| 1 |
package Whatpm::SWML::Parser; |
| 2 |
use strict; |
| 3 |
|
| 4 |
sub AA_NS () { q<http://pc5.2ch.net/test/read.cgi/hp/1096723178/aavocab#> } |
| 5 |
sub HTML_NS () { q<http://www.w3.org/1999/xhtml> } |
| 6 |
sub SW09_NS () { q<urn:x-suika-fam-cx:markup:suikawiki:0:9:> } |
| 7 |
sub SW10_NS () { q<urn:x-suika-fam-cx:markup:suikawiki:0:10:> } |
| 8 |
sub XML_NS () { q<http://www.w3.org/XML/1998/namespace> } |
| 9 |
|
| 10 |
sub IN_SECTION_IM () { 0 } |
| 11 |
sub IN_TABLE_ROW_IM () { 1 } |
| 12 |
sub IN_PARAGRAPH_IM () { 2 } |
| 13 |
|
| 14 |
sub BLOCK_START_TAG_TOKEN () { 1 } |
| 15 |
sub BLOCK_END_TAG_TOKEN () { 2 } |
| 16 |
sub CHARACTER_TOKEN () { 3 } |
| 17 |
sub COMMENT_PARAGRAPH_START_TOKEN () { 4 } |
| 18 |
sub EDITORIAL_NOTE_START_TOKEN () { 5 } |
| 19 |
sub ELEMENT_TOKEN () { 6 } |
| 20 |
sub EMPHASIS_TOKEN () { 7 } |
| 21 |
sub EMPTY_LINE_TOKEN () { 8 } |
| 22 |
sub END_OF_FILE_TOKEN () { 9 } |
| 23 |
sub FORM_TOKEN () { 10 } |
| 24 |
sub HEADING_START_TOKEN () { 11 } |
| 25 |
sub HEADING_END_TOKEN () { 12 } |
| 26 |
sub INLINE_START_TAG_TOKEN () { 13 } |
| 27 |
sub INLINE_MIDDLE_TAG_TOKEN () { 14 } |
| 28 |
sub INLINE_END_TAG_TOKEN () { 15 } |
| 29 |
sub LABELED_LIST_START_TOKEN () { 16 } |
| 30 |
sub LABELED_LIST_MIDDLE_TOKEN () { 17 } |
| 31 |
sub LIST_START_TOKEN () { 18 } |
| 32 |
sub PREFORMATTED_START_TOKEN () { 19 } |
| 33 |
sub PREFORMATTED_END_TOKEN () { 20 } |
| 34 |
sub QUOTATION_START_TOKEN () { 21 } |
| 35 |
sub STRONG_TOKEN () { 22 } |
| 36 |
sub TABLE_ROW_START_TOKEN () { 23 } |
| 37 |
sub TABLE_ROW_END_TOKEN () { 24 } |
| 38 |
sub TABLE_CELL_START_TOKEN () { 25 } |
| 39 |
sub TABLE_CELL_END_TOKEN () { 26 } |
| 40 |
sub TABLE_COLSPAN_CELL_TOKEN () { 27 } |
| 41 |
|
| 42 |
sub new ($) { |
| 43 |
my $self = bless { |
| 44 |
}, $_[0]; |
| 45 |
return $self; |
| 46 |
} # new |
| 47 |
|
| 48 |
sub parse_char_string ($$$;$) { |
| 49 |
my $self = shift; |
| 50 |
my @s = split /\x0D\x0A?|\x0A/, ref $_[0] ? ${$_[0]} : $_[0], -1; |
| 51 |
|
| 52 |
my $doc = $_[1]; |
| 53 |
@{$doc->child_nodes} = (); |
| 54 |
my $html_el = $doc->create_element_ns (HTML_NS, [undef, 'html']); |
| 55 |
$doc->append_child ($html_el); |
| 56 |
$html_el->set_attribute_ns |
| 57 |
('http://www.w3.org/2000/xmlns/', [undef, 'xmlns'] => HTML_NS); |
| 58 |
my $head_el = $doc->create_element_ns (HTML_NS, [undef, 'head']); |
| 59 |
$html_el->append_child ($head_el); |
| 60 |
my $body_el = $doc->create_element_ns (HTML_NS, [undef, 'body']); |
| 61 |
$html_el->append_child ($body_el); |
| 62 |
for ($doc, $html_el, $head_el, $body_el) { |
| 63 |
$_->set_user_data (manakai_source_line => 1); |
| 64 |
$_->set_user_data (manakai_source_column => 1); |
| 65 |
} |
| 66 |
|
| 67 |
my $_onerror = $_[2] || sub { |
| 68 |
my %opt = @_; |
| 69 |
my $r = 'Line ' . $opt{line} . ' column ' . $opt{column} . ': '; |
| 70 |
|
| 71 |
if ($opt{token}) { |
| 72 |
$r .= 'Token ' . (defined $opt{token}->{value} |
| 73 |
? $opt{token}->{value} : $opt{token}->{type}) . ': '; |
| 74 |
} |
| 75 |
|
| 76 |
$r .= $opt{type} . ';' . $opt{level}; |
| 77 |
|
| 78 |
warn $r . "\n"; |
| 79 |
}; # $_onerror |
| 80 |
|
| 81 |
my $line = 0; |
| 82 |
my $column = 0; |
| 83 |
my $token; |
| 84 |
my @nt; |
| 85 |
|
| 86 |
my $onerror = sub { |
| 87 |
$_onerror->(line => $line, column => $column, token => $token, @_); |
| 88 |
}; # $onerror |
| 89 |
|
| 90 |
my $continuous_line; |
| 91 |
|
| 92 |
my $tokenize_text = sub { |
| 93 |
my $s = shift; # ref |
| 94 |
|
| 95 |
my $nest_level = 0; |
| 96 |
|
| 97 |
if ($$s =~ s/^\[([0-9]+)\]//) { |
| 98 |
push @nt, {type => ELEMENT_TOKEN, |
| 99 |
local_name => 'anchor-end', namespace => SW09_NS, |
| 100 |
anchor => $1, content => '[' . $1 . ']'}; |
| 101 |
$column += $+[0] - $-[0]; |
| 102 |
} |
| 103 |
|
| 104 |
while (length $$s) { |
| 105 |
if ($$s =~ s/^\[\[#([a-z-]+)//) { |
| 106 |
$column = $+[0] - $-[0]; |
| 107 |
my $t = {type => FORM_TOKEN, name => $1, |
| 108 |
line => $line, column => $column}; |
| 109 |
if ($$s =~ s/^\(([^()\\]*)\)//) { |
| 110 |
$t->{id} = $1; |
| 111 |
$column += $+[0] - $-[0]; |
| 112 |
} |
| 113 |
my @param; |
| 114 |
while ($$s =~ s/^://) { |
| 115 |
if ($$s =~ s/^'((?>[^'\\]|\\.)*)//) { |
| 116 |
$column += 1 + $+[0] - $-[0]; |
| 117 |
my $n = $1; |
| 118 |
$n =~ tr/\\//d; |
| 119 |
push @param, $n; |
| 120 |
$column++ if $$s =~ s/\A\\\z//; |
| 121 |
$column++ if $$s =~ s/^'//; |
| 122 |
} elsif ($$s =~ s/^([^':\]][^:\]]*)//) { |
| 123 |
$column += 1 + $+[0] - $-[0]; |
| 124 |
push @param, $1; |
| 125 |
} |
| 126 |
} |
| 127 |
$t->{parameters} = \@param; |
| 128 |
$column += 2 if $$s =~ s/^\]\]//; |
| 129 |
push @nt, $t; |
| 130 |
} elsif ($$s =~ s/^\[\[//) { |
| 131 |
push @nt, {type => INLINE_START_TAG_TOKEN}; |
| 132 |
$column += 2; |
| 133 |
$nest_level++; |
| 134 |
} elsif ($$s =~ s/^\[([A-Z]+)(?>\(([^()\\]*)\))?(?>\@([0-9A-Za-z-]*))?\[//) { |
| 135 |
push @nt, {type => INLINE_START_TAG_TOKEN, |
| 136 |
tag_name => $1, classes => $2, language => $3, |
| 137 |
line => $line, column => $column}; |
| 138 |
$column += $+[0] - $-[0]; |
| 139 |
$nest_level++; |
| 140 |
} elsif ($$s =~ s/^\]\]//) { |
| 141 |
push @nt, {type => INLINE_END_TAG_TOKEN, |
| 142 |
line => $line, column => $column}; |
| 143 |
$column += 2; |
| 144 |
$nest_level-- if $nest_level > 0; |
| 145 |
} elsif ($$s =~ s/^(\]?)<([0-9A-Za-z%+._-]+)://) { |
| 146 |
my $t = {type => $1 ? INLINE_END_TAG_TOKEN : ELEMENT_TOKEN, |
| 147 |
res_scheme => $2, res_parameter => '', |
| 148 |
