| 1 | <?php
|
|---|
| 2 | # Copyright by Magnus Manske (2005)
|
|---|
| 3 | # Released under GPL
|
|---|
| 4 |
|
|---|
| 5 | $wiki2xml_authors = array () ;
|
|---|
| 6 |
|
|---|
| 7 | class wiki2xml
|
|---|
| 8 | {
|
|---|
| 9 | var $cnt = 0 ; # For debugging purposes
|
|---|
| 10 | var $protocols = array ( "http" , "https" , "news" , "ftp" , "irc" , "mailto" ) ;
|
|---|
| 11 | var $errormessage = "ERROR!" ;
|
|---|
| 12 | var $compensate_markup_errors = true;
|
|---|
| 13 | var $auto_fill_templates = 'all' ; # Will try and replace templates right inline, instead of using <template> tags; requires global $content_provider
|
|---|
| 14 | var $use_space_tag = true ; # Use <space/> instead of spaces before and after tags
|
|---|
| 15 | var $allowed = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ01234567890 +-#:;.,%="\'\\' ;
|
|---|
| 16 | var $directhtmltags = array (
|
|---|
| 17 | "b" => "xhtml:b",
|
|---|
| 18 | "i" => "xhtml:i",
|
|---|
| 19 | "u" => "xhtml:u",
|
|---|
| 20 | "s" => "xhtml:s",
|
|---|
| 21 | "p" => "xhtml:p",
|
|---|
| 22 | "br" => "xhtml:br",
|
|---|
| 23 | "em" => "xhtml:em",
|
|---|
| 24 | "div" => "xhtml:div",
|
|---|
| 25 | "span" => "xhtml:span",
|
|---|
| 26 | "big" => "xhtml:big",
|
|---|
| 27 | "small" => "xhtml:small",
|
|---|
| 28 | "sub" => "xhtml:sub",
|
|---|
| 29 | "sup" => "xhtml:sup",
|
|---|
| 30 | "font" => "xhtml:font",
|
|---|
| 31 | "center" => "xhtml:center",
|
|---|
| 32 | "table" => "xhtml:table",
|
|---|
| 33 | "tr" => "xhtml:tr",
|
|---|
| 34 | "th" => "xhtml:th",
|
|---|
| 35 | "td" => "xhtml:td",
|
|---|
| 36 | "pre" => "xhtml:pre",
|
|---|
| 37 | "code" => "xhtml:code",
|
|---|
| 38 | "caption" => "xhtml:caption",
|
|---|
| 39 | "cite" => "xhtml:cite",
|
|---|
| 40 | "ul" => "xhtml:ul",
|
|---|
| 41 | "ol" => "xhtml:ol",
|
|---|
| 42 | "li" => "xhtml:li",
|
|---|
| 43 | "tt" => "xhtml:tt",
|
|---|
| 44 | "h1" => "xhtml:h1",
|
|---|
| 45 | "h2" => "xhtml:h2",
|
|---|
| 46 | "h3" => "xhtml:h3",
|
|---|
| 47 | "h4" => "xhtml:h4",
|
|---|
| 48 | "h5" => "xhtml:h5",
|
|---|
| 49 | "h6" => "xhtml:h6",
|
|---|
| 50 | "h7" => "xhtml:h7",
|
|---|
| 51 | "h8" => "xhtml:h8",
|
|---|
| 52 | "h9" => "xhtml:h9",
|
|---|
| 53 | ) ;
|
|---|
| 54 |
|
|---|
| 55 | var $w ; # The wiki text
|
|---|
| 56 | var $wl ; # The wiki text length
|
|---|
| 57 | var $bold_italics ;
|
|---|
| 58 | var $tables = array () ; # List of open tables
|
|---|
| 59 | var $profile = array () ;
|
|---|
| 60 |
|
|---|
| 61 | # Some often used functions
|
|---|
| 62 |
|
|---|
| 63 | /**
|
|---|
| 64 | * Inherit settings from an existing parser
|
|---|
| 65 | */
|
|---|
| 66 | function inherit ( &$base )
|
|---|
| 67 | {
|
|---|
| 68 | $this->protocols = $base->protocols ;
|
|---|
| 69 | $this->errormessage = $base->errormessage ;
|
|---|
| 70 | $this->compensate_markup_errors = $base->compensate_markup_errors ;
|
|---|
| 71 | $this->auto_fill_templates = $base->auto_fill_templates ;
|
|---|
| 72 | $this->use_space_tag = $base->use_space_tag ;
|
|---|
| 73 | $this->compensate_markup_errors = $base->compensate_markup_errors ;
|
|---|
| 74 | $this->allowed = $base->allowed ;
|
|---|
| 75 | $this->directhtmltags = $base->directhtmltags ;
|
|---|
| 76 | }
|
|---|
| 77 |
|
|---|
| 78 | /**
|
|---|
| 79 | * Matches a function to the current text (default:once)
|
|---|
| 80 | */
|
|---|
| 81 | function once ( &$a , &$xml , $f , $atleastonce = true , $many = false )
|
|---|
| 82 | {
|
|---|
| 83 | $f = "p_{$f}" ;
|
|---|
| 84 | $cnt = 0 ;
|
|---|
| 85 | # print $f . " : " . htmlentities ( substr ( $this->w , $a , 20 ) ) . "<br/>" ; flush () ;
|
|---|
| 86 | # if ( !isset ( $this->profile[$f] ) ) $this->profile[$f] = 0 ; # PROFILING
|
|---|
| 87 | do {
|
|---|
| 88 | # $this->profile[$f]++ ; # PROFILING
|
|---|
| 89 | $matched = $this->$f ( $a , $xml ) ;
|
|---|
| 90 | if ( $matched && $many ) $again = true ;
|
|---|
| 91 | else $again = false ;
|
|---|
| 92 | if ( $matched ) $cnt++ ;
|
|---|
| 93 | } while ( $again ) ;
|
|---|
| 94 | if ( !$atleastonce ) return true ;
|
|---|
| 95 | if ( $cnt > 0 ) return true ;
|
|---|
| 96 | return false ;
|
|---|
| 97 | }
|
|---|
| 98 |
|
|---|
| 99 | function onceormore ( &$a , &$xml , $f )
|
|---|
| 100 | {
|
|---|
| 101 | return $this->once ( $a , $xml , $f , true , true ) ;
|
|---|
| 102 | }
|
|---|
| 103 |
|
|---|
| 104 | function nextis ( &$a , $t , $movecounter = true )
|
|---|
| 105 | {
|
|---|
| 106 | if ( substr ( $this->w , $a , strlen ( $t ) ) != $t ) return false ;
|
|---|
| 107 | if ( $movecounter ) $a += strlen ( $t ) ;
|
|---|
| 108 | return true ;
|
|---|
| 109 | }
|
|---|
| 110 |
|
|---|
| 111 | function nextchar ( &$a , &$x )
|
|---|
| 112 | {
|
|---|
| 113 | if ( $a >= $this->wl ) return false ;
|
|---|
| 114 | $x .= htmlspecialchars ( $this->w[$a] ) ;
|
|---|
| 115 | $a++ ;
|
|---|
| 116 | return true ;
|
|---|
| 117 | }
|
|---|
| 118 |
|
|---|
| 119 | function ischaracter ( $c )
|
|---|
| 120 | {
|
|---|
| 121 | if ( $c >= 'A' && $c <= 'Z' ) return true ;
|
|---|
| 122 | if ( $c >= 'a' && $c <= 'z' ) return true ;
|
|---|
| 123 | return false ;
|
|---|
| 124 | }
|
|---|
| 125 |
|
|---|
| 126 | function skipblanks ( &$a , $blank = " " )
|
|---|
| 127 | {
|
|---|
| 128 | while ( $a < $this->wl )
|
|---|
| 129 | {
|
|---|
| 130 | if ( $this->w[$a] != $blank ) return ;
|
|---|
| 131 | $a++ ;
|
|---|
| 132 | }
|
|---|
| 133 | }
|
|---|
| 134 |
|
|---|
| 135 | ##############
|
|---|
| 136 |
|
|---|
| 137 |
|
|---|
| 138 | function p_internal_link_target ( &$a , &$xml , $closeit = "]]" )
|
|---|
| 139 | {
|
|---|
| 140 | return $this->p_internal_link_text ( $a , $xml , true , $closeit ) ;
|
|---|
| 141 | }
|
|---|
| 142 |
|
|---|
| 143 | function p_internal_link_text2 ( &$a , &$xml , $closeit )
|
|---|
| 144 | {
|
|---|
| 145 | $bi = $this->bold_italics ;
|
|---|
| 146 | $ret = $this->p_internal_link_text ( $a , $xml , false , $closeit , false ) ;
|
|---|
| 147 | if ( $closeit == ']]' && '' != $this->bold_italics ) $ret = false ; # Dirty hack to ensure good XML; FIXME!!!
