba 1.3 = $self->{open_elements}->[$_ - 1]->[0];
4152 wakaba 1.1 }
4153     last OE;
4154     }
4155     } # OE
4156 wakaba 1.3 $foster_parent_element = $self->{open_elements}->[0]->[0]
4157 wakaba 1.1 unless defined $foster_parent_element;
4158     $foster_parent_element->insert_before
4159     ($child, $next_sibling);
4160 wakaba 1.95 $open_tables->[-1]->[1] = 1; # tainted
4161     } else {
4162    
4163     $self->{open_elements}->[-1]->[0]->append_child ($child);
4164     }
4165 wakaba 1.1 }; # $insert_to_foster
4166    
4167 wakaba 1.54 B: {
4168 wakaba 1.57 if ($token->{type} == DOCTYPE_TOKEN) {
4169 wakaba 1.79
4170 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'DOCTYPE in the middle');
4171 wakaba 1.54 ## Ignore the token
4172     ## Stay in the phase
4173     $token = $self->_get_next_token;
4174     redo B;
4175 wakaba 1.57 } elsif ($token->{type} == START_TAG_TOKEN and
4176 wakaba 1.54 $token->{tag_name} eq 'html') {
4177 wakaba 1.56 if ($self->{insertion_mode} == AFTER_HTML_BODY_IM) {
4178 wakaba 1.79
4179 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after html:html');
4180 wakaba 1.56 $self->{insertion_mode} = AFTER_BODY_IM;
4181     } elsif ($self->{insertion_mode} == AFTER_HTML_FRAMESET_IM) {
4182 wakaba 1.79
4183 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after html:html');
4184 wakaba 1.56 $self->{insertion_mode} = AFTER_FRAMESET_IM;
4185 wakaba 1.79 } else {
4186    
4187 wakaba 1.54 }
4188    
4189 wakaba 1.84
4190 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not first start tag');
4191 wakaba 1.54 my $top_el = $self->{open_elements}->[0]->[0];
4192     for my $attr_name (keys %{$token->{attributes}}) {
4193     unless ($top_el->has_attribute_ns (undef, $attr_name)) {
4194 wakaba 1.79
4195 wakaba 1.54 $top_el->set_attribute_ns
4196     (undef, [undef, $attr_name],
4197     $token->{attributes}->{$attr_name}->{value});
4198     }
4199     }
4200     $token = $self->_get_next_token;
4201     redo B;
4202 wakaba 1.57 } elsif ($token->{type} == COMMENT_TOKEN) {
4203 wakaba 1.54 my $comment = $self->{document}->create_comment ($token->{data});
4204 wakaba 1.58 if ($self->{insertion_mode} & AFTER_HTML_IMS) {
4205 wakaba 1.79
4206 wakaba 1.54 $self->{document}->append_child ($comment);
4207 wakaba 1.56 } elsif ($self->{insertion_mode} == AFTER_BODY_IM) {
4208 wakaba 1.79
4209 wakaba 1.54 $self->{open_elements}->[0]->[0]->append_child ($comment);
4210     } else {
4211 wakaba 1.79
4212 wakaba 1.54 $self->{open_elements}->[-1]->[0]->append_child ($comment);
4213     }
4214     $token = $self->_get_next_token;
4215     redo B;
4216 wakaba 1.58 } elsif ($self->{insertion_mode} & HEAD_IMS) {
4217 wakaba 1.57 if ($token->{type} == CHARACTER_TOKEN) {
4218 wakaba 1.54 if ($token->{data} =~ s/^([\x09\x0A\x0B\x0C\x20]+)//) {
4219 wakaba 1.99 unless ($self->{insertion_mode} == BEFORE_HEAD_IM) {
4220    
4221     $self->{open_elements}->[-1]->[0]->manakai_append_text ($1);
4222     } else {
4223    
4224     ## Ignore the token.
4225     $token = $self->_get_next_token;
4226     redo B;
4227     }
4228 wakaba 1.54 unless (length $token->{data}) {
4229 wakaba 1.79
4230 wakaba 1.54 $token = $self->_get_next_token;
4231     redo B;
4232     }
4233     }
4234    
4235 wakaba 1.56 if ($self->{insertion_mode} == BEFORE_HEAD_IM) {
4236 wakaba 1.79
4237 wakaba 1.54 ## As if <head>
4238    
4239     $self->{head_element} = $self->{document}->create_element_ns
4240     (q<http://www.w3.org/1999/xhtml>, [undef, 'head']);
4241    
4242     $self->{open_elements}->[-1]->[0]->append_child ($self->{head_element});
4243     push @{$self->{open_elements}}, [$self->{head_element}, 'head'];
4244    
4245     ## Reprocess in the "in head" insertion mode...
4246     pop @{$self->{open_elements}};
4247    
4248     ## Reprocess in the "after head" insertion mode...
4249 wakaba 1.56 } elsif ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) {
4250 wakaba 1.79
4251 wakaba 1.54 ## As if </noscript>
4252     pop @{$self->{open_elements}};
4253 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in noscript:#character');
4254 wakaba 1.54
4255     ## Reprocess in the "in head" insertion mode...
4256     ## As if </head>
4257     pop @{$self->{open_elements}};
4258    
4259     ## Reprocess in the "after head" insertion mode...
4260 wakaba 1.56 } elsif ($self->{insertion_mode} == IN_HEAD_IM) {
4261 wakaba 1.79
4262 wakaba 1.54 pop @{$self->{open_elements}};
4263    
4264     ## Reprocess in the "after head" insertion mode...
4265 wakaba 1.79 } else {
4266    
4267 wakaba 1.54 }
4268    
4269     ## "after head" insertion mode
4270     ## As if <body>
4271    
4272     {
4273     my $el;
4274    
4275     $el = $self->{document}->create_element_ns
4276     (q<http://www.w3.org/1999/xhtml>, [undef, 'body']);
4277    
4278     $self->{open_elements}->[-1]->[0]->append_child ($el);
4279     push @{$self->{open_elements}}, [$el, 'body'];
4280     }
4281    
4282 wakaba 1.56 $self->{insertion_mode} = IN_BODY_IM;
4283 wakaba 1.54 ## reprocess
4284     redo B;
4285 wakaba 1.57 } elsif ($token->{type} == START_TAG_TOKEN) {
4286 wakaba 1.54 if ($token->{tag_name} eq 'head') {
4287 wakaba 1.56 if ($self->{insertion_mode} == BEFORE_HEAD_IM) {
4288 wakaba 1.54
4289 wakaba 1.79
4290 wakaba 1.54 $self->{head_element} = $self->{document}->create_element_ns
4291     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
4292    
4293     for my $attr_name (keys %{ $token->{attributes}}) {
4294     $self->{head_element}->set_attribute_ns (undef, [undef, $attr_name],
4295     $token->{attributes} ->{$attr_name}->{value});
4296     }
4297    
4298     $self->{open_elements}->[-1]->[0]->append_child ($self->{head_element});
4299     push @{$self->{open_elements}}, [$self->{head_element}, $token->{tag_name}];
4300 wakaba 1.56 $self->{insertion_mode} = IN_HEAD_IM;
4301 wakaba 1.54 $token = $self->_get_next_token;
4302     redo B;
4303 wakaba 1.56 } elsif ($self->{insertion_mode} == AFTER_HEAD_IM) {
4304 wakaba 1.79
4305 wakaba 1.56 #
4306     } else {
4307 wakaba 1.79
4308 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in head:head'); # or in head noscript
4309 wakaba 1.54 ## Ignore the token
4310     $token = $self->_get_next_token;
4311     redo B;
4312     }
4313 wakaba 1.56 } elsif ($self->{insertion_mode} == BEFORE_HEAD_IM) {
4314 wakaba 1.79
4315 wakaba 1.54 ## As if <head>
4316    
4317     $self->{head_element} = $self->{document}->create_element_ns
4318     (q<http://www.w3.org/1999/xhtml>, [undef, 'head']);
4319    
4320     $self->{open_elements}->[-1]->[0]->append_child ($self->{head_element});
4321     push @{$self->{open_elements}}, [$self->{head_element}, 'head'];
4322    
4323 wakaba 1.56 $self->{insertion_mode} = IN_HEAD_IM;
4324 wakaba 1.54 ## Reprocess in the "in head" insertion mode...
4325 wakaba 1.79 } else {
4326    
4327 wakaba 1.54 }
4328    
4329     if ($token->{tag_name} eq 'base') {
4330 wakaba 1.56 if ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) {
4331 wakaba 1.79
4332 wakaba 1.54 ## As if </noscript>
4333     pop @{$self->{open_elements}};
4334 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in noscript:base');
4335 wakaba 1.54
4336 wakaba 1.56 $self->{insertion_mode} = IN_HEAD_IM;
4337 wakaba 1.54 ## Reprocess in the "in head" insertion mode...
4338 wakaba 1.79 } else {
4339    
4340 wakaba 1.54 }
4341    
4342     ## NOTE: There is a "as if in head" code clone.
4343 wakaba 1.56 if ($self->{insertion_mode} == AFTER_HEAD_IM) {
4344 wakaba 1.79
4345 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after head:'.$token->{tag_name});
4346 wakaba 1.54 push @{$self->{open_elements}}, [$self->{head_element}, 'head'];
4347 wakaba 1.79 } else {
4348    
4349 wakaba 1.54 }
4350    
4351     {
4352     my $el;
4353    
4354     $el = $self->{document}->create_element_ns
4355     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
4356    
4357     for my $attr_name (keys %{ $token->{attributes}}) {
4358     $el->set_attribute_ns (undef, [undef, $attr_name],
4359     $token->{attributes} ->{$attr_name}->{value});
4360     }
4361    
4362     $self->{open_elements}->[-1]->[0]->append_child ($el);
4363     push @{$self->{open_elements}}, [$el, $token->{tag_name}];
4364     }
4365    
4366     pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec.
4367 wakaba 1.100 pop @{$self->{open_elements}} # <head>
4368 wakaba 1.56 if $self->{insertion_mode} == AFTER_HEAD_IM;
4369 wakaba 1.54 $token = $self->_get_next_token;
4370     redo B;
4371     } elsif ($token->{tag_name} eq 'link') {
4372     ## NOTE: There is a "as if in head" code clone.
4373 wakaba 1.56 if ($self->{insertion_mode} == AFTER_HEAD_IM) {
4374 wakaba 1.79
4375 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after head:'.$token->{tag_name});
4376 wakaba 1.54 push @{$self->{open_elements}}, [$self->{head_element}, 'head'];
4377 wakaba 1.79 } else {
4378    
4379 wakaba 1.54 }
4380    
4381 wakaba 1.25 {
4382     my $el;
4383    
4384 wakaba 1.1 $el = $self->{document}->create_element_ns
4385     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
4386    
4387 wakaba 1.25 for my $attr_name (keys %{ $token->{attributes}}) {
4388 wakaba 1.1 $el->set_attribute_ns (undef, [undef, $attr_name],
4389 wakaba 1.25 $token->{attributes} ->{$attr_name}->{value});
4390 wakaba 1.1 }
4391    
4392 wakaba 1.54 $self->{open_elements}->[-1]->[0]->append_child ($el);
4393 wakaba 1.25 push @{$self->{open_elements}}, [$el, $token->{tag_name}];
4394     }
4395    
4396 wakaba 1.54 pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec.
4397 wakaba 1.100 pop @{$self->{open_elements}} # <head>
4398 wakaba 1.56 if $self->{insertion_mode} == AFTER_HEAD_IM;
4399 wakaba 1.54 $token = $self->_get_next_token;
4400     redo B;
4401     } elsif ($token->{tag_name} eq 'meta') {
4402     ## NOTE: There is a "as if in head" code clone.
4403 wakaba 1.56 if ($self->{insertion_mode} == AFTER_HEAD_IM) {
4404 wakaba 1.79
4405 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after head:'.$token->{tag_name});
4406 wakaba 1.54 push @{$self->{open_elements}}, [$self->{head_element}, 'head'];
4407 wakaba 1.79 } else {
4408    
4409 wakaba 1.54 }
4410    
4411 wakaba 1.34 {
4412     my $el;
4413    
4414     $el = $self->{document}->create_element_ns
4415     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
4416    
4417     for my $attr_name (keys %{ $token->{attributes}}) {
4418     $el->set_attribute_ns (undef, [undef, $attr_name],
4419     $token->{attributes} ->{$attr_name}->{value});
4420     }
4421    
4422 wakaba 1.54 $self->{open_elements}->[-1]->[0]->append_child ($el);
4423 wakaba 1.34 push @{$self->{open_elements}}, [$el, $token->{tag_name}];
4424     }
4425    
4426 wakaba 1.68 my $meta_el = pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec.
4427 wakaba 1.34
4428 wakaba 1.54 unless ($self->{confident}) {
4429     if ($token->{attributes}->{charset}) { ## TODO: And if supported
4430 wakaba 1.79
4431 wakaba 1.65 $self->{change_encoding}
4432     ->($self, $token->{attributes}->{charset}->{value});
4433 wakaba 1.68
4434     $meta_el->[0]->get_attribute_node_ns (undef, 'charset')
4435     ->set_user_data (manakai_has_reference =>
4436     $token->{attributes}->{charset}
4437     ->{has_reference});
4438 wakaba 1.65 } elsif ($token->{attributes}->{content}) {
4439 wakaba 1.54 ## ISSUE: Algorithm name in the spec was incorrect so that not linked to the definition.
4440 wakaba 1.65 if ($token->{attributes}->{content}->{value}
4441 wakaba 1.71 =~ /\A[^;]*;[\x09-\x0D\x20]*[Cc][Hh][Aa][Rr][Ss][Ee][Tt]
4442     [\x09-\x0D\x20]*=
4443 wakaba 1.54 [\x09-\x0D\x20]*(?>"([^"]*)"|'([^']*)'|
4444     ([^"'\x09-\x0D\x20][^\x09-\x0D\x20]*))/x) {
4445 wakaba 1.79
4446 wakaba 1.65 $self->{change_encoding}
4447     ->($self, defined $1 ? $1 : defined $2 ? $2 : $3);
4448 wakaba 1.69 $meta_el->[0]->get_attribute_node_ns (undef, 'content')
4449     ->set_user_data (manakai_has_reference =>
4450     $token->{attributes}->{content}
4451     ->{has_reference});
4452 wakaba 1.79 } else {
4453    
4454 wakaba 1.65 }
4455 wakaba 1.54 }
4456 wakaba 1.68 } else {
4457     if ($token->{attributes}->{charset}) {
4458 wakaba 1.79
4459 wakaba 1.68 $meta_el->[0]->get_attribute_node_ns (undef, 'charset')
4460     ->set_user_data (manakai_has_reference =>
4461     $token->{attributes}->{charset}
4462     ->{has_reference});
4463     }
4464 wakaba 1.69 if ($token->{attributes}->{content}) {
4465 wakaba 1.79
4466 wakaba 1.69 $meta_el->[0]->get_attribute_node_ns (undef, 'content')
4467     ->set_user_data (manakai_has_reference =>
4468     $token->{attributes}->{content}
4469     ->{has_reference});
4470     }
4471 wakaba 1.54 }
4472 wakaba 1.34
4473 wakaba 1.100 pop @{$self->{open_elements}} # <head>
4474 wakaba 1.56 if $self->{insertion_mode} == AFTER_HEAD_IM;
4475 wakaba 1.54 $token = $self->_get_next_token;
4476     redo B;
4477     } elsif ($token->{tag_name} eq 'title') {
4478 wakaba 1.56 if ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) {
4479 wakaba 1.79
4480 wakaba 1.54 ## As if </noscript>
4481     pop @{$self->{open_elements}};
4482 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in noscript:title');
4483 wakaba 1.1
4484 wakaba 1.56 $self->{insertion_mode} = IN_HEAD_IM;
4485 wakaba 1.54 ## Reprocess in the "in head" insertion mode...
4486 wakaba 1.56 } elsif ($self->{insertion_mode} == AFTER_HEAD_IM) {
4487 wakaba 1.79
4488 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after head:'.$token->{tag_name});
4489 wakaba 1.54 push @{$self->{open_elements}}, [$self->{head_element}, 'head'];
4490 wakaba 1.79 } else {
4491    
4492 wakaba 1.54 }
4493    
4494     ## NOTE: There is a "as if in head" code clone.
4495     my $parent = defined $self->{head_element} ? $self->{head_element}
4496     : $self->{open_elements}->[-1]->[0];
4497 wakaba 1.96 $parse_rcdata->(RCDATA_CONTENT_MODEL);
4498 wakaba 1.100 pop @{$self->{open_elements}} # <head>
4499 wakaba 1.56 if $self->{insertion_mode} == AFTER_HEAD_IM;
4500 wakaba 1.54 redo B;
4501     } elsif ($token->{tag_name} eq 'style') {
4502     ## NOTE: Or (scripting is enabled and tag_name eq 'noscript' and
4503 wakaba 1.56 ## insertion mode IN_HEAD_IM)
4504 wakaba 1.54 ## NOTE: There is a "as if in head" code clone.
4505 wakaba 1.56 if ($self->{insertion_mode} == AFTER_HEAD_IM) {
4506 wakaba 1.79
4507 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after head:'.$token->{tag_name});
4508 wakaba 1.54 push @{$self->{open_elements}}, [$self->{head_element}, 'head'];
4509 wakaba 1.79 } else {
4510    
4511 wakaba 1.54 }
4512 wakaba 1.96 $parse_rcdata->(CDATA_CONTENT_MODEL);
4513 wakaba 1.100 pop @{$self->{open_elements}} # <head>
4514 wakaba 1.56 if $self->{insertion_mode} == AFTER_HEAD_IM;
4515 wakaba 1.54 redo B;
4516     } elsif ($token->{tag_name} eq 'noscript') {
4517 wakaba 1.56 if ($self->{insertion_mode} == IN_HEAD_IM) {
4518 wakaba 1.79
4519 wakaba 1.54 ## NOTE: and scripting is disalbed
4520    
4521 wakaba 1.1 {
4522     my $el;
4523    
4524     $el = $self->{document}->create_element_ns
4525     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
4526    
4527     for my $attr_name (keys %{ $token->{attributes}}) {
4528     $el->set_attribute_ns (undef, [undef, $attr_name],
4529     $token->{attributes} ->{$attr_name}->{value});
4530     }
4531    
4532 wakaba 1.54 $self->{open_elements}->[-1]->[0]->append_child ($el);
4533 wakaba 1.3 push @{$self->{open_elements}}, [$el, $token->{tag_name}];
4534 wakaba 1.1 }
4535    
4536 wakaba 1.56 $self->{insertion_mode} = IN_HEAD_NOSCRIPT_IM;
4537 wakaba 1.54 $token = $self->_get_next_token;
4538     redo B;
4539 wakaba 1.56 } elsif ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) {
4540 wakaba 1.79
4541 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in noscript:noscript');
4542 wakaba 1.54 ## Ignore the token
4543     $token = $self->_get_next_token;
4544     redo B;
4545     } else {
4546 wakaba 1.79
4547 wakaba 1.54 #
4548     }
4549     } elsif ($token->{tag_name} eq 'script') {
4550 wakaba 1.56 if ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) {
4551 wakaba 1.79
4552 wakaba 1.54 ## As if </noscript>
4553     pop @{$self->{open_elements}};
4554 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in noscript:script');
4555 wakaba 1.54
4556 wakaba 1.56 $self->{insertion_mode} = IN_HEAD_IM;
4557 wakaba 1.54 ## Reprocess in the "in head" insertion mode...