line => $line, column => $column}; |
| 149 |
$column += $+[0] - $-[0]; |
| 150 |
|
| 151 |
while (length $$s) { |
| 152 |
if ($$s =~ s/^([^>"]+)//) { |
| 153 |
$t->{res_parameter} .= $1; |
| 154 |
$column += $+[0] - $-[0]; |
| 155 |
} elsif ($$s =~ s/^("(?>[^"\\]|\\.)*)//) { |
| 156 |
$t->{res_parameter} .= $1; |
| 157 |
$column += $+[0] - $-[0]; |
| 158 |
$column++ if $$s =~ s/\A\\\z//; |
| 159 |
$t->{res_parameter} .= '"' and $column++ if $$s =~ s/^"//; |
| 160 |
} else { |
| 161 |
last; |
| 162 |
} |
| 163 |
} |
| 164 |
|
| 165 |
$column++ if $$s =~ s/^>//; |
| 166 |
|
| 167 |
$t->{content} = $t->{res_scheme} . ':' . $t->{res_parameter}; |
| 168 |
if ($t->{res_scheme} !~ /[A-Z]/) { |
| 169 |
$t->{res_parameter} = $t->{content}; |
| 170 |
$t->{res_scheme} = 'URI'; |
| 171 |
} |
| 172 |
|
| 173 |
if ($t->{type} == INLINE_END_TAG_TOKEN) { |
| 174 |
$column++ if $$s =~ s/^\]//; |
| 175 |
$nest_level-- if $nest_level > 0; |
| 176 |
} else { |
| 177 |
$t->{local_name} = 'anchor-external'; |
| 178 |
$t->{namespace} = SW09_NS; |
| 179 |
} |
| 180 |
push @nt, $t; |
| 181 |
} elsif ($$s =~ s/^\]>>([0-9]+)\]//) { |
| 182 |
push @nt, {type => INLINE_END_TAG_TOKEN, |
| 183 |
anchor => $1, |
| 184 |
line => $line, column => $column}; |
| 185 |
$column += $+[0] - $-[0]; |
| 186 |
$nest_level-- if $nest_level > 0; |
| 187 |
} elsif ($nest_level > 0 and |
| 188 |
$$s =~ s/^\][\x09\x20]*(?>\@([0-9a-zA-Z-]*))?\[//) { |
| 189 |
push @nt, {type => INLINE_MIDDLE_TAG_TOKEN, |
| 190 |
language => $1, |
| 191 |
line => $line, column => $column}; |
| 192 |
$column += $+[0] - $-[0]; |
| 193 |
} elsif ($$s =~ s/^''('?)//) { |
| 194 |
push @nt, {type => $1 ? STRONG_TOKEN : EMPHASIS_TOKEN, |
| 195 |
line => $line, column => $column}; |
| 196 |
$column += $+[0] - $-[0]; |
| 197 |
} elsif ($$s =~ s/^>>([0-9]+)//) { |
| 198 |
push @nt, {type => ELEMENT_TOKEN, |
| 199 |
local_name => 'anchor-internal', namespace => SW09_NS, |
| 200 |
anchor => $1, content => '>>' . $1, |
| 201 |
line => $line, column => $column}; |
| 202 |
$column += $+[0] - $-[0]; |
| 203 |
} elsif ($$s =~ s/^__&&//) { |
| 204 |
if ($$s =~ s/^(.+?)&&__//) { |
| 205 |
push @nt, {type => ELEMENT_TOKEN, |
| 206 |
local_name => 'replace', namespace => SW09_NS, |
| 207 |
by => $1, |
| 208 |
line => $line, column => $column}; |
| 209 |
$column += 4 + $+[0] - $-[0]; |
| 210 |
} else { |
| 211 |
push @nt, {type => CHARACTER_TOKEN, |
| 212 |
data => '__&&', |
| 213 |
line => $line, column => $column}; |
| 214 |
$column += 4; |
| 215 |
} |
| 216 |
} elsif ($$s =~ s/^([^<>\[\]'_]+)//) { |
| 217 |
push @nt, {type => CHARACTER_TOKEN, data => $1, |
| 218 |
line => $line, column => $column}; |
| 219 |
$column += $+[0] - $-[0]; |
| 220 |
} else { |
| 221 |
push @nt, {type => CHARACTER_TOKEN, data => substr ($$s, 0, 1), |
| 222 |
line => $line, column => $column}; |
| 223 |
substr ($$s, 0, 1) = ''; |
| 224 |
$column++; |
| 225 |
} |
| 226 |
} |
| 227 |
}; # $tokenize_text |
| 228 |
|
| 229 |
my $get_next_token = sub { |
| 230 |
if (@nt) { |
| 231 |
return shift @nt; |
| 232 |
} |
| 233 |
|
| 234 |
if (not @s) { |
| 235 |
return {type => END_OF_FILE_TOKEN, line => $line, column => $column}; |
| 236 |
} |
| 237 |
|
| 238 |
my $s = shift @s; |
| 239 |
($line, $column) = ($line + 1, 1); |
| 240 |
if ($s eq '') { |
| 241 |
undef $continuous_line; |
| 242 |
return {type => EMPTY_LINE_TOKEN, line => $line, column => $column}; |
| 243 |
} elsif ($s =~ /^[\x09\x20]/) { |
| 244 |
push @nt, {type => PREFORMATTED_START_TOKEN, |
| 245 |
line => $line, column => $column}; |
| 246 |
$tokenize_text->(\$s); |
| 247 |
while (@s) { |
| 248 |
my $s = shift @s; |
| 249 |
($line, $column) = ($line + 1, 1); |
| 250 |
if ($s eq '') { |
| 251 |
push @nt, {type => PREFORMATTED_END_TOKEN, |
| 252 |
line => $line, column => $column}; |
| 253 |
unshift @s, $s; |
| 254 |
$line--; |
| 255 |
last; |
| 256 |
} elsif ($s =~ /\A\](INS|DEL)\][\x09\x20]*\z/) { |
| 257 |
push @nt, {type => PREFORMATTED_END_TOKEN, |
| 258 |
line => $line, column => $column}; |
| 259 |
push @nt, {type => BLOCK_END_TAG_TOKEN, tag_name => $1, |
| 260 |
line => $line, column => $column}; |
| 261 |
last; |
| 262 |
} else { |
| 263 |
push @nt, {type => CHARACTER_TOKEN, data => "\x0A", |
| 264 |
line => $line, column => $column}; |
| 265 |
$tokenize_text->(\$s); |
| 266 |
} |
| 267 |
} |
| 268 |
return shift @nt; |
| 269 |
} elsif ($s =~ s/^(\*+)[\x09\x20]*//) { |
| 270 |
push @nt, {type => HEADING_START_TOKEN, depth => length $1, |
| 271 |
line => $line, column => $column}; |
| 272 |
$column += $+[0] - $-[0]; |
| 273 |
$tokenize_text->(\$s); |
| 274 |
push @nt, {type => HEADING_END_TOKEN, |
| 275 |
line => $line, column => $column}; |
| 276 |
undef $continuous_line; |
| 277 |
return shift @nt; |
| 278 |
} elsif ($s =~ s/^([-=]+)[\x09\x20]*//) { |
| 279 |
push @nt, {type => LIST_START_TOKEN, depth => $1, |
| 280 |
line => $line, column => $column}; |
| 281 |
$column += $+[0] - $-[0]; |
| 282 |
$tokenize_text->(\$s); |
| 283 |
$continuous_line = 1; |
| 284 |
return shift @nt; |
| 285 |
} elsif ($s =~ s/^:([^:]*)//) { |
| 286 |
my $name = $1; |
| 287 |
if ($s eq '') { |
| 288 |
push @nt, {type => CHARACTER_TOKEN, data => ':', |
| 289 |
line => $line, column => $column}; |
| 290 |
$column++; |
| 291 |
$tokenize_text->(\$name); |
| 292 |
} else { |
| 293 |
my $real_column = $column + 1 + length $name; |
| 294 |
push @nt, {type => LABELED_LIST_START_TOKEN, |
| 295 |
line => $line, column => $column}; |
| 296 |