|
|---|
| 148 | return $ret ;
|
|---|
| 149 | }
|
|---|
| 150 |
|
|---|
| 151 | function p_internal_link_text ( &$a , &$xml , $istarget = false , $closeit = "]]" , $mark = true )
|
|---|
| 152 | {
|
|---|
| 153 | $b = $a ;
|
|---|
| 154 | $x = "" ;
|
|---|
| 155 | if ( $b >= $this->wl ) return false ;
|
|---|
| 156 | $bi = $this->bold_italics ;
|
|---|
| 157 | $this->bold_italics = '' ;
|
|---|
| 158 | $closeit1 = $closeit[0] ;
|
|---|
| 159 | while ( 1 )
|
|---|
| 160 | {
|
|---|
| 161 | if ( $b >= $this->wl ) {
|
|---|
| 162 | $this->bold_italics = $bi ;
|
|---|
| 163 | return false ;
|
|---|
| 164 | }
|
|---|
| 165 | $c = $this->w[$b] ;
|
|---|
| 166 | if ( $closeit != "}}" && $c == "\n" ) {
|
|---|
| 167 | $this->bold_italics = $bi ;
|
|---|
| 168 | return false ;
|
|---|
| 169 | }
|
|---|
| 170 | if ( $c == "|" ) break ;
|
|---|
| 171 | if ( $c == $closeit1 && $this->nextis ( $b , $closeit , false ) ) break ;
|
|---|
| 172 | if ( !$istarget ) {
|
|---|
| 173 | if ( $c == "[" && $this->once ( $b , $x , "internal_link" ) ) continue ;
|
|---|
| 174 | if ( $c == "[" && $this->once ( $b , $x , "external_link" ) ) continue ;
|
|---|
| 175 | if ( $c == "{" && $this->once ( $b , $x , "template_variable" ) ) continue ;
|
|---|
| 176 | if ( $c == "{" && $this->once ( $b , $x , "template" ) ) continue ;
|
|---|
| 177 | if ( $c == "<" && $this->once ( $b , $x , "html" ) ) continue ;
|
|---|
| 178 | if ( $c == "'" && $this->p_bold ( $b , $x , "internal_link_text2" , $closeit ) ) { break ; }
|
|---|
| 179 | if ( $c == "'" && $this->p_italics ( $b , $x , "internal_link_text2" , $closeit ) ) { break ; }
|
|---|
| 180 | if ( $b + 10 < $this->wl &&
|
|---|
| 181 | ( $this->w[$a+5] == '/' && $this->w[$a+7] == '/' ) &&
|
|---|
| 182 | $this->once ( $b , $x , "external_freelink" ) ) continue ;
|
|---|
| 183 | } else {
|
|---|
| 184 | if ( $c == "{" && $this->once ( $b , $x , "template" ) ) continue ;
|
|---|
| 185 | }
|
|---|
| 186 | $x .= htmlspecialchars ( $c ) ;
|
|---|
| 187 | $b++ ;
|
|---|
| 188 | /* if ( $b >= $this->wl ) {
|
|---|
| 189 | $this->bold_italics = $bi ;
|
|---|
| 190 | return false ;
|
|---|
| 191 | }*/
|
|---|
| 192 | }
|
|---|
| 193 |
|
|---|
| 194 | if ( $closeit == "}}" && !$istarget ) {
|
|---|
| 195 | $xml .= substr ( $this->w , $a , $b - $a ) ;
|
|---|
| 196 | $a = $b ;
|
|---|
| 197 | $this->bold_italics = $bi ;
|
|---|
| 198 | return true ;
|
|---|
| 199 | }
|
|---|
| 200 |
|
|---|
| 201 | $x = trim ( str_replace ( "\n" , "" , $x ) ) ;
|
|---|
| 202 | if ( $mark )
|
|---|
| 203 | {
|
|---|
| 204 | if ( $istarget ) $xml .= "<target>{$x}</target>" ;
|
|---|
| 205 | else $xml .= "<part>{$x}</part>" ;
|
|---|
| 206 |
|
|---|
| 207 | }
|
|---|
| 208 | else $xml .= $x ;
|
|---|
| 209 | $a = $b ;
|
|---|
| 210 | $this->bold_italics = $bi ;
|
|---|
| 211 | return true ;
|
|---|
| 212 | }
|
|---|
| 213 |
|
|---|
| 214 | function p_internal_link_trail ( &$a , &$xml )
|
|---|
| 215 | {
|
|---|
| 216 | $b = $a ;
|
|---|
| 217 | $x = "" ;
|
|---|
| 218 | while ( 1 )
|
|---|
| 219 | {
|
|---|
| 220 | $c = "" ;
|
|---|
| 221 |
|
|---|
| 222 | if ( !$this->nextchar ( $b , $c ) ) break ;
|
|---|
| 223 |
|
|---|
| 224 | if ( $this->ischaracter ( $c ) )
|
|---|
| 225 | {
|
|---|
| 226 | $x .= $c ;
|
|---|
| 227 | }
|
|---|
| 228 | else
|
|---|
| 229 | {
|
|---|
| 230 | $b-- ;
|
|---|
| 231 | break ;
|
|---|
| 232 | }
|
|---|
| 233 | }
|
|---|
| 234 | if ( $x == "" ) return false ; # No link trail
|
|---|
| 235 | $xml .= "<trail>{$x}</trail>" ;
|
|---|
| 236 | $a = $b ;
|
|---|
| 237 | return true ;
|
|---|
| 238 | }
|
|---|
| 239 |
|
|---|
| 240 | function p_internal_link ( &$a , &$xml )
|
|---|
| 241 | {
|
|---|
| 242 | $x = "" ;
|
|---|
| 243 | $b = $a ;
|
|---|
| 244 | if ( !$this->nextis ( $b , "[[" ) ) return false ;
|
|---|
| 245 | if ( !$this->p_internal_link_target ( $b , $x , "]]" ) ) return false ;
|
|---|
| 246 | while ( 1 )
|
|---|
| 247 | {
|
|---|
| 248 | if ( $this->nextis ( $b , "]]" ) ) break ;
|
|---|
| 249 | if ( !$this->nextis ( $b , "|" ) ) return false ;
|
|---|
| 250 | if ( !$this->p_internal_link_text ( $b , $x , false , "]]" ) ) return false ;
|
|---|
| 251 | }
|
|---|
| 252 | $this->p_internal_link_trail ( $b , $x ) ;
|
|---|
| 253 | $xml .= "<link>{$x}</link>" ;
|
|---|
| 254 | $a = $b ;
|
|---|
| 255 | return true ;
|
|---|
| 256 | }
|
|---|
| 257 |
|
|---|
| 258 | function p_magic_variable ( &$a , &$xml )
|
|---|
| 259 | {
|
|---|
| 260 | $x = "" ;
|
|---|
| 261 | $b = $a ;
|
|---|
| 262 | if ( !$this->nextis ( $b , "__" ) ) return false ;
|
|---|
| 263 | $varname = "" ;
|
|---|
| 264 | for ( $c = $b ; $c < $this->wl && $this->w[$c] != '_' ; $c++ )
|
|---|
| 265 | $varname .= $this->w[$c] ;
|
|---|
| 266 | if ( !$this->nextis ( $c , "__" ) ) return false ;
|
|---|
| 267 | $xml .= "<magic_variable>{$varname}</magic_variable>" ;
|
|---|
| 268 | $a = $c ;
|
|---|
| 269 | return true ;
|
|---|
| 270 | }
|
|---|
| 271 |
|
|---|
| 272 | # Template and template variable, utilizing parts of the internal link methods
|
|---|
| 273 | function p_template ( &$a , &$xml )
|
|---|
| 274 | {
|
|---|
| 275 | global $content_provider , $xmlg ;
|
|---|
| 276 | if ( $xmlg["useapi"] ) return false ; # API already resolved templates
|
|---|
| 277 |
|
|---|
| 278 | $x = "" ;
|
|---|
| 279 | $b = $a ;
|
|---|
| 280 | if ( !$this->nextis ( $b , "{{" ) ) return false ;
|
|---|
| 281 | # if ( $this->nextis ( $b , "{" , false ) ) return false ; # Template names may not start with "{"
|
|---|
| 282 | if ( !$this->p_internal_link_target ( $b , $x , "}}" ) ) return false ;
|
|---|
| 283 | $target = $x ;
|
|---|
| 284 | $variables = array () ;
|
|---|
| 285 | $varnames = array () ;
|
|---|
| 286 | $vcount = 1 ;
|
|---|
| 287 | while ( 1 )
|
|---|
| 288 | {
|
|---|
| 289 | if ( $this->nextis ( $b , "}}" ) ) break ;
|
|---|
| 290 | if ( !$this->nextis ( $b , "|" ) ) return false ;
|
|---|
| 291 | $l1 = strlen ( $x ) ;
|
|---|
| 292 | if ( !$this->p_internal_link_text ( $b , $x , false , "}}" ) ) return false ;
|
|---|
| 293 | $v = substr ( $x , $l1 ) ;
|
|---|
| 294 | $v = explode ( "=" , $v ) ;
|
|---|
| 295 | if ( count ( $v ) < 2 ) $vk = $vcount ;
|
|---|
| 296 | else {
|
|---|
| 297 | $vk = trim ( array_shift ( $v ) ) ;
|
|---|
| 298 | $varnames[$vcount] = $vk;
|
|---|
| 299 | }
|
|---|
| 300 | $vv = array_shift ( $v ) ;
|
|---|
| 301 | $variables[$vk] = $vv ;
|
|---|
| 302 | if ( !isset ( $variables[$vcount] ) ) $variables[$vcount] = $vv ;
|
|---|
| 303 | $vcount++ ;
|
|---|
| 304 | }
|
|---|
| 305 |
|
|---|
| 306 | $target = array_pop ( @explode ( ">" , $target , 2 ) ) ;
|
|---|
| 307 | $target = array_shift ( @explode ( "<" , $target , 2 ) ) ;
|
|---|
| 308 | if ( $this->auto_fill_templates == 'all' ) $replace_template = true ;
|
|---|
| 309 | else if ( $this->auto_fill_templates == 'none' ) $replace_template = false ;
|
|---|
| 310 | else {
|
|---|
| 311 | $found = in_array ( ucfirst ( $target ) , $this->template_list ) ;
|
|---|
| 312 | if ( $found AND $this->auto_fill_templates == 'these' ) $replace_template = true ;
|
|---|
| 313 | else if ( !$found AND $this->auto_fill_templates == 'notthese' ) $replace_template = true ;
|
|---|
| 314 | else $replace_template = false ;
|
|---|
| 315 | }
|
|---|
| 316 |
|
|---|
| 317 | if ( substr ( $target , 0 , 1 ) == '#' ) { # Try template logic
|
|---|
| 318 | $between = $this->process_template_logic ( $target , $variables ) ;
|
|---|
| 319 | # Change source (!)
|
|---|
| 320 | $w1 = substr ( $this->w , 0 , $a ) ;
|
|---|
| 321 | $w2 = substr ( $this->w , $b ) ;
|
|---|
| 322 | $this->w = $w1 . $between . $w2 ;
|
|---|
| 323 | $this->wl = strlen ( $this->w ) ;
|
|---|
| 324 | } else if ( $replace_template ) { # Do not generate <template> sections, but rather replace the template call with the template text
|
|---|
| 325 |
|
|---|
| 326 | # Get template text
|
|---|
| 327 | $between = trim ( $content_provider->get_template_text ( $target ) ) ;
|
|---|
| 328 | add_authors ( $content_provider->authors ) ;
|
|---|
| 329 |
|
|---|
| 330 | # Removing <noinclude> stuff
|
|---|
| 331 | $between = preg_replace( '?<noinclude>.*</noinclude>?msU', '', $between);
|
|---|
| 332 | $between = str_replace ( "<include>" , "" , $between ) ;
|
|---|
| 333 | $between = str_replace ( "</include>" , "" , $between ) ;
|
|---|
| 334 | $between = str_replace ( "<includeonly>" , "" , $between ) ;
|
|---|
| 335 | $between = str_replace ( "</includeonly>" , "" , $between ) ;
|
|---|
| 336 |
|
|---|
| 337 | # Remove HTML comments
|
|---|
| 338 | $between = str_replace ( "-->\n" , "-->" , $between ) ;
|
|---|
| 339 | $between = preg_replace( '?<!--.*-->?msU', '', $between) ;
|
|---|
| 340 |
|
|---|
| 341 | # Replacing template variables.
|
|---|
| 342 | # ATTENTION: Template variables within <nowiki> sections of templates will be replaced as well!
|
|---|
| 343 |
|
|---|
| 344 | if ( $a > 0 && substr ( $between , 0 , 2 ) == '{|' )
|
|---|
| 345 | $between = "\n" . $between ;
|
|---|
| 346 |
|
|---|
| 347 | $this->replace_template_variables ( $between , $variables ) ;
|
|---|
| 348 |
|
|---|
| 349 | # Change source (!)
|
|---|
| 350 | $w1 = substr ( $this->w , 0 , $a ) ;
|
|---|
| 351 | $w2 = substr ( $this->w , $b ) ;
|
|---|
| 352 | $this->w = $w1 . $between . $w2 ;
|
|---|
| 353 | $this->wl = strlen ( $this->w ) ;
|
|---|
| 354 | } else {
|
|---|
| 355 | $xml .= "<template><target>{$target}</target>";
|
|---|
| 356 | for ( $i = 1 ; $i < $vcount ; $i++ ) {
|
|---|
| 357 | if ( isset( $varnames[$i] ) ) $xml .= "<arg name=\"{$varnames[$i]}\">{$variables[$i]}</arg>";
|
|---|
| 358 | else $xml .= "<arg>{$variables[$i]}</arg>";
|
|---|
| 359 | }
|
|---|
| 360 | $xml .= "</template>" ;
|
|---|
| 361 | $a = $b ;
|
|---|
| 362 | }
|
|---|
| 363 | return true ;
|
|---|
| 364 | }
|
|---|
| 365 |
|
|---|
| 366 | function process_template_logic ( $title , $variables ) {
|
|---|
| 367 |
|
|---|
| 368 | # TODO : Process title and variables for sub-template-replacements
|
|---|
| 369 |
|
|---|
| 370 | if ( substr ( $title , 0 , 4 ) == "#if:" ) {
|
|---|
| 371 | $title = trim ( substr ( $title , 4 ) ) ;
|
|---|
| 372 | if ( $title == '' ) return array_pop ( $variables ) ; # ELSE
|
|---|
| 373 | return array_shift ( $variables ) ; # THEN
|
|---|
| 374 | }
|
|---|
| 375 |
|
|---|
| 376 | if ( substr ( $title , 0 , 8 ) == "#switch:" ) {
|
|---|
| 377 | $title = trim ( array_pop ( explode ( ':' , $title , 2 ) ) ) ;
|
|---|
| 378 | foreach ( $variables AS $v ) {
|
|---|
| 379 | $v = explode ( '=' , $v , 2 ) ;
|
|---|
| 380 | $key = trim ( array_shift ( $v ) ) ;
|
|---|
| 381 | if ( $key != $title ) continue ; # Wrong key
|
|---|
| 382 | return array_pop ( $v ) ; # Correct key, return value
|
|---|
| 383 | }
|
|---|
| 384 | }
|
|---|
| 385 |
|
|---|
| 386 | # BAD PARSER FUNCTION! Ignoring...
