dojo.provide("dojox.data.CsvStore");

dojo.require("dojo.data.util.filter");
dojo.require("dojo.data.util.simpleFetch");

dojo.declare("dojox.data.CsvStore", null, {
	//	summary:
	//		The CsvStore implements the dojo.data.api.Read API and reads
	//		data from files in CSV (Comma Separated Values) format.
	//		All values are simple string values. References to other items
	//		are not supported as attribute values in this datastore.
	//
	//		Example data file:
	//		name, color, age, tagline
	//		Kermit, green, 12, "Hi, I'm Kermit the Frog."
	//		Fozzie Bear, orange, 10, "Wakka Wakka Wakka!"
	//		Miss Piggy, pink, 11, "Kermie!"
	//
	//		Note that values containing a comma must be enclosed with quotes ("")
	//		Also note that values containing quotes must be escaped with two consecutive quotes (""quoted"")
	
	/* examples:
	 *   var csvStore = new dojox.data.CsvStore({url:"movies.csv");
	 *   var csvStore = new dojox.data.CsvStore({url:"http://example.com/movies.csv");
	 */

	constructor: function(/* Object */ keywordParameters){
		// summary: initializer
		// keywordParameters: {url: String}
		// keywordParameters: {data: String}
		// keywordParameters: {label: String} The column label for the column to use for the label returned by getLabel.
		// keywordParameters: {identifier: String} The column label for the column to use for the identity.  Optional.  If not set, the identity is the row number.
		
		this._attributes = [];			// e.g. ["Title", "Year", "Producer"]
		this._attributeIndexes = {};	// e.g. {Title: 0, Year: 1, Producer: 2}
 		this._dataArray = [];			// e.g. [[<Item0>],[<Item1>],[<Item2>]]
 		this._arrayOfAllItems = [];		// e.g. [{_csvId:0,_csvStore:store},...]
		this._loadFinished = false;
		if(keywordParameters.url){
			this.url = keywordParameters.url;
		}
		this._csvData = keywordParameters.data;
		if(keywordParameters.label){
			this.label = keywordParameters.label;
		}else if(this.label === ""){
			this.label = undefined;
		}
		this._storeProp = "_csvStore";	// Property name for the store reference on every item.
		this._idProp = "_csvId"; 		// Property name for the Item Id on every item.
		this._features = {	
			'dojo.data.api.Read': true,
			'dojo.data.api.Identity': true 
		};
		this._loadInProgress = false;	//Got to track the initial load to prevent duelling loads of the dataset.
		this._queuedFetches = [];
		this.identifier = keywordParameters.identifier;
		if (this.identifier === "") {
			delete this.identifier;
		}else{
			this._idMap = {};
		}
	},

	url: "", //Declarative hook for setting Csv source url.

	label: "", //Declarative hook for setting the label attribute. 

	identifier: "", //Declarative hook for setting the identifier.

	_assertIsItem: function(/* item */ item){
		//	summary:
		//      This function tests whether the item passed in is indeed an item in the store.
		//	item: 
		//		The item to test for being contained by the store.
		if(!this.isItem(item)){ 
			throw new Error(this.declaredClass + ": a function was passed an item argument that was not an item");
		}
	},
	
	_assertIsAttribute: function(/* item || String */ attribute){
		//	summary:
		//      This function tests whether the item passed in is indeed a valid 'attribute' like type for the store.
		//	attribute: 
		//		The attribute to test for being contained by the store.
		if(!dojo.isString(attribute)){ 
			throw new Error(this.declaredClass + ": a function was passed an attribute argument that was not an attribute object nor an attribute name string");
		}
	},

	_getIndex: function(item){
		//	summary:
		//		Internal function to get the internal index to the item data from the item handle
		//	item:
		//		The idem handle to get the index for.
		var idx = this.getIdentity(item);
		if(this.identifier){
			idx = this._idMap[idx];
		}
		return idx;
	},