4558 wakaba 1.56 } elsif ($self->{insertion_mode} == AFTER_HEAD_IM) {
4559 wakaba 1.79
4560 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after head:'.$token->{tag_name});
4561 wakaba 1.54 push @{$self->{open_elements}}, [$self->{head_element}, 'head'];
4562 wakaba 1.79 } else {
4563    
4564 wakaba 1.54 }
4565    
4566     ## NOTE: There is a "as if in head" code clone.
4567 wakaba 1.100 $script_start_tag->();
4568     pop @{$self->{open_elements}} # <head>
4569 wakaba 1.56 if $self->{insertion_mode} == AFTER_HEAD_IM;
4570 wakaba 1.54 redo B;
4571     } elsif ($token->{tag_name} eq 'body' or
4572     $token->{tag_name} eq 'frameset') {
4573 wakaba 1.56 if ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) {
4574 wakaba 1.79
4575 wakaba 1.54 ## As if </noscript>
4576     pop @{$self->{open_elements}};
4577 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in noscript:'.$token->{tag_name});
4578 wakaba 1.54
4579     ## Reprocess in the "in head" insertion mode...
4580     ## As if </head>
4581     pop @{$self->{open_elements}};
4582    
4583     ## Reprocess in the "after head" insertion mode...
4584 wakaba 1.56 } elsif ($self->{insertion_mode} == IN_HEAD_IM) {
4585 wakaba 1.79
4586 wakaba 1.54 pop @{$self->{open_elements}};
4587    
4588     ## Reprocess in the "after head" insertion mode...
4589 wakaba 1.79 } else {
4590    
4591 wakaba 1.54 }
4592    
4593     ## "after head" insertion mode
4594    
4595 wakaba 1.1 {
4596     my $el;
4597    
4598     $el = $self->{document}->create_element_ns
4599     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
4600    
4601     for my $attr_name (keys %{ $token->{attributes}}) {
4602     $el->set_attribute_ns (undef, [undef, $attr_name],
4603     $token->{attributes} ->{$attr_name}->{value});
4604     }
4605    
4606 wakaba 1.54 $self->{open_elements}->[-1]->[0]->append_child ($el);
4607 wakaba 1.3 push @{$self->{open_elements}}, [$el, $token->{tag_name}];
4608 wakaba 1.1 }
4609    
4610 wakaba 1.56 if ($token->{tag_name} eq 'body') {
4611 wakaba 1.79
4612 wakaba 1.56 $self->{insertion_mode} = IN_BODY_IM;
4613     } elsif ($token->{tag_name} eq 'frameset') {
4614 wakaba 1.79
4615 wakaba 1.56 $self->{insertion_mode} = IN_FRAMESET_IM;
4616     } else {
4617     die "$0: tag name: $self->{tag_name}";
4618     }
4619 wakaba 1.54 $token = $self->_get_next_token;
4620     redo B;
4621     } else {
4622 wakaba 1.79
4623 wakaba 1.54 #
4624 wakaba 1.8 }
4625 wakaba 1.54
4626 wakaba 1.56 if ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) {
4627 wakaba 1.79
4628 wakaba 1.54 ## As if </noscript>
4629     pop @{$self->{open_elements}};
4630 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in noscript:/'.$token->{tag_name});
4631 wakaba 1.54
4632     ## Reprocess in the "in head" insertion mode...
4633     ## As if </head>
4634     pop @{$self->{open_elements}};
4635    
4636     ## Reprocess in the "after head" insertion mode...
4637 wakaba 1.56 } elsif ($self->{insertion_mode} == IN_HEAD_IM) {
4638 wakaba 1.79
4639 wakaba 1.54 ## As if </head>
4640     pop @{$self->{open_elements}};
4641    
4642     ## Reprocess in the "after head" insertion mode...
4643 wakaba 1.79 } else {
4644    
4645 wakaba 1.54 }
4646    
4647     ## "after head" insertion mode
4648     ## As if <body>
4649    
4650 wakaba 1.1 {
4651     my $el;
4652    
4653     $el = $self->{document}->create_element_ns
4654 wakaba 1.54 (q<http://www.w3.org/1999/xhtml>, [undef, 'body']);
4655 wakaba 1.1
4656 wakaba 1.54 $self->{open_elements}->[-1]->[0]->append_child ($el);
4657     push @{$self->{open_elements}}, [$el, 'body'];
4658 wakaba 1.1 }
4659    
4660 wakaba 1.56 $self->{insertion_mode} = IN_BODY_IM;
4661 wakaba 1.54 ## reprocess
4662     redo B;
4663 wakaba 1.57 } elsif ($token->{type} == END_TAG_TOKEN) {
4664 wakaba 1.54 if ($token->{tag_name} eq 'head') {
4665 wakaba 1.56 if ($self->{insertion_mode} == BEFORE_HEAD_IM) {
4666 wakaba 1.79
4667 wakaba 1.54 ## As if <head>
4668    
4669     $self->{head_element} = $self->{document}->create_element_ns
4670     (q<http://www.w3.org/1999/xhtml>, [undef, 'head']);
4671    
4672     $self->{open_elements}->[-1]->[0]->append_child ($self->{head_element});
4673     push @{$self->{open_elements}}, [$self->{head_element}, 'head'];
4674    
4675     ## Reprocess in the "in head" insertion mode...
4676     pop @{$self->{open_elements}};
4677 wakaba 1.56 $self->{insertion_mode} = AFTER_HEAD_IM;
4678 wakaba 1.54 $token = $self->_get_next_token;
4679     redo B;
4680 wakaba 1.56 } elsif ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) {
4681 wakaba 1.79
4682 wakaba 1.54 ## As if </noscript>
4683     pop @{$self->{open_elements}};
4684 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in noscript:/head');
4685 wakaba 1.54
4686     ## Reprocess in the "in head" insertion mode...
4687     pop @{$self->{open_elements}};
4688 wakaba 1.56 $self->{insertion_mode} = AFTER_HEAD_IM;
4689 wakaba 1.54 $token = $self->_get_next_token;
4690     redo B;
4691 wakaba 1.56 } elsif ($self->{insertion_mode} == IN_HEAD_IM) {
4692 wakaba 1.79
4693 wakaba 1.54 pop @{$self->{open_elements}};
4694 wakaba 1.56 $self->{insertion_mode} = AFTER_HEAD_IM;
4695 wakaba 1.54 $token = $self->_get_next_token;
4696     redo B;
4697     } else {
4698 wakaba 1.79
4699 wakaba 1.54 #
4700     }
4701     } elsif ($token->{tag_name} eq 'noscript') {
4702 wakaba 1.56 if ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) {
4703 wakaba 1.79
4704 wakaba 1.54 pop @{$self->{open_elements}};
4705 wakaba 1.56 $self->{insertion_mode} = IN_HEAD_IM;
4706 wakaba 1.54 $token = $self->_get_next_token;
4707     redo B;
4708 wakaba 1.56 } elsif ($self->{insertion_mode} == BEFORE_HEAD_IM) {
4709 wakaba 1.79
4710 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:noscript');
4711 wakaba 1.54 ## Ignore the token ## ISSUE: An issue in the spec.
4712     $token = $self->_get_next_token;
4713     redo B;
4714     } else {
4715 wakaba 1.79
4716 wakaba 1.54 #
4717     }
4718     } elsif ({
4719     body => 1, html => 1,
4720     }->{$token->{tag_name}}) {
4721 wakaba 1.56 if ($self->{insertion_mode} == BEFORE_HEAD_IM) {
4722 wakaba 1.79
4723 wakaba 1.54 ## As if <head>
4724    
4725     $self->{head_element} = $self->{document}->create_element_ns
4726     (q<http://www.w3.org/1999/xhtml>, [undef, 'head']);
4727    
4728     $self->{open_elements}->[-1]->[0]->append_child ($self->{head_element});
4729     push @{$self->{open_elements}}, [$self->{head_element}, 'head'];
4730    
4731 wakaba 1.56 $self->{insertion_mode} = IN_HEAD_IM;
4732 wakaba 1.54 ## Reprocess in the "in head" insertion mode...
4733 wakaba 1.56 } elsif ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) {
4734 wakaba 1.79
4735 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
4736 wakaba 1.54 ## Ignore the token
4737     $token = $self->_get_next_token;
4738     redo B;
4739 wakaba 1.79 } else {
4740    
4741 wakaba 1.54 }
4742    
4743     #
4744     } elsif ({
4745     p => 1, br => 1,
4746     }->{$token->{tag_name}}) {
4747 wakaba 1.56 if ($self->{insertion_mode} == BEFORE_HEAD_IM) {
4748 wakaba 1.79
4749 wakaba 1.54 ## As if <head>
4750    
4751     $self->{head_element} = $self->{document}->create_element_ns
4752     (q<http://www.w3.org/1999/xhtml>, [undef, 'head']);
4753    
4754     $self->{open_elements}->[-1]->[0]->append_child ($self->{head_element});
4755     push @{$self->{open_elements}}, [$self->{head_element}, 'head'];
4756    
4757 wakaba 1.56 $self->{insertion_mode} = IN_HEAD_IM;
4758 wakaba 1.54 ## Reprocess in the "in head" insertion mode...
4759 wakaba 1.79 } else {
4760    
4761 wakaba 1.54 }
4762    
4763     #
4764     } else {
4765 wakaba 1.56 if ($self->{insertion_mode} == AFTER_HEAD_IM) {
4766 wakaba 1.79
4767 wakaba 1.56 #
4768     } else {
4769 wakaba 1.79
4770 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
4771 wakaba 1.54 ## Ignore the token
4772     $token = $self->_get_next_token;
4773     redo B;
4774     }
4775     }
4776    
4777 wakaba 1.56 if ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) {
4778 wakaba 1.79
4779 wakaba 1.54 ## As if </noscript>
4780     pop @{$self->{open_elements}};
4781 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in noscript:/'.$token->{tag_name});
4782 wakaba 1.54
4783     ## Reprocess in the "in head" insertion mode...
4784     ## As if </head>
4785     pop @{$self->{open_elements}};
4786    
4787     ## Reprocess in the "after head" insertion mode...
4788 wakaba 1.56 } elsif ($self->{insertion_mode} == IN_HEAD_IM) {
4789 wakaba 1.79
4790 wakaba 1.54 ## As if </head>
4791     pop @{$self->{open_elements}};
4792    
4793     ## Reprocess in the "after head" insertion mode...
4794 wakaba 1.56 } elsif ($self->{insertion_mode} == BEFORE_HEAD_IM) {
4795 wakaba 1.82 ## ISSUE: This case cannot be reached?
4796 wakaba 1.79
4797 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
4798 wakaba 1.54 ## Ignore the token ## ISSUE: An issue in the spec.
4799     $token = $self->_get_next_token;
4800     redo B;
4801 wakaba 1.79 } else {
4802    
4803 wakaba 1.8 }
4804 wakaba 1.54
4805     ## "after head" insertion mode
4806     ## As if <body>
4807    
4808 wakaba 1.1 {
4809     my $el;
4810    
4811     $el = $self->{document}->create_element_ns
4812 wakaba 1.54 (q<http://www.w3.org/1999/xhtml>, [undef, 'body']);
4813 wakaba 1.1
4814 wakaba 1.54 $self->{open_elements}->[-1]->[0]->append_child ($el);
4815     push @{$self->{open_elements}}, [$el, 'body'];
4816 wakaba 1.1 }
4817    
4818 wakaba 1.56 $self->{insertion_mode} = IN_BODY_IM;
4819 wakaba 1.54 ## reprocess
4820     redo B;
4821 wakaba 1.103 } elsif ($token->{type} == END_OF_FILE_TOKEN) {
4822     if ($self->{insertion_mode} == BEFORE_HEAD_IM) {
4823    
4824    
4825     ## NOTE: As if <head>
4826    
4827     $self->{head_element} = $self->{document}->create_element_ns
4828     (q<http://www.w3.org/1999/xhtml>, [undef, 'head']);
4829    
4830     $self->{open_elements}->[-1]->[0]->append_child
4831     ($self->{head_element});
4832     #push @{$self->{open_elements}}, [$self->{head_element}, 'head'];
4833     #$self->{insertion_mode} = IN_HEAD_IM;
4834     ## NOTE: Reprocess.
4835    
4836     ## NOTE: As if </head>
4837     #pop @{$self->{open_elements}};
4838     #$self->{insertion_mode} = IN_AFTER_HEAD_IM;
4839     ## NOTE: Reprocess.
4840    
4841     #
4842     } elsif ($self->{insertion_mode} == IN_HEAD_IM) {
4843    
4844    
4845     ## NOTE: As if </head>
4846     pop @{$self->{open_elements}};
4847     #$self->{insertion_mode} = IN_AFTER_HEAD_IM;
4848     ## NOTE: Reprocess.
4849    
4850     #
4851     } elsif ($self->{insertion_mode} == IN_HEAD_NOSCRIPT_IM) {
4852    
4853    
4854 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in noscript:#eof');
4855 wakaba 1.103
4856     ## As if </noscript>
4857     pop @{$self->{open_elements}};
4858     #$self->{insertion_mode} = IN_HEAD_IM;
4859     ## NOTE: Reprocess.
4860    
4861     ## NOTE: As if </head>
4862     pop @{$self->{open_elements}};
4863     #$self->{insertion_mode} = IN_AFTER_HEAD_IM;
4864     ## NOTE: Reprocess.
4865    
4866     #
4867     } else {
4868    
4869     #
4870     }
4871    
4872     ## NOTE: As if <body>
4873    
4874     {
4875     my $el;
4876    
4877     $el = $self->{document}->create_element_ns
4878     (q<http://www.w3.org/1999/xhtml>, [undef, 'body']);
4879    
4880     $self->{open_elements}->[-1]->[0]->append_child ($el);
4881     push @{$self->{open_elements}}, [$el, 'body'];
4882     }
4883    
4884     $self->{insertion_mode} = IN_BODY_IM;
4885     ## NOTE: Reprocess.
4886     redo B;
4887     } else {
4888     die "$0: $token->{type}: Unknown token type";
4889     }
4890 wakaba 1.54
4891     ## ISSUE: An issue in the spec.
4892 wakaba 1.58 } elsif ($self->{insertion_mode} & BODY_IMS) {
4893 wakaba 1.57 if ($token->{type} == CHARACTER_TOKEN) {
4894 wakaba 1.79
4895 wakaba 1.54 ## NOTE: There is a code clone of "character in body".
4896     $reconstruct_active_formatting_elements->($insert_to_current);
4897    
4898     $self->{open_elements}->[-1]->[0]->manakai_append_text ($token->{data});
4899    
4900     $token = $self->_get_next_token;
4901     redo B;
4902 wakaba 1.57 } elsif ($token->{type} == START_TAG_TOKEN) {
4903 wakaba 1.54 if ({
4904     caption => 1, col => 1, colgroup => 1, tbody => 1,
4905     td => 1, tfoot => 1, th => 1, thead => 1, tr => 1,
4906     }->{$token->{tag_name}}) {
4907 wakaba 1.56 if ($self->{insertion_mode} == IN_CELL_IM) {
4908 wakaba 1.54 ## have an element in table scope
4909 wakaba 1.106 for (reverse 0..$#{$self->{open_elements}}) {
4910 wakaba 1.54 my $node = $self->{open_elements}->[$_];
4911     if ($node->[1] eq 'td' or $node->[1] eq 'th') {
4912 wakaba 1.79
4913 wakaba 1.106
4914     ## Close the cell
4915     unshift @{$self->{token}}, $token; # <?>
4916     $token = {type => END_TAG_TOKEN, tag_name => $node->[1]};
4917     redo B;
4918 wakaba 1.54 } elsif ({
4919     table => 1, html => 1,
4920     }->{$node->[1]}) {
4921 wakaba 1.79
4922 wakaba 1.106 ## ISSUE: This case can never be reached, maybe.
4923     last;
4924 wakaba 1.54 }
4925 wakaba 1.106 }
4926    
4927 wakaba 1.54
4928 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'start tag not allowed',
4929     value => $token->{tag_name});
4930     ## Ignore the token
4931     $token = $self->_get_next_token;
4932 wakaba 1.54 redo B;
4933 wakaba 1.56 } elsif ($self->{insertion_mode} == IN_CAPTION_IM) {
4934 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not closed:caption');
4935 wakaba 1.54
4936 wakaba 1.106 ## NOTE: As if </caption>.
