3 * Original code by Erik John Resig (ejohn.org)
\r
4 * http://ejohn.org/blog/pure-javascript-html-parser/
\r
8 var X_HTMLParser_CHARS = {
\r
9 A:1,B:1,C:1,D:1,E:1,F:1,G:1,H:1,I:1,J:1,K:1,L:1,M:1,N:1,O:1,P:1,Q:1,R:1,S:1,T:1,U:1,V:1,W:1,X:1,Y:1,Z:1,
\r
10 a:2,b:2,c:2,d:2,e:2,f:2,g:2,h:2,i:2,j:2,k:2,l:2,m:2,n:2,o:2,p:2,q:2,r:2,s:2,t:2,u:2,v:2,w:2,x:2,y:2,z:2,
\r
11 // "0" : 4, "1" : 4, "2" : 4, "3" : 4, "4" : 4, "5" : 4, "6" : 4, "7" : 4, "8" : 4, "9" : 4, closure compiler で minify すると ie4 で error、eval使う
\r
13 '\t' : 16, '\r\n' : 16, '\r' : 16, '\n' : 16, '\f' : 16, '\b' : 16, ' ' : 16
\r
15 X_HTMLParser_alphabets = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz',
\r
16 X_HTMLParser_whiteSpace = '\t\r\n\f\b ',
\r
18 // Empty Elements - HTML 4.01
\r
19 X_HTMLParser_empty = X_Dom_DTD_EMPTY,
\r
21 // Block Elements - HTML 4.01
\r
22 X_HTMLParser_block = {ADDRESS:1,APPLET:1,BLOCKQUOTE:1,BUTTON:1,CENTER:1,DD:1,DEL:1,DIR:1,DIV:1,DL:1,DT:1,FIELDSET:1,FORM:1,FRAMESET:1,HR:1,IFRAME:1,INS:1,
\r
23 ISINDEX:1,LI:1,MAP:1,MENU:1,NOFRAMES:1,NOSCRIPT:1,OBJECT:1,OL:1,P:1,PRE:1,SCRIPT:1,TABLE:1,TBODY:1,TD:1,TFOOT:1,TH:1,THEAD:1,TR:1,UL:1 },
\r
24 // Inline Elements - HTML 4.01
\r
25 X_HTMLParser_inline = {A:1,ABBR:1,ACRONYM:1,APPLET:1,B:1,BASEFONT:1,BDO:1,BIG:1,BR:1,BUTTON:1,CITE:1,CODE:1,DEL:1,DFN:1,EM:1,FONT:1,I:1,IFRAME:1,IMG:1,
\r
26 INPUT:1,INS:1,KBD:1,LABEL:1,MAP:1,OBJECT:1,Q:1,S:1,SAMP:1,SCRIPT:1,SELECT:1,SMALL:1,SPAN:1,STRIKE:1,STRONG:1,SUB:1,SUP:1,TEXTAREA:1,TT:1,U:1,VAR:1},
\r
27 // Elements that you can, intentionally, leave open
\r
28 // (and which close themselves)
\r
29 X_HTMLParser_closeSelf = {OLGROUP:1,DD:1,DT:1,LI:1,OPTIONS:1,P:1,TBODY:1,TD:1,TFOOT:1,TH:1,THEAD:1,TR:1}, // add tbody
\r
31 X_HTMLParser_sisters = {
\r
36 COLGROUP : { CAPTION : 1 },
\r
37 THEAD : { CAPTION : 1, COLGROUP : 1 },
\r
38 TFOOT : { CAPTION : 1, COLGROUP : 1, THEAD : 1, TBODY : 1 },
\r
39 TBODY : { CAPTION : 1, COLGROUP : 1, THEAD : 1, TFOOT : 1 }
\r
42 * http://www.tohoho-web.com/html/tbody.htm
\r
43 * HTML4.01では、ヘッダとフッタを先読みして表示するために、<tbody> よりも <tfoot> の方を先に記述しなくてはならないと定義されています。
\r
44 * IE5.0 などでは HEAD → BODY → FOOT の順に表示するのですが、
\r
45 * <tfoot> に未対応の古いブラウザでは、HEAD → FOOT → BODY の順に表示されてしまいます。
\r
46 * また、HTML5 では、<tfoot> と <tbody> の順番はどちらでもよいことになりました。
\r
49 // Attributes that have their values filled in disabled="disabled"
\r
50 X_HTMLParser_fillAttrs = X_Node_Attr_noValue, //{checked:1,compact:1,declare:1,defer:1,disabled:1,ismap:1,multiple:1,nohref:1,noresize:1,noshade:1,nowrap:1,readonly:1,selected:1};
\r
52 // Special Elements (can contain anything)
\r
53 X_HTMLParser_special = { SCRIPT : 1, STYLE : 1, PLAINTEXT : 1, XMP : 1, TEXTAREA : 1 },
\r
55 X_HTMLParser_skipFixNesting = false;