/***************************************
     dojo.data.api.Read API
***************************************/
	getValue: function(	/* item */ item, 
						/* attribute || attribute-name-string */ attribute, 
						/* value? */ defaultValue){
		//	summary: 
		//      See dojo.data.api.Read.getValue()
		//		Note that for the CsvStore, an empty string value is the same as no value, 
		// 		so the defaultValue would be returned instead of an empty string.
		this._assertIsItem(item);
		this._assertIsAttribute(attribute);
		var itemValue = defaultValue;
		if(this.hasAttribute(item, attribute)){
			var itemData = this._dataArray[this._getIndex(item)];
			itemValue = itemData[this._attributeIndexes[attribute]];
		}
		return itemValue; //String
	},


	getValues: function(/* item */ item, 
						/* attribute || attribute-name-string */ attribute){
		//	summary: 
		//		See dojo.data.api.Read.getValues()
		// 		CSV syntax does not support multi-valued attributes, so this is just a
		// 		wrapper function for getValue().
		var value = this.getValue(item, attribute);
		return (value ? [value] : []); //Array
	},

	getAttributes: function(/* item */ item){
		//	summary: 
		//		See dojo.data.api.Read.getAttributes()
		this._assertIsItem(item);
		var attributes = [];
		var itemData = this._dataArray[this._getIndex(item)];
		for(var i=0; i<itemData.length; i++){
			// Check for empty string values. CsvStore treats empty strings as no value.
			if(itemData[i] !== ""){
				attributes.push(this._attributes[i]);
			}
		}
		return attributes; //Array
	},

	hasAttribute: function(	/* item */ item,
							/* attribute || attribute-name-string */ attribute){
		//	summary: 
		//		See dojo.data.api.Read.hasAttribute()
		// 		The hasAttribute test is true if attribute has an index number within the item's array length
		// 		AND if the item has a value for that attribute. Note that for the CsvStore, an
		// 		empty string value is the same as no value.
		this._assertIsItem(item);
		this._assertIsAttribute(attribute);
		var attributeIndex = this._attributeIndexes[attribute];
		var itemData = this._dataArray[this._getIndex(item)];
		return (typeof attributeIndex !== "undefined" && attributeIndex < itemData.length && itemData[attributeIndex] !== ""); //Boolean
	},

	containsValue: function(/* item */ item, 
							/* attribute || attribute-name-string */ attribute, 
							/* anything */ value){
		//	summary: 
		//		See dojo.data.api.Read.containsValue()
		var regexp = undefined;
		if(typeof value === "string"){
		   regexp = dojo.data.util.filter.patternToRegExp(value, false);
		}
		return this._containsValue(item, attribute, value, regexp); //boolean.
	},

	_containsValue: function(	/* item */ item, 
								/* attribute || attribute-name-string */ attribute, 
								/* anything */ value,
								/* RegExp?*/ regexp){
		//	summary: 
		//		Internal function for looking at the values contained by the item.
		//	description: 
		//		Internal function for looking at the values contained by the item.  This 
		//		function allows for denoting if the comparison should be case sensitive for
		//		strings or not (for handling filtering cases where string case should not matter)
		//	
		//	item:
		//		The data item to examine for attribute values.
		//	attribute:
		//		The attribute to inspect.
		//	value:	
		//		The value to match.
		//	regexp:
		//		Optional regular expression generated off value if value was of string type to handle wildcarding.
		//		If present and attribute values are string, then it can be used for comparison instead of 'value'
		var values = this.getValues(item, attribute);
		for(var i = 0; i < values.length; ++i){
			var possibleValue = values[i];
			if(typeof possibleValue === "string" && regexp){
				return (possibleValue.match(regexp) !== null);
			}else{
				//Non-string matching.
				if(value === possibleValue){
					return true; // Boolean
				}
			}
		}
		return false; // Boolean
	},

	isItem: function(/* anything */ something){
		//	summary: 
		//		See dojo.data.api.Read.isItem()
		if(something && something[this._storeProp] === this){
			var identity = something[this._idProp];
			//If an identifier was specified, we have to look it up via that and the mapping,
			//otherwise, just use row number.
			if(this.identifier){
				var data = this._dataArray[this._idMap[identity]];
				if(data){
					return true;
				}
			}else{
				if(identity >= 0 && identity < this._dataArray.length){
					return true; //Boolean
				}
			}
		}
		return false; //Boolean
	},

	isItemLoaded: function(/* anything */ something){
		//	summary: 
		//		See dojo.data.api.Read.isItemLoaded()
		//		The CsvStore always loads all items, so if it's an item, then it's loaded.
		return this.isItem(something); //Boolean
	},

	loadItem: function(/* item */ item){
		//	summary: 
		//		See dojo.data.api.Read.loadItem()
		//	description:
		//		The CsvStore always loads all items, so if it's an item, then it's loaded.
		//		From the dojo.data.api.Read.loadItem docs:
		//			If a call to isItemLoaded() returns true before loadItem() is even called,
		//			then loadItem() need not do any work at all and will not even invoke
		//			the callback handlers.
	},

	getFeatures: function(){
		//	summary: 
		//		See dojo.data.api.Read.getFeatures()
		return this._features; //Object
	},

	getLabel: function(/* item */ item){
		//	summary: 
		//		See dojo.data.api.Read.getLabel()
		if(this.label && this.isItem(item)){
			return this.getValue(item,this.label); //String
		}
		return undefined; //undefined
	},

	getLabelAttributes: function(/* item */ item){
		//	summary: 
		//		See dojo.data.api.Read.getLabelAttributes()
		if(this.label){
			return [this.label]; //array
		}
		return null; //null
	},


