closes #216, support for non-ascii characters in word integration

chrome/chromeFiles/content/scholar/xpcom/cite.js

@@ -94,6 +94,8 @@ CSL = function(csl) {
 CSL.prototype.preprocessItems = function(items) {
 	Scholar.debug("CSL: preprocessing items");
 	
+	this._ignore = null;
+	
 	// get data necessary to generate citations before sorting
 	for(var i in items) {
 		var item = items[i];
@@ -563,13 +565,6 @@ CSL.prototype._parseFields = function(ref, position, type, bibCitElement, inheri
 				if(bibCitElement && inheritFormat) {
 					itemDesc = this._merge(bibCitElement.inheritFormat, itemDesc);
 				}
-			
-				// create serialized representation
-				itemDesc._serialized = this._serializeElement(itemDesc.name, itemDesc);
-				// add to serialization for type
-				if(bibCitElement) {
-					bibCitElement._serializations[position][type][itemDesc._serialized] = itemDesc;
-				}
 			}
 			
 			// parse group children
@@ -586,6 +581,15 @@ CSL.prototype._parseFields = function(ref, position, type, bibCitElement, inheri
 				this._parseFieldAttrChildren(element, itemDesc);
 			}
 			
+			if(type != undefined) {
+				// create serialized representation
+				itemDesc._serialized = this._serializeElement(itemDesc.name, itemDesc);
+				// add to serialization for type
+				if(bibCitElement) {
+					bibCitElement._serializations[position][type][itemDesc._serialized] = itemDesc;
+				}
+			}
+			
 			typeDesc.push(itemDesc);
 		}
 	}
@@ -809,7 +813,6 @@ CSL.prototype._getFieldDefaults = function(elementName) {
  * gets a term, in singular or plural form
  */
 CSL.prototype._getTerm = function(term, plural, form) {
-	Scholar.debug("CSL: looking up term "+term);
 	if(!form) {
 		form = "long";
 	}
@@ -1039,7 +1042,7 @@ CSL.prototype._serializeElement = function(name, element) {
 		string += " relation:"+element.relation;
 	}
 	if(element.role) {
-		string += " role"+element.role;
+		string += " role:"+element.role;
 	}
 	return string;
 }
@@ -1214,6 +1217,8 @@ CSL.prototype._getCitation = function(item, position, format, bibCitElement) {
 	}
 	Scholar.debug("CSL: using CSL type "+typeName);
 	
+	// remove previous ignore entries from list
+	this._ignore = new Array();
 	var string = "";
 	for(var j in type) {
 		var value = this._getFieldValue(type[j].name, type[j], item, format,
@@ -1230,7 +1235,8 @@ CSL.prototype._getCitation = function(item, position, format, bibCitElement) {
 CSL.prototype._getFieldValue = function(name, element, item, format, bibCitElement, position, typeName) {
 	var data = "";
 	
-	if(element._serialized && item._csl.ignore[element._serialized]) {
+	var itemID = item.getID();
+	if(element._serialized && this._ignore && this._ignore[itemID] && this._ignore[itemID][element._serialized]) {
 		return "";
 	}
 	
@@ -1403,13 +1409,17 @@ CSL.prototype._getFieldValue = function(name, element, item, format, bibCitEleme
 			// clear substitute element off of the element we're substituting
 			substituteElement.substitute = undefined;
 			
-			// ignore elements with the same serialization
-			item._csl.ignore[serialization] = true;
-			
 			// get field value
 			data = this._getFieldValue(substituteElement.name,
 			                           substituteElement, item, format,
-			                           bibCitElement);
+			                           bibCitElement, position, typeName);
+			
+			// ignore elements with the same serialization
+			if(this._ignore) {	// array might not exist if doing disambiguation
+				if(!this._ignore[itemID]) this._ignore[itemID] = new Array();
+				this._ignore[itemID][substituteElement._serialized] = true;
+			}
+			
 			// return field value, if there is one; otherwise, keep processing
 			// the data
 			if(data) {

chrome/chromeFiles/content/scholar/xpcom/integration.js

@@ -50,7 +50,7 @@ Scholar.Integration = new function() {
 	/*
 	 * handles a SOAP envelope
 	 */
-	function handleEnvelope(envelope) {
+	function handleEnvelope(envelope, encoding) {
 		Scholar.debug("Integration: SOAP Request\n"+envelope);
 		envelope = envelope.replace(_XMLRe, "");
 		
