Obey length limits for wikidata import.
[redakcja.git] / src / redakcja / static / js / wiki / xslt.js
index 9da4e11..99af3c3 100644 (file)
@@ -17,21 +17,20 @@ function withStylesheets(code_block, onError)
     if (!xml2htmlStylesheet) {
        $.blockUI({message: 'Ładowanie arkuszy stylów...'});
        $.ajax({
     if (!xml2htmlStylesheet) {
        $.blockUI({message: 'Ładowanie arkuszy stylów...'});
        $.ajax({
-               url: STATIC_URL + 'xsl/wl2html_client.xsl?20201110',
-               dataType: 'xml',
-               timeout: 10000,
-               success: function(data) {
+            url: '/wlxml/wl2html.xsl',
+            dataType: 'xml',
+            timeout: 10000,
+            success: function(data) {
                xml2htmlStylesheet = createXSLT(data);
                 $.unblockUI();
                xml2htmlStylesheet = createXSLT(data);
                 $.unblockUI();
-                               code_block();
-
+               code_block();
             },
             },
-                       error: onError
+           error: onError
         })
     }
         })
     }
-       else {
-               code_block();
-       }
+    else {
+       code_block();
+    }
 }
 
 
 }
 
 
@@ -61,6 +60,20 @@ function xml2html(options) {
             source.text('');
             options.error(error.text(), source_text);
         } else {
             source.text('');
             options.error(error.text(), source_text);
         } else {
+            let galleryUrl = new URL(
+                options.base,
+                window.location.href
+            );
+            $("img", $(doc.childNodes)).each(function() {
+                $(this).attr(
+                    'src',
+                    new URL(
+                        $(this).attr('src'),
+                        galleryUrl
+                    )
+                );
+            })
+
             options.success(doc.childNodes);
 
             $.themes.withCanon(function(canonThemes) {
             options.success(doc.childNodes);
 
             $.themes.withCanon(function(canonThemes) {
@@ -81,7 +94,7 @@ function xml2html(options) {
 }
 
 /* USEFULL CONSTANTS */
 }
 
 /* USEFULL CONSTANTS */
-const ELEMENT_NODE                                      = 1;
+const ELEMENT_NODE                   = 1;
 const ATTRIBUTE_NODE                 = 2;
 const TEXT_NODE                      = 3;
 const CDATA_SECTION_NODE             = 4;
 const ATTRIBUTE_NODE                 = 2;
 const TEXT_NODE                      = 3;
 const CDATA_SECTION_NODE             = 4;
@@ -94,6 +107,7 @@ const DOCUMENT_TYPE_NODE             = 10;
 const DOCUMENT_FRAGMENT_NODE         = 11;
 const NOTATION_NODE                  = 12;
 const XATTR_RE = /^x-attr-name-(.*)$/;
 const DOCUMENT_FRAGMENT_NODE         = 11;
 const NOTATION_NODE                  = 12;
 const XATTR_RE = /^x-attr-name-(.*)$/;
+const XATTR_KNOWN_RE = /^x-a-([a-z]+)-(.*)$/;
 
 const ELEM_START = 1;
 const ELEM_END = 2;
 
 const ELEM_START = 1;
 const ELEM_END = 2;
@@ -106,13 +120,15 @@ const NAMESPACES = {
        "http://www.w3.org/XML/1998/namespace": "xml"
 };
 
        "http://www.w3.org/XML/1998/namespace": "xml"
 };
 
-function HTMLSerializer() {
-       // empty constructor
-}
-
-
+NS_PREFIXES = {
+    'wl': ''
+};
+for (prefix in NAMESPACES) {
+    NS_PREFIXES[NAMESPACES[prefix]] = prefix
+};
 
 
-HTMLSerializer.prototype._prepare = function() {
+class HTMLSerializer {
+    _prepare() {
        this.stack = [];
 
        // XML namespace is implicit
        this.stack = [];
 
        // XML namespace is implicit
@@ -120,250 +136,279 @@ HTMLSerializer.prototype._prepare = function() {
 
        this.result = "";
        this.nsCounter = 1;
 
        this.result = "";
        this.nsCounter = 1;
-}
+    }
 
 
-HTMLSerializer.prototype._pushElement = function(element) {
+    _pushElement(element) {
        this.stack.push({
        this.stack.push({
-               "type": ELEM_START,
-               "node": element
+           "type": ELEM_START,
+           "node": element
        });
        });
-}
+    }
 
 
-HTMLSerializer.prototype._pushChildren = function(element) {
+    _pushChildren(element) {
        for(var i = element.childNodes.length-1; i >= 0; i--)
        for(var i = element.childNodes.length-1; i >= 0; i--)
-               this._pushElement(element.childNodes.item(i));
-}
+           this._pushElement(element.childNodes.item(i));
+    }
 
