src/redakcja/static/js/wiki/xslt.js

   1 /*
   2  *
   3  * XSLT STUFF
   4  *
   5  */
   6 function createXSLT(xsl) {
   7     var p = new XSLTProcessor();
   8     p.importStylesheet(xsl);
   9     return p;
  10 }
  11
  12 var xml2htmlStylesheet = null;
  13
  14 // Wykonuje block z załadowanymi arkuszami stylów
  15 function withStylesheets(code_block, onError)
  16 {
  17     if (!xml2htmlStylesheet) {
  18         $.blockUI({message: 'Ładowanie arkuszy stylów...'});
  19         $.ajax({
  20             url: '/wlxml/wl2html.xsl',
  21             dataType: 'xml',
  22             timeout: 10000,
  23             success: function(data) {
  24                 xml2htmlStylesheet = createXSLT(data);
  25                 $.unblockUI();
  26                 code_block();
  27             },
  28             error: onError
  29         })
  30     }
  31     else {
  32         code_block();
  33     }
  34 }
  35
  36
  37 function xml2html(options) {
  38     withStylesheets(function() {
  39         var xml = options.xml.replace(/\/(\s+)/g, '<br />$1');
  40         xml = xml.replace(/([^a-zA-Z0-9ąćęłńóśźżĄĆĘŁŃÓŚŹŻ\s<>«»\\*_!,:;?&%."'=#()\/-]+)/g, '<alien>$1</alien>');
  41         var parser = new DOMParser();
  42         var serializer = new XMLSerializer();
  43         var doc = parser.parseFromString(xml, 'text/xml');
  44         var error = $('parsererror', doc);
  45
  46         if (error.length == 0) {
  47             doc = xml2htmlStylesheet.transformToFragment(doc, document);
  48
  49         if(doc.firstChild === null) {
  50             options.error("Błąd w przetwarzaniu XML.");
  51                 return;
  52             }
  53
  54             error = $('parsererror', doc);
  55         }
  56
  57         if (error.length > 0 && options.error) {
  58             source = $('sourcetext', doc);
  59             source_text = source.text();
  60             source.text('');
  61             options.error(error.text(), source_text);
  62         } else {
  63             let galleryUrl = new URL(
  64                 options.base,
  65                 window.location.href
  66             );
  67             $("img", $(doc.childNodes)).each(function() {
  68                 $(this).attr(
  69                     'src',
  70                     new URL(
  71                         $(this).attr('src'),
  72                         galleryUrl
  73                     )
  74                 );
  75             })
  76
  77             options.success(doc.childNodes);
  78
  79             $.themes.withCanon(function(canonThemes) {
  80                 if (canonThemes != null) {
  81                     $('.theme-text-list').addClass('canon').each(function(){
  82                         var themes = $(this).html().split(',');
  83                         for (i in themes) {
  84                             themes[i] = $.trim(themes[i]);
  85                             if (canonThemes.indexOf(themes[i]) == -1)
  86                                 themes[i] = '<span x-pass-thru="true" class="noncanon">' + themes[i] + "</span>"
  87                         }
  88                         $(this).html(themes.join(', '));
  89                     });
  90                 }
  91             });
  92         }
  93     }, function() { options.error && options.error('Nie udało się załadować XSLT'); });
  94 }
  95
  96 /* USEFULL CONSTANTS */
  97 const ELEMENT_NODE                   = 1;
  98 const ATTRIBUTE_NODE                 = 2;
  99 const TEXT_NODE                      = 3;
 100 const CDATA_SECTION_NODE             = 4;
 101 const ENTITY_REFERENCE_NODE          = 5;
 102 const ENTITY_NODE                    = 6;
 103 const PROCESSING_INSTRUCTION_NODE    = 7;
 104 const COMMENT_NODE                   = 8;
 105 const DOCUMENT_NODE                  = 9;
 106 const DOCUMENT_TYPE_NODE             = 10;
 107 const DOCUMENT_FRAGMENT_NODE         = 11;
 108 const NOTATION_NODE                  = 12;
 109 const XATTR_RE = /^x-attr-name-(.*)$/;
 110 const XATTR_KNOWN_RE = /^x-a-([a-z]+)-(.*)$/;
 111
 112 const ELEM_START = 1;
 113 const ELEM_END = 2;
 114 const NS_END = 3;
 115
 116 const NAMESPACES = {
 117         // namespaces not listed here will be assigned random names
 118         "http://www.w3.org/1999/02/22-rdf-syntax-ns#": "rdf",
 119         "http://purl.org/dc/elements/1.1/": "dc",
 120         "http://www.w3.org/XML/1998/namespace": "xml"
 121 };
 122
 123 NS_PREFIXES = {
 124     'wl': ''
 125 };
 126 for (prefix in NAMESPACES) {
 127     NS_PREFIXES[NAMESPACES[prefix]] = prefix
 128 };
 129
 130 class HTMLSerializer {
 131     _prepare() {
 132         this.stack = [];
 133
 134         // XML namespace is implicit
 135         this.nsMap = {"http://www.w3.org/XML/1998/namespace": "xml"};