@@ -115,11 +115,11 @@ Scholar.Integration = new function() {
 					</SOAP-ENV:Body>
 				</SOAP-ENV:Envelope>;
 				
-				var response = responseEnvelope.toXMLString();
+				var response = '<?xml version="1.0" encoding="'+encoding+'"?>\n'+responseEnvelope.toXMLString();
 				Scholar.debug("Integration: SOAP Response\n"+response);
 				
 				// return OK
-				return _generateResponse("200 OK", 'text/xml; charset="utf-8"',
+				return _generateResponse("200 OK", 'text/xml; charset="'+encoding+'"',
 				                         response);
 			} else {
 				Scholar.debug("Integration: SOAP method not supported");
@@ -275,20 +275,54 @@ Scholar.Integration.DataListener.prototype._bodyData = function() {
 			this.body = this.body.substr(0, this.bodyLength);
 		}
 		
-		var output = Scholar.Integration.handleEnvelope(this.body);
-		this._requestFinished(output);
+		// UTF-8 crashes AppleScript
+		var encoding = (this.header.indexOf("\nUser-Agent: Mac OS X") !== -1 ? "macintosh" : "UTF-8");
+		var output = Scholar.Integration.handleEnvelope(this.body, encoding);
+		this._requestFinished(output, encoding);
 	}
 }
 
 /*
  * returns HTTP data from a request
  */
-Scholar.Integration.DataListener.prototype._requestFinished = function(response) {
+Scholar.Integration.DataListener.prototype._requestFinished = function(response, encoding) {
 	// close input stream
 	this.iStream.close();
 	
-	// write response
+	if(encoding == "macintosh") {
+		// double percent signs
+		response = response.replace(/%/g, "%%");
+		// replace line endings with percent signs
+		response = response.replace(/\n/g, " %!");
+		response = response.replace(/\r/g, "");
+
+		// convert Unicode to Mac Roman
+		var converter = Components.classes["@mozilla.org/intl/scriptableunicodeconverter"] 
+								  .createInstance(Components.interfaces.nsIScriptableUnicodeConverter); 
+		converter.charset = "macintosh";
+		// convert text
+		response = converter.ConvertFromUnicode(response);
+		// fix returns
+		response = response.replace(/ %!/g, "\n");
+		// fix percent signs
+		response = response.replace(/%%/g, "%"); 
+		response = response + converter.Finish();
+	
+		// write
 		this.oStream.write(response, response.length); 
+	} else if(encoding) {
+		// open UTF-8 converter for output stream	
+		var intlStream = Components.classes["@mozilla.org/intl/converter-output-stream;1"]
+								   .createInstance(Components.interfaces.nsIConverterOutputStream);
+		intlStream.init(this.oStream, encoding, 1024, "?".charCodeAt(0));
+		
+		// write response
+		intlStream.writeString(response);
+		intlStream.close();
+	} else {
+		// write
+		this.oStream.write(response, response.length);
+	}
 	
 	// close output stream
 	this.oStream.close();
@@ -458,8 +492,10 @@ Scholar.Integration.SOAP = new function() {
 		} else {
 			// session ID exists
 			var sessionID = vars[0];
-			Scholar.debug(vars.toSource());
 			var session = _sessions[sessionID];
+			if(!session) {
+				return "ERROR:sessionExpired";
+			}
 			var originalStyle = session.styleID;
 			io.style = originalStyle;
 		}

chrome/chromeFiles/content/scholar/xpcom/translate.js

@@ -745,6 +745,10 @@ Scholar.Translate.prototype._canTranslate = function(translator, ignoreExtension
 	
 	return canTranslate;
 }
+
+/*
+ * parses translator detect code
+ */
 Scholar.Translate.prototype._parseDetectCode = function(translator) {
 	this._configOptions = new Array();
 	this._displayOptions = new Array();