\r
57 function X_HTMLParser_exec( html, handler, async ){
\r
58 var special = X_HTMLParser_special,
\r
59 //plainText = X_HTMLParser_plainText,
\r
60 startTime = async && X_Timer_now(),
\r
61 _parseStartTag = X_HTMLParser__parseStartTag,
\r
62 _parseEndTag = X_HTMLParser__parseEndTag,
\r
63 stack = async ? async[ 1 ] : [],
\r
65 chars, last, text, index;
\r
69 last = stack[ stack.length - 1 ];
\r
71 // Make sure we're not in a script or style element
\r
72 if ( last && special[ last ] === 1 ) {
\r
73 if( 0 <= ( index = html.toUpperCase().indexOf( '</' + last ) ) ){
\r
74 handler.chars( html.substring( 0, index ) );
\r
75 if( index = _parseEndTag( stack, handler, html ) ){
\r
76 html = html.substring( index );
\r
78 handler.chars( html );
\r
82 handler.chars( html );
\r
87 if ( html.indexOf("<!--") === 0 ) {
\r
88 if ( 0 < ( index = html.indexOf("-->") ) ) {
\r
89 handler.comment( html.substring( 4, index ) );
\r
90 html = html.substring( index + 3 );
\r
95 } else if ( html.indexOf("</") === 0 ) {
\r
96 if ( 2 < ( index = _parseEndTag( stack, handler, html ) ) ) {
\r
97 html = html.substring( index );
\r
102 } else if ( html.indexOf("<") === 0 ) {
\r
103 if( index = _parseStartTag( stack, last, handler, html ) ){
\r
104 html = html.substring( index );
\r
107 if( index === false ){
\r
113 index = html.indexOf("<");
\r
115 text = index < 0 ? html : html.substring( 0, index );
\r
116 html = index < 0 ? '' : html.substring( index );
\r
118 handler.chars( text );
\r
123 if( html === lastHtml ){
\r
124 handler.err( html );
\r
128 if( async && startTime + 15 <= X_Timer_now() && html ){
\r
129 handler.progress( 1 - html.length / async[ 0 ] );
\r
130 X.Timer.once( 0, X_HTMLParser_exec, [ html, handler, async ] );
\r
137 // Clean up any remaining tags
\r
138 X_HTMLParser_parseEndTag( stack, handler );
\r
140 async && handler.complete();
\r
143 function X_HTMLParser__parseStartTag( stack, last, handler, html ){
\r
144 var alphabets = X_HTMLParser_CHARS,
\r
145 whiteSpace = X_HTMLParser_CHARS,
\r
146 saveAttr = X_HTMLParser_saveAttr,
\r
147 uri = X_Dom_DTD_ATTR_VAL_IS_URI,
\r
152 tagName, empty = false,
\r
153 chr, start, attrName, quot, escape;
\r
155 while( i < l && phase < 9 ){
\r
156 chr = html.charAt( i );
\r
159 chr === '<' && ( ++phase );
\r
161 case 1 : // タグ名の開始を待つ
\r
162 ( alphabets[ chr ] & 3 ) && ( ++phase && ( start = i ) );
\r
164 case 2 : // タグ名の終わりの空白文字を待つ
\r
165 ( whiteSpace[ chr ] & 16 ) ?
\r
166 ( ++phase && ( tagName = html.substring( start, i ) ) ) :
\r
167 ( chr === '>' || ( empty = html.substr( i, 2 ) === '/>' ) ) &&
\r
168 ( ( tagName = html.substring( start, i ) ) && ( phase = 9 ) );
\r
170 case 3 : // 属性名の開始を待つ
\r
171 ( alphabets[ chr ] & 3 ) ?