$name =~ s/\A[\x09\x20]*//; |
| 297 |
$column += 1 + $+[0] - $-[0]; |
| 298 |
$name =~ s/[\x09\x20]+\z//; |
| 299 |
$tokenize_text->(\$name); |
| 300 |
$column = $real_column; |
| 301 |
push @nt, {type => LABELED_LIST_MIDDLE_TOKEN, |
| 302 |
line => $line, column => $column}; |
| 303 |
$column += $+[0] - $-[0] if $s =~ s/^:[\x09\x20]*//; |
| 304 |
$tokenize_text->(\$s); |
| 305 |
} |
| 306 |
$continuous_line = 1; |
| 307 |
return shift @nt; |
| 308 |
} elsif ($s =~ s/^(>+)//) { |
| 309 |
my $depth = length $1; |
| 310 |
if ($depth == 2 and $s =~ /^[0-9]/) { |
| 311 |
push @nt, {type => CHARACTER_TOKEN, data => "\x0A", |
| 312 |
line => $line, column => $column} |
| 313 |
if $continuous_line; |
| 314 |
$s = '>>' . $s; |
| 315 |
$tokenize_text->(\$s); |
| 316 |
$continuous_line = 1; |
| 317 |
} else { |
| 318 |
push @nt, {type => QUOTATION_START_TOKEN, depth => $depth, |
| 319 |
line => $line, column => $column}; |
| 320 |
$column += $depth; |
| 321 |
$column += $+[0] - $-[0] if $s =~ s/^[\x09\x20]+//; |
| 322 |
if ($s =~ s/^\@\@[\x09\x20]*//) { |
| 323 |
push @nt, {type => EDITORIAL_NOTE_START_TOKEN, |
| 324 |
line => $line, column => $column}; |
| 325 |
$column += $+[0] - $-[0]; |
| 326 |
$continuous_line = 1; |
| 327 |
} elsif ($s =~ s/^;;[\x09\x20]*//) { |
| 328 |
push @nt, {type => COMMENT_PARAGRAPH_START_TOKEN, |
| 329 |
line => $line, column => $column}; |
| 330 |
$column += $+[0] - $-[0]; |
| 331 |
$continuous_line = 1; |
| 332 |
} elsif (length $s) { |
| 333 |
$continuous_line = 1; |
| 334 |
} else { |
| 335 |
$continuous_line = 0; |
| 336 |
} |
| 337 |
$tokenize_text->(\$s); |
| 338 |
} |
| 339 |
return shift @nt; |
| 340 |
} elsif ($s =~ /\A\[(INS|DEL)(?>\(([^()\\]*)\))?\[[\x09\x20]*\z/) { |
| 341 |
undef $continuous_line; |
| 342 |
return {type => BLOCK_START_TAG_TOKEN, tag_name => $1, |
| 343 |
classes => $2, |
| 344 |
line => $line, column => $column}; |
| 345 |
} elsif ($s =~ /\A\[PRE(?>\(([^()\\]*)\))?\[[\x09\x20]*\z/) { |
| 346 |
undef $continuous_line; |
| 347 |
push @nt, {type => BLOCK_START_TAG_TOKEN, tag_name => 'PRE', |
| 348 |
classes => $1, |
| 349 |
line => $line, column => $column}; |
| 350 |
while (@s) { |
| 351 |
my $s = shift @s; |
| 352 |
($line, $column) = ($line + 1, 1); |
| 353 |
if ($s =~ /\A\]PRE\][\x09\x20]*\z/) { |
| 354 |
push @nt, {type => BLOCK_END_TAG_TOKEN, tag_name => 'PRE', |
| 355 |
line => $line, column => $column}; |
| 356 |
undef $continuous_line; |
| 357 |
last; |
| 358 |
} else { |
| 359 |
push @nt, {type => CHARACTER_TOKEN, data => "\x0A", |
| 360 |
line => $line, column => $column} |
| 361 |
if $continuous_line; |
| 362 |
$tokenize_text->(\$s); |
| 363 |
$continuous_line = 1; |
| 364 |
} |
| 365 |
} |
| 366 |
return shift @nt; |
| 367 |
} elsif ($s =~ s/^\@\@[\x09\x20]*//) { |
| 368 |
push @nt, {type => EDITORIAL_NOTE_START_TOKEN, |
| 369 |
line => $line, column => $column}; |
| 370 |
$column += $+[0] - $-[0]; |
| 371 |
$tokenize_text->(\$s); |
| 372 |
$continuous_line = 1; |
| 373 |
return shift @nt; |
| 374 |
} elsif ($s =~ s/^;;[\x09\x20]*//) { |
| 375 |
push @nt, {type => COMMENT_PARAGRAPH_START_TOKEN, |
| 376 |
line => $line, column => $column}; |
| 377 |
$column += $+[0] - $-[0]; |
| 378 |
$tokenize_text->(\$s); |
| 379 |
$continuous_line = 1; |
| 380 |
return shift @nt; |
| 381 |
} elsif ($s =~ /\A\](INS|DEL)\][\x09\x20]*\z/) { |
| 382 |
$continuous_line = 1; |
| 383 |
return {type => BLOCK_END_TAG_TOKEN, tag_name => $1, |
| 384 |
line => $line, column => $column}; |
| 385 |
} elsif ($s =~ /^,/) { |
| 386 |
push @nt, {type => TABLE_ROW_START_TOKEN, |
| 387 |
line => $line, column => $column}; |
| 388 |
while ($s =~ s/^,[\x09\x20]*//) { |
| 389 |
$column += $+[0] - $-[0]; |
| 390 |
my $cell; |
| 391 |
my $cell_quoted; |
| 392 |
my $column_quoted = $column; |
| 393 |
my $column_cell = $column; |
| 394 |
if ($s =~ s/^"//) { |
| 395 |
$s =~ s/^((?>[^"\\]|\\.)*)//; |
| 396 |
$cell_quoted = $1; |
| 397 |
$column += 1 + length $cell_quoted; |
| 398 |
$cell_quoted =~ s/\\(.)/$1/sg; |
| 399 |
$column++ if $s =~ s/\A\\\z//; |
| 400 |
$column++ if $s =~ s/^"//; |
| 401 |
} |
| 402 |
if ($s =~ s/^([^,]+)//) { |
| 403 |
$cell = $1; |
| 404 |
$column += length $cell; |
| 405 |
$cell =~ s/[\x09\x20]+\z//; |
| 406 |
} |
| 407 |
if (not defined $cell_quoted and defined $cell and |
| 408 |
$cell eq '==') { |
| 409 |
push @nt, {type => TABLE_COLSPAN_CELL_TOKEN, |
| 410 |
line => $line, column => $column_cell}; |
| 411 |
} else { |
| 412 |
push @nt, {type => TABLE_CELL_START_TOKEN, |
| 413 |
line => $line, |
| 414 |
column => defined $column_quoted ? $column_quoted: $column_cell}; |
| 415 |
my $real_column = $column; |
| 416 |
$column = $column_quoted + 1; |
| 417 |
$tokenize_text->(\$cell_quoted) if defined $cell_quoted; |
| 418 |
## NOTE: When a quoted-pair is used, column numbers |
| 419 |
## reported in this $tokenize_text call might be wrong. |
| 420 |
$column = $column_cell; |
| 421 |
$tokenize_text->(\$cell) if defined $cell; |
| 422 |
$column = $column_quoted; |
| 423 |
push @nt, {type => TABLE_CELL_END_TOKEN, |
| 424 |
line => $line, |
| 425 |
column => $column}; |
| 426 |
} |
| 427 |
} |
| 428 |
push @nt, {type => TABLE_ROW_END_TOKEN, |
| 429 |
line => $line, column => $column}; |
| 430 |
undef $continuous_line; |
| 431 |
return shift @nt; |
| 432 |
} elsif ($s eq '__IMAGE__') { |
| 433 |
my $image = $doc->create_element_ns (SW09_NS, [undef, 'image']); |
| 434 |
$image->set_user_data (manakai_source_line => $line); |
| 435 |