|
|---|
| 387 | return $title ;
|
|---|
| 388 | }
|
|---|
| 389 |
|
|---|
| 390 | function replace_template_variables ( &$text , &$variables ) {
|
|---|
| 391 | global $xmlg ;
|
|---|
| 392 | if ( $xmlg["useapi"] ) return false ; # API already resolved templates
|
|---|
| 393 | for ( $a = 0 ; $a+3 < strlen ( $text ) ; $a++ ) {
|
|---|
| 394 | if ( $text[$a] != '{' ) continue ;
|
|---|
| 395 | while ( $this->p_template_replace_single_variable ( $text , $a , $variables ) ) ;
|
|---|
| 396 | }
|
|---|
| 397 | }
|
|---|
| 398 |
|
|---|
| 399 | function p_template_replace_single_variable ( &$text , $a , &$variables ) {
|
|---|
| 400 | if ( substr ( $text , $a , 3 ) != '{{{' ) return false ;
|
|---|
| 401 | $b = $a + 3 ;
|
|---|
| 402 |
|
|---|
| 403 | # Name
|
|---|
| 404 | $start = $b ;
|
|---|
| 405 | $count = 0 ;
|
|---|
| 406 | while ( $b < strlen ( $text ) && ( $text[$b] != '|' || $count > 0 ) && ( substr ( $text , $b , 3 ) != '}}}' || $count > 0 ) ) {
|
|---|
| 407 | if ( $this->p_template_replace_single_variable ( $text , $b , $variables ) ) continue ;
|
|---|
| 408 | if ( $text[$b] == '{' ) $count++ ;
|
|---|
| 409 | if ( $text[$b] == '}' ) $count-- ;
|
|---|
| 410 | $b++ ;
|
|---|
| 411 | }
|
|---|
| 412 | if ( $b >= strlen ( $text ) ) return false ;
|
|---|
| 413 | $name = trim ( substr ( $text , $start , $b - $start ) ) ;
|
|---|
| 414 |
|
|---|
| 415 | # Default value
|
|---|
| 416 | $value = "" ;
|
|---|
| 417 | if ( $text[$b] == '|' ) {
|
|---|
| 418 | $b++ ;
|
|---|
| 419 | $start = $b ;
|
|---|
| 420 | $count = 0 ;
|
|---|
| 421 | while ( $b < strlen ( $text ) && ( substr ( $text , $b , 3 ) != '}}}' || $count > 0 ) ) {
|
|---|
| 422 | if ( $this->p_template_replace_single_variable ( $text , $b , $variables ) ) continue ;
|
|---|
| 423 | if ( $text[$b] == '{' ) $count++ ;
|
|---|
| 424 | if ( $text[$b] == '}' ) $count-- ;
|
|---|
| 425 | $b++ ;
|
|---|
| 426 | }
|
|---|
| 427 | if ( $b >= strlen ( $text ) ) return false ;
|
|---|
| 428 | $value = trim ( substr ( $text , $start , $b - $start ) ) ;#$start - $b - 1 ) ) ;
|
|---|
| 429 | }
|
|---|
| 430 |
|
|---|
| 431 | // Replace
|
|---|
| 432 | $b += 3 ; # }}}
|
|---|
| 433 | if ( isset ( $variables[$name] ) ) {
|
|---|
| 434 | $value = $variables[$name] ;
|
|---|
| 435 | }
|
|---|
| 436 | $text = substr ( $text , 0 , $a ) . $value . substr ( $text , $b ) ;
|
|---|
| 437 |
|
|---|
| 438 | return true ;
|
|---|
| 439 | }
|
|---|
| 440 |
|
|---|
| 441 | function p_template_variable ( &$a , &$xml )
|
|---|
| 442 | {
|
|---|
| 443 | $x = "" ;
|
|---|
| 444 | $b = $a ;
|
|---|
| 445 | if ( !$this->nextis ( $b , "{{{" ) ) return false ;
|
|---|
| 446 | if ( !$this->p_internal_link_text ( $b , $x , false , "}}}" ) ) return false ;
|
|---|
| 447 | if ( !$this->nextis ( $b , "}}}" ) ) return false ;
|
|---|
| 448 | $xml .= "<templatevar>{$x}</templatevar>" ;
|
|---|
| 449 | $a = $b ;
|
|---|
| 450 | return true ;
|
|---|
| 451 | }
|
|---|
| 452 |
|
|---|
| 453 | # Bold / italics
|
|---|
| 454 | function p_bold ( &$a , &$xml , $recurse = "restofline" , $end = "" )
|
|---|
| 455 | {
|
|---|
| 456 | return $this->p_intwined ( $a , $xml , "bold" , "'''" , $recurse , $end ) ;
|
|---|
| 457 | }
|
|---|
| 458 |
|
|---|
| 459 | function p_italics ( &$a , &$xml , $recurse = "restofline" , $end = "" )
|
|---|
| 460 | {
|
|---|
| 461 | return $this->p_intwined ( $a , $xml , "italics" , "''" , $recurse , $end ) ;
|
|---|
| 462 | }
|
|---|
| 463 |
|
|---|
| 464 | function p_intwined ( &$a , &$xml , $tag , $markup , $recurse , $end )
|
|---|
| 465 | {
|
|---|
| 466 | $b = $a ;
|
|---|
| 467 | if ( !$this->nextis ( $b , $markup ) ) return false ;
|
|---|
| 468 | $id = substr ( ucfirst ( $tag ) , 0 , 1 ) ;
|
|---|
| 469 | $bi = $this->bold_italics ;
|
|---|
| 470 | $open = false ;
|
|---|
| 471 | if ( substr ( $this->bold_italics , -1 ) == $id )
|
|---|
| 472 | {
|
|---|
| 473 | $x = "</{$tag}>" ;
|
|---|
| 474 | $this->bold_italics = substr ( $this->bold_italics , 0 , -1 ) ;
|
|---|
| 475 | }
|
|---|
| 476 | else
|
|---|
| 477 | {
|
|---|
| 478 | $pos = strpos ( $this->bold_italics , $id ) ;
|
|---|
| 479 | if ( false !== $pos ) return false ; # Can't close a tag that ain't open
|
|---|
| 480 | $open = true ;
|
|---|
| 481 | $x = "<{$tag}>" ;
|
|---|
| 482 | $this->bold_italics .= $id ;
|
|---|
| 483 | }
|
|---|
| 484 |
|
|---|
| 485 | if ( $end == "" )
|
|---|
| 486 | {
|
|---|
| 487 | $res = $this->once ( $b , $x , $recurse ) ;
|
|---|
| 488 | }
|
|---|
| 489 | else
|
|---|
| 490 | {
|
|---|
| 491 | $r = "p_{$recurse}" ;
|
|---|
| 492 | $res = $this->$r ( $b , $x , $end ) ;
|
|---|
| 493 | }
|
|---|
| 494 |
|
|---|
| 495 | $this->bold_italics = $bi ;
|
|---|
| 496 | if ( !$res )
|
|---|
| 497 | {
|
|---|
| 498 | return false ;
|
|---|
| 499 | }
|
|---|
| 500 | $xml .= $x ;
|
|---|
| 501 | $a = $b ;
|
|---|
| 502 | return true ;
|
|---|
| 503 | }
|
|---|
| 504 |
|
|---|
| 505 | function scanplaintext ( &$a , &$xml , $goodstop , $badstop )
|
|---|
| 506 | {
|
|---|
| 507 | $b = $a ;
|
|---|
| 508 | $x = "" ;
|
|---|
| 509 | while ( $b < $this->wl )
|
|---|
| 510 | {
|
|---|
| 511 | if ( $this->w[$b] == "{" && $this->once ( $b , $x , "template" ) ) continue ;
|
|---|
| 512 | foreach ( $goodstop AS $s )
|
|---|
| 513 | if ( $this->nextis ( $b , $s , false ) ) break 2 ;
|
|---|
| 514 | foreach ( $badstop AS $s )
|
|---|
| 515 | if ( $this->nextis ( $b , $s , false ) ) return false ;
|
|---|
| 516 | $c = $this->w[$b] ;
|
|---|
| 517 | $x .= htmlspecialchars ( $c ) ;
|
|---|
| 518 | $b++ ;
|
|---|
| 519 | }
|
|---|
| 520 | if ( count ( $goodstop ) > 0 && $b >= $this->wl ) return false ; # Reached end; not good
|
|---|
| 521 | $a = $b ;
|
|---|
| 522 | $xml .= $x ;
|
|---|
| 523 | return true ;
|
|---|
| 524 | }
|
|---|
| 525 |
|
|---|
| 526 | # External link
|
|---|
| 527 | function p_external_freelink ( &$a , &$xml , $mark = true )
|
|---|
| 528 | {
|
|---|
| 529 | if ( $this->wl <= $a + 10 ) return false ; # Can't have an URL shorter than that
|
|---|
| 530 | if ( $this->w[$a+5] != '/' && $this->w[$a+7] != '/' ) return false ; # Shortcut for protocols 3-6 chars length
|
|---|
| 531 | $protocol = "" ;
|
|---|
| 532 | $b = $a ;
|
|---|
| 533 | # while ( $this->w[$b] == "{" && $this->once ( $b , $x , "template" ) ) $b = $a ;
|
|---|
| 534 | foreach ( $this->protocols AS $p )
|
|---|
| 535 | {
|
|---|
| 536 | if ( $this->nextis ( $b , $p . "://" ) )
|
|---|
| 537 | {
|
|---|
| 538 | $protocol = $p ;
|
|---|
| 539 | break ;
|
|---|
| 540 | }
|
|---|
| 541 | }
|
|---|
| 542 | if ( $protocol == "" ) return false ;
|
|---|
| 543 | $x = "{$protocol}://" ;
|
|---|
| 544 | while ( $b < $this->wl )
|
|---|
| 545 | {
|
|---|
| 546 | $c = $this->w[$b] ;
|
|---|
| 547 | if ( $c == "{" && $this->once ( $b , $x , "template" ) ) continue ;
|
|---|
| 548 | if ( $c == "\n" || $c == " " || $c == '|' ) break ;
|
|---|
| 549 | if ( !$mark && $c == "]" ) break ;
|
|---|
| 550 | $x .= htmlspecialchars ( $c ) ;
|
|---|
| 551 | $b++ ;
|
|---|
| 552 | }
|
|---|
| 553 | if ( substr ( $x , -1 ) == "." || substr ( $x , -1 ) == "," )
|
|---|
| 554 | {
|
|---|
| 555 | $x = substr ( $x , 0 , -1 ) ;
|
|---|
| 556 | $b-- ;
|
|---|
| 557 | }
|
|---|
| 558 | $a = $b ;
|
|---|
| 559 | $x = htmlspecialchars ( $x , ENT_QUOTES ) ;
|
|---|
| 560 | if ( $mark ) $xml .= "<link type='external' href='$x'/>" ;
|
|---|
| 561 | else $xml .= $x ;
|
|---|
| 562 | return true ;
|
|---|
| 563 | }
|
|---|
| 564 |
|
|---|
| 565 | function p_external_link ( &$a , &$xml , $mark = true )
|
|---|
| 566 | {
|
|---|
| 567 | $b = $a ;
|
|---|
| 568 | if ( !$this->nextis ( $b , "[" ) ) return false ;
|
|---|
| 569 | $url = "" ;
|
|---|
| 570 | $c = $b ;
|
|---|
| 571 | $x = "" ;
|
|---|
| 572 | while ( $c < $this->wl && $this->w[$c] == "{" && $this->once ( $c , $x , "template" ) ) $c = $b ;
|
|---|
| 573 | if ( $c >= $this->wl ) return false ;
|
|---|
| 574 | $x = "" ;
|
|---|
| 575 | if ( !$this->p_external_freelink ( $b , $url , false ) ) return false ;