	// The dojo.data.api.Read.fetch() function is implemented as
	// a mixin from dojo.data.util.simpleFetch.
	// That mixin requires us to define _fetchItems().
	_fetchItems: function(	/* Object */ keywordArgs, 
							/* Function */ findCallback, 
							/* Function */ errorCallback){
		//	summary: 
		//		See dojo.data.util.simpleFetch.fetch()
		
		var self = this;

		var filter = function(requestArgs, arrayOfAllItems){
			var items = null;
			if(requestArgs.query){
				var key, value;
				items = [];
				var ignoreCase = requestArgs.queryOptions ? requestArgs.queryOptions.ignoreCase : false; 

				//See if there are any string values that can be regexp parsed first to avoid multiple regexp gens on the
				//same value for each item examined.  Much more efficient.
				var regexpList = {};
				for(key in requestArgs.query){
					value = requestArgs.query[key];
					if(typeof value === "string"){
						regexpList[key] = dojo.data.util.filter.patternToRegExp(value, ignoreCase);
					}
				}

				for(var i = 0; i < arrayOfAllItems.length; ++i){
					var match = true;
					var candidateItem = arrayOfAllItems[i];
					for(key in requestArgs.query){
						value = requestArgs.query[key];
						if(!self._containsValue(candidateItem, key, value, regexpList[key])){
							match = false;
						}
					}
					if(match){
						items.push(candidateItem);
					}
				}
			}else{
				// We want a copy to pass back in case the parent wishes to sort the array.  We shouldn't allow resort 
				// of the internal list so that multiple callers can get lists and sort without affecting each other.
				if(arrayOfAllItems.length> 0){
					items = arrayOfAllItems.slice(0,arrayOfAllItems.length); 
				}
			}
			findCallback(items, requestArgs);
		};

		if(this._loadFinished){
			filter(keywordArgs, this._arrayOfAllItems);
		}else{
			if(this.url !== ""){
				//If fetches come in before the loading has finished, but while
				//a load is in progress, we have to defer the fetching to be 
				//invoked in the callback.
				if(this._loadInProgress){
					this._queuedFetches.push({args: keywordArgs, filter: filter});
				}else{
					this._loadInProgress = true;
					var getArgs = {
							url: self.url, 
							handleAs: "text"
						};
					var getHandler = dojo.xhrGet(getArgs);
					getHandler.addCallback(function(data){
						try{
							self._processData(data);
							filter(keywordArgs, self._arrayOfAllItems);
							self._handleQueuedFetches();
						}catch(e){
							errorCallback(e, keywordArgs);
						}
					});
					getHandler.addErrback(function(error){
						self._loadInProgress = false;
						if(errorCallback){
							errorCallback(error, keywordArgs);
						}else{
							throw error;
						}
					});
				}
			}else if(this._csvData){
				try{
					this._processData(this._csvData);
					this._csvData = null;
					filter(keywordArgs, this._arrayOfAllItems);
				}catch(e){
					errorCallback(e, keywordArgs);
				}
			}else{
				var error = new Error(this.declaredClass + ": No CSV source data was provided as either URL or String data input.");
				if(errorCallback){
					errorCallback(error, keywordArgs);
				}else{
					throw error;
				}
			}
		}
	},
	
	close: function(/*dojo.data.api.Request || keywordArgs || null */ request){
		 //	summary: 
		 //		See dojo.data.api.Read.close()
	},
	
	
	// -------------------------------------------------------------------
	// Private methods
	_getArrayOfArraysFromCsvFileContents: function(/* string */ csvFileContents){
		/* summary:
		 *   Parses a string of CSV records into a nested array structure.
		 * description:
		 *   Given a string containing CSV records, this method parses
		 *   the string and returns a data structure containing the parsed
		 *   content.  The data structure we return is an array of length
		 *   R, where R is the number of rows (lines) in the CSV data.  The 
		 *   return array contains one sub-array for each CSV line, and each 
		 *   sub-array contains C string values, where C is the number of 
		 *   columns in the CSV data.
		 */
		 