\r
172 ( ++phase && ( start = i ) ) :
\r
173 ( chr === '>' || ( empty = html.substr( i, 2 ) === '/>' ) ) &&
\r
176 case 4 : // 属性名の終わりを待つ
\r
178 ( ( phase = 6 ) && ( attrName = html.substring( start, i ) ) ) :
\r
179 ( whiteSpace[ chr ] & 16 ) &&
\r
180 ( ( phase = 5 ) && ( attrName = html.substring( start, i ) ) );
\r
182 case 5 : // 属性の = または次の属性または htmlタグの閉じ
\r
183 ( whiteSpace[ chr ] & 16 ) ?// ie4 未対応の属性には cite = http:// となる
\r
185 ( alphabets[ chr ] & 3 ) ?
\r
186 ( ( phase = 4 ) && ( attrs[ attrs.length ] = attrName ) && ( start = i ) ) :
\r
189 ( chr === '>' || ( empty = html.substr( i, 2 ) === '/>' ) ) &&
\r
190 ( ( phase = 9 ) && ( attrs[ attrs.length ] = attrName ) );
\r
192 case 6 : // 属性値の開始 quot を待つ
\r
193 ( chr === '"' || chr === "'" ) ?
\r
194 ( ( phase = 7 ) && ( quot = chr ) && ( start = i + 1 ) ):
\r
195 !( whiteSpace[ chr ] & 16 ) &&
\r
196 ( ( phase = 8 ) && ( start = i ) ); // no quot
\r
198 case 7 : //属性値の閉じ quot を待つ
\r
199 !escape && ( chr === quot ) && ( phase = 3 ) && saveAttr( attrs, attrName, html.substring( start, i ) );
\r
201 case 8 : //閉じ quot のない属性の値
\r
202 ( whiteSpace[ chr ] & 16 ) ?
\r
203 ( ( phase = 3 ) && saveAttr( attrs, attrName, html.substring( start, i ) ) ) :
\r
205 ( ( phase = 9 ) && saveAttr( attrs, attrName, html.substring( start, i ) ) ) :
\r
206 !escape && !uri[ attrName ] && ( empty = html.substr( i, 2 ) === '/>' ) && // attr の val が uri で / で終わりかつ、未対応属性の場合
\r
210 escape = chr === '\\' && !escape; // \\\\ is not escape for "
\r
215 if( X_HTMLParser_parseStartTag( stack, last, handler, tagName.toUpperCase(), attrs, empty, i ) === false ) return false;
\r
221 function X_HTMLParser__parseEndTag( stack, handler, html ){
\r
222 var alphabets = X_HTMLParser_CHARS,
\r
223 whiteSpace = X_HTMLParser_CHARS,
\r
230 while( i < l && phase < 9 ){
\r
231 chr = html.charAt( i );
\r
234 html.substr( i, 2 ) === '</' && ( ++phase && ++i );
\r
236 case 1 : // タグ名の開始を待つ
\r
237 ( alphabets[ chr ] & 3 ) && ( ++phase && ( start = i ) );
\r
239 case 2 : // タグ名の終わりの空白文字を待つ
\r
240 ( whiteSpace[ chr ] & 16 ) && ( ++phase );
\r
241 ( chr === '>' ) && ( phase = 9 );
\r
242 ( phase !== 2 ) && ( tagName = html.substring( start, i ) );
\r
244 case 3 : // タグの終了を待つ
\r
245 chr === '>' && ( phase = 9 );
\r
251 X_HTMLParser_parseEndTag( stack, handler, tagName.toUpperCase() );
\r
257 function X_HTMLParser_saveAttr( attrs, name, value ){
\r
258 name = name.toLowerCase();
\r
259 value = X_HTMLParser_fillAttrs[ name ] === 1 ? name : value;
\r
260 attrs[ attrs.length ] = {
\r
264 value.indexOf( '"' ) !== -1 ?