 136
 137         this.result = "";
 138         this.nsCounter = 1;
 139     }
 140
 141     _pushElement(element) {
 142         this.stack.push({
 143             "type": ELEM_START,
 144             "node": element
 145         });
 146     }
 147
 148     _pushChildren(element) {
 149         for(var i = element.childNodes.length-1; i >= 0; i--)
 150             this._pushElement(element.childNodes.item(i));
 151     }
 152
 153     _pushTagEnd(tagName) {
 154         this.stack.push({
 155             "type": ELEM_END,
 156             "tagName": tagName
 157         });
 158     }
 159
 160     _verseBefore(node) {
 161         /* true if previous element is a previous verse of a stanza */
 162         var parent = node.parentNode;
 163         if (!parent || !parent.hasAttribute('x-node') || parent.getAttribute('x-node') != 'strofa')
 164             return false;
 165
 166         var prev = node.previousSibling;
 167
 168         while((prev !== null) && (prev.nodeType != ELEMENT_NODE)) {
 169             prev = prev.previousSibling;
 170         }
 171
 172         return (prev !== null) && prev.hasAttribute('x-verse');
 173     }
 174
 175     _nodeIgnored(node) {
 176         return node.getAttribute('x-auto-node') == 'true';
 177     }
 178
 179     _ignoredWithWhitespace(node) {
 180         while (node.nodeType == ELEMENT_NODE && this._nodeIgnored(node) && node.childNodes.length > 0)
 181             node = node.childNodes[0];
 182         if (node.nodeType == TEXT_NODE)
 183             return node.nodeValue.match(/^\s/)
 184         else return false;
 185     }
 186
 187
 188     serialize(rootElement, stripOuter)
 189     {
 190         var self = this;
 191         self._prepare();
 192
 193         if(!stripOuter)
 194             self._pushElement(rootElement);
 195         else
 196             self._pushChildren(rootElement);
 197
 198         var text_buffer = '';
 199
 200         while(self.stack.length > 0) {
 201             var token = self.stack.pop();
 202
 203             if(token.type === ELEM_END) {
 204                 self.result += text_buffer;
 205                 text_buffer = '';
 206                 if (token.tagName != '')
 207                     self.result += "</" + token.tagName + ">";
 208                 continue;
 209             };
 210
 211             if(token.type === NS_END) {
 212                 self._unassignNamespace(token.namespace);
 213                 continue;
 214             }
 215
 216             switch(token.node.nodeType) {
 217             case ELEMENT_NODE:
 218                 if(token.node.hasAttribute('x-pass-thru')
 219                    || token.node.hasAttribute('data-pass-thru')) {
 220                     self._pushChildren(token.node);
 221                     break;
 222                 }
 223
 224                 if(!token.node.hasAttribute('x-node'))
 225                     break;
 226
 227                 var xnode = token.node.getAttribute('x-node');
 228
 229                 if(xnode === 'out-of-flow-text') {
 230                     self._pushChildren(token.node);
 231                     break;
 232                 }
 233
 234                 if(token.node.hasAttribute('x-verse') && self._verseBefore(token.node)) {
 235                     self.result += '/';
 236                     // add whitespace if there's none
 237                     if (!(text_buffer.match(/^\s/) || self._ignoredWithWhitespace(token.node)))
 238                         self.result += ' ';
 239                 }
 240
 241                 self.result += text_buffer;
 242                 text_buffer = '';
 243                 self._serializeElement(token.node);
 244                 break;
 245             case TEXT_NODE:
 246                 self.result += text_buffer;
 247                 text_buffer = token.node.nodeValue.replace(/&/g, '&amp;').replace(/</g, '&lt;');
 248                 break;
 249             case COMMENT_NODE:
 250                 self.result += text_buffer;
 251                 text_buffer = '';
 252                 self.result += '<!--' + token.node.nodeValue + '-->';
 253                 break;
 254             };
 255         };
 256         self.result += text_buffer;
 257
 258         return this.result;
 259     }
 260
 261     /*
 262      * TODO: this doesn't support prefix redefinitions
 263      */
 264     _unassignNamespace(nsData) {
 265         this.nsMap[nsData.uri] = undefined;
 266     }
 267
 268     _assignNamespace(uri) {
 269         if(uri === null) {
 270             // default namespace
 271             return ({"prefix": "", "uri": "", "fresh": false});
 272         }
 273
 274         if(this.nsMap[uri] === undefined) {
 275             // this prefix hasn't been defined yet in current context