		/* example:
		 *   For example, given this CSV string as input:
		 *     "Title, Year, Producer \n Alien, 1979, Ridley Scott \n Blade Runner, 1982, Ridley Scott"
		 *   this._dataArray will be set to:
		 *     [["Alien", "1979", "Ridley Scott"],
		 *      ["Blade Runner", "1982", "Ridley Scott"]]
		 *   And this._attributes will be set to:
		 *     ["Title", "Year", "Producer"]
		 *   And this._attributeIndexes will be set to:
		 *     { "Title":0, "Year":1, "Producer":2 }
		 */
		if(dojo.isString(csvFileContents)){
			var lineEndingCharacters = new RegExp("\r\n|\n|\r");
			var leadingWhiteSpaceCharacters = new RegExp("^\\s+",'g');
			var trailingWhiteSpaceCharacters = new RegExp("\\s+$",'g');
			var doubleQuotes = new RegExp('""','g');
			var arrayOfOutputRecords = [];
			var i;
			
			var arrayOfInputLines = this._splitLines(csvFileContents);
			for(i = 0; i < arrayOfInputLines.length; ++i){
				var singleLine = arrayOfInputLines[i];
				if(singleLine.length > 0){
					var listOfFields = singleLine.split(',');
					var j = 0;
					while(j < listOfFields.length){
						var space_field_space = listOfFields[j];
						var field_space = space_field_space.replace(leadingWhiteSpaceCharacters, ''); // trim leading whitespace
						var field = field_space.replace(trailingWhiteSpaceCharacters, ''); // trim trailing whitespace
						var firstChar = field.charAt(0);
						var lastChar = field.charAt(field.length - 1);
						var secondToLastChar = field.charAt(field.length - 2);
						var thirdToLastChar = field.charAt(field.length - 3);
						if(field.length === 2 && field == "\"\""){
							listOfFields[j] = "";  //Special case empty string field.
						}else if((firstChar == '"') && 
								((lastChar != '"') || 
								 ((lastChar == '"') && (secondToLastChar == '"') && (thirdToLastChar != '"')))){
							if(j+1 === listOfFields.length){
								// alert("The last field in record " + i + " is corrupted:\n" + field);
								return null; //null
							}
							var nextField = listOfFields[j+1];
							listOfFields[j] = field_space + ',' + nextField;
							listOfFields.splice(j+1, 1); // delete element [j+1] from the list
						}else{
							if((firstChar == '"') && (lastChar == '"')){
								field = field.slice(1, (field.length - 1)); // trim the " characters off the ends
								field = field.replace(doubleQuotes, '"');   // replace "" with "
							}
							listOfFields[j] = field;
							j += 1;
						}
					}
					arrayOfOutputRecords.push(listOfFields);
				}
			}
			
			// The first item of the array must be the header row with attribute names.
			this._attributes = arrayOfOutputRecords.shift();
			for(i = 0; i<this._attributes.length; i++){
				// Store the index of each attribute 
				this._attributeIndexes[this._attributes[i]] = i;
			}
			this._dataArray = arrayOfOutputRecords; //Array
		}
	},

	_splitLines: function(csvContent){
		//	summary:
		//		Function to split the CSV file contents into separate lines.
		//		Since line breaks can occur inside quotes, a Regexp didn't
		//		work as well.  A quick passover parse should be just as efficient.
		var split = [];
		var i;
		var line = "";
		var inQuotes = false;
		for(i = 0; i < csvContent.length; i++){
			var c = csvContent.charAt(i);
			switch(c){
				case '\"':
					inQuotes = !inQuotes;
					line += c;
					break;
				case '\r':
					if(inQuotes){
						line += c;
					}else{
						split.push(line);
						line = "";
						if (i < (csvContent.length - 1) && csvContent.charAt(i + 1) == '\n') {
							i++; //Skip it, it's CRLF
						}
					}
					break;
				case '\n':
					if(inQuotes){
						line += c;
					}else{
						split.push(line);
						line = "";
					}
					break;
				default:
					line +=c;
			}
		}
		if(line !== ""){
			split.push(line);
		}
		return split;
	},
	
	_processData: function(/* String */ data){
		this._getArrayOfArraysFromCsvFileContents(data);
		this._arrayOfAllItems = [];