4937 wakaba 1.54 ## have a table element in table scope
4938     my $i;
4939 wakaba 1.106 INSCOPE: {
4940     for (reverse 0..$#{$self->{open_elements}}) {
4941     my $node = $self->{open_elements}->[$_];
4942     if ($node->[1] eq 'caption') {
4943    
4944     $i = $_;
4945     last INSCOPE;
4946     } elsif ({
4947     table => 1, html => 1,
4948     }->{$node->[1]}) {
4949    
4950     last;
4951     }
4952 wakaba 1.54 }
4953 wakaba 1.106
4954    
4955     $self->{parse_error}->(level => $self->{must_level}, type => 'start tag not allowed',
4956     value => $token->{tag_name});
4957     ## Ignore the token
4958     $token = $self->_get_next_token;
4959     redo B;
4960 wakaba 1.54 } # INSCOPE
4961    
4962     ## generate implied end tags
4963 wakaba 1.86 while ({
4964     dd => 1, dt => 1, li => 1, p => 1,
4965     }->{$self->{open_elements}->[-1]->[1]}) {
4966 wakaba 1.79
4967 wakaba 1.86 pop @{$self->{open_elements}};
4968 wakaba 1.54 }
4969    
4970     if ($self->{open_elements}->[-1]->[1] ne 'caption') {
4971 wakaba 1.79
4972 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not closed:'.$self->{open_elements}->[-1]->[1]);
4973 wakaba 1.79 } else {
4974    
4975 wakaba 1.54 }
4976    
4977     splice @{$self->{open_elements}}, $i;
4978    
4979     $clear_up_to_marker->();
4980    
4981 wakaba 1.56 $self->{insertion_mode} = IN_TABLE_IM;
4982 wakaba 1.54
4983     ## reprocess
4984     redo B;
4985     } else {
4986 wakaba 1.79
4987 wakaba 1.54 #
4988     }
4989     } else {
4990 wakaba 1.79
4991 wakaba 1.54 #
4992     }
4993 wakaba 1.57 } elsif ($token->{type} == END_TAG_TOKEN) {
4994 wakaba 1.54 if ($token->{tag_name} eq 'td' or $token->{tag_name} eq 'th') {
4995 wakaba 1.56 if ($self->{insertion_mode} == IN_CELL_IM) {
4996 wakaba 1.54 ## have an element in table scope
4997     my $i;
4998     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
4999     my $node = $self->{open_elements}->[$_];
5000     if ($node->[1] eq $token->{tag_name}) {
5001 wakaba 1.79
5002 wakaba 1.54 $i = $_;
5003     last INSCOPE;
5004     } elsif ({
5005     table => 1, html => 1,
5006     }->{$node->[1]}) {
5007 wakaba 1.79
5008 wakaba 1.54 last INSCOPE;
5009     }
5010     } # INSCOPE
5011     unless (defined $i) {
5012 wakaba 1.79
5013 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
5014 wakaba 1.54 ## Ignore the token
5015     $token = $self->_get_next_token;
5016     redo B;
5017     }
5018    
5019     ## generate implied end tags
5020 wakaba 1.86 while ({
5021     dd => 1, dt => 1, li => 1, p => 1,
5022     }->{$self->{open_elements}->[-1]->[1]}) {
5023 wakaba 1.79
5024 wakaba 1.86 pop @{$self->{open_elements}};
5025 wakaba 1.54 }
5026 wakaba 1.86
5027 wakaba 1.54 if ($self->{open_elements}->[-1]->[1] ne $token->{tag_name}) {
5028 wakaba 1.79
5029 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not closed:'.$self->{open_elements}->[-1]->[1]);
5030 wakaba 1.79 } else {
5031    
5032 wakaba 1.54 }
5033    
5034     splice @{$self->{open_elements}}, $i;
5035    
5036     $clear_up_to_marker->();
5037    
5038 wakaba 1.56 $self->{insertion_mode} = IN_ROW_IM;
5039 wakaba 1.54
5040     $token = $self->_get_next_token;
5041     redo B;
5042 wakaba 1.56 } elsif ($self->{insertion_mode} == IN_CAPTION_IM) {
5043 wakaba 1.79
5044 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
5045 wakaba 1.54 ## Ignore the token
5046     $token = $self->_get_next_token;
5047     redo B;
5048     } else {
5049 wakaba 1.79
5050 wakaba 1.54 #
5051     }
5052     } elsif ($token->{tag_name} eq 'caption') {
5053 wakaba 1.56 if ($self->{insertion_mode} == IN_CAPTION_IM) {
5054 wakaba 1.54 ## have a table element in table scope
5055     my $i;
5056 wakaba 1.106 INSCOPE: {
5057     for (reverse 0..$#{$self->{open_elements}}) {
5058     my $node = $self->{open_elements}->[$_];
5059     if ($node->[1] eq $token->{tag_name}) {
5060    
5061     $i = $_;
5062     last INSCOPE;
5063     } elsif ({
5064     table => 1, html => 1,
5065     }->{$node->[1]}) {
5066    
5067     last;
5068     }
5069 wakaba 1.54 }
5070 wakaba 1.106
5071    
5072     $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag',
5073     value => $token->{tag_name});
5074     ## Ignore the token
5075     $token = $self->_get_next_token;
5076     redo B;
5077 wakaba 1.54 } # INSCOPE
5078    
5079     ## generate implied end tags
5080 wakaba 1.86 while ({
5081     dd => 1, dt => 1, li => 1, p => 1,
5082     }->{$self->{open_elements}->[-1]->[1]}) {
5083 wakaba 1.79
5084 wakaba 1.86 pop @{$self->{open_elements}};
5085 wakaba 1.54 }
5086    
5087     if ($self->{open_elements}->[-1]->[1] ne 'caption') {
5088 wakaba 1.79
5089 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not closed:'.$self->{open_elements}->[-1]->[1]);
5090 wakaba 1.79 } else {
5091    
5092 wakaba 1.54 }
5093    
5094     splice @{$self->{open_elements}}, $i;
5095    
5096     $clear_up_to_marker->();
5097    
5098 wakaba 1.56 $self->{insertion_mode} = IN_TABLE_IM;
5099 wakaba 1.54
5100     $token = $self->_get_next_token;
5101     redo B;
5102 wakaba 1.56 } elsif ($self->{insertion_mode} == IN_CELL_IM) {
5103 wakaba 1.79
5104 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
5105 wakaba 1.54 ## Ignore the token
5106     $token = $self->_get_next_token;
5107     redo B;
5108     } else {
5109 wakaba 1.79
5110 wakaba 1.54 #
5111 wakaba 1.1 }
5112 wakaba 1.54 } elsif ({
5113     table => 1, tbody => 1, tfoot => 1,
5114     thead => 1, tr => 1,
5115     }->{$token->{tag_name}} and
5116 wakaba 1.56 $self->{insertion_mode} == IN_CELL_IM) {
5117 wakaba 1.54 ## have an element in table scope
5118     my $i;
5119     my $tn;
5120 wakaba 1.106 INSCOPE: {
5121     for (reverse 0..$#{$self->{open_elements}}) {
5122     my $node = $self->{open_elements}->[$_];
5123     if ($node->[1] eq $token->{tag_name}) {
5124    
5125     $i = $_;
5126    
5127     ## Close the cell
5128     unshift @{$self->{token}}, $token; # </?>
5129     $token = {type => END_TAG_TOKEN, tag_name => $tn};
5130     redo B;
5131     } elsif ($node->[1] eq 'td' or $node->[1] eq 'th') {
5132    
5133     $tn = $node->[1];
5134     ## NOTE: There is exactly one |td| or |th| element
5135     ## in scope in the stack of open elements by definition.
5136     } elsif ({
5137     table => 1, html => 1,
5138     }->{$node->[1]}) {
5139     ## ISSUE: Can this be reached?
5140    
5141     last;
5142     }
5143 wakaba 1.54 }
5144 wakaba 1.106
5145 wakaba 1.79
5146 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag',
5147     value => $token->{tag_name});
5148 wakaba 1.54 ## Ignore the token
5149     $token = $self->_get_next_token;
5150     redo B;
5151 wakaba 1.106 } # INSCOPE
5152 wakaba 1.54 } elsif ($token->{tag_name} eq 'table' and
5153 wakaba 1.56 $self->{insertion_mode} == IN_CAPTION_IM) {
5154 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not closed:caption');
5155 wakaba 1.31
5156 wakaba 1.54 ## As if </caption>
5157     ## have a table element in table scope
5158     my $i;
5159     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
5160     my $node = $self->{open_elements}->[$_];
5161     if ($node->[1] eq 'caption') {
5162 wakaba 1.79
5163 wakaba 1.54 $i = $_;
5164     last INSCOPE;
5165     } elsif ({
5166     table => 1, html => 1,
5167     }->{$node->[1]}) {
5168 wakaba 1.79
5169 wakaba 1.54 last INSCOPE;
5170     }
5171     } # INSCOPE
5172     unless (defined $i) {
5173 wakaba 1.79
5174 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:caption');
5175 wakaba 1.54 ## Ignore the token
5176     $token = $self->_get_next_token;
5177     redo B;
5178     }
5179    
5180     ## generate implied end tags
5181 wakaba 1.86 while ({
5182     dd => 1, dt => 1, li => 1, p => 1,
5183     }->{$self->{open_elements}->[-1]->[1]}) {
5184 wakaba 1.79
5185 wakaba 1.86 pop @{$self->{open_elements}};
5186 wakaba 1.54 }
5187 wakaba 1.20
5188 wakaba 1.54 if ($self->{open_elements}->[-1]->[1] ne 'caption') {
5189 wakaba 1.79
5190 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not closed:'.$self->{open_elements}->[-1]->[1]);
5191 wakaba 1.79 } else {
5192    
5193 wakaba 1.54 }
5194 wakaba 1.12
5195 wakaba 1.54 splice @{$self->{open_elements}}, $i;
5196 wakaba 1.31
5197 wakaba 1.54 $clear_up_to_marker->();
5198 wakaba 1.1
5199 wakaba 1.56 $self->{insertion_mode} = IN_TABLE_IM;
5200 wakaba 1.3
5201 wakaba 1.54 ## reprocess
5202     redo B;
5203     } elsif ({
5204     body => 1, col => 1, colgroup => 1, html => 1,
5205     }->{$token->{tag_name}}) {
5206 wakaba 1.58 if ($self->{insertion_mode} & BODY_TABLE_IMS) {
5207 wakaba 1.79
5208 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
5209 wakaba 1.54 ## Ignore the token
5210     $token = $self->_get_next_token;
5211     redo B;
5212     } else {
5213 wakaba 1.79
5214 wakaba 1.54 #
5215     }
5216     } elsif ({
5217     tbody => 1, tfoot => 1,
5218     thead => 1, tr => 1,
5219     }->{$token->{tag_name}} and
5220 wakaba 1.56 $self->{insertion_mode} == IN_CAPTION_IM) {
5221 wakaba 1.79
5222 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
5223 wakaba 1.1 ## Ignore the token
5224     $token = $self->_get_next_token;
5225 wakaba 1.54 redo B;
5226     } else {
5227 wakaba 1.79
5228 wakaba 1.54 #
5229 wakaba 1.1 }
5230 wakaba 1.103 } elsif ($token->{type} == END_OF_FILE_TOKEN) {
5231     for my $entry (@{$self->{open_elements}}) {
5232     if (not {
5233     dd => 1, dt => 1, li => 1, p => 1, tbody => 1, td => 1, tfoot => 1,
5234     th => 1, thead => 1, tr => 1, body => 1, html => 1,
5235     }->{$entry->[1]}) {
5236    
5237 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in body:#eof');
5238 wakaba 1.103 last;
5239     }
5240     }
5241    
5242     ## Stop parsing.
5243     last B;
5244 wakaba 1.54 } else {
5245     die "$0: $token->{type}: Unknown token type";
5246 wakaba 1.1 }
5247    
5248 wakaba 1.54 $insert = $insert_to_current;
5249     #
5250 wakaba 1.58 } elsif ($self->{insertion_mode} & TABLE_IMS) {
5251 wakaba 1.60 if ($token->{type} == CHARACTER_TOKEN) {
5252 wakaba 1.95 if (not $open_tables->[-1]->[1] and # tainted
5253     $token->{data} =~ s/^([\x09\x0A\x0B\x0C\x20]+)//) {
5254     $self->{open_elements}->[-1]->[0]->manakai_append_text ($1);
5255 wakaba 1.54
5256 wakaba 1.95 unless (length $token->{data}) {
5257    
5258     $token = $self->_get_next_token;
5259     redo B;
5260     } else {
5261    
5262     }
5263     }
5264 wakaba 1.1
5265 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in table:#character');
5266 wakaba 1.1
5267 wakaba 1.54 ## As if in body, but insert into foster parent element
5268     ## ISSUE: Spec says that "whenever a node would be inserted
5269     ## into the current node" while characters might not be
5270     ## result in a new Text node.
5271     $reconstruct_active_formatting_elements->($insert_to_foster);
5272    
5273     if ({
5274     table => 1, tbody => 1, tfoot => 1,
5275     thead => 1, tr => 1,
5276     }->{$self->{open_elements}->[-1]->[1]}) {
5277     # MUST
5278     my $foster_parent_element;
5279     my $next_sibling;
5280     my $prev_sibling;
5281     OE: for (reverse 0..$#{$self->{open_elements}}) {
5282     if ($self->{open_elements}->[$_]->[1] eq 'table') {
5283     my $parent = $self->{open_elements}->[$_]->[0]->parent_node;
5284     if (defined $parent and $parent->node_type == 1) {
5285 wakaba 1.79
5286 wakaba 1.54 $foster_parent_element = $parent;
5287     $next_sibling = $self->{open_elements}->[$_]->[0];
5288     $prev_sibling = $next_sibling->previous_sibling;
5289     } else {
5290 wakaba 1.79
5291 wakaba 1.54 $foster_parent_element = $self->{open_elements}->[$_ - 1]->[0];
5292     $prev_sibling = $foster_parent_element->last_child;
5293     }
5294     last OE;
5295     }
5296     } # OE
5297     $foster_parent_element = $self->{open_elements}->[0]->[0] and
5298     $prev_sibling = $foster_parent_element->last_child
5299     unless defined $foster_parent_element;
5300     if (defined $prev_sibling and
5301     $prev_sibling->node_type == 3) {
5302 wakaba 1.79
5303 wakaba 1.54 $prev_sibling->manakai_append_text ($token->{data});
5304     } else {
5305 wakaba 1.79
5306 wakaba 1.54 $foster_parent_element->insert_before
5307     ($self->{document}->create_text_node ($token->{data}),
5308     $next_sibling);
5309     }
5310 wakaba 1.95 $open_tables->[-1]->[1] = 1; # tainted
5311     } else {
5312    
5313     $self->{open_elements}->[-1]->[0]->manakai_append_text ($token->{data});
5314     }
5315 wakaba 1.54
5316 wakaba 1.95 $token = $self->_get_next_token;
5317     redo B;
5318 wakaba 1.60 } elsif ($token->{type} == START_TAG_TOKEN) {
5319 wakaba 1.54 if ({
5320 wakaba 1.56 tr => ($self->{insertion_mode} != IN_ROW_IM),
5321 wakaba 1.54 th => 1, td => 1,
5322     }->{$token->{tag_name}}) {
5323 wakaba 1.56 if ($self->{insertion_mode} == IN_TABLE_IM) {
5324 wakaba 1.54 ## Clear back to table context
5325     while ($self->{open_elements}->[-1]->[1] ne 'table' and
5326     $self->{open_elements}->[-1]->[1] ne 'html') {
5327 wakaba 1.79
5328 wakaba 1.54 pop @{$self->{open_elements}};
5329     }
5330    
5331    
5332     {
5333     my $el;
5334    
5335     $el = $self->{document}->create_element_ns
5336     (q<http://www.w3.org/1999/xhtml>, [undef, 'tbody']);
5337    
5338     $self->{open_elements}->[-1]->[0]->append_child ($el);
5339     push @{$self->{open_elements}}, [$el, 'tbody'];
5340     }
5341    
5342 wakaba 1.56 $self->{insertion_mode} = IN_TABLE_BODY_IM;
5343 wakaba 1.54 ## reprocess in the "in table body" insertion mode...
5344     }
5345    
5346 wakaba 1.56 if ($self->{insertion_mode} == IN_TABLE_BODY_IM) {
5347 wakaba 1.54 unless ($token->{tag_name} eq 'tr') {
5348 wakaba 1.79
5349 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'missing start tag:tr');
5350 wakaba 1.54 }
5351    
5352     ## Clear back to table body context
5353     while (not {
5354     tbody => 1, tfoot => 1, thead => 1, html => 1,
5355     }->{$self->{open_elements}->[-1]->[1]}) {
5356 wakaba 1.79
5357 wakaba 1.83 ## ISSUE: Can this case be reached?
5358 wakaba 1.54 pop @{$self->{open_elements}};
5359     }
5360    
5361 wakaba 1.56 $self->{insertion_mode} = IN_ROW_IM;
5362 wakaba 1.54 if ($token->{tag_name} eq 'tr') {
5363    
5364 wakaba 1.79
5365 wakaba 1.54 {
5366     my $el;
5367    
5368     $el = $self->{document}->create_element_ns
5369     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
5370    
5371     for my $attr_name (keys %{ $token->{attributes}}) {
5372     $el->set_attribute_ns (undef, [undef, $attr_name],
5373     $token->{attributes} ->{$attr_name}->{value});
5374 wakaba 1.1 }
5375 wakaba 1.54
5376     $self->{open_elements}->[-1]->[0]->append_child ($el);
5377     push @{$self->{open_elements}}, [$el, $token->{tag_name}];
5378     }
5379    
5380     $token = $self->_get_next_token;
5381     redo B;
5382     } else {
5383    
5384 wakaba 1.79
5385 wakaba 1.54 {
5386     my $el;
5387    
5388     $el = $self->{document}->create_element_ns
5389     (q<http://www.w3.org/1999/xhtml>, [undef, 'tr']);
5390    
5391     $self->{open_elements}->[-1]->[0]->append_child ($el);
5392     push @{$self->{open_elements}}, [$el, 'tr'];
5393     }
5394    
5395     ## reprocess in the "in row" insertion mode
5396     }
5397 wakaba 1.79 } else {
5398    
5399 wakaba 1.54 }
5400 wakaba 1.52
5401 wakaba 1.54 ## Clear back to table row context
5402     while (not {
5403     tr => 1, html => 1,
5404     }->{$self->{open_elements}->[-1]->[1]}) {
5405 wakaba 1.79
5406 wakaba 1.54 pop @{$self->{open_elements}};
5407     }
5408    
5409    
5410     {
5411     my $el;
5412    
5413     $el = $self->{document}->create_element_ns
5414     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
5415 wakaba 1.1
5416 wakaba 1.54 for my $attr_name (keys %{ $token->{attributes}}) {
5417     $el->set_attribute_ns (undef, [undef, $attr_name],
5418     $token->{attributes} ->{$attr_name}->{value});
5419     }
5420    
5421     $self->{open_elements}->[-1]->[0]->append_child ($el);
5422     push @{$self->{open_elements}}, [$el, $token->{tag_name}];
5423     }
5424    
5425 wakaba 1.56 $self->{insertion_mode} = IN_CELL_IM;
5426 wakaba 1.54
5427     push @$active_formatting_elements, ['#marker', ''];
5428    
5429     $token = $self->_get_next_token;
5430     redo B;
5431     } elsif ({
5432     caption => 1, col => 1, colgroup => 1,
5433     tbody => 1, tfoot => 1, thead => 1,
5434 wakaba 1.56 tr => 1, # $self->{insertion_mode} == IN_ROW_IM
5435 wakaba 1.54 }->{$token->{tag_name}}) {
5436 wakaba 1.56 if ($self->{insertion_mode} == IN_ROW_IM) {
5437 wakaba 1.54 ## As if </tr>
5438     ## have an element in table scope
5439     my $i;
5440     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
5441     my $node = $self->{open_elements}->[$_];
5442     if ($node->[1] eq 'tr') {
5443 wakaba 1.79
5444 wakaba 1.54 $i = $_;
5445     last INSCOPE;
5446     } elsif ({
5447 wakaba 1.83 html => 1,
5448    
5449     ## NOTE: This element does not appear here, maybe.
5450     table => 1,
5451 wakaba 1.54 }->{$node->[1]}) {
5452 wakaba 1.79
5453 wakaba 1.54 last INSCOPE;
5454     }
5455     } # INSCOPE
5456 wakaba 1.79 unless (defined $i) {
5457    
5458 wakaba 1.83 ## TODO: This type is wrong.
5459 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmacthed end tag:'.$token->{tag_name});
5460 wakaba 1.54 ## Ignore the token
5461     $token = $self->_get_next_token;
5462     redo B;
5463     }
5464    
5465     ## Clear back to table row context
5466     while (not {
5467     tr => 1, html => 1,
5468     }->{$self->{open_elements}->[-1]->[1]}) {
5469 wakaba 1.79
5470 wakaba 1.83 ## ISSUE: Can this case be reached?
5471 wakaba 1.54 pop @{$self->{open_elements}};
5472     }
5473    
5474     pop @{$self->{open_elements}}; # tr
5475 wakaba 1.56 $self->{insertion_mode} = IN_TABLE_BODY_IM;
5476 wakaba 1.54 if ($token->{tag_name} eq 'tr') {
5477 wakaba 1.79
5478 wakaba 1.54 ## reprocess
5479     redo B;
5480     } else {
5481 wakaba 1.79
5482 wakaba 1.54 ## reprocess in the "in table body" insertion mode...
5483     }
5484     }
5485 wakaba 1.52
5486 wakaba 1.56 if ($self->{insertion_mode} == IN_TABLE_BODY_IM) {
5487 wakaba 1.54 ## have an element in table scope
5488     my $i;
5489     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
5490     my $node = $self->{open_elements}->[$_];
5491     if ({
5492     tbody => 1, thead => 1, tfoot => 1,
5493     }->{$node->[1]}) {
5494 wakaba 1.79
5495 wakaba 1.54 $i = $_;
5496     last INSCOPE;
5497     } elsif ({
5498     table => 1, html => 1,
5499     }->{$node->[1]}) {
5500 wakaba 1.79
5501 wakaba 1.54 last INSCOPE;
5502     }
5503     } # INSCOPE
5504     unless (defined $i) {
5505 wakaba 1.79
5506 wakaba 1.83 ## TODO: This erorr type ios wrong.
5507 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
5508 wakaba 1.54 ## Ignore the token
5509     $token = $self->_get_next_token;
5510     redo B;
5511     }
5512 wakaba 1.52
5513 wakaba 1.54 ## Clear back to table body context
5514     while (not {
5515     tbody => 1, tfoot => 1, thead => 1, html => 1,
5516     }->{$self->{open_elements}->[-1]->[1]}) {
5517 wakaba 1.79
5518 wakaba 1.83 ## ISSUE: Can this state be reached?