\r
265 value.split( '"' ).join( '\\"' ).split( '\\\\"' ).join( '\\"' ) :
\r
270 function X_HTMLParser_parseStartTag( stack, last, handler, tagName, attrs, empty, index ) {
\r
271 var inline = X_HTMLParser_inline,
\r
272 parseEndTag = X_HTMLParser_parseEndTag,
\r
273 sisters = X_HTMLParser_sisters;
\r
274 if( !X_HTMLParser_skipFixNesting && X_HTMLParser_block[ tagName ] === 1 ){
\r
275 while( last && inline[ last ] === 1 ){
\r
276 parseEndTag( stack, handler, last );
\r
277 last = stack[ stack.length - 1 ];
\r
280 last && X_HTMLParser_closeSelf[ tagName ] === 1 && ( last === tagName || ( sisters[ tagName ] && sisters[ tagName ][ last ] === 1 ) ) && parseEndTag( stack, handler, last );
\r
281 empty = empty || X_HTMLParser_empty[ tagName ];
\r
282 !empty && ( stack[ stack.length ] = tagName );
\r
284 return handler.start( tagName, attrs, empty, index );
\r
287 function X_HTMLParser_parseEndTag( stack, handler, tagName ) {
\r
288 var pos = 0, i = stack.length;
\r
289 // If no tag name is provided, clean shop
\r
291 // Find the closest opened tag of the same type
\r
293 for ( pos = i; 0 <= pos; )
\r
294 if ( stack[ --pos ] === tagName )
\r
298 // Close all the open elements, up the stack
\r
300 handler.end( stack[ --i ] );
\r
302 // Remove the open elements from the stack
\r
303 stack.length = pos;
\r
307 var X_HTMLParser_htmlStringToXNode = {
\r
310 err : function( html ){
\r
311 X_HTMLParser_htmlStringToXNode.flat.length = 0;
\r
312 !X_HTMLParser_htmlStringToXNode.ignoreError && X.Logger.warn( 'X_Dom_Parser() error ' + html );
\r
314 start : function( tagName, attrs, noChild, length ){
\r
316 nest = X_HTMLParser_htmlStringToXNode.nest,
\r
317 flat = X_HTMLParser_htmlStringToXNode.flat,
\r
319 attr, name, i, _attrs; //, toIndex;
\r
321 xnode = nest[ l - 1 ].create( tagName );
\r
323 xnode = flat[ flat.length ] = X_Doc_create( tagName );
\r
325 if( !noChild ) nest[ l ] = xnode;
\r
326 if( i = attrs.length ){
\r
329 if( attr = attrs[ --i ] ){
\r
330 if( X.Type.isString( attr ) ){
\r
332 _attrs[ name ] = true;
\r
335 _attrs[ name ] = attr.escaped;
\r
339 xnode.attr( _attrs );
\r
343 0 < X_HTMLParser_htmlStringToXNode.nest.length && ( --X_HTMLParser_htmlStringToXNode.nest.length );
\r
345 chars : function( text ){
\r
346 if( X_HTMLParser_htmlStringToXNode.nest.length ){
\r
347 X_HTMLParser_htmlStringToXNode.nest[ X_HTMLParser_htmlStringToXNode.nest.length - 1 ].createText( text );
\r
349 X_HTMLParser_htmlStringToXNode.flat[ X_HTMLParser_htmlStringToXNode.flat.length ] = X_Doc_createText( text );
\r
352 comment : X.emptyFunction
\r
355 function X_HtmlParser_parse( html, ignoreError ){
\r
356 var worker = X_HTMLParser_htmlStringToXNode, ret;
\r
358 worker.nest.length = 0;
\r
359 worker.ignoreError = ignoreError;
\r
360 X_HTMLParser_exec( html, worker );
\r
362 delete worker.flat;
\r
366 var X_HTMLParser_asyncHtmlStringToXNode = {
\r
367 err : function( html ){
\r
368 X_HTMLParser_htmlStringToXNode.err( html );
\r
369 this.asyncDispatch( X.Event.ERROR );
\r
371 start : X_HTMLParser_htmlStringToXNode.start,
\r
372 end : X_HTMLParser_htmlStringToXNode.end,
\r
373 chars : X_HTMLParser_htmlStringToXNode.chars,
\r
374 comment : X.emptyFunction,
\r
376 progress : function( pct ){
\r
377 this.asyncDispatch( { type : X.Event.PROGRESS, percent : pct } );
\r
379 complete : function(){
\r
380 var ret = X_HTMLParser_htmlStringToXNode.flat;
\r
381 delete X_HTMLParser_htmlStringToXNode.flat;
\r
382 this.asyncDispatch( { type : X.Event.SUCCESS, xnodes : ret } );
\r
386 function X_HTMLParser_asyncParse( html, ignoreError ){
\r
387 var dispatcher = X_Class_override( new X.EventDispatcher(), X_HTMLParser_asyncHtmlStringToXNode ),
\r
388 worker = X_HTMLParser_htmlStringToXNode;
\r
389 dispatcher.listenOnce( X.Event.SUCCESS, dispatcher, dispatcher.kill );
\r
391 worker.nest.length = 0;
\r
392 worker.ignoreError = ignoreError;
\r
393 X_HTMLParser_exec( html, dispatcher, [ html.length, [] ] );
\r