$image->set_user_data (manakai_source_column => 1); |
| 436 |
$image->text_content (join "\x0A", @s); |
| 437 |
($line, $column) = ($line + @s, 1); |
| 438 |
@s = (); |
| 439 |
$doc->document_element->append_child ($image); |
| 440 |
return {type => END_OF_FILE_TOKEN, |
| 441 |
line => $line, column => $column}; |
| 442 |
} else { |
| 443 |
push @nt, {type => CHARACTER_TOKEN, data => "\x0A", |
| 444 |
line => $line, column => $column} if $continuous_line; |
| 445 |
$tokenize_text->(\$s); |
| 446 |
$continuous_line = 1; |
| 447 |
return shift @nt; |
| 448 |
} |
| 449 |
}; # $get_next_token |
| 450 |
|
| 451 |
## NOTE: The "initial" mode. |
| 452 |
if (@s and $s[0] =~ s/^#\?//) { |
| 453 |
## NOTE: "Parse a magic line". |
| 454 |
|
| 455 |
my $s = shift @s; |
| 456 |
if ($s =~ s/^([^\x09\x20]+)//) { |
| 457 |
$column += $+[0] - $-[0]; |
| 458 |
my ($name, $version) = split m#/#, $1, 2; |
| 459 |
my $el = $doc->document_element; |
| 460 |
$el->set_attribute_ns (SW09_NS, ['sw', 'Name'] => $name); |
| 461 |
$el->set_attribute_ns (SW09_NS, ['sw', 'Version'] => $version) |
| 462 |
if defined $version; |
| 463 |
} |
| 464 |
|
| 465 |
while (length $s) { |
| 466 |
$column += $+[0] - $-[0] and next if $s =~ s/^[\x09\x20]+//; |
| 467 |
my $name = ''; |
| 468 |
if ($s =~ s/^([^=]*)=//) { |
| 469 |
$name = $1; |
| 470 |
$column += length $name + 1; |
| 471 |
} |
| 472 |
my $param = $doc->create_element_ns (SW09_NS, [undef, 'parameter']); |
| 473 |
$param->set_attribute_ns (undef, [undef, 'name'] => $name); |
| 474 |
$param->set_user_data (manakai_source_line => $line); |
| 475 |
$param->set_user_data (manakai_source_column => $column); |
| 476 |
$head_el->append_child ($param); |
| 477 |
|
| 478 |
$column++ if $s =~ s/^"//; |
| 479 |
if ($s =~ s/^([^"]*)//) { |
| 480 |
my $values = $1; |
| 481 |
$column += length $values; |
| 482 |
$values =~ tr/\\//d; |
| 483 |
my @values = split /,/, $values, -1; |
| 484 |
push @values, '' unless length $values; |
| 485 |
for (@values) { |
| 486 |
my $value = $doc->create_element_ns (SW09_NS, [undef, 'value']); |
| 487 |
$value->text_content ($_); |
| 488 |
$value->set_user_data (manakai_source_line => $line); |
| 489 |
$value->set_user_data (manakai_source_column => $column); |
| 490 |
$param->append_child ($value); |
| 491 |
} |
| 492 |
} |
| 493 |
$column++ if $s =~ s/^"//; |
| 494 |
} |
| 495 |
|
| 496 |
$line = 2; |
| 497 |
$column = 1; |
| 498 |
} |
| 499 |
|
| 500 |
## NOTE: Switched to the "body" mode. |
| 501 |
|
| 502 |
my $oe = [{node => $body_el, |
| 503 |
section_depth => 0, |
| 504 |
quotation_depth => 0, |
| 505 |
list_depth => 0}]; |
| 506 |
my $structural_elements = { |
| 507 |
body => 1, section => 1, insert => 1, delete => 1, blockquote => 1, |
| 508 |
h1 => 1, ul => 1, ol => 1, dl => 1, li => 1, dt => 1, dd => 1, |
| 509 |
table => 1, tbody => 1, tr => 1, td => 1, p => 1, 'comment-p' => 1, |
| 510 |
ed => 1, pre => 1, |
| 511 |
}; |
| 512 |
|
| 513 |
my $im = IN_SECTION_IM; |
| 514 |
$token = $get_next_token->(); |
| 515 |
|
| 516 |
A: { |
| 517 |
if ($im == IN_PARAGRAPH_IM) { |
| 518 |
if ($token->{type} == CHARACTER_TOKEN) { |
| 519 |
$oe->[-1]->{node}->manakai_append_text ($token->{data}); |
| 520 |
$token = $get_next_token->(); |
| 521 |
redo A; |
| 522 |
} elsif ($token->{type} == INLINE_START_TAG_TOKEN) { |
| 523 |
if (not defined $token->{tag_name}) { |
| 524 |
my $el = $doc->create_element_ns (SW09_NS, [undef, 'anchor']); |
| 525 |
$oe->[-1]->{node}->append_child ($el); |
| 526 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 527 |
$el->set_user_data (manakai_source_line => $token->{line}); |
| 528 |
$el->set_user_data (manakai_source_column => $token->{column}); |
| 529 |
|
| 530 |
$token = $get_next_token->(); |
| 531 |
redo A; |
| 532 |
} else { |
| 533 |
my $type = { |
| 534 |
AA => [AA_NS, 'aa'], |
| 535 |
ABBR => [HTML_NS, 'abbr'], |
| 536 |
CITE => [HTML_NS, 'cite'], |
| 537 |
CODE => [HTML_NS, 'code'], |
| 538 |
CSECTION => [SW10_NS, 'csection'], |
| 539 |
DEL => [HTML_NS, 'del'], |
| 540 |
DFN => [HTML_NS, 'dfn'], |
| 541 |
INS => [HTML_NS, 'ins'], |
| 542 |
KBD => [HTML_NS, 'kbd'], |
| 543 |
KEY => [SW10_NS, 'key'], |
| 544 |
Q => [HTML_NS, 'q'], |
| 545 |
QN => [SW10_NS, 'qn'], |
| 546 |
RUBY => [HTML_NS, 'ruby'], |
| 547 |
RUBYB => [SW09_NS, 'rubyb'], |
| 548 |
SAMP => [HTML_NS, 'samp'], |
| 549 |
SPAN => [HTML_NS, 'span'], |
| 550 |
SRC => [SW10_NS, 'src'], |
| 551 |
SUB => [HTML_NS, 'sub'], |
| 552 |
SUP => [HTML_NS, 'sup'], |
| 553 |
TIME => [HTML_NS, 'time'], |
| 554 |
VAR => [HTML_NS, 'var'], |
| 555 |
WEAK => [SW09_NS, 'weak'], |
| 556 |
}->{$token->{tag_name}} || [SW10_NS, $token->{tag_name}]; |
| 557 |
my $el = $doc->create_element_ns ($type->[0], [undef, $type->[1]]); |
| 558 |
$oe->[-1]->{node}->append_child ($el); |
| 559 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 560 |
$el->set_user_data (manakai_source_line => $token->{line}); |
| 561 |
$el->set_user_data (manakai_source_column => $token->{column}); |
| 562 |
|
| 563 |
$el->set_attribute_ns (undef, [undef, 'class'] => $token->{classes}) |
| 564 |
if defined $token->{classes}; |
| 565 |
$el->set_attribute_ns (XML_NS, ['xml', 'lang'] => $token->{language}) |
| 566 |
if defined $token->{language}; |
| 567 |
|
| 568 |
$token = $get_next_token->(); |
| 569 |
redo A; |
| 570 |
} |
| 571 |
} elsif ($token->{type} == INLINE_MIDDLE_TAG_TOKEN) { |
| 572 |
my ($ns, $ln, $pop) = @{{ |
| 573 |
rt => [HTML_NS, 'rt', 1], |
| 574 |