 
-HTMLSerializer.prototype._pushTagEnd = function(tagName) {
+    _pushTagEnd(tagName) {
        this.stack.push({
        this.stack.push({
-               "type": ELEM_END,
-               "tagName": tagName
+           "type": ELEM_END,
+           "tagName": tagName
        });
        });
-}
+    }
 
 
-HTMLSerializer.prototype._verseBefore = function(node) {
-    /* true if previous element is a previous verse of a stanza */
-    var parent = node.parentNode;
-    if (!parent || !parent.hasAttribute('x-node') || parent.getAttribute('x-node') != 'strofa')
-        return false;
+    _verseBefore(node) {
+        /* true if previous element is a previous verse of a stanza */
+        var parent = node.parentNode;
+        if (!parent || !parent.hasAttribute('x-node') || parent.getAttribute('x-node') != 'strofa')
+            return false;
 
        var prev = node.previousSibling;
 
        while((prev !== null) && (prev.nodeType != ELEMENT_NODE)) {
 
        var prev = node.previousSibling;
 
        while((prev !== null) && (prev.nodeType != ELEMENT_NODE)) {
-               prev = prev.previousSibling;
+           prev = prev.previousSibling;
        }
 
        return (prev !== null) && prev.hasAttribute('x-verse');
        }
 
        return (prev !== null) && prev.hasAttribute('x-verse');
-}
+    }
 
 
-HTMLSerializer.prototype._nodeIgnored = function(node) {
-    return node.getAttribute('x-auto-node') == 'true';
-}
+    _nodeIgnored(node) {
+        return node.getAttribute('x-auto-node') == 'true';
+    }
 
 
-HTMLSerializer.prototype._ignoredWithWhitespace = function(node) {
-    while (node.nodeType == ELEMENT_NODE && this._nodeIgnored(node) && node.childNodes.length > 0)
-        node = node.childNodes[0];
-    if (node.nodeType == TEXT_NODE)
-        return node.nodeValue.match(/^\s/)
-    else return false;
-}
+    _ignoredWithWhitespace(node) {
+        while (node.nodeType == ELEMENT_NODE && this._nodeIgnored(node) && node.childNodes.length > 0)
+            node = node.childNodes[0];
+        if (node.nodeType == TEXT_NODE)
+            return node.nodeValue.match(/^\s/)
+        else return false;
+    }
 
 
 
 
-HTMLSerializer.prototype.serialize = function(rootElement, stripOuter)
-{
+    serialize(rootElement, stripOuter)
+    {
        var self = this;
        self._prepare();
        var self = this;
        self._prepare();
-
+        
        if(!stripOuter)
        if(!stripOuter)
-               self._pushElement(rootElement);
+           self._pushElement(rootElement);
        else
        else
-               self._pushChildren(rootElement);
-
-    var text_buffer = '';
+           self._pushChildren(rootElement);
+        
+        var text_buffer = '';
 
        while(self.stack.length > 0) {
 
        while(self.stack.length > 0) {
-               var token = self.stack.pop();
-
-        if(token.type === ELEM_END) {
-            self.result += text_buffer;
-            text_buffer = '';
-            if (token.tagName != '')
-                self.result += "</" + token.tagName + ">";
-            continue;
-        };
-
-               if(token.type === NS_END) {
-                       self._unassignNamespace(token.namespace);
-                       continue;
-               }
-
-
-               switch(token.node.nodeType) {
-                       case ELEMENT_NODE:
-                               if(token.node.hasAttribute('x-pass-thru')
-                                || token.node.hasAttribute('data-pass-thru')) {
-                                       self._pushChildren(token.node);
-                                       break;
-                               }
-
-                               if(!token.node.hasAttribute('x-node'))
-                                       break;
-
-                               var xnode = token.node.getAttribute('x-node');
-
-                               if(xnode === 'out-of-flow-text') {
-                                       self._pushChildren(token.node);
-                                       break;
-                               }
+           var token = self.stack.pop();
 