|
|---|
| 576 | $this->skipblanks ( $b ) ;
|
|---|
| 577 | if ( !$this->scanplaintext ( $b , $x , array ( "]" ) , array ( "\n" ) ) ) return false ;
|
|---|
| 578 | $a = $b + 1 ;
|
|---|
| 579 | $xml .= "<link type='external' href='{$url}'>{$x}</link>" ;
|
|---|
| 580 | return true ;
|
|---|
| 581 | }
|
|---|
| 582 |
|
|---|
| 583 | # Heading
|
|---|
| 584 | function p_heading ( &$a , &$xml )
|
|---|
| 585 | {
|
|---|
| 586 | if ( $a >= $this->wl || $this->w[$a] != '=' ) return false ;
|
|---|
| 587 | $b = $a ;
|
|---|
| 588 | $level = 0 ;
|
|---|
| 589 | $h = "" ;
|
|---|
| 590 | $x = "" ;
|
|---|
| 591 | while ( $this->nextis ( $b , "=" ) )
|
|---|
| 592 | {
|
|---|
| 593 | $level++ ;
|
|---|
| 594 | $h .= "=" ;
|
|---|
| 595 | }
|
|---|
| 596 | $this->skipblanks ( $b ) ;
|
|---|
| 597 | if ( !$this->once ( $b , $x , "restofline" ) ) return false ;
|
|---|
| 598 | if ( $this->compensate_markup_errors ) $x = trim ( $x ) ;
|
|---|
| 599 | else if ( $x != trim ( $x ) ) $xml .= "<error type='heading' reason='trailing blank'/>" ;
|
|---|
| 600 | if ( substr ( $x , -$level ) != $h ) return false ; # No match
|
|---|
| 601 |
|
|---|
| 602 | $x = trim ( substr ( $x , 0 , -$level ) ) ;
|
|---|
| 603 | $level -= 1 ;
|
|---|
| 604 | $a = $b ;
|
|---|
| 605 | $xml .= "<heading level='" . ($level+1) . "'>{$x}</heading>" ;
|
|---|
| 606 | return true ;
|
|---|
| 607 | }
|
|---|
| 608 |
|
|---|
| 609 | # Line
|
|---|
| 610 | # Often used function for parsing the rest of a text line
|
|---|
| 611 | function p_restofline ( &$a , &$xml , $closeit = array() )
|
|---|
| 612 | {
|
|---|
| 613 | $b = $a ;
|
|---|
| 614 | $x = "" ;
|
|---|
| 615 | $override = false ;
|
|---|
| 616 | while ( $b < $this->wl && !$override )
|
|---|
| 617 | {
|
|---|
| 618 | $c = $this->w[$b] ;
|
|---|
| 619 | if ( $c == "\n" ) { $b++ ; break ; }
|
|---|
| 620 | foreach ( $closeit AS $z )
|
|---|
| 621 | if ( $this->nextis ( $b , $z , false ) ) break ;
|
|---|
| 622 | if ( $c == "_" && $this->once ( $b , $x , "magic_variable" ) ) continue ;
|
|---|
| 623 | if ( $c == "[" && $this->once ( $b , $x , "internal_link" ) ) continue ;
|
|---|
| 624 | if ( $c == "[" && $this->once ( $b , $x , "external_link" ) ) continue ;
|
|---|
| 625 | if ( $c == "{" && $this->once ( $b , $x , "template_variable" ) ) continue ;
|
|---|
| 626 | if ( $c == "{" && $this->once ( $b , $x , "template" ) ) continue ;
|
|---|
| 627 | if ( $c == "{" && $this->p_table ( $b , $x ) ) continue ;
|
|---|
| 628 | if ( $c == "<" && $this->once ( $b , $x , "html" ) ) continue ;
|
|---|
| 629 | if ( $c == "'" && $this->once ( $b , $x , "bold" ) ) { $override = true ; break ; }
|
|---|
| 630 | if ( $c == "'" && $this->once ( $b , $x , "italics" ) ) { $override = true ; break ; }
|
|---|
| 631 | if ( $b + 10 < $this->wl &&
|
|---|
| 632 | ( $this->w[$a+5] == '/' && $this->w[$a+7] == '/' ) &&
|
|---|
| 633 | $this->once ( $b , $x , "external_freelink" ) ) continue ;
|
|---|
| 634 |
|
|---|
| 635 | # Just an ordinary character
|
|---|
| 636 | $x .= htmlspecialchars ( $c ) ;
|
|---|
| 637 | $b++ ;
|
|---|
| 638 | if ( $b >= $this->wl ) break ;
|
|---|
| 639 | }
|
|---|
| 640 | if ( !$override && $this->bold_italics != "" )
|
|---|
| 641 | {
|
|---|
| 642 | return false ;
|
|---|
| 643 | }
|
|---|
| 644 | $xml .= $x ;
|
|---|
| 645 | $a = $b ;
|
|---|
| 646 | return true ;
|
|---|
| 647 | }
|
|---|
| 648 |
|
|---|
| 649 | function p_line ( &$a , &$xml , $force )
|
|---|
| 650 | {
|
|---|
| 651 | if ( $a >= $this->wl ) return false ; # Already at the end of the text
|
|---|
| 652 | $c = $this->w[$a] ;
|
|---|
| 653 | if ( !$force )
|
|---|
| 654 | {
|
|---|
| 655 | if ( $c == '*' || $c == ':' || $c == '#' || $c == ';' || $c == ' ' || $c == "\n" ) return false ; # Not a suitable beginning
|
|---|
| 656 | if ( $this->nextis ( $a , "{|" , false ) ) return false ; # Table
|
|---|
| 657 | if ( count ( $this->tables ) > 0 && $this->nextis ( $a , "|" , false ) ) return false ; # Table
|
|---|
| 658 | if ( count ( $this->tables ) > 0 && $this->nextis ( $a , "!" , false ) ) return false ; # Table
|
|---|
| 659 | if ( $this->nextis ( $a , "=" , false ) ) return false ; # Heading
|
|---|
| 660 | if ( $this->nextis ( $a , "----" , false ) ) return false ; # <hr>
|
|---|
| 661 | }
|
|---|
| 662 | $this->bold_italics = "" ;
|
|---|
| 663 | return $this->once ( $a , $xml , "restofline" ) ;
|
|---|
| 664 | }
|
|---|
| 665 |
|
|---|
| 666 | function p_blankline ( &$a , &$xml )
|
|---|
| 667 | {
|
|---|
| 668 | if ( $this->nextis ( $a , "\n" ) ) return true ;
|
|---|
| 669 | return false ;
|
|---|
| 670 | }
|
|---|
| 671 |
|
|---|
| 672 | function p_block_lines ( &$a , &$xml , $force = false )
|
|---|
| 673 | {
|
|---|
| 674 | $x = "" ;
|
|---|
| 675 | $b = $a ;
|
|---|
| 676 | if ( !$this->p_line ( $b , $x , $force ) ) return false ;
|
|---|
| 677 | while ( $this->p_line ( $b , $x , false ) ) ;
|
|---|
| 678 | while ( $this->p_blankline ( $b , $x ) ) ; # Skip coming blank lines
|
|---|
| 679 | $xml .= "<paragraph>{$x}</paragraph>" ;
|
|---|
| 680 | $a = $b ;
|
|---|
| 681 | return true ;
|
|---|
| 682 | }
|
|---|
| 683 |
|
|---|
| 684 |
|
|---|
| 685 |
|
|---|
| 686 | # PRE block
|
|---|
| 687 | # Parses a line starting with ' '
|
|---|
| 688 | function p_preline ( &$a , &$xml )
|
|---|
| 689 | {
|
|---|
| 690 | if ( $a >= $this->wl ) return false ; # Already at the end of the text
|
|---|
| 691 | if ( $this->w[$a]!= ' ' ) return false ; # Not a preline
|
|---|
| 692 |
|
|---|
| 693 | $a++ ;
|
|---|
| 694 | $this->bold_italics = "" ;
|
|---|
| 695 | $x = "" ;
|
|---|
| 696 | $ret = $this->once ( $a , $x , "restofline" ) ;
|
|---|
| 697 | if ( $ret ) {
|
|---|
| 698 | $xml .= "<preline>" . $x . "</preline>" ;
|
|---|
| 699 | }
|
|---|
| 700 | return $ret ;
|
|---|
| 701 | }
|
|---|
| 702 |
|
|---|
| 703 | # Parses a block of lines each starting with ' '
|
|---|
| 704 | function p_block_pre ( &$a , &$xml )
|
|---|
| 705 | {
|
|---|
| 706 | $x = "" ;
|
|---|
| 707 | $b = $a ;
|
|---|
| 708 | if ( !$this->once ( $b , $x , "preline" , true , true ) ) return false ;
|
|---|
| 709 | $this->once ( $b , $x , "blankline" , false , true ) ;
|
|---|
| 710 | $xml .= "<preblock>{$x}</preblock>" ;
|
|---|
| 711 | $a = $b ;
|
|---|
| 712 | return true ;
|
|---|
| 713 | }
|
|---|
| 714 |
|
|---|
| 715 | # LIST block
|
|---|
| 716 | # Returns a list tag depending on the wiki markup
|
|---|
| 717 | function listtag ( $c , $open = true )
|
|---|
| 718 | {
|
|---|
| 719 | if ( !$open ) return "</list>" ;
|
|---|
| 720 | $r = "" ;
|
|---|
| 721 | if ( $c == '#' ) $r = "numbered" ;
|
|---|
| 722 | if ( $c == '*' ) $r = "bullet" ;
|
|---|
| 723 | if ( $c == ':' ) $r = "ident" ;
|
|---|
| 724 | if ( $c == ';' ) $r = "def" ;
|
|---|
| 725 | if ( $r != "" ) $r = " type='{$r}'" ;
|
|---|
| 726 | $r = "<list{$r}>" ;
|
|---|
| 727 | return $r ;
|
|---|
| 728 | }
|
|---|
| 729 |
|
|---|
| 730 | # Opens/closes list tags
|
|---|
| 731 | function fixlist ( $last , $cur )
|
|---|
| 732 | {
|
|---|
| 733 | $r = "" ;
|
|---|
| 734 | $olast = $last ;
|
|---|
| 735 | $ocur = $cur ;
|
|---|
| 736 | $ocommon = "" ;
|
|---|
| 737 |
|
|---|
| 738 | # Remove matching parts
|
|---|
| 739 | while ( $last != "" && $cur != "" && $last[0] == $cur[0] )
|
|---|
| 740 | {
|
|---|
| 741 | $ocommon = $cur[0] ;
|
|---|
| 742 | $cur = substr ( $cur , 1 ) ;
|
|---|
| 743 | $last = substr ( $last , 1 ) ;
|
|---|
| 744 | }
|
|---|
| 745 |
|
|---|
| 746 | # Close old tags
|
|---|
| 747 | $fixitemtag = false ;
|
|---|
| 748 | if ( $last != "" && $ocommon != "" ) $fixitemtag = true ;
|
|---|
| 749 | while ( $last != "" )
|
|---|
| 750 | {
|
|---|
| 751 | $r .= "</listitem>" . $this->listtag ( substr ( $last , -1 ) , false ) ;
|
|---|
| 752 | $last = substr ( $last , 0 , -1 ) ;
|
|---|
| 753 | }
|
|---|
| 754 | if ( $fixitemtag ) $r .= "</listitem><listitem>" ;
|
|---|
| 755 |
|
|---|
| 756 | # Open new tags
|
|---|
| 757 | while ( $cur != "" )
|
|---|
| 758 | {
|
|---|