title => [SW10_NS, 'attrvalue', 1], |
| 575 |
nsuri => [SW10_NS, 'attrvalue', 1], |
| 576 |
qn => [SW10_NS, 'nsuri'], |
| 577 |
ruby => [HTML_NS, 'rt'], |
| 578 |
rubyb => [HTML_NS, 'rt'], |
| 579 |
}->{$oe->[-1]->{node}->manakai_local_name} || [SW10_NS, 'title']}; |
| 580 |
pop @$oe if $pop; |
| 581 |
|
| 582 |
my $el = $doc->create_element_ns ($ns, [undef, $ln]); |
| 583 |
$oe->[-1]->{node}->append_child ($el); |
| 584 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 585 |
$el->set_user_data (manakai_source_line => $token->{line}); |
| 586 |
$el->set_user_data (manakai_source_column => $token->{column}); |
| 587 |
|
| 588 |
$el->set_attribute_ns (XML_NS, ['xml', 'lang'] => $token->{language}) |
| 589 |
if defined $token->{language}; |
| 590 |
|
| 591 |
$token = $get_next_token->(); |
| 592 |
redo A; |
| 593 |
} elsif ($token->{type} == INLINE_END_TAG_TOKEN) { |
| 594 |
pop @$oe if { |
| 595 |
rt => 1, title => 1, nsuri => 1, attrvalue => 1, |
| 596 |
}->{$oe->[-1]->{node}->manakai_local_name}; |
| 597 |
|
| 598 |
if ({%$structural_elements, |
| 599 |
strong => 1, em => 1}->{$oe->[-1]->{node}->manakai_local_name}) { |
| 600 |
unless (defined $token->{res_scheme} or defined $token->{anchor}) { |
| 601 |
$oe->[-1]->{node}->manakai_append_text (']]'); |
| 602 |
push @$oe, $oe->[-1]; |
| 603 |
} else { |
| 604 |
my $el = $doc->create_element_ns |
| 605 |
(SW09_NS, |
| 606 |
[undef, defined $token->{res_scheme} |
| 607 |
? 'anchor-external' : 'anchor-internal']); |
| 608 |
$oe->[-1]->{node}->append_child ($el); |
| 609 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 610 |
$el->set_user_data (manakai_source_line => $token->{line}); |
| 611 |
$el->set_user_data (manakai_source_column => $token->{column}); |
| 612 |
$el->text_content (']]'); |
| 613 |
} |
| 614 |
} |
| 615 |
|
| 616 |
$oe->[-1]->{node}->set_attribute_ns (SW09_NS, ['sw', 'anchor'], |
| 617 |
$token->{anchor}) |
| 618 |
if defined $token->{anchor}; |
| 619 |
$oe->[-1]->{node}->set_attribute_ns (SW09_NS, ['sw', 'resScheme'], |
| 620 |
$token->{res_scheme}) |
| 621 |
if defined $token->{res_scheme}; |
| 622 |
$oe->[-1]->{node}->set_attribute_ns (SW09_NS, ['sw', 'resParameter'], |
| 623 |
$token->{res_parameter}) |
| 624 |
if defined $token->{res_parameter}; |
| 625 |
|
| 626 |
pop @$oe; |
| 627 |
|
| 628 |
$token = $get_next_token->(); |
| 629 |
redo A; |
| 630 |
} elsif ($token->{type} == STRONG_TOKEN) { |
| 631 |
if ($oe->[-1]->{node}->manakai_local_name eq 'strong') { |
| 632 |
pop @$oe; |
| 633 |
} else { |
| 634 |
my $el = $doc->create_element_ns (HTML_NS, [undef, 'strong']); |
| 635 |
$oe->[-1]->{node}->append_child ($el); |
| 636 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 637 |
$el->set_user_data (manakai_source_line => $token->{line}); |
| 638 |
$el->set_user_data (manakai_source_column => $token->{column}); |
| 639 |
} |
| 640 |
$token = $get_next_token->(); |
| 641 |
redo A; |
| 642 |
} elsif ($token->{type} == EMPHASIS_TOKEN) { |
| 643 |
if ($oe->[-1]->{node}->manakai_local_name eq 'em') { |
| 644 |
pop @$oe; |
| 645 |
} else { |
| 646 |
my $el = $doc->create_element_ns (HTML_NS, [undef, 'em']); |
| 647 |
$oe->[-1]->{node}->append_child ($el); |
| 648 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 649 |
$el->set_user_data (manakai_source_line => $token->{line}); |
| 650 |
$el->set_user_data (manakai_source_column => $token->{column}); |
| 651 |
} |
| 652 |
$token = $get_next_token->(); |
| 653 |
redo A; |
| 654 |
} elsif ($token->{type} == FORM_TOKEN) { |
| 655 |
## There is an exact code clone. |
| 656 |
if ($token->{name} eq 'form') { |
| 657 |
my $el = $doc->create_element_ns (SW09_NS, [undef, 'form']); |
| 658 |
$oe->[-1]->{node}->append_child ($el); |
| 659 |
$el->set_user_data (manakai_source_line => $token->{line}); |
| 660 |
$el->set_user_data (manakai_source_column => $token->{column}); |
| 661 |
|
| 662 |
$el->set_attribute_ns (undef, [undef, 'id'] |
| 663 |
=> $token->{id}) if defined $token->{id}; |
| 664 |
$el->set_attribute_ns (undef, [undef, 'input'] |
| 665 |
=> shift @{$token->{parameters}}) |
| 666 |
if @{$token->{parameters}}; |
| 667 |
$el->set_attribute_ns (undef, [undef, 'template'] |
| 668 |
=> shift @{$token->{parameters}}) |
| 669 |
if @{$token->{parameters}}; |
| 670 |
$el->set_attribute_ns (undef, [undef, 'option'] |
| 671 |
=> shift @{$token->{parameters}}) |
| 672 |
if @{$token->{parameters}}; |
| 673 |
$el->set_attribute_ns (undef, [undef, 'parameter'] |
| 674 |
=> join ':', @{$token->{parameters}}) |
| 675 |
if @{$token->{parameters}}; |
| 676 |
|
| 677 |
$token = $get_next_token->(); |
| 678 |
redo A; |
| 679 |
} else { |
| 680 |
my $el = $doc->create_element_ns (SW09_NS, [undef, 'form']); |
| 681 |
$oe->[-1]->{node}->append_child ($el); |
| 682 |
$el->set_user_data (manakai_source_line => $token->{line}); |
| 683 |
$el->set_user_data (manakai_source_column => $token->{column}); |
| 684 |
|
| 685 |
$el->set_attribute_ns (undef, [undef, 'ref'] |
| 686 |
=> $token->{name}); |
| 687 |
$el->set_attribute_ns (undef, [undef, 'id'] |
| 688 |
=> $token->{id}) if defined $token->{id}; |
| 689 |
$el->set_attribute_ns (undef, [undef, 'parameter'] |
| 690 |
=> join ':', @{$token->{parameters}}) |
| 691 |
if @{$token->{parameters}}; |
| 692 |
|
| 693 |
$token = $get_next_token->(); |
| 694 |
redo A; |
| 695 |
} |
| 696 |
} elsif ($token->{type} == ELEMENT_TOKEN) { |
| 697 |
## NOTE: There is an exact code clone. |
| 698 |
my $el = $doc->create_element_ns |
| 699 |