 
+            if(token.type === ELEM_END) {
+                self.result += text_buffer;
+                text_buffer = '';
+                if (token.tagName != '')
+                    self.result += "</" + token.tagName + ">";
+                continue;
+            };
+
+           if(token.type === NS_END) {
+               self._unassignNamespace(token.namespace);
+               continue;
+           }
+
+           switch(token.node.nodeType) {
+           case ELEMENT_NODE:
+               if(token.node.hasAttribute('x-pass-thru')
+                  || token.node.hasAttribute('data-pass-thru')) {
+                   self._pushChildren(token.node);
+                   break;
+               }
+                
+               if(!token.node.hasAttribute('x-node'))
+                   break;
+                
+               var xnode = token.node.getAttribute('x-node');
+                
+               if(xnode === 'out-of-flow-text') {
+                   self._pushChildren(token.node);
+                   break;
+               }
+                
                 if(token.node.hasAttribute('x-verse') && self._verseBefore(token.node)) {
                     self.result += '/';
                     // add whitespace if there's none
                     if (!(text_buffer.match(/^\s/) || self._ignoredWithWhitespace(token.node)))
                         self.result += ' ';
                 }
                 if(token.node.hasAttribute('x-verse') && self._verseBefore(token.node)) {
                     self.result += '/';
                     // add whitespace if there's none
                     if (!(text_buffer.match(/^\s/) || self._ignoredWithWhitespace(token.node)))
                         self.result += ' ';
                 }
-
+                
                 self.result += text_buffer;
                 text_buffer = '';
                 self.result += text_buffer;
                 text_buffer = '';
-                               self._serializeElement(token.node);
-                               break;
-                       case TEXT_NODE:
-                               self.result += text_buffer;
-                               text_buffer = token.node.nodeValue.replace(/&/g, '&amp;').replace(/</g, '&lt;');
-                               break;
+               self._serializeElement(token.node);
+               break;
+           case TEXT_NODE:
+               self.result += text_buffer;
+               text_buffer = token.node.nodeValue.replace(/&/g, '&amp;').replace(/</g, '&lt;');
+               break;
             case COMMENT_NODE:
                 self.result += text_buffer;
                 text_buffer = '';
                 self.result += '<!--' + token.node.nodeValue + '-->';
                 break;
             case COMMENT_NODE:
                 self.result += text_buffer;
                 text_buffer = '';
                 self.result += '<!--' + token.node.nodeValue + '-->';
                 break;
-               };
+           };
        };
        };
-    self.result += text_buffer;
+        self.result += text_buffer;
 
        return this.result;
 
        return this.result;
-}
+    }
 
 
-/*
- * TODO: this doesn't support prefix redefinitions
- */
-HTMLSerializer.prototype._unassignNamespace = function(nsData) {
+    /*
    * TODO: this doesn't support prefix redefinitions
    */
+    _unassignNamespace(nsData) {
        this.nsMap[nsData.uri] = undefined;
        this.nsMap[nsData.uri] = undefined;
-};
+    }
 
 
-HTMLSerializer.prototype._assignNamespace = function(uri) {
+    _assignNamespace(uri) {
        if(uri === null) {
        if(uri === null) {
-               // default namespace
-               return ({"prefix": "", "uri": "", "fresh": false});
+           // default namespace
+           return ({"prefix": "", "uri": "", "fresh": false});
        }
        }
-
+        
        if(this.nsMap[uri] === undefined) {
        if(this.nsMap[uri] === undefined) {
-               // this prefix hasn't been defined yet in current context
-               var prefix = NAMESPACES[uri];
-
-               if (prefix === undefined) { // not predefined
-                       prefix = "ns" + this.nsCounter;
-                       this.nsCounter += 1;
-               }
-
-               this.nsMap[uri] = prefix;
-               return ({
-                       "prefix": prefix,
-                       "uri": uri,
-                       "fresh": true
-               });
+           // this prefix hasn't been defined yet in current context
+           var prefix = NAMESPACES[uri];
+            
+           if (prefix === undefined) { // not predefined
+               prefix = "ns" + this.nsCounter;
+               this.nsCounter += 1;
+           }
+            
+           this.nsMap[uri] = prefix;
+           return ({
+               "prefix": prefix,
+               "uri": uri,
+               "fresh": true
+           });
        }
        }
-
+        
        return ({"prefix": this.nsMap[uri], "uri": uri, "fresh": false});
        return ({"prefix": this.nsMap[uri], "uri": uri, "fresh": false});
-};
-
-HTMLSerializer.prototype._join = function(prefix, name) {
+    }
+    
+    _join(prefix, name) {
        if(!!prefix)
        if(!!prefix)
-               return prefix + ":" + name;
+           return prefix + ":" + name;
        return name;
        return name;
-};
+    }
 
 
-HTMLSerializer.prototype._rjoin = function(prefix, name) {
+    _rjoin(prefix, name) {
        if(!!name)
        if(!!name)
-               return prefix + ":" + name;
+           return prefix + ":" + name;
        return prefix;
        return prefix;
-};
+    }
 
 
-HTMLSerializer.prototype._serializeElement = function(node) {
-    var self = this;
+    _serializeElement(node) {
+        var self = this;
 
 
-    if (self._nodeIgnored(node)) {
-        self._pushTagEnd('');
-        self._pushChildren(node);
-    }
-    else {
-       var ns = node.getAttribute('x-ns');
-       var nsPrefix = null;
-       var newNamespaces = [];
+        if (self._nodeIgnored(node)) {
+            self._pushTagEnd('');
+            self._pushChildren(node);
+        }
+        else {
+           var ns = node.getAttribute('x-ns');
+           var nsPrefix = null;
+           var newNamespaces = [];
 
