Date: Loose Matching

Fixes #292

src/build/intro-date-runtime.js

@@ -42,6 +42,7 @@
 "use strict";
 
 var createErrorUnsupportedFeature = Globalize._createErrorUnsupportedFeature,
+	looseMatching = Globalize._looseMatching,
 	regexpEscape = Globalize._regexpEscape,
 	removeLiteralQuotes = Globalize._removeLiteralQuotes,
 	runtimeKey = Globalize._runtimeKey,

src/build/intro-date.js

@@ -40,6 +40,8 @@
 var createError = Globalize._createError,
 	createErrorUnsupportedFeature = Globalize._createErrorUnsupportedFeature,
 	formatMessage = Globalize._formatMessage,
+	isPlainObject = Globalize._isPlainObject,
+	looseMatching = Globalize._looseMatching,
 	numberSymbol = Globalize._numberSymbol,
 	regexpEscape = Globalize._regexpEscape,
 	removeLiteralQuotes = Globalize._removeLiteralQuotes,

src/date/tokenizer-properties.js

@@ -6,9 +6,12 @@ define([
 	"../common/create-error/unsupported-feature",
 	"../common/format-message",
 	"../number/symbol",
+	"../util/is-plain-object",
+	"../util/loose-matching",
 	"../util/object/filter"
 ], function( dateGetTimeZoneName, datePatternRe, dateTimezoneHourFormatH, dateTimezoneHourFormatHm,
-	createErrorUnsupportedFeature, formatMessage, numberSymbol, objectFilter ) {
+	createErrorUnsupportedFeature, formatMessage, numberSymbol, isPlainObject, looseMatching,
+	objectFilter ) {
 
 /**
  * tokenizerProperties( pattern, cldr )
@@ -21,7 +24,7 @@ define([
  */
 return function( pattern, cldr, timeZone ) {
 	var properties = {
-			pattern: pattern,
+			pattern: looseMatching( pattern ),
 			timeSeparator: numberSymbol( "timeSeparator", cldr )
 		},
 		widths = [ "abbreviated", "wide", "narrow" ];
@@ -34,8 +37,29 @@ return function( pattern, cldr, timeZone ) {
 			return;
 		}
 
+		if ( !value ) {
+			return;
+		}
+
 		// The `dates` and `calendars` trim's purpose is to reduce properties' key size only.
-		properties[ path.replace( /^.*\/dates\//, "" ).replace( /calendars\//, "" ) ] = value;
+		path = path.replace( /^.*\/dates\//, "" ).replace( /calendars\//, "" );
+
+		// Specific filter for "gregorian/dayPeriods/format/wide".
+		if ( path === "gregorian/dayPeriods/format/wide" ) {
+			value = objectFilter( value, /^am|^pm/ );
+		}
+
+		// Transform object into array of pairs [key, /value/].
+		if ( isPlainObject( value ) ) {
+			value = Object.keys( value ).map(function( key ) {
+				return [ key, new RegExp( "^" + looseMatching( value[ key ] ) ) ];
+			});
+
+		// If typeof value === "string".
+		} else {
+			value = looseMatching( value );
+		}
+		properties[ path ] = value;
 	}
 
 	cldr.on( "get", populateProperties );
@@ -59,10 +83,10 @@ return function( pattern, cldr, timeZone ) {
 			standardTzName = dateGetTimeZoneName( length, "standard", timeZone, cldr );
 			daylightTzName = dateGetTimeZoneName( length, "daylight", timeZone, cldr );
 			if ( standardTzName ) {
-				properties.standardTzName = standardTzName;
+				properties.standardTzName = looseMatching( standardTzName );
 			}
 			if ( daylightTzName ) {
-				properties.daylightTzName = daylightTzName;
+				properties.daylightTzName = looseMatching( daylightTzName );
 			}
 
 			// Fall through the "O" format in case one name is missing.
@@ -86,7 +110,7 @@ return function( pattern, cldr, timeZone ) {
 			}
 			var genericTzName = dateGetTimeZoneName( length, "generic", timeZone, cldr );
 			if ( genericTzName ) {
-				properties.genericTzName = genericTzName;
+				properties.genericTzName = looseMatching( genericTzName );
 				chr = "O";
 
 			// Fall back to "V" format.
@@ -186,10 +210,6 @@ return function( pattern, cldr, timeZone ) {
 				cldr.main(
 					"dates/calendars/gregorian/dayPeriods/format/wide"
 				);
-				properties[ "gregorian/dayPeriods/format/wide" ] = objectFilter(
-					properties[ "gregorian/dayPeriods/format/wide" ],
-					/^am|^pm/
-				);
 				break;
 
 			// Zone
@@ -203,6 +223,8 @@ return function( pattern, cldr, timeZone ) {
 
 				if ( timeZone ) {
 					if ( length === 2 ) {
+
+						// Skip looseMatching processing since timeZone is a canonical posix value.
 						properties.timeZoneName = timeZone;
 						break;
 					}
@@ -231,7 +253,7 @@ return function( pattern, cldr, timeZone ) {
 					}
 
 					if ( timeZoneName ) {
-						properties.timeZoneName = timeZoneName;
+						properties.timeZoneName = looseMatching( timeZoneName );
 					}
 				}
 