| 759 | $r .= $this->listtag ( $cur[0] ) . "<listitem>" ;
|
|---|
| 760 | $cur = substr ( $cur , 1 ) ;
|
|---|
| 761 | }
|
|---|
| 762 |
|
|---|
| 763 | return $r ;
|
|---|
| 764 | }
|
|---|
| 765 |
|
|---|
| 766 | # Parses a single list line
|
|---|
| 767 | function p_list_line ( &$a , &$xml , &$last )
|
|---|
| 768 | {
|
|---|
| 769 | $cur = "" ;
|
|---|
| 770 | do {
|
|---|
| 771 | $lcur = $cur ;
|
|---|
| 772 | while ( $this->nextis ( $a , "*" ) ) $cur .= "*" ;
|
|---|
| 773 | while ( $this->nextis ( $a , "#" ) ) $cur .= "#" ;
|
|---|
| 774 | while ( $this->nextis ( $a , ":" ) ) $cur .= ":" ;
|
|---|
| 775 | while ( $this->nextis ( $a , ";" ) ) $cur .= ";" ;
|
|---|
| 776 | } while ( $cur != $lcur ) ;
|
|---|
| 777 |
|
|---|
| 778 | $unchanged = false ;
|
|---|
| 779 | # if ( substr ( $cur , 0 , strlen ( $last ) ) == $last ) $unchanged = true ;
|
|---|
| 780 | if ( $last == $cur ) $unchanged = true ;
|
|---|
| 781 | $xml .= $this->fixlist ( $last , $cur ) ;
|
|---|
| 782 |
|
|---|
| 783 | if ( $cur == "" ) return false ; # Not a list line
|
|---|
| 784 | $last = $cur ;
|
|---|
| 785 | $this->skipblanks ( $a ) ;
|
|---|
| 786 |
|
|---|
| 787 | if ( $unchanged ) $xml .= "</listitem><listitem>" ;
|
|---|
| 788 | if ( $cur == ";" ) # Definition
|
|---|
| 789 | {
|
|---|
| 790 | $b = $a ;
|
|---|
| 791 | while ( $b < $this->wl && $this->w[$b] != "\n" && $this->w[$b] != ':' ) $b++ ;
|
|---|
| 792 | if ( $b >= $this->wl || $this->w[$b] == "\n" )
|
|---|
| 793 | {
|
|---|
| 794 | $xml .= "<defkey>" ;
|
|---|
| 795 | $this->p_restofline ( $a , $xml ) ;
|
|---|
| 796 | $xml .= "</defkey>" ;
|
|---|
| 797 | }
|
|---|
| 798 | else
|
|---|
| 799 | {
|
|---|
| 800 | $xml .= "<defkey>" ;
|
|---|
| 801 | $this->w[$b] = "\n" ;
|
|---|
| 802 | $this->p_restofline ( $a , $xml ) ;
|
|---|
| 803 | $xml .= "</defkey>" ;
|
|---|
| 804 | $xml .= "<defval>" ;
|
|---|
| 805 | $this->p_restofline ( $a , $xml ) ;
|
|---|
| 806 | $xml .= "</defval>" ;
|
|---|
| 807 | }
|
|---|
| 808 | }
|
|---|
| 809 | else $this->p_restofline ( $a , $xml ) ;
|
|---|
| 810 | return true ;
|
|---|
| 811 | }
|
|---|
| 812 |
|
|---|
| 813 | # Checks for a list block ( those nasty things starting with '*', '#', or the like...
|
|---|
| 814 | function p_block_list ( &$a , &$xml )
|
|---|
| 815 | {
|
|---|
| 816 | $last = "" ;
|
|---|
| 817 | $found = false ;
|
|---|
| 818 | while ( $this->p_list_line ( $a , $xml , $last ) ) $found = true ;
|
|---|
| 819 | return $found ;
|
|---|
| 820 | }
|
|---|
| 821 |
|
|---|
| 822 | # HTML
|
|---|
| 823 | # This function detects a HTML tag, finds the matching close tag,
|
|---|
| 824 | # parses everything in between, and returns everything as an extension.
|
|---|
| 825 | # Returns false otherwise.
|
|---|
| 826 | function p_html ( &$a , &$xml )
|
|---|
| 827 | {
|
|---|
| 828 | if ( !$this->nextis ( $a , "<" , false ) ) return false ;
|
|---|
| 829 |
|
|---|
| 830 | $b = $a ;
|
|---|
| 831 | $x = "" ;
|
|---|
| 832 | $tag = "" ;
|
|---|
| 833 | $closing = false ;
|
|---|
| 834 | $selfclosing = false ;
|
|---|
| 835 |
|
|---|
| 836 | if ( !$this->p_html_tag ( $b , $x , $tag , $closing , $selfclosing ) ) return false ;
|
|---|
| 837 |
|
|---|
| 838 | if ( isset ( $this->directhtmltags[$tag] ) )
|
|---|
| 839 | {
|
|---|
| 840 | $tag_open = "<" . $this->directhtmltags[$tag] ;
|
|---|
| 841 | $tag_close = "</" . $this->directhtmltags[$tag] . ">" ;
|
|---|
| 842 | }
|
|---|
| 843 | else
|
|---|
| 844 | {
|
|---|
| 845 | $tag_open = "<extension extension_name='{$tag}'" ;
|
|---|
| 846 | $tag_close = "</extension>" ;
|
|---|
| 847 | }
|
|---|
| 848 |
|
|---|
| 849 | # Is this tag self-closing?
|
|---|
| 850 | if ( $selfclosing )
|
|---|
| 851 | {
|
|---|
| 852 | $a = $b ;
|
|---|
| 853 | $xml .= $tag_open . $x . ">" . $tag_close ;
|
|---|
| 854 | return true ;
|
|---|
| 855 | }
|
|---|
| 856 |
|
|---|
| 857 | # Find the matching close tag
|
|---|
| 858 | # TODO : The simple open/close counter should be replaced with a
|
|---|
| 859 | # stack to allow for tolerating half-broken HTML,
|
|---|
| 860 | # such as unclosed <li> tags
|
|---|
| 861 | $begin = $b ;
|
|---|
| 862 | $cnt = 1 ;
|
|---|
| 863 | $tag2 = "" ;
|
|---|
| 864 | while ( $cnt > 0 && $b < $this->wl )
|
|---|
| 865 | {
|
|---|
| 866 | $x2 = "" ;
|
|---|
| 867 | $last = $b ;
|
|---|
| 868 | if ( !$this->p_html_tag ( $b , $x2 , $tag2 , $closing , $selfclosing ) )
|
|---|
| 869 | {
|
|---|
| 870 | $dummy = "";
|
|---|
| 871 | if ( $tag != "nowiki" && $this->w[$b] == '{' && $this->p_template ( $b , $dummy ) )
|
|---|
| 872 | continue ;
|
|---|
| 873 | $b++ ;
|
|---|
| 874 | continue ;
|
|---|
| 875 | }
|
|---|
| 876 | if ( $tag != $tag2 ) continue ;
|
|---|
| 877 | if ( $selfclosing ) continue ;
|
|---|
| 878 | if ( $closing ) $cnt-- ;
|
|---|
| 879 | else $cnt++ ;
|
|---|
| 880 | }
|
|---|
| 881 |
|
|---|
| 882 | if ( $cnt > 0 ) return false ; # Tag was never closed
|
|---|
| 883 |
|
|---|
| 884 | # What happens in between?
|
|---|
| 885 | $between = substr ( $this->w , $begin , $last - $begin ) ;
|
|---|
| 886 |
|
|---|
| 887 | if ( $tag != "nowiki" && $tag != "math" )
|
|---|
| 888 | {
|
|---|
| 889 | if ( $tag == 'gallery' ) {
|
|---|
| 890 | $this->gallery2wiki ( $between ) ;
|
|---|
| 891 | $tag_open = "" ;
|
|---|
| 892 | $tag_close = "" ;
|
|---|
| 893 | }
|
|---|
| 894 |
|
|---|
| 895 | # Parse the part in between the tags
|
|---|
| 896 | $subparser = new wiki2xml ;
|
|---|
| 897 | $subparser->inherit ( $this ) ;
|
|---|
| 898 | $between2 = $subparser->parse ( $between ) ;
|
|---|
| 899 |
|
|---|
| 900 | # Was the parsing correct?
|
|---|
| 901 | if ( $between2 != $this->errormessage )
|
|---|
| 902 | $between = $this->strip_single_paragraph ( $between2 ) ; # No <paragraph> for inline HTML tags
|
|---|
| 903 | else
|
|---|
| 904 | $between = htmlspecialchars ( $between ) ; # Incorrect markup, use safe wiki source instead
|
|---|
| 905 | }
|
|---|
| 906 | else $between = htmlspecialchars ( $between ) ; # No wiki parsing in here
|
|---|
| 907 |
|
|---|
| 908 | $a = $b ;
|
|---|
| 909 | if ( $tag_open != "" ) $xml .= $tag_open . $x . ">" ;
|
|---|
| 910 | $xml .= $between ;
|
|---|
| 911 | if ( $tag_close != "" ) $xml .= $tag_close ;
|
|---|
| 912 | return true ;
|
|---|
| 913 | }
|
|---|
| 914 |
|
|---|
| 915 | /**
|
|---|
| 916 | * Converts the lines within a <gallery> to wiki tables
|
|---|
| 917 | */
|
|---|
| 918 | function gallery2wiki ( &$text ) {
|
|---|
| 919 | $lines = explode ( "\n" , trim ( $text ) ) ;
|
|---|
| 920 | $text = "{| style='border-collapse: collapse; border: 1px solid grey;'\n" ;
|
|---|
| 921 | $cnt = 0 ;
|
|---|
| 922 | foreach ( $lines AS $line ) {
|
|---|
| 923 | if ( $cnt >= 4 ) {
|
|---|
| 924 | $cnt = 0 ;
|
|---|
| 925 | $text .= "|--\n" ;
|
|---|
| 926 | }
|
|---|
| 927 | $a = explode ( "|" , $line , 2 ) ;
|
|---|
| 928 | if ( count ( $a ) == 1 ) { # Generate caption from file name
|
|---|
| 929 | $b = $a[0] ;
|
|---|
| 930 | $b = explode ( ":" , $b , 2 ) ;
|
|---|
| 931 | $b = array_pop ( $b ) ;
|
|---|
| 932 | $b = explode ( "." , $b ) ;
|
|---|
| 933 | array_pop ( $b ) ;
|
|---|
| 934 | $a[] = implode ( "." , $b ) ;
|
|---|
| 935 | }
|
|---|
| 936 | $link = array_shift ( $a ) ;
|
|---|
| 937 | $caption = array_pop ( $a ) ;
|
|---|
| 938 | $text .= "|valign=top align=left|[[{$link}|thumb|center|]]<br/>{$caption}\n" ;
|
|---|
| 939 | $cnt++ ;
|
|---|
| 940 | }
|
|---|
| 941 | $text .= "|}\n" ;
|
|---|
| 942 | }
|
|---|
| 943 |
|
|---|
| 944 | function strip_single_paragraph ( $s )
|
|---|
| 945 | {
|
|---|
| 946 | if ( substr_count ( $s , "paragraph>" ) == 2 &&
|
|---|
| 947 | substr ( $s , 0 , 11 ) == "<paragraph>" &&
|
|---|
| 948 | substr ( $s , -12 ) == "</paragraph>" )
|
|---|
| 949 | $s = substr ( $s , 11 , -12 ) ;
|
|---|
| 950 | return $s ;
|
|---|
| 951 | }
|
|---|
| 952 |
|
|---|
| 953 | # This function checks for and parses a HTML tag
|
|---|
| 954 | # Only to be called from p_html, as it returns only a partial extension tag!