 276             var prefix = NAMESPACES[uri];
 277
 278             if (prefix === undefined) { // not predefined
 279                 prefix = "ns" + this.nsCounter;
 280                 this.nsCounter += 1;
 281             }
 282
 283             this.nsMap[uri] = prefix;
 284             return ({
 285                 "prefix": prefix,
 286                 "uri": uri,
 287                 "fresh": true
 288             });
 289         }
 290
 291         return ({"prefix": this.nsMap[uri], "uri": uri, "fresh": false});
 292     }
 293
 294     _join(prefix, name) {
 295         if(!!prefix)
 296             return prefix + ":" + name;
 297         return name;
 298     }
 299
 300     _rjoin(prefix, name) {
 301         if(!!name)
 302             return prefix + ":" + name;
 303         return prefix;
 304     }
 305
 306     _serializeElement(node) {
 307         var self = this;
 308
 309         if (self._nodeIgnored(node)) {
 310             self._pushTagEnd('');
 311             self._pushChildren(node);
 312         }
 313         else {
 314             var ns = node.getAttribute('x-ns');
 315             var nsPrefix = null;
 316             var newNamespaces = [];
 317
 318             var nsData = self._assignNamespace(node.getAttribute('x-ns'));
 319
 320             if(nsData.fresh) {
 321                 newNamespaces.push(nsData);
 322                 self.stack.push({
 323                     "type": NS_END,
 324                     "namespace": nsData
 325                 });
 326             }
 327
 328             var tagName = self._join(nsData.prefix, node.getAttribute('x-node'));
 329
 330             /* retrieve attributes */
 331             var attributeIDs = [];
 332             var attributes = [];
 333             for (var i = 0; i < node.attributes.length; i++) {
 334                 var attr = node.attributes.item(i);
 335
 336                 m = attr.name.match(XATTR_KNOWN_RE);
 337                 if (m !== null) {
 338                     prefix = m[1];
 339                     let tag = m[2];
 340                     attributes.push([
 341                         NS_PREFIXES[prefix],
 342                         tag,
 343                         attr.value
 344                     ]);
 345                 } else {
 346                     // check if name starts with "x-attr-name"
 347                     var m = attr.name.match(XATTR_RE);
 348                     if (m !== null) {
 349                         attributeIDs.push(m[1]);
 350                     }
 351                 }
 352             }
 353
 354             /* print out */
 355
 356             self.result += '<' + tagName;
 357
 358             function writeAttr(ns, tag, value) {
 359                 if (ns) {
 360                     var nsData = self._assignNamespace(ns);
 361                     if(nsData.fresh) {
 362                         newNamespaces.push(nsData);
 363                         self.stack.push({
 364                             "type": NS_END,
 365                             "namespace": nsData
 366                         });
 367                     };
 368                     tag = self._join(nsData.prefix, tag);
 369                 }
 370
 371                 self.result += ' ' + tag;
 372                 self.result += '="' + value.replace(/&/g, '&amp;').replace(/"/g, '&quot;') + '"';
 373             }
 374
 375             $.each(attributes, function() {
 376                 writeAttr(
 377                     this[0], this[1], this[2]
 378                 );
 379             });
 380
 381             $.each(attributeIDs, function() {
 382                 writeAttr(
 383                     node.getAttribute('x-attr-ns-'+this),
 384                     node.getAttribute('x-attr-name-'+this),
 385                     node.getAttribute('x-attr-value-'+this)
 386                 );
 387             });
 388
 389             /* print new namespace declarations */
 390             $.each(newNamespaces, function() {
 391                 self.result += " " + self._rjoin("xmlns", this.prefix);
 392                 self.result += '="' + this.uri + '"';
 393             });
 394
 395             if (node.childNodes.length > 0) {
 396                 self.result += ">";
 397                 self._pushTagEnd(tagName);
 398                 self._pushChildren(node);
 399             }
 400             else {
 401                 self.result += "/>";
 402             };
 403         }
 404     }
 405 }
 406
 407 function html2text(params) {
 408     try {
 409         var s = new HTMLSerializer();
 410         params.success( s.serialize(params.element, params.stripOuter) );
 411     } catch(e) {
 412         params.error("Nie udało się zserializować tekstu:" + e)
 413     }
 414 }