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
 111 const ELEM_START = 1;
 112 const ELEM_END = 2;
 113 const NS_END = 3;
 114
 115 const NAMESPACES = {
 116         // namespaces not listed here will be assigned random names
 117         "http://www.w3.org/1999/02/22-rdf-syntax-ns#": "rdf",
 118         "http://purl.org/dc/elements/1.1/": "dc",
 119         "http://www.w3.org/XML/1998/namespace": "xml"
 120 };
 121
 122 function HTMLSerializer() {
 123         // empty constructor
 124 }
 125
 126
 127
 128 HTMLSerializer.prototype._prepare = function() {
 129         this.stack = [];
 130
 131         // XML namespace is implicit
 132         this.nsMap = {"http://www.w3.org/XML/1998/namespace": "xml"};
 133
 134         this.result = "";
 135         this.nsCounter = 1;
 136 }
 137
 138 HTMLSerializer.prototype._pushElement = function(element) {
 139         this.stack.push({
 140                 "type": ELEM_START,
 141                 "node": element
 142         });
 143 }
 144
 145 HTMLSerializer.prototype._pushChildren = function(element) {
 146         for(var i = element.childNodes.length-1; i >= 0; i--)
 147                 this._pushElement(element.childNodes.item(i));
 148 }
 149
 150 HTMLSerializer.prototype._pushTagEnd = function(tagName) {
 151         this.stack.push({
 152                 "type": ELEM_END,
 153                 "tagName": tagName
 154         });
 155 }
 156
 157 HTMLSerializer.prototype._verseBefore = function(node) {
 158     /* true if previous element is a previous verse of a stanza */
 159     var parent = node.parentNode;
 160     if (!parent || !parent.hasAttribute('x-node') || parent.getAttribute('x-node') != 'strofa')
 161         return false;
 162
 163         var prev = node.previousSibling;
 164
 165         while((prev !== null) && (prev.nodeType != ELEMENT_NODE)) {
 166                 prev = prev.previousSibling;
 167         }
 168
 169         return (prev !== null) && prev.hasAttribute('x-verse');
 170 }
 171
 172 HTMLSerializer.prototype._nodeIgnored = function(node) {
 173     return node.getAttribute('x-auto-node') == 'true';
 174 }
 175
 176 HTMLSerializer.prototype._ignoredWithWhitespace = function(node) {
 177     while (node.nodeType == ELEMENT_NODE && this._nodeIgnored(node) && node.childNodes.length > 0)
 178         node = node.childNodes[0];
 179     if (node.nodeType == TEXT_NODE)
 180         return node.nodeValue.match(/^\s/)
 181     else return false;
 182 }
 183
 184
 185 HTMLSerializer.prototype.serialize = function(rootElement, stripOuter)
 186 {
 187         var self = this;
 188         self._prepare();
 189
 190         if(!stripOuter)
 191                 self._pushElement(rootElement);
 192         else
 193                 self._pushChildren(rootElement);
 194
 195     var text_buffer = '';
 196
 197         while(self.stack.length > 0) {
 198                 var token = self.stack.pop();
 199
 200         if(token.type === ELEM_END) {
 201             self.result += text_buffer;
 202             text_buffer = '';
 203             if (token.tagName != '')
 204                 self.result += "</" + token.tagName + ">";
 205             continue;
 206         };
 207
 208                 if(token.type === NS_END) {
 209                         self._unassignNamespace(token.namespace);
 210                         continue;
 211                 }
 212
 213
 214                 switch(token.node.nodeType) {
 215                         case ELEMENT_NODE:
 216                                 if(token.node.hasAttribute('x-pass-thru')
 217                                  || token.node.hasAttribute('data-pass-thru')) {
 218                                         self._pushChildren(token.node);
 219                                         break;
 220                                 }
 221
 222                                 if(!token.node.hasAttribute('x-node'))
 223                                         break;
 224
 225                                 var xnode = token.node.getAttribute('x-node');
 226
 227                                 if(xnode === 'out-of-flow-text') {
 228                                         self._pushChildren(token.node);
 229                                         break;
 230                                 }
 231
 232                 if(token.node.hasAttribute('x-verse') && self._verseBefore(token.node)) {
 233                     self.result += '/';
 234                     // add whitespace if there's none
 235                     if (!(text_buffer.match(/^\s/) || self._ignoredWithWhitespace(token.node)))
 236                         self.result += ' ';
 237                 }
 238
 239                 self.result += text_buffer;
 240                 text_buffer = '';
 241                                 self._serializeElement(token.node);
 242                                 break;
 243                         case TEXT_NODE:
 244                                 self.result += text_buffer;