|
|---|
| 955 | function p_html_tag ( &$a , &$xml , &$tag , &$closing , &$selfclosing )
|
|---|
| 956 | {
|
|---|
| 957 | if ( $this->w[$a] != '<' ) return false ;
|
|---|
| 958 | $b = $a + 1 ;
|
|---|
| 959 | $this->skipblanks ( $b ) ;
|
|---|
| 960 | $tag = "" ;
|
|---|
| 961 | $attrs = array () ;
|
|---|
| 962 | if ( !$this->scanplaintext ( $b , $tag , array ( " " , ">" ) , array ( "\n" ) ) ) return false ;
|
|---|
| 963 |
|
|---|
| 964 | $this->skipblanks ( $b ) ;
|
|---|
| 965 | if ( $b >= $this->wl ) return false ;
|
|---|
| 966 |
|
|---|
| 967 | $tag = trim ( strtolower ( $tag ) ) ;
|
|---|
| 968 | $closing = false ;
|
|---|
| 969 | $selfclosing = false ;
|
|---|
| 970 |
|
|---|
| 971 | # Is closing tag?
|
|---|
| 972 | if ( substr ( $tag , 0 , 1 ) == "/" )
|
|---|
| 973 | {
|
|---|
| 974 | $tag = substr ( $tag , 1 ) ;
|
|---|
| 975 | $closing = true ;
|
|---|
| 976 | $this->skipblanks ( $b ) ;
|
|---|
| 977 | if ( $b >= $this->wl ) return false ;
|
|---|
| 978 | }
|
|---|
| 979 |
|
|---|
| 980 | if ( substr ( $tag , -1 ) == "/" )
|
|---|
| 981 | {
|
|---|
| 982 | $tag = substr ( $tag , 0 , -1 ) ;
|
|---|
| 983 | $selfclosing = true ;
|
|---|
| 984 | }
|
|---|
| 985 |
|
|---|
| 986 | # Parsing attributes
|
|---|
| 987 | $ob = $b ;
|
|---|
| 988 | $q = "" ;
|
|---|
| 989 | while ( $b < $this->wl && ( $q != "" || ( $this->w[$b] != '>' && $this->w[$b] != '/' ) ) ) {
|
|---|
| 990 | if ( $this->w[$b] == '"' || $this->w[$b] == "'" ) {
|
|---|
| 991 | if ( $q == "" ) $q = $this->w[$b] ;
|
|---|
| 992 | else if ( $this->w[$b] == $q ) $q = "" ;
|
|---|
| 993 | }
|
|---|
| 994 | $b++ ;
|
|---|
| 995 | }
|
|---|
| 996 | if ( $b >= $this->wl ) return false ;
|
|---|
| 997 | $attrs = $this->preparse_attributes ( substr ( $this->w , $ob , $b - $ob + 1 ) ) ;
|
|---|
| 998 |
|
|---|
| 999 | # Is self closing?
|
|---|
| 1000 | if ( $tag == 'br' ) $selfclosing = true ; # Always regard <br> as <br/>
|
|---|
| 1001 | if ( $this->w[$b] == '/' )
|
|---|
| 1002 | {
|
|---|
| 1003 | $b++ ;
|
|---|
| 1004 | $this->skipblanks ( $b ) ;
|
|---|
| 1005 | $selfclosing = true ;
|
|---|
| 1006 | }
|
|---|
| 1007 |
|
|---|
| 1008 | $this->skipblanks ( $b ) ;
|
|---|
| 1009 | if ( $b >= $this->wl ) return false ;
|
|---|
| 1010 | if ( $this->w[$b] != '>' ) return false ;
|
|---|
| 1011 |
|
|---|
| 1012 | $a = $b + 1 ;
|
|---|
| 1013 | if ( count ( $attrs ) > 0 )
|
|---|
| 1014 | {
|
|---|
| 1015 | $xml = " " . implode ( " " , $attrs ) ;
|
|---|
| 1016 | }
|
|---|
| 1017 | return true ;
|
|---|
| 1018 | }
|
|---|
| 1019 |
|
|---|
| 1020 | # This function replaces templates and separates HTML attributes.
|
|---|
| 1021 | # It is used for both HTML tags and wiki tables
|
|---|
| 1022 | function preparse_attributes ( $x )
|
|---|
| 1023 | {
|
|---|
| 1024 | # Creating a temporary new parser to run the attribute list in
|
|---|
| 1025 | $np = new wiki2xml ;
|
|---|
| 1026 | $np->inherit ( $this ) ;
|
|---|
| 1027 | $np->w = $x ;
|
|---|
| 1028 | $np->wl = strlen ( $x ) ;
|
|---|
| 1029 |
|
|---|
| 1030 | # Replacing templates, and '<' and '>' in parameters
|
|---|
| 1031 | $c = 0 ;
|
|---|
| 1032 | $q = "" ;
|
|---|
| 1033 | while ( $q != "" || ( $c < $np->wl && $np->w[$c] != '>' && $np->w[$c] != '/' ) )
|
|---|
| 1034 | {
|
|---|
| 1035 | $y = $np->w[$c] ;
|
|---|
| 1036 | if ( $np->nextis ( $c , "{{" , false ) ) {
|
|---|
| 1037 | $xx = "" ;
|
|---|
| 1038 | if ( $np->p_template ( $c , $xx ) ) continue ;
|
|---|
| 1039 | else $c++ ;
|
|---|
| 1040 | } else if ( $y == "'" || $y == '"' ) {
|
|---|
| 1041 | if ( $q == "" ) $q = $y ;
|
|---|
| 1042 | else if ( $y == $q ) $q = "" ;
|
|---|
| 1043 | $c++ ;
|
|---|
| 1044 | } else if ( $q != "" && ( $y == '<' || $y == '>' ) ) {
|
|---|
| 1045 | $y = htmlentities ( $y ) ;
|
|---|
| 1046 | $np->w = substr ( $np->w , 0 , $c ) . $y . substr ( $np->w , $c + 1 ) ;
|
|---|
| 1047 | $np->wl += strlen ( $y ) - 1 ;
|
|---|
| 1048 | } else $c++ ;
|
|---|
| 1049 | if ( $c >= $np->wl ) return array () ;
|
|---|
| 1050 | }
|
|---|
| 1051 |
|
|---|
| 1052 | $attrs = array () ;
|
|---|
| 1053 | $c = 0 ;
|
|---|
| 1054 |
|
|---|
| 1055 | # Seeking attributes
|
|---|
| 1056 | while ( $np->w[$c] != '>' && $np->w[$c] != '/' )
|
|---|
| 1057 | {
|
|---|
| 1058 | $attr = "" ;
|
|---|
| 1059 | if ( !$np->p_html_attr ( $c , $attr ) ) break ;
|
|---|
| 1060 | if ( $attr != "" ) {
|
|---|
| 1061 | $key = array_shift ( explode ( "=" , $attr , 2 ) ) ;
|
|---|
| 1062 | if ( !isset ( $attrs[$key] ) && substr ( $attr , -3 , 3 ) != '=""' ) {
|
|---|
| 1063 | $attrs[$key] = $attr ;
|
|---|
| 1064 | }
|
|---|
| 1065 | }
|
|---|
| 1066 | $np->skipblanks ( $c ) ;
|
|---|
| 1067 | if ( $c >= $np->wl ) return array () ;
|
|---|
| 1068 | }
|
|---|
| 1069 | if ( substr ( $np->w , $c ) != ">" AND substr ( $np->w , $c ) != "/" ) return array() ;
|
|---|
| 1070 |
|
|---|
| 1071 | return $attrs ;
|
|---|
| 1072 | }
|
|---|
| 1073 |
|
|---|
| 1074 |
|
|---|
| 1075 | # This function scans a single HTML tag attribute and returns it as <attr name='key'>value</attr>
|
|---|
| 1076 | function p_html_attr ( &$a , &$xml )
|
|---|
| 1077 | {
|
|---|
| 1078 | $b = $a ;
|
|---|
| 1079 | $this->skipblanks ( $b ) ;
|
|---|
| 1080 | if ( $b >= $this->wl ) return false ;
|
|---|
| 1081 | $name = "" ;
|
|---|
| 1082 | if ( !$this->scanplaintext ( $b , $name , array ( " " , "=" , ">" , "/" ) , array ( "\n" ) ) ) return false ;
|
|---|
| 1083 |
|
|---|
| 1084 | $this->skipblanks ( $b ) ;
|
|---|
| 1085 | if ( $b >= $this->wl ) return false ;
|
|---|
| 1086 | $name = trim ( strtolower ( $name ) ) ;
|
|---|
| 1087 |
|
|---|
| 1088 | # Trying to catch illegal names; should be replaced with regexp
|
|---|
| 1089 | $n2 = "" ;
|
|---|
| 1090 | for ( $q = 0 ; $q < strlen ( $name ) ; $q++ ) {
|
|---|
| 1091 | if ( $name[$q] == '_' OR ( $name[$q] >= 'a' AND $name[$q] <= 'z' ) )
|
|---|
| 1092 | $n2 .= $name[$q] ;
|
|---|
| 1093 | }
|
|---|
| 1094 | $name = trim ( $n2 ) ;
|
|---|
| 1095 | if ( $name == 'extension_name' ) return false ; # Not allowed, because used internally
|
|---|
| 1096 | if ( $name == '' ) return false ;
|
|---|
| 1097 |
|
|---|
| 1098 | # Determining value
|
|---|
| 1099 | $value = "" ;
|
|---|
| 1100 | if ( $this->w[$b] == "=" )
|
|---|
| 1101 | {
|
|---|
| 1102 | $b++ ;
|
|---|
| 1103 | $this->skipblanks ( $b ) ;
|
|---|
| 1104 | if ( $b >= $this->wl ) return false ;
|
|---|
| 1105 | $q = "" ;
|
|---|
| 1106 | $is_q = false ;
|
|---|
| 1107 | if ( $this->w[$b] == '"' || $this->w[$b] == "'" )
|
|---|
| 1108 | {
|
|---|
| 1109 | $q = $this->w[$b] ;
|
|---|
| 1110 | $b++ ;
|
|---|
| 1111 | if ( $b >= $this->wl ) return false ;
|
|---|
| 1112 | $is_q = true ;
|
|---|
| 1113 | }
|
|---|
| 1114 | while ( $b < $this->wl )
|
|---|
| 1115 | {
|
|---|
| 1116 | $c = $this->w[$b] ;
|
|---|
| 1117 | if ( $c == $q )
|
|---|
| 1118 | {
|
|---|
| 1119 | $b++ ;
|
|---|
| 1120 | if ( $is_q ) break ;
|
|---|
| 1121 | return false ; # Broken attribute value
|
|---|
| 1122 | }
|
|---|
| 1123 | if ( $this->nextis ( $b , "\\{$q}" ) ) # Ignore escaped quotes
|
|---|
| 1124 | {
|
|---|
| 1125 | $value .= "\\{$q}" ;
|
|---|
| 1126 | continue ;
|
|---|
| 1127 | }
|
|---|
| 1128 | if ( $c == "\n" ) return false ; # Line break before value end
|
|---|
| 1129 | if ( !$is_q && ( $c == ' ' || $c == '>' || $c == '/' ) ) break ;
|
|---|
| 1130 | $value .= htmlspecialchars ( $c ) ;
|
|---|
| 1131 | $b++ ;
|
|---|
| 1132 | }
|
|---|
| 1133 | }
|
|---|
| 1134 | if ( $name == "" ) return true ;
|
|---|
| 1135 |
|
|---|
| 1136 | $a = $b ;
|
|---|
| 1137 | if ( $q == "'" ) $q = "'" ;
|
|---|
| 1138 | else $q = '"' ;
|
|---|
| 1139 | $xml = "{$name}={$q}{$value}{$q}" ;
|
|---|
| 1140 | #$xml .= "<attr name='{$name}'>{$value}</attr>" ;
|
|---|
| 1141 | return true ;
|
|---|
| 1142 | }
|
|---|
| 1143 |
|
|---|
| 1144 | # Horizontal ruler (<hr> / ----)
|
|---|