@@ -244,7 +266,8 @@ return function( pattern, cldr, timeZone ) {
 			case "O":
 				cldr.main( "dates/timeZoneNames/gmtFormat" );
 				cldr.main( "dates/timeZoneNames/gmtZeroFormat" );
-				aux = cldr.main( "dates/timeZoneNames/hourFormat" );
+				cldr.main( "dates/timeZoneNames/hourFormat" );
+				aux = properties[ "timeZoneNames/hourFormat" ];
 				properties[ "timeZoneNames/hourFormat" ] = length < 4 ?
 					[ dateTimezoneHourFormatH( aux ), dateTimezoneHourFormatHm( aux, "H" ) ] :
 					[ dateTimezoneHourFormatHm( aux, "HH" ) ];

src/date/tokenizer.js

@@ -1,15 +1,18 @@
 define([
 	"./pattern-re",
+	"../util/loose-matching",
 	"../util/regexp/escape",
 	"../util/regexp/n",
 	"../util/remove-literal-quotes"
-], function( datePatternRe, regexpEscape, regexpN, removeLiteralQuotes ) {
+], function( datePatternRe, looseMatching, regexpEscape, regexpN, removeLiteralQuotes ) {
 
 /**
- * tokenizer( value, pattern, properties )
+ * tokenizer( value, numberParser, properties )
  *
  * @value [String] string date.
  *
+ * @numberParser [Function]
+ *
  * @properties [Object] output returned by date/tokenizer-properties.
  *
  * Returns an Array of tokens, eg. value "5 o'clock PM", pattern "h 'o''clock' a":
@@ -50,6 +53,8 @@ return function( value, numberParser, properties ) {
 		}) + ")";
 	}
 
+	value = looseMatching( value );
+
 	valid = properties.pattern.match( datePatternRe ).every(function( current ) {
 		var aux, chr, length, numeric, tokenRe,
 			token = {};
@@ -153,16 +158,21 @@ return function( value, numberParser, properties ) {
 		// Brute-force test every locale entry in an attempt to match the given value.
 		// Return the first found one (and set token accordingly), or null.
 		function lookup( path ) {
-			var i, re,
-				data = properties[ path.join( "/" ) ];
-
-			for ( i in data ) {
-				re = new RegExp( "^" + data[ i ] );
-				if ( re.test( value ) ) {
-					token.value = i;
-					return tokenRe = new RegExp( data[ i ] );
-				}
+			var array = properties[ path.join( "/" ) ];
+
+			if ( !array ) {
+				return null;
 			}
+
+			// array of pairs [key, value].
+			array.some(function( item ) {
+				var valueRe = item[ 1 ];
+				if ( valueRe.test( value ) ) {
+					token.value = item[ 0 ];
+					tokenRe = item[ 1 ];
+					return true;
+				}
+			});
 			return null;
 		}
 

src/number-runtime.js

@@ -9,13 +9,16 @@ define([
 	"./number/formatter-fn",
 	"./number/parse",
 	"./number/parser-fn",
+	"./util/loose-matching",
 	"./util/number/round",
 	"./util/remove-literal-quotes"
 ], function( runtimeKey, createErrorUnsupportedFeature, validateParameterPresence,
 	validateParameterTypeNumber, validateParameterTypeString, Globalize, numberFormat,
-	numberFormatterFn, numberParse, numberParserFn, numberRound, removeLiteralQuotes ) {
+	numberFormatterFn, numberParse, numberParserFn, looseMatching, numberRound,
+	removeLiteralQuotes ) {
 
 Globalize._createErrorUnsupportedFeature = createErrorUnsupportedFeature;
+Globalize._looseMatching = looseMatching;
 Globalize._numberFormat = numberFormat;
 Globalize._numberFormatterFn = numberFormatterFn;
 Globalize._numberParse = numberParse;

src/number.js

@@ -16,6 +16,7 @@ define([
 	"./number/parse-properties",
 	"./number/pattern",
 	"./number/symbol",
+	"./util/loose-matching",
 	"./util/remove-literal-quotes",
 	"./util/string/pad",
 