5519 wakaba 1.54 pop @{$self->{open_elements}};
5520     }
5521    
5522     ## As if <{current node}>
5523     ## have an element in table scope
5524     ## true by definition
5525    
5526     ## Clear back to table body context
5527     ## nop by definition
5528    
5529     pop @{$self->{open_elements}};
5530 wakaba 1.56 $self->{insertion_mode} = IN_TABLE_IM;
5531 wakaba 1.54 ## reprocess in "in table" insertion mode...
5532 wakaba 1.79 } else {
5533    
5534 wakaba 1.54 }
5535 wakaba 1.51
5536 wakaba 1.54 if ($token->{tag_name} eq 'col') {
5537     ## Clear back to table context
5538     while ($self->{open_elements}->[-1]->[1] ne 'table' and
5539     $self->{open_elements}->[-1]->[1] ne 'html') {
5540 wakaba 1.79
5541 wakaba 1.83 ## ISSUE: Can this state be reached?
5542 wakaba 1.54 pop @{$self->{open_elements}};
5543     }
5544    
5545    
5546 wakaba 1.51 {
5547     my $el;
5548    
5549     $el = $self->{document}->create_element_ns
5550 wakaba 1.54 (q<http://www.w3.org/1999/xhtml>, [undef, 'colgroup']);
5551 wakaba 1.51
5552     $self->{open_elements}->[-1]->[0]->append_child ($el);
5553 wakaba 1.54 push @{$self->{open_elements}}, [$el, 'colgroup'];
5554 wakaba 1.51 }
5555    
5556 wakaba 1.56 $self->{insertion_mode} = IN_COLUMN_GROUP_IM;
5557 wakaba 1.54 ## reprocess
5558     redo B;
5559     } elsif ({
5560     caption => 1,
5561     colgroup => 1,
5562     tbody => 1, tfoot => 1, thead => 1,
5563     }->{$token->{tag_name}}) {
5564     ## Clear back to table context
5565     while ($self->{open_elements}->[-1]->[1] ne 'table' and
5566     $self->{open_elements}->[-1]->[1] ne 'html') {
5567 wakaba 1.79
5568 wakaba 1.83 ## ISSUE: Can this state be reached?
5569 wakaba 1.54 pop @{$self->{open_elements}};
5570     }
5571    
5572     push @$active_formatting_elements, ['#marker', '']
5573     if $token->{tag_name} eq 'caption';
5574    
5575 wakaba 1.52
5576 wakaba 1.54 {
5577     my $el;
5578    
5579     $el = $self->{document}->create_element_ns
5580 wakaba 1.52 (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
5581    
5582 wakaba 1.54 for my $attr_name (keys %{ $token->{attributes}}) {
5583     $el->set_attribute_ns (undef, [undef, $attr_name],
5584     $token->{attributes} ->{$attr_name}->{value});
5585 wakaba 1.52 }
5586    
5587 wakaba 1.54 $self->{open_elements}->[-1]->[0]->append_child ($el);
5588     push @{$self->{open_elements}}, [$el, $token->{tag_name}];
5589     }
5590    
5591     $self->{insertion_mode} = {
5592 wakaba 1.56 caption => IN_CAPTION_IM,
5593     colgroup => IN_COLUMN_GROUP_IM,
5594     tbody => IN_TABLE_BODY_IM,
5595     tfoot => IN_TABLE_BODY_IM,
5596     thead => IN_TABLE_BODY_IM,
5597 wakaba 1.54 }->{$token->{tag_name}};
5598 wakaba 1.52 $token = $self->_get_next_token;
5599     redo B;
5600 wakaba 1.54 } else {
5601     die "$0: in table: <>: $token->{tag_name}";
5602     }
5603     } elsif ($token->{tag_name} eq 'table') {
5604 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not closed:'.$self->{open_elements}->[-1]->[1]);
5605 wakaba 1.54
5606     ## As if </table>
5607     ## have a table element in table scope
5608     my $i;
5609     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
5610     my $node = $self->{open_elements}->[$_];
5611     if ($node->[1] eq 'table') {
5612 wakaba 1.79
5613 wakaba 1.54 $i = $_;
5614     last INSCOPE;
5615     } elsif ({
5616 wakaba 1.83 #table => 1,
5617     html => 1,
5618 wakaba 1.54 }->{$node->[1]}) {
5619 wakaba 1.79
5620 wakaba 1.54 last INSCOPE;
5621     }
5622     } # INSCOPE
5623     unless (defined $i) {
5624 wakaba 1.79
5625 wakaba 1.83 ## TODO: The following is wrong, maybe.
5626 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:table');
5627 wakaba 1.54 ## Ignore tokens </table><table>
5628 wakaba 1.52 $token = $self->_get_next_token;
5629     redo B;
5630     }
5631    
5632 wakaba 1.105 ## TODO: Followings are removed from the latest spec.
5633 wakaba 1.54 ## generate implied end tags
5634 wakaba 1.86 while ({
5635     dd => 1, dt => 1, li => 1, p => 1,
5636     }->{$self->{open_elements}->[-1]->[1]}) {
5637 wakaba 1.79
5638 wakaba 1.86 pop @{$self->{open_elements}};
5639 wakaba 1.54 }
5640    
5641     if ($self->{open_elements}->[-1]->[1] ne 'table') {
5642 wakaba 1.79
5643 wakaba 1.83 ## ISSUE: Can this case be reached?
5644 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not closed:'.$self->{open_elements}->[-1]->[1]);
5645 wakaba 1.79 } else {
5646    
5647 wakaba 1.54 }
5648    
5649     splice @{$self->{open_elements}}, $i;
5650 wakaba 1.95 pop @{$open_tables};
5651 wakaba 1.54
5652     $self->_reset_insertion_mode;
5653 wakaba 1.52
5654 wakaba 1.54 ## reprocess
5655     redo B;
5656 wakaba 1.100 } elsif ($token->{tag_name} eq 'style') {
5657     if (not $open_tables->[-1]->[1]) { # tainted
5658    
5659     ## NOTE: This is a "as if in head" code clone.
5660     $parse_rcdata->(CDATA_CONTENT_MODEL);
5661     redo B;
5662     } else {
5663    
5664     #
5665     }
5666     } elsif ($token->{tag_name} eq 'script') {
5667     if (not $open_tables->[-1]->[1]) { # tainted
5668    
5669     ## NOTE: This is a "as if in head" code clone.
5670     $script_start_tag->();
5671     redo B;
5672     } else {
5673    
5674     #
5675     }
5676 wakaba 1.98 } elsif ($token->{tag_name} eq 'input') {
5677     if (not $open_tables->[-1]->[1]) { # tainted
5678     if ($token->{attributes}->{type}) { ## TODO: case
5679     my $type = lc $token->{attributes}->{type}->{value};
5680     if ($type eq 'hidden') {
5681    
5682 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in table:'.$token->{tag_name});
5683 wakaba 1.98
5684    
5685     {
5686     my $el;
5687    
5688     $el = $self->{document}->create_element_ns
5689     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
5690    
5691     for my $attr_name (keys %{ $token->{attributes}}) {
5692     $el->set_attribute_ns (undef, [undef, $attr_name],
5693     $token->{attributes} ->{$attr_name}->{value});
5694     }
5695    
5696     $self->{open_elements}->[-1]->[0]->append_child ($el);
5697     push @{$self->{open_elements}}, [$el, $token->{tag_name}];
5698     }
5699    
5700    
5701     ## TODO: form element pointer
5702    
5703     pop @{$self->{open_elements}};
5704    
5705     $token = $self->_get_next_token;
5706     redo B;
5707     } else {
5708    
5709     #
5710     }
5711     } else {
5712    
5713     #
5714     }
5715     } else {
5716    
5717     #
5718     }
5719 wakaba 1.60 } else {
5720 wakaba 1.79
5721 wakaba 1.60 #
5722     }
5723 wakaba 1.98
5724 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in table:'.$token->{tag_name});
5725 wakaba 1.98
5726     $insert = $insert_to_foster;
5727     #
5728 wakaba 1.60 } elsif ($token->{type} == END_TAG_TOKEN) {
5729 wakaba 1.54 if ($token->{tag_name} eq 'tr' and
5730 wakaba 1.56 $self->{insertion_mode} == IN_ROW_IM) {
5731 wakaba 1.54 ## have an element in table scope
5732     my $i;
5733     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
5734     my $node = $self->{open_elements}->[$_];
5735     if ($node->[1] eq $token->{tag_name}) {
5736 wakaba 1.79
5737 wakaba 1.54 $i = $_;
5738     last INSCOPE;
5739     } elsif ({
5740     table => 1, html => 1,
5741     }->{$node->[1]}) {
5742 wakaba 1.79
5743 wakaba 1.54 last INSCOPE;
5744     }
5745     } # INSCOPE
5746     unless (defined $i) {
5747 wakaba 1.79
5748 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
5749 wakaba 1.54 ## Ignore the token
5750     $token = $self->_get_next_token;
5751     redo B;
5752 wakaba 1.79 } else {
5753    
5754 wakaba 1.54 }
5755    
5756     ## Clear back to table row context
5757     while (not {
5758     tr => 1, html => 1,
5759     }->{$self->{open_elements}->[-1]->[1]}) {
5760 wakaba 1.79
5761 wakaba 1.83 ## ISSUE: Can this state be reached?
5762 wakaba 1.54 pop @{$self->{open_elements}};
5763     }
5764    
5765     pop @{$self->{open_elements}}; # tr
5766 wakaba 1.56 $self->{insertion_mode} = IN_TABLE_BODY_IM;
5767 wakaba 1.54 $token = $self->_get_next_token;
5768     redo B;
5769     } elsif ($token->{tag_name} eq 'table') {
5770 wakaba 1.56 if ($self->{insertion_mode} == IN_ROW_IM) {
5771 wakaba 1.54 ## As if </tr>
5772     ## have an element in table scope
5773     my $i;
5774     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
5775     my $node = $self->{open_elements}->[$_];
5776     if ($node->[1] eq 'tr') {
5777 wakaba 1.79
5778 wakaba 1.54 $i = $_;
5779     last INSCOPE;
5780     } elsif ({
5781     table => 1, html => 1,
5782     }->{$node->[1]}) {
5783 wakaba 1.79
5784 wakaba 1.54 last INSCOPE;
5785     }
5786     } # INSCOPE
5787     unless (defined $i) {
5788 wakaba 1.79
5789 wakaba 1.83 ## TODO: The following is wrong.
5790 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{type});
5791 wakaba 1.54 ## Ignore the token
5792     $token = $self->_get_next_token;
5793     redo B;
5794     }
5795    
5796     ## Clear back to table row context
5797     while (not {
5798     tr => 1, html => 1,
5799     }->{$self->{open_elements}->[-1]->[1]}) {
5800 wakaba 1.79
5801 wakaba 1.83 ## ISSUE: Can this state be reached?
5802 wakaba 1.54 pop @{$self->{open_elements}};
5803     }
5804    
5805     pop @{$self->{open_elements}}; # tr
5806 wakaba 1.56 $self->{insertion_mode} = IN_TABLE_BODY_IM;
5807 wakaba 1.54 ## reprocess in the "in table body" insertion mode...
5808     }
5809    
5810 wakaba 1.56 if ($self->{insertion_mode} == IN_TABLE_BODY_IM) {
5811 wakaba 1.54 ## have an element in table scope
5812     my $i;
5813     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
5814     my $node = $self->{open_elements}->[$_];
5815     if ({
5816     tbody => 1, thead => 1, tfoot => 1,
5817     }->{$node->[1]}) {
5818 wakaba 1.79
5819 wakaba 1.54 $i = $_;
5820     last INSCOPE;
5821     } elsif ({
5822     table => 1, html => 1,
5823     }->{$node->[1]}) {
5824 wakaba 1.79
5825 wakaba 1.54 last INSCOPE;
5826     }
5827     } # INSCOPE
5828     unless (defined $i) {
5829 wakaba 1.79
5830 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
5831 wakaba 1.54 ## Ignore the token
5832     $token = $self->_get_next_token;
5833     redo B;
5834     }
5835    
5836     ## Clear back to table body context
5837     while (not {
5838     tbody => 1, tfoot => 1, thead => 1, html => 1,
5839     }->{$self->{open_elements}->[-1]->[1]}) {
5840 wakaba 1.79
5841 wakaba 1.54 pop @{$self->{open_elements}};
5842     }
5843    
5844     ## As if <{current node}>
5845     ## have an element in table scope
5846     ## true by definition
5847    
5848     ## Clear back to table body context
5849     ## nop by definition
5850    
5851     pop @{$self->{open_elements}};
5852 wakaba 1.56 $self->{insertion_mode} = IN_TABLE_IM;
5853 wakaba 1.54 ## reprocess in the "in table" insertion mode...
5854     }
5855 wakaba 1.52
5856 wakaba 1.94 ## NOTE: </table> in the "in table" insertion mode.
5857     ## When you edit the code fragment below, please ensure that
5858     ## the code for <table> in the "in table" insertion mode
5859     ## is synced with it.
5860    
5861 wakaba 1.54 ## have a table element in table scope
5862     my $i;
5863     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
5864     my $node = $self->{open_elements}->[$_];
5865     if ($node->[1] eq $token->{tag_name}) {
5866 wakaba 1.79
5867 wakaba 1.54 $i = $_;
5868     last INSCOPE;
5869     } elsif ({
5870     table => 1, html => 1,
5871     }->{$node->[1]}) {
5872 wakaba 1.79
5873 wakaba 1.54 last INSCOPE;
5874     }
5875     } # INSCOPE
5876     unless (defined $i) {
5877 wakaba 1.79
5878 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
5879 wakaba 1.54 ## Ignore the token
5880     $token = $self->_get_next_token;
5881     redo B;
5882 wakaba 1.51 }
5883 wakaba 1.54
5884     splice @{$self->{open_elements}}, $i;
5885 wakaba 1.95 pop @{$open_tables};
5886 wakaba 1.54
5887     $self->_reset_insertion_mode;
5888 wakaba 1.1
5889     $token = $self->_get_next_token;
5890 wakaba 1.25 redo B;
5891 wakaba 1.54 } elsif ({
5892     tbody => 1, tfoot => 1, thead => 1,
5893     }->{$token->{tag_name}} and
5894 wakaba 1.58 $self->{insertion_mode} & ROW_IMS) {
5895 wakaba 1.56 if ($self->{insertion_mode} == IN_ROW_IM) {
5896 wakaba 1.54 ## have an element in table scope
5897     my $i;
5898     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
5899     my $node = $self->{open_elements}->[$_];
5900     if ($node->[1] eq $token->{tag_name}) {
5901 wakaba 1.79
5902 wakaba 1.54 $i = $_;
5903     last INSCOPE;
5904     } elsif ({
5905     table => 1, html => 1,
5906     }->{$node->[1]}) {
5907 wakaba 1.79
5908 wakaba 1.54 last INSCOPE;
5909     }
5910     } # INSCOPE
5911     unless (defined $i) {
5912 wakaba 1.79
5913 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
5914 wakaba 1.54 ## Ignore the token
5915     $token = $self->_get_next_token;
5916     redo B;
5917     }
5918    
5919     ## As if </tr>
5920     ## have an element in table scope
5921     my $i;
5922     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
5923     my $node = $self->{open_elements}->[$_];
5924     if ($node->[1] eq 'tr') {
5925 wakaba 1.79
5926 wakaba 1.54 $i = $_;
5927     last INSCOPE;
5928     } elsif ({
5929     table => 1, html => 1,
5930     }->{$node->[1]}) {
5931 wakaba 1.79
5932 wakaba 1.54 last INSCOPE;
5933     }
5934     } # INSCOPE
5935     unless (defined $i) {
5936 wakaba 1.79
5937 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:tr');
5938 wakaba 1.54 ## Ignore the token
5939     $token = $self->_get_next_token;
5940     redo B;
5941     }
5942    
5943     ## Clear back to table row context
5944     while (not {
5945     tr => 1, html => 1,
5946     }->{$self->{open_elements}->[-1]->[1]}) {
5947 wakaba 1.79
5948 wakaba 1.83 ## ISSUE: Can this case be reached?
5949 wakaba 1.54 pop @{$self->{open_elements}};
5950     }
5951    
5952     pop @{$self->{open_elements}}; # tr
5953 wakaba 1.56 $self->{insertion_mode} = IN_TABLE_BODY_IM;
5954 wakaba 1.54 ## reprocess in the "in table body" insertion mode...
5955 wakaba 1.34 }
5956    
5957 wakaba 1.54 ## have an element in table scope
5958     my $i;
5959     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
5960     my $node = $self->{open_elements}->[$_];
5961     if ($node->[1] eq $token->{tag_name}) {
5962 wakaba 1.79
5963 wakaba 1.54 $i = $_;
5964     last INSCOPE;
5965     } elsif ({
5966     table => 1, html => 1,
5967     }->{$node->[1]}) {
5968 wakaba 1.79
5969 wakaba 1.54 last INSCOPE;
5970 wakaba 1.34 }
5971 wakaba 1.54 } # INSCOPE
5972     unless (defined $i) {
5973 wakaba 1.79
5974 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
5975 wakaba 1.54 ## Ignore the token
5976     $token = $self->_get_next_token;
5977     redo B;
5978 wakaba 1.34 }
5979    
5980 wakaba 1.54 ## Clear back to table body context
5981     while (not {
5982     tbody => 1, tfoot => 1, thead => 1, html => 1,
5983     }->{$self->{open_elements}->[-1]->[1]}) {
5984 wakaba 1.79
5985 wakaba 1.83 ## ISSUE: Can this case be reached?