| 1145 | function p_hr ( &$a , &$xml )
|
|---|
| 1146 | {
|
|---|
| 1147 | if ( !$this->nextis ( $a , "----" ) ) return false ;
|
|---|
| 1148 | $this->skipblanks ( $a , "-" ) ;
|
|---|
| 1149 | $this->skipblanks ( $a ) ;
|
|---|
| 1150 | $xml .= "<hr/>" ;
|
|---|
| 1151 | return true ;
|
|---|
| 1152 | }
|
|---|
| 1153 |
|
|---|
| 1154 | # TABLE
|
|---|
| 1155 | # Scans the rest of the line as HTML attributes and returns the usual <attrs><attr> string
|
|---|
| 1156 | function scanattributes ( &$a )
|
|---|
| 1157 | {
|
|---|
| 1158 | $x = "" ;
|
|---|
| 1159 | while ( $a < $this->wl )
|
|---|
| 1160 | {
|
|---|
| 1161 | if ( $this->w[$a] == "\n" ) break ;
|
|---|
| 1162 | $x .= $this->w[$a] ;
|
|---|
| 1163 | $a++ ;
|
|---|
| 1164 | }
|
|---|
| 1165 | $x .= ">" ;
|
|---|
| 1166 |
|
|---|
| 1167 | $attrs = $this->preparse_attributes ( $x ) ;
|
|---|
| 1168 |
|
|---|
| 1169 | $ret = "" ;
|
|---|
| 1170 | if ( count ( $attrs ) > 0 )
|
|---|
| 1171 | {
|
|---|
| 1172 | #$ret .= "<attrs>" ;
|
|---|
| 1173 | $ret .= " " . implode ( " " , $attrs ) ;
|
|---|
| 1174 | #$ret .= "</attrs>" ;
|
|---|
| 1175 | }
|
|---|
| 1176 | return $ret ;
|
|---|
| 1177 | }
|
|---|
| 1178 |
|
|---|
| 1179 | # Finds the first of the given items; does *not* alter $a
|
|---|
| 1180 | function scanahead ( $a , $matches )
|
|---|
| 1181 | {
|
|---|
| 1182 | while ( $a < $this->wl )
|
|---|
| 1183 | {
|
|---|
| 1184 | foreach ( $matches AS $x )
|
|---|
| 1185 | {
|
|---|
| 1186 | if ( $this->nextis ( $a , $x , false ) )
|
|---|
| 1187 | {
|
|---|
| 1188 | return $a ;
|
|---|
| 1189 | }
|
|---|
| 1190 | }
|
|---|
| 1191 | $a++ ;
|
|---|
| 1192 | }
|
|---|
| 1193 | return -1 ; # Not found
|
|---|
| 1194 | }
|
|---|
| 1195 |
|
|---|
| 1196 |
|
|---|
| 1197 | # The main table parsing function
|
|---|
| 1198 | function p_table ( &$a , &$xml )
|
|---|
| 1199 | {
|
|---|
| 1200 | if ( $a >= $this->wl ) return false ;
|
|---|
| 1201 | $c = $this->w[$a] ;
|
|---|
| 1202 | if ( $c == "{" && $this->nextis ( $a , "{|" , false ) )
|
|---|
| 1203 | return $this->p_table_open ( $a , $xml ) ;
|
|---|
| 1204 |
|
|---|
| 1205 | # print "p_table for " . htmlentities ( substr ( $this->w , $a ) ) . "<br/><br/>" ; flush () ;
|
|---|
| 1206 |
|
|---|
| 1207 | if ( count ( $this->tables ) == 0 ) return false ; # No tables open, nothing to do
|
|---|
| 1208 |
|
|---|
| 1209 | # Compatability for table cell lines starting with blanks; *evil MediaWiki parser!*
|
|---|
| 1210 | $b = $a ;
|
|---|
| 1211 | $this->skipblanks ( $b ) ;
|
|---|
| 1212 | if ( $b >= $this->wl ) return false ;
|
|---|
| 1213 | $c = $this->w[$b] ;
|
|---|
| 1214 |
|
|---|
| 1215 | if ( $c != "|" && $c != "!" ) return false ; # No possible table markup
|
|---|
| 1216 |
|
|---|
| 1217 | if ( $c == "|" && $this->nextis ( $b , "|}" , false ) ) return $this->p_table_close ( $b , $xml ) ;
|
|---|
| 1218 |
|
|---|
| 1219 | #if ( $this->nextis ( $a , "|" , false ) || $this->nextis ( $a , "!" , false ) )
|
|---|
| 1220 | return $this->p_table_element ( $b , $xml , true ) ;
|
|---|
| 1221 | }
|
|---|
| 1222 |
|
|---|
| 1223 | function lasttable ()
|
|---|
| 1224 | {
|
|---|
| 1225 | return $this->tables[count($this->tables)-1] ;
|
|---|
| 1226 | }
|
|---|
| 1227 |
|
|---|
| 1228 | # Returns the attributes for table cells
|
|---|
| 1229 | function tryfindparams ( &$a )
|
|---|
| 1230 | {
|
|---|
| 1231 | $n = strspn ( $this->w , $this->allowed , $a ) ; # PHP 4.3.0 and above
|
|---|
| 1232 | # $n = strspn ( substr ( $this->w , $a ) , $this->allowed ) ; # PHP < 4.3.0
|
|---|
| 1233 | if ( $n == 0 ) return "" ; # None found
|
|---|
| 1234 |
|
|---|
| 1235 | $b = $a + $n ;
|
|---|
| 1236 | if ( $b >= $this->wl ) return "" ;
|
|---|
| 1237 | if ( $this->w[$b] != "|" && $this->w[$b] != "!" ) return "" ;
|
|---|
| 1238 | if ( $this->nextis ( $b , "||" , false ) ) return "" ; # Reached a ||, so return blank string
|
|---|
| 1239 | if ( $this->nextis ( $b , "!!" , false ) ) return "" ; # Reached a ||, so return blank string
|
|---|
| 1240 | $this->w[$b] = "\n" ;
|
|---|
| 1241 | $ret = $this->scanattributes ( $a ) ;
|
|---|
| 1242 | $this->w[$b] = "|" ;
|
|---|
| 1243 | $a = $b + 1 ;
|
|---|
| 1244 | return $ret ;
|
|---|
| 1245 | }
|
|---|
| 1246 |
|
|---|
| 1247 | function p_table_element ( &$a , &$xml , $newline = false )
|
|---|
| 1248 | {
|
|---|
| 1249 | # print "p_table_element for " . htmlentities ( substr ( $this->w , $a ) ) . "<br/><br/>" ; flush () ;
|
|---|
| 1250 | $b = $a ;
|
|---|
| 1251 | $this->skipblanks ( $b ) ; # Compatability for table cells starting with blanks; *evil MediaWiki parser!*
|
|---|
| 1252 | if ( $b >= $this->wl ) return false ; # End of the game
|
|---|
| 1253 | $x = "" ;
|
|---|
| 1254 | if ( $newline && $this->nextis ( $b , "|-" ) ) # Table row
|
|---|
| 1255 | {
|
|---|
| 1256 | $this->skipblanks ( $b , "-" ) ;
|
|---|
| 1257 | $this->skipblanks ( $b ) ;
|
|---|
| 1258 |
|
|---|
| 1259 | $attrs = $this->scanattributes ( $b ) ;
|
|---|
| 1260 | if ( $this->tables[count($this->tables)-1]->is_row_open ) $x .= "</tablerow>" ;
|
|---|
| 1261 | else $this->tables[count($this->tables)-1]->is_row_open = true ;
|
|---|
| 1262 | $this->tables[count($this->tables)-1]->had_row = true ;
|
|---|
| 1263 | $x .= "<tablerow{$attrs}>" ;
|
|---|
| 1264 | $y = "" ;
|
|---|
| 1265 | $this->p_restofcell ( $b , $y ) ;
|
|---|
| 1266 | }
|
|---|
| 1267 | else if ( $newline && $this->nextis ( $b , "|+" ) ) # Table caption
|
|---|
| 1268 | {
|
|---|
| 1269 | $this->skipblanks ( $b ) ;
|
|---|
| 1270 | $attrs = $this->tryfindparams ( $b ) ;
|
|---|
| 1271 | $this->skipblanks ( $b ) ;
|
|---|
| 1272 | if ( $this->tables[count($this->tables)-1]->is_row_open ) $x .= "</tablerow>" ;
|
|---|
| 1273 | $this->tables[count($this->tables)-1]->is_row_open = false ;
|
|---|
| 1274 |
|
|---|
| 1275 | $y = "" ;
|
|---|
| 1276 | if ( !$this->p_restofcell ( $b , $y ) ) return false ;
|
|---|
| 1277 | $x .= "<tablecaption{$attrs}>{$y}</tablecaption>" ;
|
|---|
| 1278 | }
|
|---|
| 1279 | else # TD or TH
|
|---|
| 1280 | {
|
|---|
| 1281 | $c = $this->w[$b] ;
|
|---|
| 1282 | $b++ ;
|
|---|
| 1283 | $tag = "error" ;
|
|---|
| 1284 | if ( $c == '|' ) $tag = "tablecell" ;
|
|---|
| 1285 | else if ( $c == '!' ) $tag = "tablehead" ;
|
|---|
| 1286 | $attrs = $this->tryfindparams ( $b ) ;
|
|---|
| 1287 | $this->skipblanks ( $b ) ;
|
|---|
| 1288 | if ( !$this->p_restofcell ( $b , $x ) ) return false ;
|
|---|
| 1289 |
|
|---|
| 1290 | if ( substr ( $x , 0 , 1 ) == "|" ) # Crude fix to compensate for MediaWiki "tolerant" parsing
|
|---|
| 1291 | $x = substr ( $x , 1 ) ;
|
|---|
| 1292 | $x = "<{$tag}{$attrs}>{$x}</{$tag}>" ;
|
|---|
| 1293 | $this->tables[count($this->tables)-1]->had_cell = true ;
|
|---|
| 1294 | if ( !$this->tables[count($this->tables)-1]->is_row_open )
|
|---|
| 1295 | {
|
|---|
| 1296 | $this->tables[count($this->tables)-1]->is_row_open = true ;
|
|---|
| 1297 | $this->tables[count($this->tables)-1]->had_row = true ;
|
|---|
| 1298 | $x = "<tablerow>{$x}" ;
|
|---|
| 1299 | }
|
|---|
| 1300 | }
|
|---|
| 1301 |
|
|---|
| 1302 | $a = $b ;
|
|---|
| 1303 | $xml .= $x ;
|
|---|
| 1304 | return true ;
|
|---|
| 1305 | }
|
|---|
| 1306 |
|
|---|
| 1307 | # Finds the substring that composes the table cell,
|
|---|
| 1308 | # then runs a new parser on it
|
|---|
| 1309 | function p_restofcell ( &$a , &$xml )
|
|---|
| 1310 | {
|
|---|
| 1311 | # Get substring for cell
|
|---|
| 1312 | $b = $a ;
|
|---|
| 1313 | $sameline = true ;
|
|---|
| 1314 | $x = "" ;
|
|---|
| 1315 | $itables = 0 ;
|
|---|
| 1316 | while ( $b < $this->wl )
|
|---|
| 1317 | {
|
|---|
| 1318 | $c = $this->w[$b] ;
|
|---|
| 1319 | if ( $c == "<" && $this->once ( $b , $x , "html" ) ) continue ; # Up front to catch pre and nowiki
|
|---|
| 1320 | if ( $c == "\n" ) { $sameline = false ; }
|
|---|