($token->{namespace}, [undef, $token->{local_name}]); |
| 700 |
$oe->[-1]->{node}->append_child ($el); |
| 701 |
$el->set_user_data (manakai_source_line => $token->{line}); |
| 702 |
$el->set_user_data (manakai_source_column => $token->{column}); |
| 703 |
|
| 704 |
$el->set_attribute_ns (SW09_NS, ['sw', 'anchor'], $token->{anchor}) |
| 705 |
if defined $token->{anchor}; |
| 706 |
$el->set_attribute_ns (undef, [undef, 'by'] |
| 707 |
=> $token->{by}) if defined $token->{by}; |
| 708 |
$el->set_attribute_ns (SW09_NS, ['sw', 'resScheme'], |
| 709 |
$token->{res_scheme}) |
| 710 |
if defined $token->{res_scheme}; |
| 711 |
$el->set_attribute_ns (SW09_NS, ['sw', 'resParameter'], |
| 712 |
$token->{res_parameter}) |
| 713 |
if defined $token->{res_parameter}; |
| 714 |
$el->text_content ($token->{content}) if defined $token->{content}; |
| 715 |
|
| 716 |
$token = $get_next_token->(); |
| 717 |
redo A; |
| 718 |
} elsif ($token->{type} == LABELED_LIST_MIDDLE_TOKEN) { |
| 719 |
pop @$oe while not $structural_elements |
| 720 |
->{$oe->[-1]->{node}->manakai_local_name}; |
| 721 |
pop @$oe if $oe->[-1]->{node}->manakai_local_name eq 'dt'; |
| 722 |
|
| 723 |
my $el = $doc->create_element_ns (HTML_NS, [undef, 'dd']); |
| 724 |
$oe->[-1]->{node}->append_child ($el); |
| 725 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 726 |
$el->set_user_data (manakai_source_line => $token->{line}); |
| 727 |
$el->set_user_data (manakai_source_column => $token->{column}); |
| 728 |
|
| 729 |
$token = $get_next_token->(); |
| 730 |
redo A; |
| 731 |
} elsif ($token->{type} == HEADING_END_TOKEN) { |
| 732 |
pop @$oe while not $structural_elements |
| 733 |
->{$oe->[-1]->{node}->manakai_local_name}; |
| 734 |
pop @$oe if $oe->[-1]->{node}->manakai_local_name eq 'h1'; |
| 735 |
|
| 736 |
$im = IN_SECTION_IM; |
| 737 |
$token = $get_next_token->(); |
| 738 |
redo A; |
| 739 |
} elsif ($token->{type} == TABLE_CELL_END_TOKEN) { |
| 740 |
pop @$oe while not $structural_elements |
| 741 |
->{$oe->[-1]->{node}->manakai_local_name}; |
| 742 |
pop @$oe if $oe->[-1]->{node}->manakai_local_name eq 'td'; |
| 743 |
|
| 744 |
$im = IN_TABLE_ROW_IM; |
| 745 |
$token = $get_next_token->(); |
| 746 |
redo A; |
| 747 |
} elsif (($token->{type} == BLOCK_END_TAG_TOKEN and |
| 748 |
$token->{tag_name} eq 'PRE') or |
| 749 |
$token->{type} == PREFORMATTED_END_TOKEN) { |
| 750 |
pop @$oe while not $structural_elements |
| 751 |
->{$oe->[-1]->{node}->manakai_local_name}; |
| 752 |
pop @$oe if $oe->[-1]->{node}->manakai_local_name eq 'pre'; |
| 753 |
|
| 754 |
$im = IN_SECTION_IM; |
| 755 |
$token = $get_next_token->(); |
| 756 |
redo A; |
| 757 |
} else { |
| 758 |
pop @$oe while not $structural_elements |
| 759 |
->{$oe->[-1]->{node}->manakai_local_name}; |
| 760 |
|
| 761 |
$im = IN_SECTION_IM; |
| 762 |
## Reconsume. |
| 763 |
redo A; |
| 764 |
} |
| 765 |
} elsif ($im == IN_SECTION_IM) { |
| 766 |
if ($token->{type} == HEADING_START_TOKEN) { |
| 767 |
B: { |
| 768 |
pop @$oe and redo B |
| 769 |
if not {body => 1, section => 1, insert => 1, delete => 1} |
| 770 |
->{$oe->[-1]->{node}->manakai_local_name} or |
| 771 |
$token->{depth} <= $oe->[-1]->{section_depth}; |
| 772 |
if ($token->{depth} > $oe->[-1]->{section_depth} + 1) { |
| 773 |
my $el = $doc->create_element_ns (HTML_NS, [undef, 'section']); |
| 774 |
$oe->[-1]->{node}->append_child ($el); |
| 775 |
push @$oe, {node => $el, |
| 776 |
section_depth => $oe->[-1]->{section_depth} + 1, |
| 777 |
quotation_depth => 0, list_depth => 0}; |
| 778 |
redo B; |
| 779 |
} |
| 780 |
} # B |
| 781 |
|
| 782 |
my $el = $doc->create_element_ns (HTML_NS, [undef, 'section']); |
| 783 |
$oe->[-1]->{node}->append_child ($el); |
| 784 |
push @$oe, {node => $el, |
| 785 |
section_depth => $oe->[-1]->{section_depth} + 1, |
| 786 |
quotation_depth => 0, list_depth => 0}; |
| 787 |
|
| 788 |
my $el2 = $doc->create_element_ns (HTML_NS, [undef, 'h1']); |
| 789 |
$oe->[-1]->{node}->append_child ($el2); |
| 790 |
push @$oe, {%{$oe->[-1]}, node => $el2}; |
| 791 |
|
| 792 |
$im = IN_PARAGRAPH_IM; |
| 793 |
$token = $get_next_token->(); |
| 794 |
redo A; |
| 795 |
} elsif ($token->{type} == BLOCK_START_TAG_TOKEN and |
| 796 |
($token->{tag_name} eq 'INS' or |
| 797 |
$token->{tag_name} eq 'DEL')) { |
| 798 |
my $el = $doc->create_element_ns |
| 799 |
(SW09_NS, |
| 800 |
[undef, $token->{tag_name} eq 'INS' ? 'insert' : 'delete']); |
| 801 |
$oe->[-1]->{node}->append_child ($el); |
| 802 |
push @$oe, {node => $el, section_depth => 0, |
| 803 |
quotation_depth => 0, list_depth => 0}; |
| 804 |
$el->set_attribute_ns (undef, [undef, 'class'] => $token->{classes}) |
| 805 |
if defined $token->{classes}; |
| 806 |
$token = $get_next_token->(); |
| 807 |
redo A; |
| 808 |
} elsif ($token->{type} == QUOTATION_START_TOKEN) { |
| 809 |
B: { |
| 810 |
pop @$oe and redo B |
| 811 |
if not {body => 1, section => 1, insert => 1, delete => 1, |
| 812 |
blockquote => 1} |
| 813 |
->{$oe->[-1]->{node}->manakai_local_name} or |
| 814 |
$token->{depth} < $oe->[-1]->{quotation_depth}; |
| 815 |
if ($token->{depth} > $oe->[-1]->{quotation_depth}) { |
| 816 |
my $el = $doc->create_element_ns (HTML_NS, [undef, 'blockquote']); |
| 817 |
$oe->[-1]->{node}->append_child ($el); |
| 818 |
push @$oe, {node => $el, section_depth => 0, |
| 819 |
quotation_depth => $oe->[-1]->{quotation_depth} + 1, |
| 820 |
list_depth => 0}; |
| 821 |
redo B; |
| 822 |
} |
| 823 |
} # B |
| 824 |
|
| 825 |
$token = $get_next_token->(); |