5986 wakaba 1.51 pop @{$self->{open_elements}};
5987 wakaba 1.25 }
5988 wakaba 1.51
5989 wakaba 1.54 pop @{$self->{open_elements}};
5990 wakaba 1.56 $self->{insertion_mode} = IN_TABLE_IM;
5991 wakaba 1.54 $token = $self->_get_next_token;
5992     redo B;
5993     } elsif ({
5994     body => 1, caption => 1, col => 1, colgroup => 1,
5995     html => 1, td => 1, th => 1,
5996 wakaba 1.56 tr => 1, # $self->{insertion_mode} == IN_ROW_IM
5997     tbody => 1, tfoot => 1, thead => 1, # $self->{insertion_mode} == IN_TABLE_IM
5998 wakaba 1.54 }->{$token->{tag_name}}) {
5999 wakaba 1.79
6000 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
6001 wakaba 1.54 ## Ignore the token
6002     $token = $self->_get_next_token;
6003 wakaba 1.1 redo B;
6004 wakaba 1.60 } else {
6005 wakaba 1.79
6006 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in table:/'.$token->{tag_name});
6007 wakaba 1.54
6008 wakaba 1.60 $insert = $insert_to_foster;
6009     #
6010     }
6011 wakaba 1.103 } elsif ($token->{type} == END_OF_FILE_TOKEN) {
6012     unless ($self->{open_elements}->[-1]->[1] eq 'html' and
6013     @{$self->{open_elements}} == 1) { # redundant, maybe
6014 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in body:#eof');
6015 wakaba 1.103
6016 wakaba 1.104 #
6017 wakaba 1.103 } else {
6018    
6019 wakaba 1.104 #
6020 wakaba 1.103 }
6021    
6022     ## Stop parsing
6023     last B;
6024 wakaba 1.60 } else {
6025     die "$0: $token->{type}: Unknown token type";
6026     }
6027 wakaba 1.56 } elsif ($self->{insertion_mode} == IN_COLUMN_GROUP_IM) {
6028 wakaba 1.57 if ($token->{type} == CHARACTER_TOKEN) {
6029 wakaba 1.54 if ($token->{data} =~ s/^([\x09\x0A\x0B\x0C\x20]+)//) {
6030     $self->{open_elements}->[-1]->[0]->manakai_append_text ($1);
6031     unless (length $token->{data}) {
6032 wakaba 1.79
6033 wakaba 1.54 $token = $self->_get_next_token;
6034     redo B;
6035 wakaba 1.25 }
6036 wakaba 1.54 }
6037    
6038 wakaba 1.79
6039 wakaba 1.54 #
6040 wakaba 1.57 } elsif ($token->{type} == START_TAG_TOKEN) {
6041 wakaba 1.54 if ($token->{tag_name} eq 'col') {
6042    
6043 wakaba 1.79
6044 wakaba 1.25 {
6045     my $el;
6046    
6047 wakaba 1.1 $el = $self->{document}->create_element_ns
6048     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
6049    
6050 wakaba 1.25 for my $attr_name (keys %{ $token->{attributes}}) {
6051 wakaba 1.1 $el->set_attribute_ns (undef, [undef, $attr_name],
6052 wakaba 1.25 $token->{attributes} ->{$attr_name}->{value});
6053 wakaba 1.1 }
6054    
6055 wakaba 1.25 $self->{open_elements}->[-1]->[0]->append_child ($el);
6056     push @{$self->{open_elements}}, [$el, $token->{tag_name}];
6057     }
6058    
6059 wakaba 1.54 pop @{$self->{open_elements}};
6060     $token = $self->_get_next_token;
6061     redo B;
6062     } else {
6063 wakaba 1.79
6064 wakaba 1.54 #
6065     }
6066 wakaba 1.57 } elsif ($token->{type} == END_TAG_TOKEN) {
6067 wakaba 1.54 if ($token->{tag_name} eq 'colgroup') {
6068     if ($self->{open_elements}->[-1]->[1] eq 'html') {
6069 wakaba 1.79
6070 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:colgroup');
6071 wakaba 1.25 ## Ignore the token
6072 wakaba 1.42 $token = $self->_get_next_token;
6073 wakaba 1.25 redo B;
6074 wakaba 1.24 } else {
6075 wakaba 1.79
6076 wakaba 1.54 pop @{$self->{open_elements}}; # colgroup
6077 wakaba 1.56 $self->{insertion_mode} = IN_TABLE_IM;
6078 wakaba 1.54 $token = $self->_get_next_token;
6079     redo B;
6080 wakaba 1.25 }
6081 wakaba 1.54 } elsif ($token->{tag_name} eq 'col') {
6082 wakaba 1.79
6083 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:col');
6084 wakaba 1.54 ## Ignore the token
6085     $token = $self->_get_next_token;
6086     redo B;
6087     } else {
6088 wakaba 1.79
6089 wakaba 1.54 #
6090     }
6091 wakaba 1.103 } elsif ($token->{type} == END_OF_FILE_TOKEN) {
6092     if ($self->{open_elements}->[-1]->[1] eq 'html' or
6093     @{$self->{open_elements}} == 1) { # redundant, maybe
6094    
6095     ## Stop parsing.
6096     last B;
6097     } else {
6098     ## NOTE: As if </colgroup>.
6099    
6100     pop @{$self->{open_elements}}; # colgroup
6101     $self->{insertion_mode} = IN_TABLE_IM;
6102     ## Reprocess.
6103     redo B;
6104     }
6105     } else {
6106     die "$0: $token->{type}: Unknown token type";
6107     }
6108 wakaba 1.51
6109 wakaba 1.54 ## As if </colgroup>
6110     if ($self->{open_elements}->[-1]->[1] eq 'html') {
6111 wakaba 1.79
6112 wakaba 1.103 ## TODO: Wrong error type?
6113 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:colgroup');
6114 wakaba 1.54 ## Ignore the token
6115     $token = $self->_get_next_token;
6116     redo B;
6117     } else {
6118 wakaba 1.79
6119 wakaba 1.54 pop @{$self->{open_elements}}; # colgroup
6120 wakaba 1.56 $self->{insertion_mode} = IN_TABLE_IM;
6121 wakaba 1.54 ## reprocess
6122     redo B;
6123     }
6124 wakaba 1.101 } elsif ($self->{insertion_mode} & SELECT_IMS) {
6125 wakaba 1.60 if ($token->{type} == CHARACTER_TOKEN) {
6126 wakaba 1.79
6127 wakaba 1.60 $self->{open_elements}->[-1]->[0]->manakai_append_text ($token->{data});
6128     $token = $self->_get_next_token;
6129     redo B;
6130     } elsif ($token->{type} == START_TAG_TOKEN) {
6131 wakaba 1.54 if ($token->{tag_name} eq 'option') {
6132     if ($self->{open_elements}->[-1]->[1] eq 'option') {
6133 wakaba 1.79
6134 wakaba 1.54 ## As if </option>
6135 wakaba 1.51 pop @{$self->{open_elements}};
6136 wakaba 1.79 } else {
6137    
6138 wakaba 1.51 }
6139    
6140 wakaba 1.1
6141     {
6142     my $el;
6143    
6144     $el = $self->{document}->create_element_ns
6145 wakaba 1.51 (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
6146 wakaba 1.1
6147     for my $attr_name (keys %{ $token->{attributes}}) {
6148     $el->set_attribute_ns (undef, [undef, $attr_name],
6149     $token->{attributes} ->{$attr_name}->{value});
6150     }
6151    
6152 wakaba 1.3 $self->{open_elements}->[-1]->[0]->append_child ($el);
6153 wakaba 1.51 push @{$self->{open_elements}}, [$el, $token->{tag_name}];
6154 wakaba 1.1 }
6155    
6156     $token = $self->_get_next_token;
6157     redo B;
6158 wakaba 1.54 } elsif ($token->{tag_name} eq 'optgroup') {
6159     if ($self->{open_elements}->[-1]->[1] eq 'option') {
6160 wakaba 1.79
6161 wakaba 1.54 ## As if </option>
6162 wakaba 1.51 pop @{$self->{open_elements}};
6163 wakaba 1.79 } else {
6164    
6165 wakaba 1.51 }
6166 wakaba 1.54
6167     if ($self->{open_elements}->[-1]->[1] eq 'optgroup') {
6168 wakaba 1.79
6169 wakaba 1.54 ## As if </optgroup>
6170 wakaba 1.51 pop @{$self->{open_elements}};
6171 wakaba 1.79 } else {
6172    
6173 wakaba 1.51 }
6174    
6175    
6176 wakaba 1.1 {
6177     my $el;
6178    
6179     $el = $self->{document}->create_element_ns
6180 wakaba 1.54 (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
6181 wakaba 1.1
6182 wakaba 1.54 for my $attr_name (keys %{ $token->{attributes}}) {
6183     $el->set_attribute_ns (undef, [undef, $attr_name],
6184     $token->{attributes} ->{$attr_name}->{value});
6185     }
6186    
6187 wakaba 1.3 $self->{open_elements}->[-1]->[0]->append_child ($el);
6188 wakaba 1.54 push @{$self->{open_elements}}, [$el, $token->{tag_name}];
6189 wakaba 1.1 }
6190    
6191 wakaba 1.54 $token = $self->_get_next_token;
6192     redo B;
6193 wakaba 1.101 } elsif ($token->{tag_name} eq 'select' or
6194     $token->{tag_name} eq 'input' or
6195     ($self->{insertion_mode} == IN_SELECT_IN_TABLE_IM and
6196     {
6197     caption => 1, table => 1,
6198     tbody => 1, tfoot => 1, thead => 1,
6199     tr => 1, td => 1, th => 1,
6200     }->{$token->{tag_name}})) {
6201     ## TODO: The type below is not good - <select> is replaced by </select>
6202 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not closed:select');
6203 wakaba 1.101 ## NOTE: As if the token were </select> (<select> case) or
6204     ## as if there were </select> (otherwise).
6205 wakaba 1.54 ## have an element in table scope
6206     my $i;
6207     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
6208     my $node = $self->{open_elements}->[$_];
6209 wakaba 1.101 if ($node->[1] eq 'select') {
6210 wakaba 1.79
6211 wakaba 1.54 $i = $_;
6212     last INSCOPE;
6213     } elsif ({
6214     table => 1, html => 1,
6215     }->{$node->[1]}) {
6216 wakaba 1.79
6217 wakaba 1.54 last INSCOPE;
6218 wakaba 1.44 }
6219 wakaba 1.54 } # INSCOPE
6220     unless (defined $i) {
6221 wakaba 1.79
6222 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:select');
6223 wakaba 1.54 ## Ignore the token
6224     $token = $self->_get_next_token;
6225 wakaba 1.44 redo B;
6226     }
6227 wakaba 1.54
6228 wakaba 1.79
6229 wakaba 1.54 splice @{$self->{open_elements}}, $i;
6230    
6231     $self->_reset_insertion_mode;
6232    
6233 wakaba 1.101 if ($token->{tag_name} eq 'select') {
6234    
6235     $token = $self->_get_next_token;
6236     redo B;
6237     } else {
6238    
6239     ## Reprocess the token.
6240     redo B;
6241     }
6242 wakaba 1.60 } else {
6243 wakaba 1.79
6244 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in select:'.$token->{tag_name});
6245 wakaba 1.60 ## Ignore the token
6246     $token = $self->_get_next_token;
6247     redo B;
6248     }
6249     } elsif ($token->{type} == END_TAG_TOKEN) {
6250 wakaba 1.54 if ($token->{tag_name} eq 'optgroup') {
6251     if ($self->{open_elements}->[-1]->[1] eq 'option' and
6252     $self->{open_elements}->[-2]->[1] eq 'optgroup') {
6253 wakaba 1.79
6254 wakaba 1.54 ## As if </option>
6255     splice @{$self->{open_elements}}, -2;
6256     } elsif ($self->{open_elements}->[-1]->[1] eq 'optgroup') {
6257 wakaba 1.79
6258 wakaba 1.54 pop @{$self->{open_elements}};
6259     } else {
6260 wakaba 1.79
6261 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
6262 wakaba 1.43 ## Ignore the token
6263 wakaba 1.54 }
6264     $token = $self->_get_next_token;
6265     redo B;
6266     } elsif ($token->{tag_name} eq 'option') {
6267     if ($self->{open_elements}->[-1]->[1] eq 'option') {
6268 wakaba 1.79
6269 wakaba 1.54 pop @{$self->{open_elements}};
6270 wakaba 1.44 } else {
6271 wakaba 1.79
6272 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
6273 wakaba 1.54 ## Ignore the token
6274 wakaba 1.43 }
6275 wakaba 1.54 $token = $self->_get_next_token;
6276     redo B;
6277     } elsif ($token->{tag_name} eq 'select') {
6278     ## have an element in table scope
6279     my $i;
6280     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
6281     my $node = $self->{open_elements}->[$_];
6282     if ($node->[1] eq $token->{tag_name}) {
6283 wakaba 1.79
6284 wakaba 1.54 $i = $_;
6285     last INSCOPE;
6286     } elsif ({
6287     table => 1, html => 1,
6288     }->{$node->[1]}) {
6289 wakaba 1.79
6290 wakaba 1.54 last INSCOPE;
6291 wakaba 1.44 }
6292 wakaba 1.54 } # INSCOPE
6293     unless (defined $i) {
6294 wakaba 1.79
6295 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
6296 wakaba 1.44 ## Ignore the token
6297     $token = $self->_get_next_token;
6298 wakaba 1.43 redo B;
6299     }
6300 wakaba 1.54
6301 wakaba 1.79
6302 wakaba 1.54 splice @{$self->{open_elements}}, $i;
6303    
6304     $self->_reset_insertion_mode;
6305    
6306     $token = $self->_get_next_token;
6307     redo B;
6308 wakaba 1.101 } elsif ($self->{insertion_mode} == IN_SELECT_IN_TABLE_IM and
6309     {
6310     caption => 1, table => 1, tbody => 1,
6311     tfoot => 1, thead => 1, tr => 1, td => 1, th => 1,
6312     }->{$token->{tag_name}}) {
6313 wakaba 1.83 ## TODO: The following is wrong?
6314 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
6315 wakaba 1.54
6316 wakaba 1.44 ## have an element in table scope
6317 wakaba 1.43 my $i;
6318     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
6319     my $node = $self->{open_elements}->[$_];
6320     if ($node->[1] eq $token->{tag_name}) {
6321 wakaba 1.79
6322 wakaba 1.43 $i = $_;
6323     last INSCOPE;
6324     } elsif ({
6325     table => 1, html => 1,
6326     }->{$node->[1]}) {
6327 wakaba 1.79
6328 wakaba 1.43 last INSCOPE;
6329     }
6330     } # INSCOPE
6331     unless (defined $i) {
6332 wakaba 1.79
6333 wakaba 1.43 ## Ignore the token
6334     $token = $self->_get_next_token;
6335     redo B;
6336     }
6337 wakaba 1.54
6338     ## As if </select>
6339     ## have an element in table scope
6340     undef $i;
6341 wakaba 1.43 INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
6342     my $node = $self->{open_elements}->[$_];
6343 wakaba 1.54 if ($node->[1] eq 'select') {
6344 wakaba 1.79
6345 wakaba 1.43 $i = $_;
6346     last INSCOPE;
6347     } elsif ({
6348     table => 1, html => 1,
6349     }->{$node->[1]}) {
6350 wakaba 1.83 ## ISSUE: Can this state be reached?
6351 wakaba 1.79
6352 wakaba 1.43 last INSCOPE;
6353     }
6354     } # INSCOPE
6355     unless (defined $i) {
6356 wakaba 1.79
6357 wakaba 1.83 ## TODO: The following error type is correct?
6358 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:select');
6359 wakaba 1.54 ## Ignore the </select> token
6360     $token = $self->_get_next_token; ## TODO: ok?
6361 wakaba 1.43 redo B;
6362     }
6363 wakaba 1.54
6364 wakaba 1.79
6365 wakaba 1.54 splice @{$self->{open_elements}}, $i;
6366    
6367     $self->_reset_insertion_mode;
6368    
6369     ## reprocess
6370     redo B;
6371 wakaba 1.60 } else {
6372 wakaba 1.79
6373 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in select:/'.$token->{tag_name});
6374 wakaba 1.54 ## Ignore the token
6375     $token = $self->_get_next_token;
6376     redo B;
6377 wakaba 1.60 }
6378 wakaba 1.103 } elsif ($token->{type} == END_OF_FILE_TOKEN) {
6379     unless ($self->{open_elements}->[-1]->[1] eq 'html' and
6380     @{$self->{open_elements}} == 1) { # redundant, maybe
6381    
6382 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in body:#eof');
6383 wakaba 1.103 } else {
6384    
6385     }
6386    
6387     ## Stop parsing.
6388     last B;
6389 wakaba 1.60 } else {
6390     die "$0: $token->{type}: Unknown token type";
6391     }
6392 wakaba 1.58 } elsif ($self->{insertion_mode} & BODY_AFTER_IMS) {
6393 wakaba 1.57 if ($token->{type} == CHARACTER_TOKEN) {
6394 wakaba 1.54 if ($token->{data} =~ s/^([\x09\x0A\x0B\x0C\x20]+)//) {
6395     my $data = $1;
6396     ## As if in body
6397     $reconstruct_active_formatting_elements->($insert_to_current);
6398    
6399     $self->{open_elements}->[-1]->[0]->manakai_append_text ($1);
6400    
6401     unless (length $token->{data}) {
6402 wakaba 1.79
6403 wakaba 1.54 $token = $self->_get_next_token;
6404     redo B;
6405     }
6406     }
6407    
6408 wakaba 1.56 if ($self->{insertion_mode} == AFTER_HTML_BODY_IM) {
6409 wakaba 1.79
6410 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after html:#character');
6411 wakaba 1.54
6412 wakaba 1.85 ## Reprocess in the "after body" insertion mode.
6413 wakaba 1.79 } else {
6414    
6415 wakaba 1.54 }
6416    
6417     ## "after body" insertion mode
6418 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after body:#character');
6419 wakaba 1.54
6420 wakaba 1.56 $self->{insertion_mode} = IN_BODY_IM;
6421 wakaba 1.54 ## reprocess
6422     redo B;
6423 wakaba 1.57 } elsif ($token->{type} == START_TAG_TOKEN) {
6424 wakaba 1.56 if ($self->{insertion_mode} == AFTER_HTML_BODY_IM) {
6425 wakaba 1.79
6426 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after html:'.$token->{tag_name});
6427 wakaba 1.54
6428 wakaba 1.85 ## Reprocess in the "after body" insertion mode.
6429 wakaba 1.79 } else {
6430    
6431 wakaba 1.54 }
6432    
6433     ## "after body" insertion mode
6434 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after body:'.$token->{tag_name});
6435 wakaba 1.54
6436 wakaba 1.56 $self->{insertion_mode} = IN_BODY_IM;
6437 wakaba 1.54 ## reprocess
6438     redo B;
6439 wakaba 1.57 } elsif ($token->{type} == END_TAG_TOKEN) {
6440 wakaba 1.56 if ($self->{insertion_mode} == AFTER_HTML_BODY_IM) {
6441 wakaba 1.79
6442 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after html:/'.$token->{tag_name});
6443 wakaba 1.54
6444 wakaba 1.56 $self->{insertion_mode} = AFTER_BODY_IM;
6445 wakaba 1.85 ## Reprocess in the "after body" insertion mode.
6446 wakaba 1.79 } else {
6447    
6448 wakaba 1.54 }
6449    
6450     ## "after body" insertion mode
6451     if ($token->{tag_name} eq 'html') {
6452     if (defined $self->{inner_html_node}) {
6453 wakaba 1.79
6454 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:html');
6455 wakaba 1.54 ## Ignore the token
6456     $token = $self->_get_next_token;
6457     redo B;
6458     } else {
6459 wakaba 1.79
6460 wakaba 1.56 $self->{insertion_mode} = AFTER_HTML_BODY_IM;
6461 wakaba 1.54 $token = $self->_get_next_token;
6462     redo B;
6463     }
6464     } else {
6465 wakaba 1.79
6466 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after body:/'.$token->{tag_name});
6467 wakaba 1.54
6468 wakaba 1.56 $self->{insertion_mode} = IN_BODY_IM;
6469 wakaba 1.54 ## reprocess
6470     redo B;
6471     }
6472 wakaba 1.103 } elsif ($token->{type} == END_OF_FILE_TOKEN) {
6473    
6474     ## Stop parsing
6475     last B;
6476 wakaba 1.54 } else {
6477     die "$0: $token->{type}: Unknown token type";
6478     }
6479 wakaba 1.58 } elsif ($self->{insertion_mode} & FRAME_IMS) {
6480 wakaba 1.57 if ($token->{type} == CHARACTER_TOKEN) {
6481 wakaba 1.54 if ($token->{data} =~ s/^([\x09\x0A\x0B\x0C\x20]+)//) {
6482     $self->{open_elements}->[-1]->[0]->manakai_append_text ($1);
6483    
6484     unless (length $token->{data}) {
6485 wakaba 1.79
6486 wakaba 1.54 $token = $self->_get_next_token;
6487     redo B;
6488     }
6489     }
6490    
6491     if ($token->{data} =~ s/^[^\x09\x0A\x0B\x0C\x20]+//) {
6492 wakaba 1.56 if ($self->{insertion_mode} == IN_FRAMESET_IM) {
6493 wakaba 1.79
6494 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in frameset:#character');
6495 wakaba 1.56 } elsif ($self->{insertion_mode} == AFTER_FRAMESET_IM) {
6496 wakaba 1.79
6497 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after frameset:#character');
6498 wakaba 1.54 } else { # "after html frameset"
6499 wakaba 1.79
6500 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after html:#character');
6501 wakaba 1.54
6502 wakaba 1.56 $self->{insertion_mode} = AFTER_FRAMESET_IM;
6503 wakaba 1.85 ## Reprocess in the "after frameset" insertion mode.