| 1321 | if ( $c == "\n" && $this->nextis ( $b , "\n{|" ) ) { $itables++ ; continue ; }
|
|---|
| 1322 | if ( $c == "\n" && $itables > 0 && $this->nextis ( $b , "\n|}" ) ) { $itables-- ; continue ; }
|
|---|
| 1323 |
|
|---|
| 1324 | if ( ( $c == "\n" && $this->nextis ( $b , "\n|" , false ) ) OR
|
|---|
| 1325 | ( $c == "\n" && $this->nextis ( $b , "\n!" , false ) ) OR
|
|---|
| 1326 | ( $c == "\n" && $this->nextis ( $b , "\n |" , false ) ) OR # MediaWiki parser madness compensator
|
|---|
| 1327 | ( $c == "\n" && $this->nextis ( $b , "\n !" , false ) ) OR # MediaWiki parser madness compensator
|
|---|
| 1328 | ( $c == "|" && $sameline && $this->nextis ( $b , "||" , false ) ) OR
|
|---|
| 1329 | ( $c == "!" && $sameline && $this->nextis ( $b , "!!" , false ) ) )
|
|---|
| 1330 | {
|
|---|
| 1331 | if ( $itables == 0 ) break ;
|
|---|
| 1332 | $b += 2 ;
|
|---|
| 1333 | }
|
|---|
| 1334 |
|
|---|
| 1335 | if ( $c == "[" && $this->once ( $b , $x , "internal_link" ) ) continue ;
|
|---|
| 1336 | if ( $c == "{" && $this->once ( $b , $x , "template_variable" ) ) continue ;
|
|---|
| 1337 | if ( $c == "{" && $this->once ( $b , $x , "template" ) ) continue ;
|
|---|
| 1338 | $b++ ;
|
|---|
| 1339 | }
|
|---|
| 1340 |
|
|---|
| 1341 | # if ( $itables > 0 ) return false ;
|
|---|
| 1342 |
|
|---|
| 1343 | # Parse cell substring
|
|---|
| 1344 | $s = substr ( $this->w , $a , $b - $a ) ;
|
|---|
| 1345 | $p = new wiki2xml ;
|
|---|
| 1346 | $p->inherit ( $this ) ;
|
|---|
| 1347 | $x = $p->parse ( $s ) ;
|
|---|
| 1348 | if ( $x == $this->errormessage ) return false ;
|
|---|
| 1349 |
|
|---|
| 1350 | $a = $b + 1 ;
|
|---|
| 1351 | $xml .= $this->strip_single_paragraph ( $x ) ;
|
|---|
| 1352 | return true ;
|
|---|
| 1353 | }
|
|---|
| 1354 |
|
|---|
| 1355 | function p_table_close ( &$a , &$xml )
|
|---|
| 1356 | {
|
|---|
| 1357 | if ( count ( $this->tables ) == 0 ) return false ;
|
|---|
| 1358 | $b = $a ;
|
|---|
| 1359 | if ( !$this->nextis ( $b , "|}" ) ) return false ;
|
|---|
| 1360 | if ( !$this->tables[count($this->tables)-1]->had_row ) return false ; # Table but no row was used
|
|---|
| 1361 | if ( !$this->tables[count($this->tables)-1]->had_cell ) return false ; # Table but no cell was used
|
|---|
| 1362 | $x = "" ;
|
|---|
| 1363 | if ( $this->tables[count($this->tables)-1]->is_row_open ) $x .= "</tablerow>" ;
|
|---|
| 1364 | unset ( $this->tables[count($this->tables)-1] ) ;
|
|---|
| 1365 | $x .= "</table>" ;
|
|---|
| 1366 | $xml .= $x ;
|
|---|
| 1367 | $a = $b ;
|
|---|
| 1368 | while ( $this->nextis ( $a , "\n" ) ) ;
|
|---|
| 1369 | return true ;
|
|---|
| 1370 | }
|
|---|
| 1371 |
|
|---|
| 1372 | function p_table_open ( &$a , &$xml )
|
|---|
| 1373 | {
|
|---|
| 1374 | $b = $a ;
|
|---|
| 1375 | if ( !$this->nextis ( $b , "{|" ) ) return false ;
|
|---|
| 1376 |
|
|---|
| 1377 | $this->is_row_open = false ;
|
|---|
| 1378 |
|
|---|
| 1379 | # Add table to stack
|
|---|
| 1380 | $nt->is_row_open = false ;
|
|---|
| 1381 | $nt->had_row = false ;
|
|---|
| 1382 | $nt->had_cell = false ;
|
|---|
| 1383 | $this->tables[count($this->tables)] = $nt ;
|
|---|
| 1384 |
|
|---|
| 1385 | $x = "<table" ;
|
|---|
| 1386 | $x .= $this->scanattributes ( $b ) . ">" ;
|
|---|
| 1387 | while ( $this->nextis ( $b , "\n" ) ) ;
|
|---|
| 1388 |
|
|---|
| 1389 | while ( !$this->p_table_close ( $b , $x ) )
|
|---|
| 1390 | {
|
|---|
| 1391 | if ( $b >= $this->wl )
|
|---|
| 1392 | {
|
|---|
| 1393 | unset ( $this->tables[count($this->tables)-1] ) ;
|
|---|
| 1394 | return false ;
|
|---|
| 1395 | }
|
|---|
| 1396 | if ( $this->p_table_open ( $b , $x ) ) continue ;
|
|---|
| 1397 | if ( !$this->p_table_element ( $b , $x , true ) ) # No |} and no table element
|
|---|
| 1398 | {
|
|---|
| 1399 | unset ( $this->tables[count($this->tables)-1] ) ;
|
|---|
| 1400 | return false ;
|
|---|
| 1401 | }
|
|---|
| 1402 | }
|
|---|
| 1403 | $a = $b ;
|
|---|
| 1404 | $xml .= $x ;
|
|---|
| 1405 | return true ;
|
|---|
| 1406 | }
|
|---|
| 1407 |
|
|---|
| 1408 | #-----------------------------------
|
|---|
| 1409 | # Parse the article
|
|---|
| 1410 | function p_article ( &$a , &$xml )
|
|---|
| 1411 | {
|
|---|
| 1412 | $x = "" ;
|
|---|
| 1413 | $b = $a ;
|
|---|
| 1414 | while ( $b < $this->wl )
|
|---|
| 1415 | {
|
|---|
| 1416 | if ( $this->onceormore ( $b , $x , "heading" ) ) continue ;
|
|---|
| 1417 | if ( $this->onceormore ( $b , $x , "block_lines" ) ) continue ;
|
|---|
| 1418 | if ( $this->onceormore ( $b , $x , "block_pre" ) ) continue ;
|
|---|
| 1419 | if ( $this->onceormore ( $b , $x , "block_list" ) ) continue ;
|
|---|
| 1420 | if ( $this->onceormore ( $b , $x , "hr" ) ) continue ;
|
|---|
| 1421 | if ( $this->onceormore ( $b , $x , "table" ) ) continue ;
|
|---|
| 1422 | if ( $this->onceormore ( $b , $x , "blankline" ) ) continue ;
|
|---|
| 1423 | if ( $this->p_block_lines ( $b , $x , true ) ) continue ;
|
|---|
| 1424 | # The last resort! It should never come to this!
|
|---|
| 1425 | if ( !$this->compensate_markup_errors ) $xml .= "<error type='general' reason='no matching markup'/>" ;
|
|---|
| 1426 | $xml .= htmlspecialchars ( $this->w[$b] ) ;
|
|---|
| 1427 | $b++ ;
|
|---|
| 1428 | }
|
|---|
| 1429 | $a = $b ;
|
|---|
| 1430 | $xml .= $x ;
|
|---|
| 1431 |
|
|---|
| 1432 | # asort ( $this->profile ) ;
|
|---|
| 1433 | # $p = "" ;
|
|---|
| 1434 | # foreach ( $this->profile AS $k => $v ) $p .= "<p>{$k} : {$v}</p>" ;
|
|---|
| 1435 | # $xml = "<debug>{$this->cnt}{$p}</debug>" . $xml ;
|
|---|
| 1436 | return true ;
|
|---|
| 1437 | }
|
|---|
| 1438 |
|
|---|
| 1439 | # The only function to be called directly from outside the class
|
|---|
| 1440 | function parse ( &$wiki )
|
|---|
| 1441 | {
|
|---|
| 1442 | $this->w = rtrim ( $wiki ) ;
|
|---|
| 1443 |
|
|---|
| 1444 | # Fix line endings
|
|---|
| 1445 | $cc = count_chars ( $wiki , 0 ) ;
|
|---|
| 1446 | if ( $cc[10] > 0 && $cc[13] == 0 )
|
|---|
| 1447 | $this->w = str_replace ( "\r" , "\n" , $this->w ) ;
|
|---|
| 1448 | $this->w = str_replace ( "\r" , "" , $this->w ) ;
|
|---|
| 1449 |
|
|---|
| 1450 | # Remove HTML comments
|
|---|
| 1451 | # $this->w = str_replace ( "\n<!--" , "<!--" , $this->w ) ;
|
|---|
| 1452 | # Important: Do not remove leading \n, since it could be a heading delimiter
|
|---|
| 1453 | $this->w= preg_replace('/\n<!--(.|\s)*?-->\n/', "\n<!-- --> ", $this->w);
|
|---|
| 1454 | $this->w= preg_replace('/<!--(.|\s)*?-->/', '', $this->w);
|
|---|
| 1455 | $this->w= preg_replace('/<!--(.|\s)*$/', '', $this->w);
|
|---|
| 1456 |
|
|---|
| 1457 | # Run the thing!
|
|---|
| 1458 | # $this->tables = array () ;
|
|---|
| 1459 | $this->wl = strlen ( $this->w ) ;
|
|---|
| 1460 | $xml = "" ;
|
|---|
| 1461 | $a = 0 ;
|
|---|
| 1462 | if ( !$this->p_article ( $a , $xml ) ) return $this->errormessage ;
|
|---|
| 1463 |
|
|---|
| 1464 | # XML cleanup
|
|---|
| 1465 | $ol = -1 ;
|
|---|
| 1466 | while ( $ol != strlen ( $xml ) ) {
|
|---|
| 1467 | $ol = strlen ( $xml ) ;
|
|---|
| 1468 | $xml = str_replace ( "<preline> " , "<preline><space/>" , $xml ) ;
|
|---|
| 1469 | $xml = str_replace ( "<space/> " , "<space/><space/>" , $xml ) ;
|
|---|
| 1470 | }
|
|---|
| 1471 | $ol = -1 ;
|
|---|
| 1472 | while ( $ol != strlen ( $xml ) ) {
|
|---|
| 1473 | $ol = strlen ( $xml ) ;
|
|---|
| 1474 | $xml = str_replace ( " " , " " , $xml ) ;
|
|---|
| 1475 | }
|
|---|
| 1476 | $ol = -1 ;
|
|---|
| 1477 | while ( $this->use_space_tag && $ol != strlen ( $xml ) ) {
|
|---|
| 1478 | $ol = strlen ( $xml ) ;
|
|---|
| 1479 | $xml = str_replace ( "> " , "><space/>" , $xml ) ;
|
|---|
| 1480 | $xml = str_replace ( " <" , "<space/><" , $xml ) ;
|
|---|
| 1481 | }
|
|---|
| 1482 | $xml = str_replace ( '<tablerow></tablerow>' , '' , $xml ) ;
|
|---|
| 1483 |
|
|---|
| 1484 | return $xml ;
|
|---|
| 1485 | }
|
|---|
| 1486 |
|
|---|
| 1487 | }
|
|---|
| 1488 |
|
|---|
| 1489 | ?>
|
|---|