 
-       var nsData = self._assignNamespace(node.getAttribute('x-ns'));
+           var nsData = self._assignNamespace(node.getAttribute('x-ns'));
 
 
-       if(nsData.fresh) {
+           if(nsData.fresh) {
                newNamespaces.push(nsData);
                self.stack.push({
                newNamespaces.push(nsData);
                self.stack.push({
-                       "type": NS_END,
-                       "namespace": nsData
+                   "type": NS_END,
+                   "namespace": nsData
                });
                });
-       }
-
-       var tagName = self._join(nsData.prefix, node.getAttribute('x-node'));
-
-       /* retrieve attributes */
-       var attributeIDs = [];
-       for (var i = 0; i < node.attributes.length; i++) {
-               var attr = node.attributes.item(i);
-
-               // check if name starts with "x-attr-name"
-               var m = attr.name.match(XATTR_RE);
-               if (m !== null)
-                       attributeIDs.push(m[1]);
-       };
-
-       /* print out */
-
-       self.result += '<' + tagName;
-
-       $.each(attributeIDs, function() {
-               var nsData = self._assignNamespace(node.getAttribute('x-attr-ns-'+this));
-
-               if(nsData.fresh) {
-                       newNamespaces.push(nsData);
-                       self.stack.push({
-                               "type": NS_END,
-                               "namespace": nsData
-                       });
-               };
-
-               self.result += ' ' + self._join(nsData.prefix, node.getAttribute('x-attr-name-'+this));
-               self.result += '="'+node.getAttribute('x-attr-value-'+this) +'"';
-       });
-
-       /* print new namespace declarations */
-       $.each(newNamespaces, function() {
+           }
+
+           var tagName = self._join(nsData.prefix, node.getAttribute('x-node'));
+            
+           /* retrieve attributes */
+           var attributeIDs = [];
+            var attributes = [];
+           for (var i = 0; i < node.attributes.length; i++) {
+               var attr = node.attributes.item(i);
+                
+                m = attr.name.match(XATTR_KNOWN_RE);
+                if (m !== null) {
+                    prefix = m[1];
+                    let tag = m[2];
+                    attributes.push([
+                        NS_PREFIXES[prefix],
+                        tag,
+                        attr.value
+                    ]);
+                } else {
+                   // check if name starts with "x-attr-name"
+                   var m = attr.name.match(XATTR_RE);
+                   if (m !== null) {
+                       attributeIDs.push(m[1]);
+                    }
+                }
+           }
+
+           /* print out */
+
+           self.result += '<' + tagName;
+
+            function writeAttr(ns, tag, value) {
+                if (ns) {
+                   var nsData = self._assignNamespace(ns);
+                   if(nsData.fresh) {
+                       newNamespaces.push(nsData);
+                       self.stack.push({
+                           "type": NS_END,
+                           "namespace": nsData
+                       });
+                   };
+                    tag = self._join(nsData.prefix, tag);
+                }
+                
+               self.result += ' ' + tag;
+               self.result += '="' + value.replace(/&/g, '&amp;').replace(/"/g, '&quot;') + '"';
+            }
+        
+            $.each(attributes, function() {
+                writeAttr(
+                    this[0], this[1], this[2]
+                );
+            });
+        
+           $.each(attributeIDs, function() {
+                writeAttr(
+                    node.getAttribute('x-attr-ns-'+this),
+                   node.getAttribute('x-attr-name-'+this),
+                   node.getAttribute('x-attr-value-'+this)
+                );
+           });
+
+           /* print new namespace declarations */
+           $.each(newNamespaces, function() {
                self.result += " " + self._rjoin("xmlns", this.prefix);
                self.result += '="' + this.uri + '"';
                self.result += " " + self._rjoin("xmlns", this.prefix);
                self.result += '="' + this.uri + '"';
-       });
+           });
 
 
-       if (node.childNodes.length > 0) {
+           if (node.childNodes.length > 0) {
                self.result += ">";
                self._pushTagEnd(tagName);
                self._pushChildren(node);
                self.result += ">";
                self._pushTagEnd(tagName);
                self._pushChildren(node);
-       }
-       else {
+           }
+           else {
                self.result += "/>";
                self.result += "/>";
-       };
+           };
+        }
     }
     }
-};
+}
 
 function html2text(params) {
 
 function html2text(params) {
-       try {
-               var s = new HTMLSerializer();
-               params.success( s.serialize(params.element, params.stripOuter) );
-       } catch(e) {
-               params.error("Nie udało się zserializować tekstu:" + e)
-       }
+    try {
+       var s = new HTMLSerializer();
+       params.success( s.serialize(params.element, params.stripOuter) );
+    } catch(e) {
+       params.error("Nie udało się zserializować tekstu:" + e)
+    }
 }
 }