6504 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after frameset:#character');
6505 wakaba 1.54 }
6506    
6507     ## Ignore the token.
6508     if (length $token->{data}) {
6509 wakaba 1.79
6510 wakaba 1.54 ## reprocess the rest of characters
6511     } else {
6512 wakaba 1.79
6513 wakaba 1.54 $token = $self->_get_next_token;
6514     }
6515     redo B;
6516     }
6517    
6518     die qq[$0: Character "$token->{data}"];
6519 wakaba 1.57 } elsif ($token->{type} == START_TAG_TOKEN) {
6520 wakaba 1.56 if ($self->{insertion_mode} == AFTER_HTML_FRAMESET_IM) {
6521 wakaba 1.79
6522 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after html:'.$token->{tag_name});
6523 wakaba 1.54
6524 wakaba 1.79 $self->{insertion_mode} = AFTER_FRAMESET_IM;
6525 wakaba 1.85 ## Process in the "after frameset" insertion mode.
6526 wakaba 1.79 } else {
6527    
6528     }
6529    
6530 wakaba 1.54 if ($token->{tag_name} eq 'frameset' and
6531 wakaba 1.56 $self->{insertion_mode} == IN_FRAMESET_IM) {
6532 wakaba 1.54
6533 wakaba 1.79
6534 wakaba 1.54 {
6535     my $el;
6536    
6537     $el = $self->{document}->create_element_ns
6538     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
6539    
6540     for my $attr_name (keys %{ $token->{attributes}}) {
6541     $el->set_attribute_ns (undef, [undef, $attr_name],
6542     $token->{attributes} ->{$attr_name}->{value});
6543     }
6544    
6545     $self->{open_elements}->[-1]->[0]->append_child ($el);
6546     push @{$self->{open_elements}}, [$el, $token->{tag_name}];
6547     }
6548    
6549     $token = $self->_get_next_token;
6550     redo B;
6551     } elsif ($token->{tag_name} eq 'frame' and
6552 wakaba 1.56 $self->{insertion_mode} == IN_FRAMESET_IM) {
6553 wakaba 1.54
6554 wakaba 1.79
6555 wakaba 1.54 {
6556     my $el;
6557    
6558     $el = $self->{document}->create_element_ns
6559     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
6560    
6561     for my $attr_name (keys %{ $token->{attributes}}) {
6562     $el->set_attribute_ns (undef, [undef, $attr_name],
6563     $token->{attributes} ->{$attr_name}->{value});
6564     }
6565    
6566     $self->{open_elements}->[-1]->[0]->append_child ($el);
6567     push @{$self->{open_elements}}, [$el, $token->{tag_name}];
6568     }
6569    
6570     pop @{$self->{open_elements}};
6571     $token = $self->_get_next_token;
6572     redo B;
6573     } elsif ($token->{tag_name} eq 'noframes') {
6574 wakaba 1.79
6575 wakaba 1.54 ## NOTE: As if in body.
6576 wakaba 1.96 $parse_rcdata->(CDATA_CONTENT_MODEL);
6577 wakaba 1.54 redo B;
6578     } else {
6579 wakaba 1.56 if ($self->{insertion_mode} == IN_FRAMESET_IM) {
6580 wakaba 1.79
6581 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in frameset:'.$token->{tag_name});
6582 wakaba 1.54 } else {
6583 wakaba 1.79
6584 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after frameset:'.$token->{tag_name});
6585 wakaba 1.54 }
6586     ## Ignore the token
6587     $token = $self->_get_next_token;
6588     redo B;
6589     }
6590 wakaba 1.57 } elsif ($token->{type} == END_TAG_TOKEN) {
6591 wakaba 1.56 if ($self->{insertion_mode} == AFTER_HTML_FRAMESET_IM) {
6592 wakaba 1.79
6593 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after html:/'.$token->{tag_name});
6594 wakaba 1.54
6595 wakaba 1.56 $self->{insertion_mode} = AFTER_FRAMESET_IM;
6596 wakaba 1.85 ## Process in the "after frameset" insertion mode.
6597 wakaba 1.79 } else {
6598    
6599 wakaba 1.54 }
6600    
6601     if ($token->{tag_name} eq 'frameset' and
6602 wakaba 1.56 $self->{insertion_mode} == IN_FRAMESET_IM) {
6603 wakaba 1.54 if ($self->{open_elements}->[-1]->[1] eq 'html' and
6604     @{$self->{open_elements}} == 1) {
6605 wakaba 1.79
6606 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
6607 wakaba 1.54 ## Ignore the token
6608     $token = $self->_get_next_token;
6609     } else {
6610 wakaba 1.79
6611 wakaba 1.54 pop @{$self->{open_elements}};
6612     $token = $self->_get_next_token;
6613     }
6614 wakaba 1.43
6615 wakaba 1.54 if (not defined $self->{inner_html_node} and
6616     $self->{open_elements}->[-1]->[1] ne 'frameset') {
6617 wakaba 1.79
6618 wakaba 1.56 $self->{insertion_mode} = AFTER_FRAMESET_IM;
6619 wakaba 1.79 } else {
6620    
6621 wakaba 1.54 }
6622     redo B;
6623     } elsif ($token->{tag_name} eq 'html' and
6624 wakaba 1.56 $self->{insertion_mode} == AFTER_FRAMESET_IM) {
6625 wakaba 1.79
6626 wakaba 1.56 $self->{insertion_mode} = AFTER_HTML_FRAMESET_IM;
6627 wakaba 1.54 $token = $self->_get_next_token;
6628     redo B;
6629     } else {
6630 wakaba 1.56 if ($self->{insertion_mode} == IN_FRAMESET_IM) {
6631 wakaba 1.79
6632 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in frameset:/'.$token->{tag_name});
6633 wakaba 1.54 } else {
6634 wakaba 1.79
6635 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'after frameset:/'.$token->{tag_name});
6636 wakaba 1.54 }
6637     ## Ignore the token
6638     $token = $self->_get_next_token;
6639     redo B;
6640     }
6641 wakaba 1.103 } elsif ($token->{type} == END_OF_FILE_TOKEN) {
6642     unless ($self->{open_elements}->[-1]->[1] eq 'html' and
6643     @{$self->{open_elements}} == 1) { # redundant, maybe
6644    
6645 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in body:#eof');
6646 wakaba 1.103 } else {
6647    
6648     }
6649    
6650     ## Stop parsing
6651     last B;
6652 wakaba 1.54 } else {
6653     die "$0: $token->{type}: Unknown token type";
6654     }
6655 wakaba 1.43
6656 wakaba 1.54 ## ISSUE: An issue in spec here
6657     } else {
6658     die "$0: $self->{insertion_mode}: Unknown insertion mode";
6659     }
6660 wakaba 1.43
6661 wakaba 1.54 ## "in body" insertion mode
6662 wakaba 1.57 if ($token->{type} == START_TAG_TOKEN) {
6663 wakaba 1.54 if ($token->{tag_name} eq 'script') {
6664 wakaba 1.79
6665 wakaba 1.54 ## NOTE: This is an "as if in head" code clone
6666 wakaba 1.100 $script_start_tag->();
6667 wakaba 1.55 redo B;
6668 wakaba 1.54 } elsif ($token->{tag_name} eq 'style') {
6669 wakaba 1.79
6670 wakaba 1.54 ## NOTE: This is an "as if in head" code clone
6671 wakaba 1.96 $parse_rcdata->(CDATA_CONTENT_MODEL);
6672 wakaba 1.55 redo B;
6673 wakaba 1.54 } elsif ({
6674     base => 1, link => 1,
6675     }->{$token->{tag_name}}) {
6676 wakaba 1.79
6677 wakaba 1.54 ## NOTE: This is an "as if in head" code clone, only "-t" differs
6678    
6679     {
6680     my $el;
6681    
6682     $el = $self->{document}->create_element_ns
6683     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
6684    
6685     for my $attr_name (keys %{ $token->{attributes}}) {
6686     $el->set_attribute_ns (undef, [undef, $attr_name],
6687     $token->{attributes} ->{$attr_name}->{value});
6688     }
6689    
6690     $insert->($el);
6691     push @{$self->{open_elements}}, [$el, $token->{tag_name}];
6692     }
6693    
6694     pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec.
6695     $token = $self->_get_next_token;
6696 wakaba 1.55 redo B;
6697 wakaba 1.54 } elsif ($token->{tag_name} eq 'meta') {
6698     ## NOTE: This is an "as if in head" code clone, only "-t" differs
6699    
6700     {
6701     my $el;
6702    
6703     $el = $self->{document}->create_element_ns
6704     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
6705    
6706     for my $attr_name (keys %{ $token->{attributes}}) {
6707     $el->set_attribute_ns (undef, [undef, $attr_name],
6708     $token->{attributes} ->{$attr_name}->{value});
6709     }
6710    
6711     $insert->($el);
6712     push @{$self->{open_elements}}, [$el, $token->{tag_name}];
6713     }
6714    
6715 wakaba 1.68 my $meta_el = pop @{$self->{open_elements}}; ## ISSUE: This step is missing in the spec.
6716 wakaba 1.43
6717 wakaba 1.54 unless ($self->{confident}) {
6718     if ($token->{attributes}->{charset}) { ## TODO: And if supported
6719 wakaba 1.79
6720 wakaba 1.65 $self->{change_encoding}
6721     ->($self, $token->{attributes}->{charset}->{value});
6722 wakaba 1.68
6723     $meta_el->[0]->get_attribute_node_ns (undef, 'charset')
6724     ->set_user_data (manakai_has_reference =>
6725     $token->{attributes}->{charset}
6726     ->{has_reference});
6727 wakaba 1.65 } elsif ($token->{attributes}->{content}) {
6728 wakaba 1.54 ## ISSUE: Algorithm name in the spec was incorrect so that not linked to the definition.
6729 wakaba 1.65 if ($token->{attributes}->{content}->{value}
6730 wakaba 1.71 =~ /\A[^;]*;[\x09-\x0D\x20]*[Cc][Hh][Aa][Rr][Ss][Ee][Tt]
6731     [\x09-\x0D\x20]*=
6732 wakaba 1.54 [\x09-\x0D\x20]*(?>"([^"]*)"|'([^']*)'|
6733     ([^"'\x09-\x0D\x20][^\x09-\x0D\x20]*))/x) {
6734 wakaba 1.79
6735 wakaba 1.65 $self->{change_encoding}
6736     ->($self, defined $1 ? $1 : defined $2 ? $2 : $3);
6737 wakaba 1.69 $meta_el->[0]->get_attribute_node_ns (undef, 'content')
6738     ->set_user_data (manakai_has_reference =>
6739     $token->{attributes}->{content}
6740     ->{has_reference});
6741 wakaba 1.65 }
6742 wakaba 1.54 }
6743 wakaba 1.68 } else {
6744     if ($token->{attributes}->{charset}) {
6745 wakaba 1.79
6746 wakaba 1.68 $meta_el->[0]->get_attribute_node_ns (undef, 'charset')
6747     ->set_user_data (manakai_has_reference =>
6748     $token->{attributes}->{charset}
6749     ->{has_reference});
6750     }
6751 wakaba 1.69 if ($token->{attributes}->{content}) {
6752 wakaba 1.79
6753 wakaba 1.69 $meta_el->[0]->get_attribute_node_ns (undef, 'content')
6754     ->set_user_data (manakai_has_reference =>
6755     $token->{attributes}->{content}
6756     ->{has_reference});
6757     }
6758 wakaba 1.54 }
6759 wakaba 1.43
6760 wakaba 1.54 $token = $self->_get_next_token;
6761 wakaba 1.55 redo B;
6762 wakaba 1.54 } elsif ($token->{tag_name} eq 'title') {
6763 wakaba 1.79
6764 wakaba 1.54 ## NOTE: This is an "as if in head" code clone
6765 wakaba 1.96 $parse_rcdata->(RCDATA_CONTENT_MODEL);
6766 wakaba 1.55 redo B;
6767 wakaba 1.54 } elsif ($token->{tag_name} eq 'body') {
6768 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in body:body');
6769 wakaba 1.1
6770 wakaba 1.54 if (@{$self->{open_elements}} == 1 or
6771     $self->{open_elements}->[1]->[1] ne 'body') {
6772 wakaba 1.79
6773 wakaba 1.54 ## Ignore the token
6774     } else {
6775     my $body_el = $self->{open_elements}->[1]->[0];
6776     for my $attr_name (keys %{$token->{attributes}}) {
6777     unless ($body_el->has_attribute_ns (undef, $attr_name)) {
6778 wakaba 1.79
6779 wakaba 1.54 $body_el->set_attribute_ns
6780     (undef, [undef, $attr_name],
6781     $token->{attributes}->{$attr_name}->{value});
6782 wakaba 1.1 }
6783 wakaba 1.54 }
6784     }
6785     $token = $self->_get_next_token;
6786 wakaba 1.55 redo B;
6787 wakaba 1.54 } elsif ({
6788     address => 1, blockquote => 1, center => 1, dir => 1,
6789 wakaba 1.85 div => 1, dl => 1, fieldset => 1,
6790     h1 => 1, h2 => 1, h3 => 1, h4 => 1, h5 => 1, h6 => 1,
6791 wakaba 1.97 menu => 1, ol => 1, p => 1, ul => 1,
6792     pre => 1, listing => 1,
6793 wakaba 1.107 form => 1,
6794     table => 1,
6795     hr => 1,
6796 wakaba 1.54 }->{$token->{tag_name}}) {
6797 wakaba 1.107 if ($token->{tag_name} eq 'form' and defined $self->{form_element}) {
6798    
6799     $self->{parse_error}->(level => $self->{must_level}, type => 'in form:form');
6800     ## Ignore the token
6801     $token = $self->_get_next_token;
6802     redo B;
6803     }
6804    
6805 wakaba 1.54 ## has a p element in scope
6806     INSCOPE: for (reverse @{$self->{open_elements}}) {
6807     if ($_->[1] eq 'p') {
6808 wakaba 1.79
6809 wakaba 1.54 unshift @{$self->{token}}, $token;
6810 wakaba 1.57 $token = {type => END_TAG_TOKEN, tag_name => 'p'};
6811 wakaba 1.55 redo B;
6812 wakaba 1.54 } elsif ({
6813 wakaba 1.102 applet => 1, table => 1, caption => 1, td => 1, th => 1,
6814 wakaba 1.54 button => 1, marquee => 1, object => 1, html => 1,
6815     }->{$_->[1]}) {
6816 wakaba 1.79
6817 wakaba 1.54 last INSCOPE;
6818     }
6819     } # INSCOPE
6820    
6821    
6822 wakaba 1.48 {
6823     my $el;
6824    
6825     $el = $self->{document}->create_element_ns
6826 wakaba 1.54 (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
6827 wakaba 1.48
6828 wakaba 1.54 for my $attr_name (keys %{ $token->{attributes}}) {
6829     $el->set_attribute_ns (undef, [undef, $attr_name],
6830     $token->{attributes} ->{$attr_name}->{value});
6831     }
6832    
6833     $insert->($el);
6834     push @{$self->{open_elements}}, [$el, $token->{tag_name}];
6835 wakaba 1.48 }
6836    
6837 wakaba 1.97 if ($token->{tag_name} eq 'pre' or $token->{tag_name} eq 'listing') {
6838 wakaba 1.54 $token = $self->_get_next_token;
6839 wakaba 1.57 if ($token->{type} == CHARACTER_TOKEN) {
6840 wakaba 1.54 $token->{data} =~ s/^\x0A//;
6841     unless (length $token->{data}) {
6842 wakaba 1.79
6843 wakaba 1.54 $token = $self->_get_next_token;
6844 wakaba 1.79 } else {
6845    
6846 wakaba 1.54 }
6847 wakaba 1.79 } else {
6848    
6849 wakaba 1.54 }
6850 wakaba 1.107 } elsif ($token->{tag_name} eq 'form') {
6851    
6852     $self->{form_element} = $self->{open_elements}->[-1]->[0];
6853    
6854     $token = $self->_get_next_token;
6855     } elsif ($token->{tag_name} eq 'table') {
6856    
6857     push @{$open_tables}, [$self->{open_elements}->[-1]->[0]];
6858 wakaba 1.79
6859 wakaba 1.107 $self->{insertion_mode} = IN_TABLE_IM;
6860    
6861 wakaba 1.54 $token = $self->_get_next_token;
6862 wakaba 1.107 } elsif ($token->{tag_name} eq 'hr') {
6863 wakaba 1.79
6864 wakaba 1.107 pop @{$self->{open_elements}};
6865    
6866 wakaba 1.54 $token = $self->_get_next_token;
6867     } else {
6868    
6869     $token = $self->_get_next_token;
6870     }
6871 wakaba 1.55 redo B;
6872 wakaba 1.107 } elsif ({li => 1, dt => 1, dd => 1}->{$token->{tag_name}}) {
6873 wakaba 1.54 ## has a p element in scope
6874     INSCOPE: for (reverse @{$self->{open_elements}}) {
6875     if ($_->[1] eq 'p') {
6876 wakaba 1.79
6877 wakaba 1.54 unshift @{$self->{token}}, $token;
6878 wakaba 1.57 $token = {type => END_TAG_TOKEN, tag_name => 'p'};
6879 wakaba 1.55 redo B;
6880 wakaba 1.54 } elsif ({
6881 wakaba 1.102 applet => 1, table => 1, caption => 1, td => 1, th => 1,
6882 wakaba 1.54 button => 1, marquee => 1, object => 1, html => 1,
6883     }->{$_->[1]}) {
6884 wakaba 1.79
6885 wakaba 1.54 last INSCOPE;
6886     }
6887     } # INSCOPE
6888    
6889     ## Step 1
6890     my $i = -1;
6891     my $node = $self->{open_elements}->[$i];
6892 wakaba 1.107 my $li_or_dtdd = {li => {li => 1},
6893     dt => {dt => 1, dd => 1},
6894     dd => {dt => 1, dd => 1}}->{$token->{tag_name}};
6895 wakaba 1.54 LI: {
6896     ## Step 2
6897 wakaba 1.107 if ($li_or_dtdd->{$node->[1]}) {
6898 wakaba 1.54 if ($i != -1) {
6899 wakaba 1.79
6900 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'end tag missing:'.