| 826 |
redo A; |
| 827 |
} elsif ($token->{type} == LIST_START_TOKEN) { |
| 828 |
my $depth = length $token->{depth}; |
| 829 |
my $list_type = substr ($token->{depth}, -1, 1) eq '-' ? 'ul' : 'ol'; |
| 830 |
B: { |
| 831 |
pop @$oe and redo B if $oe->[-1]->{list_depth} > $depth; |
| 832 |
pop @$oe and redo B if $oe->[-1]->{list_depth} == $depth and |
| 833 |
$list_type ne $oe->[-1]->{node}->manakai_local_name; |
| 834 |
if ($oe->[-1]->{list_depth} < $depth) { |
| 835 |
my $type = substr $token->{depth}, $oe->[-1]->{list_depth}, 1; |
| 836 |
my $el = $doc->create_element_ns |
| 837 |
(HTML_NS, [undef, $type eq '-' ? 'ul' : 'ol']); |
| 838 |
$oe->[-1]->{node}->append_child ($el); |
| 839 |
push @$oe, {%{$oe->[-1]}, node => $el, |
| 840 |
list_depth => $oe->[-1]->{list_depth} + 1}; |
| 841 |
if ($oe->[-1]->{list_depth} < $depth) { |
| 842 |
my $el = $doc->create_element_ns (HTML_NS, [undef, 'li']); |
| 843 |
$oe->[-1]->{node}->append_child ($el); |
| 844 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 845 |
} |
| 846 |
redo B; |
| 847 |
} |
| 848 |
} # B |
| 849 |
|
| 850 |
my $el = $doc->create_element_ns (HTML_NS, [undef, 'li']); |
| 851 |
$oe->[-1]->{node}->append_child ($el); |
| 852 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 853 |
|
| 854 |
$im = IN_PARAGRAPH_IM; |
| 855 |
$token = $get_next_token->(); |
| 856 |
redo A; |
| 857 |
} elsif ($token->{type} == LABELED_LIST_START_TOKEN) { |
| 858 |
pop @$oe if $oe->[-1]->{node}->manakai_local_name eq 'dd'; |
| 859 |
if ($oe->[-1]->{node}->manakai_local_name ne 'dl') { |
| 860 |
my $el = $doc->create_element_ns (HTML_NS, [undef, 'dl']); |
| 861 |
$oe->[-1]->{node}->append_child ($el); |
| 862 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 863 |
} |
| 864 |
|
| 865 |
my $el = $doc->create_element_ns (HTML_NS, [undef, 'dt']); |
| 866 |
$oe->[-1]->{node}->append_child ($el); |
| 867 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 868 |
|
| 869 |
$im = IN_PARAGRAPH_IM; |
| 870 |
$token = $get_next_token->(); |
| 871 |
redo A; |
| 872 |
} elsif ($token->{type} == TABLE_ROW_START_TOKEN) { |
| 873 |
my $el = $doc->create_element_ns (HTML_NS, [undef, 'table']); |
| 874 |
$oe->[-1]->{node}->append_child ($el); |
| 875 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 876 |
|
| 877 |
$el = $doc->create_element_ns (HTML_NS, [undef, 'tbody']); |
| 878 |
$oe->[-1]->{node}->append_child ($el); |
| 879 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 880 |
|
| 881 |
$el = $doc->create_element_ns (HTML_NS, [undef, 'tr']); |
| 882 |
$oe->[-1]->{node}->append_child ($el); |
| 883 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 884 |
|
| 885 |
$im = IN_TABLE_ROW_IM; |
| 886 |
$token = $get_next_token->(); |
| 887 |
redo A; |
| 888 |
} elsif (($token->{type} == BLOCK_START_TAG_TOKEN and |
| 889 |
$token->{tag_name} eq 'PRE') or |
| 890 |
$token->{type} == PREFORMATTED_START_TOKEN) { |
| 891 |
my $el = $doc->create_element_ns (HTML_NS, [undef, 'pre']); |
| 892 |
$oe->[-1]->{node}->append_child ($el); |
| 893 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 894 |
|
| 895 |
$el->set_attribute_ns (undef, [undef, 'class'] => $token->{classes}) |
| 896 |
if defined $token->{classes}; |
| 897 |
|
| 898 |
$im = IN_PARAGRAPH_IM; |
| 899 |
$token = $get_next_token->(); |
| 900 |
redo A; |
| 901 |
} elsif ($token->{type} == COMMENT_PARAGRAPH_START_TOKEN) { |
| 902 |
my $el = $doc->create_element_ns (SW10_NS, [undef, 'comment-p']); |
| 903 |
$oe->[-1]->{node}->append_child ($el); |
| 904 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 905 |
|
| 906 |
$im = IN_PARAGRAPH_IM; |
| 907 |
$token = $get_next_token->(); |
| 908 |
redo A; |
| 909 |
} elsif ($token->{type} == EDITORIAL_NOTE_START_TOKEN) { |
| 910 |
my $el = $doc->create_element_ns (SW10_NS, [undef, 'ed']); |
| 911 |
$oe->[-1]->{node}->append_child ($el); |
| 912 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 913 |
|
| 914 |
$im = IN_PARAGRAPH_IM; |
| 915 |
$token = $get_next_token->(); |
| 916 |
redo A; |
| 917 |
} elsif ($token->{type} == EMPTY_LINE_TOKEN) { |
| 918 |
pop @$oe while not {body => 1, section => 1, insert => 1, delete => 1} |
| 919 |
->{$oe->[-1]->{node}->manakai_local_name}; |
| 920 |
$token = $get_next_token->(); |
| 921 |
redo A; |
| 922 |
} elsif ($token->{type} == BLOCK_END_TAG_TOKEN) { |
| 923 |
if ($token->{tag_name} eq 'INS') { |
| 924 |
for (reverse 1..$#$oe) { |
| 925 |
if ($oe->[$_]->{node}->manakai_local_name eq 'insert') { |
| 926 |
splice @$oe, $_; |
| 927 |
last; |
| 928 |
} |
| 929 |
} |
| 930 |
} elsif ($token->{tag_name} eq 'DEL') { |
| 931 |
for (reverse 1..$#$oe) { |
| 932 |
if ($oe->[$_]->{node}->manakai_local_name eq 'delete') { |
| 933 |
splice @$oe, $_; |
| 934 |
last; |
| 935 |
} |
| 936 |
} |
| 937 |
} else { |
| 938 |
## NOTE: Ignore the token. |
| 939 |
} |
| 940 |
undef $continuous_line; |
| 941 |
$token = $get_next_token->(); |
| 942 |
redo A; |
| 943 |
} elsif ($token->{type} == FORM_TOKEN) { |
| 944 |
## There is an exact code clone. |
| 945 |
if ($token->{name} eq 'form') { |
| 946 |
my $el = $doc->create_element_ns (SW09_NS, [undef, 'form']); |
| 947 |
$oe->[-1]->{node}->append_child ($el); |
| 948 |
$el->set_user_data (manakai_source_line => $token->{line}); |
| 949 |
$el->set_user_data (manakai_source_column => $token->{column}); |
| 950 |
|
| 951 |
$el->set_attribute_ns (undef, [undef, 'id'] |
| 952 |
=> $token->{id}) if defined $token->{id}; |
| 953 |