@@ -25,7 +26,8 @@ define([
 	validateDefaultLocale, validateParameterPresence, validateParameterRange,
 	validateParameterTypeNumber, validateParameterTypePlainObject, validateParameterTypeString,
 	numberFormatterFn, numberFormatProperties, numberNumberingSystem, numberParserFn,
-	numberParseProperties, numberPattern, numberSymbol, removeLiteralQuotes, stringPad ) {
+	numberParseProperties, numberPattern, numberSymbol, looseMatching, removeLiteralQuotes,
+	stringPad ) {
 
 function validateDigits( properties ) {
 	var minimumIntegerDigits = properties[ 2 ],
@@ -178,6 +180,7 @@ Globalize._createErrorUnsupportedFeature = createErrorUnsupportedFeature;
 Globalize._numberNumberingSystem = numberNumberingSystem;
 Globalize._numberPattern = numberPattern;
 Globalize._numberSymbol = numberSymbol;
+Globalize._looseMatching = looseMatching;
 Globalize._removeLiteralQuotes = removeLiteralQuotes;
 Globalize._stringPad = stringPad;
 Globalize._validateParameterTypeNumber = validateParameterTypeNumber;

src/number/parse-properties.js

@@ -2,14 +2,12 @@ define([
 	"./format-properties",
 	"./symbol/inverted-map",
 	"../util/always-array",
+	"../util/loose-matching",
 	"../util/object/map",
-	"../util/regexp/cf-g",
-	"../util/regexp/dash-g",
 	"../util/regexp/escape",
-	"../util/regexp/zs-g",
 	"../util/remove-literal-quotes"
-], function( numberFormatProperties, numberSymbolInvertedMap, alwaysArray, objectMap, regexpCfG,
-	regexpDashG, regexpEscape, regexpZsG, removeLiteralQuotes ) {
+], function( numberFormatProperties, numberSymbolInvertedMap, alwaysArray, looseMatching, objectMap,
+	regexpEscape, removeLiteralQuotes ) {
 
 /**
  * parseProperties( pattern, cldr )
@@ -31,18 +29,6 @@ return function( pattern, cldr, options ) {
 		numberTokenizer, prefix, primaryGroupingSize, secondaryGroupingSize, suffix, symbolMap,
 		formatProperties = numberFormatProperties( pattern, cldr, options );
 
-	// Loose Matching:
-	// - Ignore all format characters, which includes RLM, LRM or ALM used to control BIDI
-	//   formatting.
-	// - Map all characters in [:Zs:] to U+0020 SPACE;
-	// - Map all characters in [:Dash:] to U+002D HYPHEN-MINUS;
-	function looseMatching( value ) {
-		return value
-			.replace( regexpCfG, "" )
-			.replace( regexpDashG, "-" )
-			.replace( regexpZsG, " " );
-	}
-
 	prefix = looseMatching( formatProperties[ 0 ] );
 	maximumFractionDigits = formatProperties[ 4 ];
 	minimumSignificantDigits = formatProperties[ 5 ];

src/number/parse.js

@@ -1,8 +1,6 @@
 define([
-	"../util/regexp/cf-g",
-	"../util/regexp/dash-g",
-	"../util/regexp/zs-g"
-], function( regexpCfG, regexpDashG, regexpZsG ) {
+	"../util/loose-matching"
+], function( looseMatching ) {
 
 /**
  * parse( value, properties )
@@ -36,15 +34,7 @@ return function( value, properties ) {
 	invertedNuDigitsMap = properties[ 1 ] || {};
 	tokenizer = properties[ 2 ];
 
-	// Loose Matching:
-	// - Ignore all format characters, which includes RLM, LRM or ALM used to control BIDI
-	//   formatting.
-	// - Map all characters in [:Zs:] to U+0020 SPACE;
-	// - Map all characters in [:Dash:] to U+002D HYPHEN-MINUS;
-	value = value
-		.replace( regexpCfG, "" )
-		.replace( regexpDashG, "-" )
-		.replace( regexpZsG, " " );
+	value = looseMatching( value );
 
 	function parse( type ) {
 		return function( lexeme ) {

src/util/loose-matching.js

@@ -0,0 +1,21 @@
+define([
+	"./regexp/cf-g",
+	"./regexp/dash-g",
+	"./regexp/zs-g"
+], function( regexpCfG, regexpDashG, regexpZsG ) {
+
+/**
+ * Loose Matching:
+ * - Ignore all format characters, which includes RLM, LRM or ALM used to control BIDI
+ *   formatting.
+ * - Map all characters in [:Zs:] to U+0020 SPACE;
+ * - Map all characters in [:Dash:] to U+002D HYPHEN-MINUS;
+ */
+return function( value ) {
+	return value
+		.replace( regexpCfG, "" )
+		.replace( regexpDashG, "-" )
+		.replace( regexpZsG, " " );
+};
+
+});

test/functional/date/parse-date.js

@@ -150,6 +150,9 @@ QUnit.test( "should parse skeleton", function( assert ) {
 	assertParseDate( assert, "9/15/2010", { skeleton: "yMd" }, date );
 	assertParseDate( assert, "الأربعاء، ١٥ سبتمبر، ٢٠١٠ م", { skeleton: "GyMMMEd" }, date, ar );
 
+	// Loose matching: ignore control characters.
+	assertParseDate( assert, "١٥/٩/٢٠١٠", { skeleton: "yMd" }, date, ar );
+
 	date = new Date( 2010, 0 );
 	date = startOf( date, "year" );
 	assertParseDate( assert, "Q3 2010", { skeleton: "yQQQ" }, date );