6901 wakaba 1.54 $self->{open_elements}->[-1]->[1]);
6902 wakaba 1.79 } else {
6903    
6904 wakaba 1.54 }
6905     splice @{$self->{open_elements}}, $i;
6906     last LI;
6907 wakaba 1.79 } else {
6908    
6909 wakaba 1.54 }
6910    
6911     ## Step 3
6912     if (not $formatting_category->{$node->[1]} and
6913     #not $phrasing_category->{$node->[1]} and
6914     ($special_category->{$node->[1]} or
6915     $scoping_category->{$node->[1]}) and
6916     $node->[1] ne 'address' and $node->[1] ne 'div') {
6917 wakaba 1.79
6918 wakaba 1.54 last LI;
6919     }
6920    
6921 wakaba 1.79
6922 wakaba 1.54 ## Step 4
6923     $i--;
6924     $node = $self->{open_elements}->[$i];
6925     redo LI;
6926     } # LI
6927    
6928    
6929 wakaba 1.49 {
6930     my $el;
6931    
6932     $el = $self->{document}->create_element_ns
6933     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
6934    
6935     for my $attr_name (keys %{ $token->{attributes}}) {
6936     $el->set_attribute_ns (undef, [undef, $attr_name],
6937     $token->{attributes} ->{$attr_name}->{value});
6938     }
6939    
6940 wakaba 1.54 $insert->($el);
6941 wakaba 1.49 push @{$self->{open_elements}}, [$el, $token->{tag_name}];
6942     }
6943    
6944 wakaba 1.54 $token = $self->_get_next_token;
6945 wakaba 1.55 redo B;
6946 wakaba 1.54 } elsif ($token->{tag_name} eq 'plaintext') {
6947     ## has a p element in scope
6948     INSCOPE: for (reverse @{$self->{open_elements}}) {
6949     if ($_->[1] eq 'p') {
6950 wakaba 1.79
6951 wakaba 1.54 unshift @{$self->{token}}, $token;
6952 wakaba 1.57 $token = {type => END_TAG_TOKEN, tag_name => 'p'};
6953 wakaba 1.55 redo B;
6954 wakaba 1.54 } elsif ({
6955 wakaba 1.102 applet => 1, table => 1, caption => 1, td => 1, th => 1,
6956 wakaba 1.54 button => 1, marquee => 1, object => 1, html => 1,
6957     }->{$_->[1]}) {
6958 wakaba 1.79
6959 wakaba 1.54 last INSCOPE;
6960     }
6961     } # INSCOPE
6962    
6963    
6964 wakaba 1.1 {
6965     my $el;
6966    
6967     $el = $self->{document}->create_element_ns
6968 wakaba 1.54 (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
6969 wakaba 1.1
6970 wakaba 1.54 for my $attr_name (keys %{ $token->{attributes}}) {
6971     $el->set_attribute_ns (undef, [undef, $attr_name],
6972     $token->{attributes} ->{$attr_name}->{value});
6973     }
6974    
6975     $insert->($el);
6976     push @{$self->{open_elements}}, [$el, $token->{tag_name}];
6977 wakaba 1.1 }
6978    
6979 wakaba 1.54
6980     $self->{content_model} = PLAINTEXT_CONTENT_MODEL;
6981    
6982     $token = $self->_get_next_token;
6983 wakaba 1.55 redo B;
6984 wakaba 1.54 } elsif ($token->{tag_name} eq 'a') {
6985     AFE: for my $i (reverse 0..$#$active_formatting_elements) {
6986     my $node = $active_formatting_elements->[$i];
6987     if ($node->[1] eq 'a') {
6988 wakaba 1.79
6989 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in a:a');
6990 wakaba 1.54
6991     unshift @{$self->{token}}, $token;
6992 wakaba 1.57 $token = {type => END_TAG_TOKEN, tag_name => 'a'};
6993 wakaba 1.54 $formatting_end_tag->($token->{tag_name});
6994    
6995     AFE2: for (reverse 0..$#$active_formatting_elements) {
6996     if ($active_formatting_elements->[$_]->[0] eq $node->[0]) {
6997 wakaba 1.79
6998 wakaba 1.54 splice @$active_formatting_elements, $_, 1;
6999     last AFE2;
7000 wakaba 1.49 }
7001 wakaba 1.54 } # AFE2
7002     OE: for (reverse 0..$#{$self->{open_elements}}) {
7003     if ($self->{open_elements}->[$_]->[0] eq $node->[0]) {
7004 wakaba 1.79
7005 wakaba 1.54 splice @{$self->{open_elements}}, $_, 1;
7006     last OE;
7007 wakaba 1.49 }
7008 wakaba 1.54 } # OE
7009     last AFE;
7010     } elsif ($node->[0] eq '#marker') {
7011 wakaba 1.79
7012 wakaba 1.54 last AFE;
7013     }
7014     } # AFE
7015    
7016     $reconstruct_active_formatting_elements->($insert_to_current);
7017 wakaba 1.49
7018 wakaba 1.54
7019     {
7020     my $el;
7021    
7022     $el = $self->{document}->create_element_ns
7023     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
7024    
7025     for my $attr_name (keys %{ $token->{attributes}}) {
7026     $el->set_attribute_ns (undef, [undef, $attr_name],
7027     $token->{attributes} ->{$attr_name}->{value});
7028     }
7029    
7030     $insert->($el);
7031     push @{$self->{open_elements}}, [$el, $token->{tag_name}];
7032     }
7033    
7034     push @$active_formatting_elements, $self->{open_elements}->[-1];
7035 wakaba 1.48
7036 wakaba 1.54 $token = $self->_get_next_token;
7037 wakaba 1.55 redo B;
7038 wakaba 1.54 } elsif ($token->{tag_name} eq 'nobr') {
7039     $reconstruct_active_formatting_elements->($insert_to_current);
7040 wakaba 1.1
7041 wakaba 1.54 ## has a |nobr| element in scope
7042     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
7043     my $node = $self->{open_elements}->[$_];
7044     if ($node->[1] eq 'nobr') {
7045 wakaba 1.79
7046 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in nobr:nobr');
7047 wakaba 1.54 unshift @{$self->{token}}, $token;
7048 wakaba 1.57 $token = {type => END_TAG_TOKEN, tag_name => 'nobr'};
7049 wakaba 1.55 redo B;
7050 wakaba 1.54 } elsif ({
7051 wakaba 1.102 applet => 1, table => 1, caption => 1, td => 1, th => 1,
7052 wakaba 1.54 button => 1, marquee => 1, object => 1, html => 1,
7053     }->{$node->[1]}) {
7054 wakaba 1.79
7055 wakaba 1.54 last INSCOPE;
7056     }
7057     } # INSCOPE
7058    
7059    
7060     {
7061     my $el;
7062    
7063     $el = $self->{document}->create_element_ns
7064     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
7065    
7066     for my $attr_name (keys %{ $token->{attributes}}) {
7067     $el->set_attribute_ns (undef, [undef, $attr_name],
7068     $token->{attributes} ->{$attr_name}->{value});
7069     }
7070    
7071     $insert->($el);
7072     push @{$self->{open_elements}}, [$el, $token->{tag_name}];
7073     }
7074    
7075     push @$active_formatting_elements, $self->{open_elements}->[-1];
7076    
7077     $token = $self->_get_next_token;
7078 wakaba 1.55 redo B;
7079 wakaba 1.54 } elsif ($token->{tag_name} eq 'button') {
7080     ## has a button element in scope
7081     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
7082     my $node = $self->{open_elements}->[$_];
7083     if ($node->[1] eq 'button') {
7084 wakaba 1.79
7085 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in button:button');
7086 wakaba 1.54 unshift @{$self->{token}}, $token;
7087 wakaba 1.57 $token = {type => END_TAG_TOKEN, tag_name => 'button'};
7088 wakaba 1.55 redo B;
7089 wakaba 1.54 } elsif ({
7090 wakaba 1.102 applet => 1, table => 1, caption => 1, td => 1, th => 1,
7091 wakaba 1.54 button => 1, marquee => 1, object => 1, html => 1,
7092     }->{$node->[1]}) {
7093 wakaba 1.79
7094 wakaba 1.54 last INSCOPE;
7095 wakaba 1.1 }
7096 wakaba 1.54 } # INSCOPE
7097    
7098     $reconstruct_active_formatting_elements->($insert_to_current);
7099    
7100    
7101     {
7102     my $el;
7103    
7104     $el = $self->{document}->create_element_ns
7105     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
7106    
7107     for my $attr_name (keys %{ $token->{attributes}}) {
7108     $el->set_attribute_ns (undef, [undef, $attr_name],
7109     $token->{attributes} ->{$attr_name}->{value});
7110     }
7111    
7112     $insert->($el);
7113     push @{$self->{open_elements}}, [$el, $token->{tag_name}];
7114     }
7115    
7116 wakaba 1.86
7117     ## TODO: associate with $self->{form_element} if defined
7118    
7119 wakaba 1.54 push @$active_formatting_elements, ['#marker', ''];
7120 wakaba 1.1
7121 wakaba 1.54 $token = $self->_get_next_token;
7122 wakaba 1.55 redo B;
7123 wakaba 1.102 } elsif ({
7124 wakaba 1.107 xmp => 1,
7125     iframe => 1,
7126     noembed => 1,
7127     noframes => 1,
7128     noscript => 0, ## TODO: 1 if scripting is enabled
7129 wakaba 1.102 }->{$token->{tag_name}}) {
7130 wakaba 1.107 if ($token->{tag_name} eq 'xmp') {
7131    
7132     $reconstruct_active_formatting_elements->($insert_to_current);
7133     } else {
7134    
7135 wakaba 1.1 }
7136 wakaba 1.107 ## NOTE: There is an "as if in body" code clone.
7137 wakaba 1.96 $parse_rcdata->(CDATA_CONTENT_MODEL);
7138 wakaba 1.55 redo B;
7139 wakaba 1.54 } elsif ($token->{tag_name} eq 'isindex') {
7140 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'isindex');
7141 wakaba 1.54
7142     if (defined $self->{form_element}) {
7143 wakaba 1.79
7144 wakaba 1.54 ## Ignore the token
7145     $token = $self->_get_next_token;
7146 wakaba 1.55 redo B;
7147 wakaba 1.54 } else {
7148     my $at = $token->{attributes};
7149     my $form_attrs;
7150     $form_attrs->{action} = $at->{action} if $at->{action};
7151     my $prompt_attr = $at->{prompt};
7152     $at->{name} = {name => 'name', value => 'isindex'};
7153     delete $at->{action};
7154     delete $at->{prompt};
7155     my @tokens = (
7156 wakaba 1.57 {type => START_TAG_TOKEN, tag_name => 'form',
7157 wakaba 1.54 attributes => $form_attrs},
7158 wakaba 1.57 {type => START_TAG_TOKEN, tag_name => 'hr'},
7159     {type => START_TAG_TOKEN, tag_name => 'p'},
7160     {type => START_TAG_TOKEN, tag_name => 'label'},
7161 wakaba 1.54 );
7162     if ($prompt_attr) {
7163 wakaba 1.79
7164 wakaba 1.57 push @tokens, {type => CHARACTER_TOKEN, data => $prompt_attr->{value}};
7165 wakaba 1.1 } else {
7166 wakaba 1.79
7167 wakaba 1.57 push @tokens, {type => CHARACTER_TOKEN,
7168 wakaba 1.54 data => 'This is a searchable index. Insert your search keywords here: '}; # SHOULD
7169     ## TODO: make this configurable
7170 wakaba 1.1 }
7171 wakaba 1.54 push @tokens,
7172 wakaba 1.57 {type => START_TAG_TOKEN, tag_name => 'input', attributes => $at},
7173     #{type => CHARACTER_TOKEN, data => ''}, # SHOULD
7174     {type => END_TAG_TOKEN, tag_name => 'label'},
7175     {type => END_TAG_TOKEN, tag_name => 'p'},
7176     {type => START_TAG_TOKEN, tag_name => 'hr'},
7177     {type => END_TAG_TOKEN, tag_name => 'form'};
7178 wakaba 1.54 $token = shift @tokens;
7179     unshift @{$self->{token}}, (@tokens);
7180 wakaba 1.55 redo B;
7181 wakaba 1.54 }
7182     } elsif ($token->{tag_name} eq 'textarea') {
7183     my $tag_name = $token->{tag_name};
7184     my $el;
7185    
7186     $el = $self->{document}->create_element_ns
7187     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
7188    
7189     for my $attr_name (keys %{ $token->{attributes}}) {
7190     $el->set_attribute_ns (undef, [undef, $attr_name],
7191     $token->{attributes} ->{$attr_name}->{value});
7192     }
7193    
7194    
7195     ## TODO: $self->{form_element} if defined
7196     $self->{content_model} = RCDATA_CONTENT_MODEL;
7197     delete $self->{escape}; # MUST
7198    
7199     $insert->($el);
7200    
7201     my $text = '';
7202     $token = $self->_get_next_token;
7203 wakaba 1.57 if ($token->{type} == CHARACTER_TOKEN) {
7204 wakaba 1.54 $token->{data} =~ s/^\x0A//;
7205 wakaba 1.53 unless (length $token->{data}) {
7206 wakaba 1.79
7207 wakaba 1.53 $token = $self->_get_next_token;
7208 wakaba 1.79 } else {
7209    
7210 wakaba 1.53 }
7211 wakaba 1.79 } else {
7212    
7213 wakaba 1.53 }
7214 wakaba 1.57 while ($token->{type} == CHARACTER_TOKEN) {
7215 wakaba 1.79
7216 wakaba 1.54 $text .= $token->{data};
7217     $token = $self->_get_next_token;
7218     }
7219     if (length $text) {
7220 wakaba 1.79
7221 wakaba 1.54 $el->manakai_append_text ($text);
7222     }
7223    
7224     $self->{content_model} = PCDATA_CONTENT_MODEL;
7225    
7226 wakaba 1.57 if ($token->{type} == END_TAG_TOKEN and
7227 wakaba 1.54 $token->{tag_name} eq $tag_name) {
7228 wakaba 1.79
7229 wakaba 1.54 ## Ignore the token
7230     } else {
7231 wakaba 1.79
7232 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in RCDATA:#'.$token->{type});
7233 wakaba 1.54 }
7234     $token = $self->_get_next_token;
7235 wakaba 1.55 redo B;
7236 wakaba 1.54 } elsif ({
7237     caption => 1, col => 1, colgroup => 1, frame => 1,
7238     frameset => 1, head => 1, option => 1, optgroup => 1,
7239     tbody => 1, td => 1, tfoot => 1, th => 1,
7240     thead => 1, tr => 1,
7241     }->{$token->{tag_name}}) {
7242 wakaba 1.79
7243 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'in body:'.$token->{tag_name});
7244 wakaba 1.54 ## Ignore the token
7245     $token = $self->_get_next_token;
7246 wakaba 1.55 redo B;
7247 wakaba 1.54
7248     ## ISSUE: An issue on HTML5 new elements in the spec.
7249     } else {
7250 wakaba 1.108 if ($token->{tag_name} eq 'image') {
7251    
7252     $self->{parse_error}->(level => $self->{must_level}, type => 'image');
7253     $token->{tag_name} = 'img';
7254     } else {
7255    
7256     }
7257    
7258     ## NOTE: There is an "as if <br>" code clone.
7259 wakaba 1.54 $reconstruct_active_formatting_elements->($insert_to_current);
7260 wakaba 1.53
7261 wakaba 1.54
7262     {
7263     my $el;
7264    
7265     $el = $self->{document}->create_element_ns
7266     (q<http://www.w3.org/1999/xhtml>, [undef, $token->{tag_name}]);
7267    
7268     for my $attr_name (keys %{ $token->{attributes}}) {
7269     $el->set_attribute_ns (undef, [undef, $attr_name],
7270     $token->{attributes} ->{$attr_name}->{value});
7271 wakaba 1.53 }
7272 wakaba 1.54
7273     $insert->($el);
7274     push @{$self->{open_elements}}, [$el, $token->{tag_name}];
7275     }
7276    
7277 wakaba 1.107
7278 wakaba 1.108 if ({
7279     applet => 1, marquee => 1, object => 1,
7280     }->{$token->{tag_name}}) {
7281    
7282     push @$active_formatting_elements, ['#marker', ''];
7283     } elsif ({
7284     b => 1, big => 1, em => 1, font => 1, i => 1,
7285     s => 1, small => 1, strile => 1,
7286     strong => 1, tt => 1, u => 1,
7287     }->{$token->{tag_name}}) {
7288    
7289     push @$active_formatting_elements, $self->{open_elements}->[-1];
7290     } elsif ($token->{tag_name} eq 'input') {
7291    
7292     ## TODO: associate with $self->{form_element} if defined
7293     pop @{$self->{open_elements}};
7294     } elsif ({
7295     area => 1, basefont => 1, bgsound => 1, br => 1,
7296     embed => 1, img => 1, param => 1, spacer => 1, wbr => 1,
7297     #image => 1,
7298     }->{$token->{tag_name}}) {
7299    
7300     pop @{$self->{open_elements}};
7301     } elsif ($token->{tag_name} eq 'select') {
7302 wakaba 1.107 ## TODO: associate with $self->{form_element} if defined
7303    
7304     if ($self->{insertion_mode} & TABLE_IMS or
7305     $self->{insertion_mode} & BODY_TABLE_IMS or
7306     $self->{insertion_mode} == IN_COLUMN_GROUP_IM) {
7307    
7308     $self->{insertion_mode} = IN_SELECT_IN_TABLE_IM;
7309     } else {
7310    
7311     $self->{insertion_mode} = IN_SELECT_IM;
7312     }
7313 wakaba 1.108 } else {
7314    
7315 wakaba 1.107 }
7316 wakaba 1.53
7317 wakaba 1.54 $token = $self->_get_next_token;
7318 wakaba 1.55 redo B;
7319 wakaba 1.54 }
7320 wakaba 1.57 } elsif ($token->{type} == END_TAG_TOKEN) {
7321 wakaba 1.54 if ($token->{tag_name} eq 'body') {
7322 wakaba 1.105 ## has a |body| element in scope
7323     my $i;
7324 wakaba 1.109 INSCOPE: {
7325     for (reverse @{$self->{open_elements}}) {
7326     if ($_->[1] eq 'body') {
7327    
7328     $i = $_;
7329     last INSCOPE;
7330     } elsif ({
7331     applet => 1, table => 1, caption => 1, td => 1, th => 1,
7332     button => 1, marquee => 1, object => 1, html => 1,
7333     }->{$_->[1]}) {
7334    
7335     last;
7336     }
7337 wakaba 1.54 }
7338 wakaba 1.109
7339     $self->{parse_error}->(level => $self->{must_level}, type => 'start tag not allowed',
7340     value => $token->{tag_name});
7341 wakaba 1.105 ## NOTE: Ignore the token.
7342 wakaba 1.54 $token = $self->_get_next_token;
7343 wakaba 1.55 redo B;
7344 wakaba 1.109 } # INSCOPE
7345 wakaba 1.105
7346     for (@{$self->{open_elements}}) {
7347     unless ({
7348     dd => 1, dt => 1, li => 1, p => 1, td => 1,
7349     th => 1, tr => 1, body => 1, html => 1,
7350     tbody => 1, tfoot => 1, thead => 1,
7351     }->{$_->[1]}) {
7352    
7353 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not closed:'.$_->[1]);
7354 wakaba 1.105 last;
7355     } else {
7356    
7357     }
7358     }
7359    
7360     $self->{insertion_mode} = AFTER_BODY_IM;
7361     $token = $self->_get_next_token;
7362     redo B;
7363 wakaba 1.54 } elsif ($token->{tag_name} eq 'html') {
7364     if (@{$self->{open_elements}} > 1 and $self->{open_elements}->[1]->[1] eq 'body') {
7365     ## ISSUE: There is an issue in the spec.
7366     if ($self->{open_elements}->[-1]->[1] ne 'body') {
7367 wakaba 1.79
7368 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not closed:'.$self->{open_elements}->[1]->[1]);
7369 wakaba 1.79 } else {
7370    
7371 wakaba 1.54 }
7372 wakaba 1.56 $self->{insertion_mode} = AFTER_BODY_IM;
7373 wakaba 1.54 ## reprocess
7374 wakaba 1.55 redo B;
7375 wakaba 1.54 } else {
7376 wakaba 1.79
7377 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
7378 wakaba 1.54 ## Ignore the token
7379     $token = $self->_get_next_token;
7380 wakaba 1.55 redo B;
7381 wakaba 1.53 }
7382 wakaba 1.54 } elsif ({
7383     address => 1, blockquote => 1, center => 1, dir => 1,
7384     div => 1, dl => 1, fieldset => 1, listing => 1,
7385     menu => 1, ol => 1, pre => 1, ul => 1,
7386     dd => 1, dt => 1, li => 1,
7387 wakaba 1.102 applet => 1, button => 1, marquee => 1, object => 1,
7388 wakaba 1.54 }->{$token->{tag_name}}) {
7389     ## has an element in scope
7390     my $i;
7391     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
7392     my $node = $self->{open_elements}->[$_];
7393     if ($node->[1] eq $token->{tag_name}) {
7394 wakaba 1.79
7395 wakaba 1.54 $i = $_;
7396 wakaba 1.87 last INSCOPE;
7397 wakaba 1.54 } elsif ({
7398 wakaba 1.102 applet => 1, table => 1, caption => 1, td => 1, th => 1,
7399 wakaba 1.54 button => 1, marquee => 1, object => 1, html => 1,
7400     }->{$node->[1]}) {
7401 wakaba 1.79
7402 wakaba 1.54 last INSCOPE;
7403     }
7404     } # INSCOPE
7405 wakaba 1.89
7406     unless (defined $i) { # has an element in scope
7407    
7408 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
7409 wakaba 1.89 } else {
7410     ## Step 1. generate implied end tags
7411     while ({
7412     dd => ($token->{tag_name} ne 'dd'),
7413     dt => ($token->{tag_name} ne 'dt'),
7414     li => ($token->{tag_name} ne 'li'),
7415     p => 1,
7416     }->{$self->{open_elements}->[-1]->[1]}) {
7417    
7418     pop @{$self->{open_elements}};
7419     }
7420    
7421     ## Step 2.