		//Check that the specified Identifier is actually a column title, if provided.
		if(this.identifier){
			if(this._attributeIndexes[this.identifier] === undefined){
				throw new Error(this.declaredClass + ": Identity specified is not a column header in the data set.")
			}
		}

		for(var i=0; i<this._dataArray.length; i++){
			var id = i;
			//Associate the identifier to a row in this case
			//for o(1) lookup.
			if(this.identifier){
				var iData = this._dataArray[i];
				id = iData[this._attributeIndexes[this.identifier]];
				this._idMap[id] = i;
			}
			this._arrayOfAllItems.push(this._createItemFromIdentity(id));
		}
		this._loadFinished = true;
		this._loadInProgress = false;
	},
	
	_createItemFromIdentity: function(/* String */ identity){
		var item = {};
		item[this._storeProp] = this;
		item[this._idProp] = identity;
		return item; //Object
	},
	
	
/***************************************
     dojo.data.api.Identity API
***************************************/
	getIdentity: function(/* item */ item){
		//	summary: 
		//		See dojo.data.api.Identity.getIdentity()
		if(this.isItem(item)){
			return item[this._idProp]; //String
		}
		return null; //null
	},

	fetchItemByIdentity: function(/* Object */ keywordArgs){
		//	summary: 
		//		See dojo.data.api.Identity.fetchItemByIdentity()
		var item;
		var scope =  keywordArgs.scope?keywordArgs.scope:dojo.global;
		//Hasn't loaded yet, we have to trigger the load.
		if(!this._loadFinished){
			var self = this;
			if(this.url !== ""){
				//If fetches come in before the loading has finished, but while
				//a load is in progress, we have to defer the fetching to be 
				//invoked in the callback.
				if(this._loadInProgress){
					this._queuedFetches.push({args: keywordArgs});
				}else{
					this._loadInProgress = true;
					var getArgs = {
							url: self.url, 
							handleAs: "text"
						};
					var getHandler = dojo.xhrGet(getArgs);
					getHandler.addCallback(function(data){
						try{
							self._processData(data);
							var item = self._createItemFromIdentity(keywordArgs.identity);
							if(!self.isItem(item)){
								item = null;
							}
							if(keywordArgs.onItem){
								keywordArgs.onItem.call(scope, item);
							}
							self._handleQueuedFetches();
						}catch(error){
							if(keywordArgs.onError){
								keywordArgs.onError.call(scope, error);
							}
						}
					});
					getHandler.addErrback(function(error){
						this._loadInProgress = false;
						if(keywordArgs.onError){
							keywordArgs.onError.call(scope, error);
						}
					});
				}
			}else if(this._csvData){
				try{
					self._processData(self._csvData);
					self._csvData = null;
					item = self._createItemFromIdentity(keywordArgs.identity);
					if(!self.isItem(item)){
						item = null;
					}
					if(keywordArgs.onItem){
						keywordArgs.onItem.call(scope, item);
					}
				}catch(e){
					if(keywordArgs.onError){
						keywordArgs.onError.call(scope, e);
					}
				}
			}
		}else{
			//Already loaded.  We can just look it up and call back.
			item = this._createItemFromIdentity(keywordArgs.identity);
			if(!this.isItem(item)){
				item = null;
			}
			if(keywordArgs.onItem){
				keywordArgs.onItem.call(scope, item);
			}
		}
	},

	getIdentityAttributes: function(/* item */ item){
		//	summary: 
		//		See dojo.data.api.Identity.getIdentifierAttributes()
		 
		//Identity isn't a public attribute in the item, it's the row position index.
		//So, return null.
		if(this.identifier){
			return [this.identifier];
		}else{
			return null;
		}
	},

	_handleQueuedFetches: function(){
		//	summary: 
		//		Internal function to execute delayed request in the store.
		//Execute any deferred fetches now.
		if (this._queuedFetches.length > 0) {
			for(var i = 0; i < this._queuedFetches.length; i++){
				var fData = this._queuedFetches[i];
				var delayedFilter = fData.filter;
				var delayedQuery = fData.args;
				if(delayedFilter){
					delayedFilter(delayedQuery, this._arrayOfAllItems); 
				}else{
					this.fetchItemByIdentity(fData.args);
				}
			}
			this._queuedFetches = [];
		}
	}
});
//Mix in the simple fetch implementation to this class.
dojo.extend(dojox.data.CsvStore,dojo.data.util.simpleFetch);