$el->set_attribute_ns (undef, [undef, 'input'] |
| 954 |
=> shift @{$token->{parameters}}) |
| 955 |
if @{$token->{parameters}}; |
| 956 |
$el->set_attribute_ns (undef, [undef, 'template'] |
| 957 |
=> shift @{$token->{parameters}}) |
| 958 |
if @{$token->{parameters}}; |
| 959 |
$el->set_attribute_ns (undef, [undef, 'option'] |
| 960 |
=> shift @{$token->{parameters}}) |
| 961 |
if @{$token->{parameters}}; |
| 962 |
$el->set_attribute_ns (undef, [undef, 'parameter'] |
| 963 |
=> join ':', @{$token->{parameters}}) |
| 964 |
if @{$token->{parameters}}; |
| 965 |
|
| 966 |
$token = $get_next_token->(); |
| 967 |
redo A; |
| 968 |
} else { |
| 969 |
my $el = $doc->create_element_ns (SW09_NS, [undef, 'form']); |
| 970 |
$oe->[-1]->{node}->append_child ($el); |
| 971 |
$el->set_user_data (manakai_source_line => $token->{line}); |
| 972 |
$el->set_user_data (manakai_source_column => $token->{column}); |
| 973 |
|
| 974 |
$el->set_attribute_ns (undef, [undef, 'ref'] => $token->{name}); |
| 975 |
$el->set_attribute_ns (undef, [undef, 'id'] |
| 976 |
=> $token->{id}) if defined $token->{id}; |
| 977 |
$el->set_attribute_ns (undef, [undef, 'parameter'] |
| 978 |
=> join ':', @{$token->{parameters}}) |
| 979 |
if @{$token->{parameters}}; |
| 980 |
|
| 981 |
$token = $get_next_token->(); |
| 982 |
redo A; |
| 983 |
} |
| 984 |
} elsif ($token->{type} == ELEMENT_TOKEN and |
| 985 |
$token->{local_name} eq 'replace') { |
| 986 |
## NOTE: There is an exact code clone. |
| 987 |
my $el = $doc->create_element_ns |
| 988 |
($token->{namespace}, [undef, $token->{local_name}]); |
| 989 |
$oe->[-1]->{node}->append_child ($el); |
| 990 |
$el->set_user_data (manakai_source_line => $token->{line}); |
| 991 |
$el->set_user_data (manakai_source_column => $token->{column}); |
| 992 |
|
| 993 |
$el->set_attribute_ns (SW09_NS, ['sw', 'anchor'], $token->{anchor}) |
| 994 |
if defined $token->{anchor}; |
| 995 |
$el->set_attribute_ns (undef, [undef, 'by'] |
| 996 |
=> $token->{by}) if defined $token->{by}; |
| 997 |
$el->set_attribute_ns (SW09_NS, ['sw', 'resScheme'], |
| 998 |
$token->{res_scheme}) |
| 999 |
if defined $token->{res_scheme}; |
| 1000 |
$el->set_attribute_ns (SW09_NS, ['sw', 'resParameter'], |
| 1001 |
$token->{res_parameter}) |
| 1002 |
if defined $token->{res_parameter}; |
| 1003 |
$el->text_content ($token->{content}) if defined $token->{content}; |
| 1004 |
|
| 1005 |
$token = $get_next_token->(); |
| 1006 |
redo A; |
| 1007 |
} elsif ($token->{type} == END_OF_FILE_TOKEN) { |
| 1008 |
return; |
| 1009 |
} elsif ({LABELED_LIST_MIDDLE_TOKEN => 1, |
| 1010 |
HEADING_END_TOKEN => 1, |
| 1011 |
PREFORMATTED_END_TOKEN => 1, |
| 1012 |
TABLE_ROW_END_TOKEN => 1, |
| 1013 |
TABLE_CELL_START_TOKEN => 1, |
| 1014 |
TABLE_CELL_END_TOKEN => 1, |
| 1015 |
TABLE_COLSPAN_CELL_TOKEN => 1}->{$token->{type}}) { |
| 1016 |
## NOTE: Ignore the token. |
| 1017 |
} else { |
| 1018 |
unless ({dd => 1, |
| 1019 |
li => 1, |
| 1020 |
'comment-p' => 1, |
| 1021 |
ed => 1}->{$oe->[-1]->{node}->manakai_local_name}) { |
| 1022 |
my $el = $doc->create_element_ns (HTML_NS, [undef, 'p']); |
| 1023 |
$oe->[-1]->{node}->append_child ($el); |
| 1024 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 1025 |
} |
| 1026 |
|
| 1027 |
$im = IN_PARAGRAPH_IM; |
| 1028 |
## Reprocess. |
| 1029 |
redo A; |
| 1030 |
} |
| 1031 |
} elsif ($im == IN_TABLE_ROW_IM) { |
| 1032 |
if ($token->{type} == TABLE_CELL_START_TOKEN) { |
| 1033 |
my $el = $doc->create_element_ns (HTML_NS, [undef, 'td']); |
| 1034 |
$oe->[-1]->{node}->append_child ($el); |
| 1035 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 1036 |
$el->set_user_data (manakai_source_line => $token->{line}); |
| 1037 |
$el->set_user_data (manakai_source_column => $token->{column}); |
| 1038 |
|
| 1039 |
$im = IN_PARAGRAPH_IM; |
| 1040 |
$token = $get_next_token->(); |
| 1041 |
redo A; |
| 1042 |
} elsif ($token->{type} == TABLE_COLSPAN_CELL_TOKEN) { |
| 1043 |
my $lc = $oe->[-1]->{node}->last_child; |
| 1044 |
if ($lc and $lc->manakai_local_name eq 'td') { |
| 1045 |
$lc->set_attribute_ns |
| 1046 |
(undef, [undef, 'colspan'], |
| 1047 |
($lc->get_attribute_ns (undef, 'colspan') || 1) + 1); |
| 1048 |
} else { |
| 1049 |
my $el = $doc->create_element_ns (HTML_NS, [undef, 'td']); |
| 1050 |
$oe->[-1]->{node}->append_child ($el); |
| 1051 |
$el->set_user_data (manakai_source_line => $token->{line}); |
| 1052 |
$el->set_user_data (manakai_source_column => $token->{column}); |
| 1053 |
} |
| 1054 |
|
| 1055 |
$token = $get_next_token->(); |
| 1056 |
redo A; |
| 1057 |
} elsif ($token->{type} == TABLE_ROW_END_TOKEN) { |
| 1058 |
pop @$oe if $oe->[-1]->{node}->manakai_local_name eq 'tr'; |
| 1059 |
$token = $get_next_token->(); |
| 1060 |
redo A; |
| 1061 |
} elsif ($token->{type} == TABLE_ROW_START_TOKEN) { |
| 1062 |
my $el = $doc->create_element_ns (HTML_NS, [undef, 'tr']); |
| 1063 |
$oe->[-1]->{node}->append_child ($el); |
| 1064 |
push @$oe, {%{$oe->[-1]}, node => $el}; |
| 1065 |
$el->set_user_data (manakai_source_line => $token->{line}); |
| 1066 |
$el->set_user_data (manakai_source_column => $token->{column}); |
| 1067 |
|
| 1068 |
$token = $get_next_token->(); |
| 1069 |
redo A; |
| 1070 |
} else { |
| 1071 |
$im = IN_SECTION_IM; |
| 1072 |
## Reprocess. |
| 1073 |
redo A; |
| 1074 |
} |
| 1075 |
} else { |
| 1076 |
die "$0: Unknown insertion mode: $im"; |
| 1077 |
} |
| 1078 |
} # A |
| 1079 |
} # parse_char_string |
| 1080 |
|
| 1081 |
1; |
Parent Directory
| 
Revision Log