7422     if ($self->{open_elements}->[-1]->[1] ne $token->{tag_name}) {
7423 wakaba 1.79
7424 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not closed:'.$self->{open_elements}->[-1]->[1]);
7425 wakaba 1.1 } else {
7426 wakaba 1.79
7427 wakaba 1.1 }
7428 wakaba 1.89
7429     ## Step 3.
7430 wakaba 1.54 splice @{$self->{open_elements}}, $i;
7431 wakaba 1.89
7432     ## Step 4.
7433     $clear_up_to_marker->()
7434     if {
7435 wakaba 1.102 applet => 1, button => 1, marquee => 1, object => 1,
7436 wakaba 1.89 }->{$token->{tag_name}};
7437 wakaba 1.54 }
7438     $token = $self->_get_next_token;
7439 wakaba 1.55 redo B;
7440 wakaba 1.54 } elsif ($token->{tag_name} eq 'form') {
7441 wakaba 1.92 undef $self->{form_element};
7442    
7443 wakaba 1.54 ## has an element in scope
7444 wakaba 1.92 my $i;
7445 wakaba 1.54 INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
7446     my $node = $self->{open_elements}->[$_];
7447     if ($node->[1] eq $token->{tag_name}) {
7448 wakaba 1.79
7449 wakaba 1.92 $i = $_;
7450 wakaba 1.54 last INSCOPE;
7451     } elsif ({
7452 wakaba 1.102 applet => 1, table => 1, caption => 1, td => 1, th => 1,
7453 wakaba 1.54 button => 1, marquee => 1, object => 1, html => 1,
7454     }->{$node->[1]}) {
7455 wakaba 1.79
7456 wakaba 1.54 last INSCOPE;
7457 wakaba 1.36 }
7458 wakaba 1.54 } # INSCOPE
7459 wakaba 1.92
7460     unless (defined $i) { # has an element in scope
7461 wakaba 1.79
7462 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
7463 wakaba 1.54 } else {
7464 wakaba 1.92 ## Step 1. generate implied end tags
7465     while ({
7466     dd => 1, dt => 1, li => 1, p => 1,
7467     }->{$self->{open_elements}->[-1]->[1]}) {
7468    
7469     pop @{$self->{open_elements}};
7470     }
7471    
7472     ## Step 2.
7473     if ($self->{open_elements}->[-1]->[1] ne $token->{tag_name}) {
7474    
7475 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not closed:'.$self->{open_elements}->[-1]->[1]);
7476 wakaba 1.92 } else {
7477    
7478     }
7479 wakaba 1.79
7480 wakaba 1.92 ## Step 3.
7481     splice @{$self->{open_elements}}, $i;
7482 wakaba 1.36 }
7483    
7484 wakaba 1.54 $token = $self->_get_next_token;
7485 wakaba 1.55 redo B;
7486 wakaba 1.54 } elsif ({
7487     h1 => 1, h2 => 1, h3 => 1, h4 => 1, h5 => 1, h6 => 1,
7488     }->{$token->{tag_name}}) {
7489     ## has an element in scope
7490     my $i;
7491     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
7492     my $node = $self->{open_elements}->[$_];
7493     if ({
7494     h1 => 1, h2 => 1, h3 => 1, h4 => 1, h5 => 1, h6 => 1,
7495     }->{$node->[1]}) {
7496 wakaba 1.79
7497 wakaba 1.54 $i = $_;
7498     last INSCOPE;
7499     } elsif ({
7500 wakaba 1.102 applet => 1, table => 1, caption => 1, td => 1, th => 1,
7501 wakaba 1.54 button => 1, marquee => 1, object => 1, html => 1,
7502     }->{$node->[1]}) {
7503 wakaba 1.79
7504 wakaba 1.54 last INSCOPE;
7505 wakaba 1.53 }
7506 wakaba 1.54 } # INSCOPE
7507 wakaba 1.93
7508     unless (defined $i) { # has an element in scope
7509 wakaba 1.79
7510 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
7511 wakaba 1.79 } else {
7512 wakaba 1.93 ## Step 1. generate implied end tags
7513     while ({
7514     dd => 1, dt => 1, li => 1, p => 1,
7515     }->{$self->{open_elements}->[-1]->[1]}) {
7516    
7517     pop @{$self->{open_elements}};
7518     }
7519 wakaba 1.79
7520 wakaba 1.93 ## Step 2.
7521     if ($self->{open_elements}->[-1]->[1] ne $token->{tag_name}) {
7522    
7523 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
7524 wakaba 1.93 } else {
7525    
7526     }
7527    
7528     ## Step 3.
7529     splice @{$self->{open_elements}}, $i;
7530 wakaba 1.53 }
7531    
7532 wakaba 1.54 $token = $self->_get_next_token;
7533 wakaba 1.55 redo B;
7534 wakaba 1.87 } elsif ($token->{tag_name} eq 'p') {
7535     ## has an element in scope
7536     my $i;
7537     INSCOPE: for (reverse 0..$#{$self->{open_elements}}) {
7538     my $node = $self->{open_elements}->[$_];
7539     if ($node->[1] eq $token->{tag_name}) {
7540    
7541     $i = $_;
7542 wakaba 1.88 last INSCOPE;
7543 wakaba 1.87 } elsif ({
7544 wakaba 1.102 applet => 1, table => 1, caption => 1, td => 1, th => 1,
7545 wakaba 1.87 button => 1, marquee => 1, object => 1, html => 1,
7546     }->{$node->[1]}) {
7547    
7548     last INSCOPE;
7549     }
7550     } # INSCOPE
7551 wakaba 1.91
7552     if (defined $i) {
7553     if ($self->{open_elements}->[-1]->[1] ne $token->{tag_name}) {
7554 wakaba 1.87
7555 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not closed:'.$self->{open_elements}->[-1]->[1]);
7556 wakaba 1.87 } else {
7557    
7558     }
7559 wakaba 1.91
7560 wakaba 1.87 splice @{$self->{open_elements}}, $i;
7561     } else {
7562    
7563 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
7564 wakaba 1.91
7565    
7566 wakaba 1.87 ## As if <p>, then reprocess the current token
7567     my $el;
7568    
7569     $el = $self->{document}->create_element_ns
7570     (q<http://www.w3.org/1999/xhtml>, [undef, 'p']);
7571    
7572     $insert->($el);
7573 wakaba 1.91 ## NOTE: Not inserted into |$self->{open_elements}|.
7574 wakaba 1.87 }
7575 wakaba 1.91
7576 wakaba 1.87 $token = $self->_get_next_token;
7577     redo B;
7578 wakaba 1.54 } elsif ({
7579     a => 1,
7580     b => 1, big => 1, em => 1, font => 1, i => 1,
7581     nobr => 1, s => 1, small => 1, strile => 1,
7582     strong => 1, tt => 1, u => 1,
7583     }->{$token->{tag_name}}) {
7584 wakaba 1.79
7585 wakaba 1.54 $formatting_end_tag->($token->{tag_name});
7586 wakaba 1.55 redo B;
7587 wakaba 1.54 } elsif ($token->{tag_name} eq 'br') {
7588 wakaba 1.79
7589 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:br');
7590 wakaba 1.53
7591 wakaba 1.54 ## As if <br>
7592     $reconstruct_active_formatting_elements->($insert_to_current);
7593    
7594     my $el;
7595    
7596 wakaba 1.1 $el = $self->{document}->create_element_ns
7597 wakaba 1.54 (q<http://www.w3.org/1999/xhtml>, [undef, 'br']);
7598 wakaba 1.1
7599 wakaba 1.54 $insert->($el);
7600    
7601     ## Ignore the token.
7602     $token = $self->_get_next_token;
7603 wakaba 1.55 redo B;
7604 wakaba 1.54 } elsif ({
7605     caption => 1, col => 1, colgroup => 1, frame => 1,
7606     frameset => 1, head => 1, option => 1, optgroup => 1,
7607     tbody => 1, td => 1, tfoot => 1, th => 1,
7608     thead => 1, tr => 1,
7609     area => 1, basefont => 1, bgsound => 1,
7610     embed => 1, hr => 1, iframe => 1, image => 1,
7611     img => 1, input => 1, isindex => 1, noembed => 1,
7612     noframes => 1, param => 1, select => 1, spacer => 1,
7613     table => 1, textarea => 1, wbr => 1,
7614     noscript => 0, ## TODO: if scripting is enabled
7615     }->{$token->{tag_name}}) {
7616 wakaba 1.79
7617 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
7618 wakaba 1.54 ## Ignore the token
7619     $token = $self->_get_next_token;
7620 wakaba 1.55 redo B;
7621 wakaba 1.54
7622     ## ISSUE: Issue on HTML5 new elements in spec
7623    
7624     } else {
7625     ## Step 1
7626     my $node_i = -1;
7627     my $node = $self->{open_elements}->[$node_i];
7628    
7629     ## Step 2
7630     S2: {
7631     if ($node->[1] eq $token->{tag_name}) {
7632     ## Step 1
7633     ## generate implied end tags
7634 wakaba 1.86 while ({
7635     dd => 1, dt => 1, li => 1, p => 1,
7636     }->{$self->{open_elements}->[-1]->[1]}) {
7637 wakaba 1.79
7638 wakaba 1.83 ## ISSUE: Can this case be reached?
7639 wakaba 1.86 pop @{$self->{open_elements}};
7640 wakaba 1.54 }
7641    
7642     ## Step 2
7643     if ($token->{tag_name} ne $self->{open_elements}->[-1]->[1]) {
7644 wakaba 1.79
7645 wakaba 1.60 ## NOTE: <x><y></x>
7646 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'not closed:'.$self->{open_elements}->[-1]->[1]);
7647 wakaba 1.79 } else {
7648    
7649 wakaba 1.54 }
7650    
7651     ## Step 3
7652     splice @{$self->{open_elements}}, $node_i;
7653 wakaba 1.53
7654 wakaba 1.36 $token = $self->_get_next_token;
7655 wakaba 1.54 last S2;
7656 wakaba 1.1 } else {
7657 wakaba 1.54 ## Step 3
7658     if (not $formatting_category->{$node->[1]} and
7659     #not $phrasing_category->{$node->[1]} and
7660     ($special_category->{$node->[1]} or
7661     $scoping_category->{$node->[1]})) {
7662 wakaba 1.79
7663 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'unmatched end tag:'.$token->{tag_name});
7664 wakaba 1.54 ## Ignore the token
7665     $token = $self->_get_next_token;
7666     last S2;
7667     }
7668 wakaba 1.79
7669    
7670 wakaba 1.1 }
7671 wakaba 1.54
7672     ## Step 4
7673     $node_i--;
7674     $node = $self->{open_elements}->[$node_i];
7675    
7676     ## Step 5;
7677     redo S2;
7678     } # S2
7679 wakaba 1.55 redo B;
7680 wakaba 1.1 }
7681     }
7682 wakaba 1.54 redo B;
7683 wakaba 1.1 } # B
7684    
7685     ## Stop parsing # MUST
7686    
7687     ## TODO: script stuffs
7688 wakaba 1.3 } # _tree_construct_main
7689    
7690     sub set_inner_html ($$$) {
7691     my $class = shift;
7692     my $node = shift;
7693     my $s = \$_[0];
7694     my $onerror = $_[1];
7695    
7696 wakaba 1.65 ## ISSUE: Should {confident} be true?
7697    
7698 wakaba 1.3 my $nt = $node->node_type;
7699     if ($nt == 9) {
7700     # MUST
7701    
7702     ## Step 1 # MUST
7703     ## TODO: If the document has an active parser, ...
7704     ## ISSUE: There is an issue in the spec.
7705    
7706     ## Step 2 # MUST
7707     my @cn = @{$node->child_nodes};
7708     for (@cn) {
7709     $node->remove_child ($_);
7710     }
7711    
7712     ## Step 3, 4, 5 # MUST
7713     $class->parse_string ($$s => $node, $onerror);
7714     } elsif ($nt == 1) {
7715     ## TODO: If non-html element
7716    
7717     ## NOTE: Most of this code is copied from |parse_string|
7718    
7719     ## Step 1 # MUST
7720 wakaba 1.14 my $this_doc = $node->owner_document;
7721     my $doc = $this_doc->implementation->create_document;
7722 wakaba 1.18 $doc->manakai_is_html (1);
7723 wakaba 1.3 my $p = $class->new;
7724     $p->{document} = $doc;
7725    
7726 wakaba 1.84 ## Step 8 # MUST
7727 wakaba 1.3 my $i = 0;
7728     my $line = 1;
7729     my $column = 0;
7730 wakaba 1.76 $p->{set_next_char} = sub {
7731 wakaba 1.3 my $self = shift;
7732 wakaba 1.14
7733 wakaba 1.76 pop @{$self->{prev_char}};
7734     unshift @{$self->{prev_char}}, $self->{next_char};
7735 wakaba 1.14
7736 wakaba 1.76 $self->{next_char} = -1 and return if $i >= length $$s;
7737     $self->{next_char} = ord substr $$s, $i++, 1;
7738 wakaba 1.3 $column++;
7739 wakaba 1.4
7740 wakaba 1.76 if ($self->{next_char} == 0x000A) { # LF
7741 wakaba 1.4 $line++;
7742     $column = 0;
7743 wakaba 1.79
7744 wakaba 1.76 } elsif ($self->{next_char} == 0x000D) { # CR
7745 wakaba 1.15 $i++ if substr ($$s, $i, 1) eq "\x0A";
7746 wakaba 1.76 $self->{next_char} = 0x000A; # LF # MUST
7747 wakaba 1.3 $line++;
7748 wakaba 1.4 $column = 0;
7749 wakaba 1.79
7750 wakaba 1.76 } elsif ($self->{next_char} > 0x10FFFF) {
7751     $self->{next_char} = 0xFFFD; # REPLACEMENT CHARACTER # MUST
7752 wakaba 1.79
7753 wakaba 1.76 } elsif ($self->{next_char} == 0x0000) { # NULL
7754 wakaba 1.79
7755 wakaba 1.106 $self->{parse_error}->(level => $self->{must_level}, type => 'NULL');
7756 wakaba 1.76 $self->{next_char} = 0xFFFD; # REPLACEMENT CHARACTER # MUST
7757 wakaba 1.3 }
7758     };
7759 wakaba 1.76 $p->{prev_char} = [-1, -1, -1];
7760     $p->{next_char} = -1;
7761 wakaba 1.3
7762     my $ponerror = $onerror || sub {
7763     my (%opt) = @_;
7764     warn "Parse error ($opt{type}) at line $opt{line} column $opt{column}\n";
7765     };
7766     $p->{parse_error} = sub {
7767     $ponerror->(@_, line => $line, column => $column);
7768     };
7769    
7770     $p->_initialize_tokenizer;
7771     $p->_initialize_tree_constructor;
7772    
7773     ## Step 2
7774 wakaba 1.71 my $node_ln = $node->manakai_local_name;
7775 wakaba 1.41 $p->{content_model} = {
7776     title => RCDATA_CONTENT_MODEL,
7777     textarea => RCDATA_CONTENT_MODEL,
7778     style => CDATA_CONTENT_MODEL,
7779     script => CDATA_CONTENT_MODEL,
7780     xmp => CDATA_CONTENT_MODEL,
7781     iframe => CDATA_CONTENT_MODEL,
7782     noembed => CDATA_CONTENT_MODEL,
7783     noframes => CDATA_CONTENT_MODEL,
7784     noscript => CDATA_CONTENT_MODEL,
7785     plaintext => PLAINTEXT_CONTENT_MODEL,
7786     }->{$node_ln};
7787     $p->{content_model} = PCDATA_CONTENT_MODEL
7788     unless defined $p->{content_model};
7789     ## ISSUE: What is "the name of the element"? local name?
7790 wakaba 1.3
7791     $p->{inner_html_node} = [$node, $node_ln];
7792    
7793 wakaba 1.84 ## Step 3
7794 wakaba 1.3 my $root = $doc->create_element_ns
7795     ('http://www.w3.org/1999/xhtml', [undef, 'html']);
7796    
7797 wakaba 1.84 ## Step 4 # MUST
7798 wakaba 1.3 $doc->append_child ($root);
7799    
7800 wakaba 1.84 ## Step 5 # MUST
7801 wakaba 1.3 push @{$p->{open_elements}}, [$root, 'html'];
7802    
7803     undef $p->{head_element};
7804    
7805 wakaba 1.84 ## Step 6 # MUST
7806 wakaba 1.3 $p->_reset_insertion_mode;
7807    
7808 wakaba 1.84 ## Step 7 # MUST
7809 wakaba 1.3 my $anode = $node;
7810     AN: while (defined $anode) {
7811     if ($anode->node_type == 1) {
7812     my $nsuri = $anode->namespace_uri;
7813     if (defined $nsuri and $nsuri eq 'http://www.w3.org/1999/xhtml') {
7814 wakaba 1.71 if ($anode->manakai_local_name eq 'form') {
7815 wakaba 1.79
7816 wakaba 1.3 $p->{form_element} = $anode;
7817     last AN;
7818     }
7819     }
7820     }
7821     $anode = $anode->parent_node;
7822     } # AN
7823    
7824 wakaba 1.84 ## Step 9 # MUST
7825 wakaba 1.3 {
7826     my $self = $p;
7827     $token = $self->_get_next_token;
7828     }
7829     $p->_tree_construction_main;
7830    
7831 wakaba 1.84 ## Step 10 # MUST
7832 wakaba 1.3 my @cn = @{$node->child_nodes};
7833     for (@cn) {
7834     $node->remove_child ($_);
7835     }
7836     ## ISSUE: mutation events? read-only?
7837    
7838 wakaba 1.84 ## Step 11 # MUST
7839 wakaba 1.3 @cn = @{$root->child_nodes};
7840     for (@cn) {
7841 wakaba 1.14 $this_doc->adopt_node ($_);
7842 wakaba 1.3 $node->append_child ($_);
7843     }
7844 wakaba 1.14 ## ISSUE: mutation events?
7845 wakaba 1.3
7846     $p->_terminate_tree_constructor;
7847     } else {
7848     die "$0: |set_inner_html| is not defined for node of type $nt";
7849     }
7850     } # set_inner_html
7851    
7852     } # tree construction stage
7853 wakaba 1.1
7854 wakaba 1.65 package Whatpm::HTML::RestartParser;
7855     push our @ISA, 'Error';
7856    
7857 wakaba 1.1 1;
7858 wakaba 1.109 # $Date: 2008/03/11 01:23:50 $

[email protected]
ViewVC Help
Powered by ViewVC 1.1.24