 245                                 text_buffer = token.node.nodeValue.replace(/&/g, '&amp;').replace(/</g, '&lt;');
 246                                 break;
 247             case COMMENT_NODE:
 248                 self.result += text_buffer;
 249                 text_buffer = '';
 250                 self.result += '<!--' + token.node.nodeValue + '-->';
 251                 break;
 252                 };
 253         };
 254     self.result += text_buffer;
 255
 256         return this.result;
 257 }
 258
 259 /*
 260  * TODO: this doesn't support prefix redefinitions
 261  */
 262 HTMLSerializer.prototype._unassignNamespace = function(nsData) {
 263         this.nsMap[nsData.uri] = undefined;
 264 };
 265
 266 HTMLSerializer.prototype._assignNamespace = function(uri) {
 267         if(uri === null) {
 268                 // default namespace
 269                 return ({"prefix": "", "uri": "", "fresh": false});
 270         }
 271
 272         if(this.nsMap[uri] === undefined) {
 273                 // this prefix hasn't been defined yet in current context
 274                 var prefix = NAMESPACES[uri];
 275
 276                 if (prefix === undefined) { // not predefined
 277                         prefix = "ns" + this.nsCounter;
 278                         this.nsCounter += 1;
 279                 }
 280
 281                 this.nsMap[uri] = prefix;
 282                 return ({
 283                         "prefix": prefix,
 284                         "uri": uri,
 285                         "fresh": true
 286                 });
 287         }
 288
 289         return ({"prefix": this.nsMap[uri], "uri": uri, "fresh": false});
 290 };
 291
 292 HTMLSerializer.prototype._join = function(prefix, name) {
 293         if(!!prefix)
 294                 return prefix + ":" + name;
 295         return name;
 296 };
 297
 298 HTMLSerializer.prototype._rjoin = function(prefix, name) {
 299         if(!!name)
 300                 return prefix + ":" + name;
 301         return prefix;
 302 };
 303
 304 HTMLSerializer.prototype._serializeElement = function(node) {
 305     var self = this;
 306
 307     if (self._nodeIgnored(node)) {
 308         self._pushTagEnd('');
 309         self._pushChildren(node);
 310     }
 311     else {
 312         var ns = node.getAttribute('x-ns');
 313         var nsPrefix = null;
 314         var newNamespaces = [];
 315
 316         var nsData = self._assignNamespace(node.getAttribute('x-ns'));
 317
 318         if(nsData.fresh) {
 319                 newNamespaces.push(nsData);
 320                 self.stack.push({
 321                         "type": NS_END,
 322                         "namespace": nsData
 323                 });
 324         }
 325
 326         var tagName = self._join(nsData.prefix, node.getAttribute('x-node'));
 327
 328         /* retrieve attributes */
 329         var attributeIDs = [];
 330         for (var i = 0; i < node.attributes.length; i++) {
 331                 var attr = node.attributes.item(i);
 332
 333                 // check if name starts with "x-attr-name"
 334                 var m = attr.name.match(XATTR_RE);
 335                 if (m !== null)
 336                         attributeIDs.push(m[1]);
 337         };
 338
 339         /* print out */
 340
 341         self.result += '<' + tagName;
 342
 343         $.each(attributeIDs, function() {
 344                 var nsData = self._assignNamespace(node.getAttribute('x-attr-ns-'+this));
 345
 346                 if(nsData.fresh) {
 347                         newNamespaces.push(nsData);
 348                         self.stack.push({
 349                                 "type": NS_END,
 350                                 "namespace": nsData
 351                         });
 352                 };
 353
 354                 self.result += ' ' + self._join(nsData.prefix, node.getAttribute('x-attr-name-'+this));
 355                 self.result += '="' + node.getAttribute('x-attr-value-'+this).replace(/&/g, '&amp;').replace(/"/g, '&quot;') + '"';
 356         });
 357
 358         /* print new namespace declarations */
 359         $.each(newNamespaces, function() {
 360                 self.result += " " + self._rjoin("xmlns", this.prefix);
 361                 self.result += '="' + this.uri + '"';
 362         });
 363
 364         if (node.childNodes.length > 0) {
 365                 self.result += ">";
 366                 self._pushTagEnd(tagName);
 367                 self._pushChildren(node);
 368         }
 369         else {
 370                 self.result += "/>";
 371         };
 372     }
 373 };
 374
 375 function html2text(params) {
 376         try {
 377                 var s = new HTMLSerializer();
 378                 params.success( s.serialize(params.element, params.stripOuter) );
 379         } catch(e) {
 380                 params.error("Nie udało się zserializować tekstu:" + e)
 381         }
 382 }