almost finished first draft, untested

Author: kevb34ns

README.md

@@ -8,4 +8,7 @@ The function `pinyinify()` takes a string parameter, which can be one of the fol
 + A pinyin word or phrase using tone numbers. Examples: `mao1`, `chi1fan4`, `wo3 men5`. For simplicity's sake, the function will process most words with tone numbers, including ones that may not be valid pinyin.
 + A string containing both pinyin words and non-pinyin words, characters, or symbols. Example: `My Chinese name is yang2kai3wen2.` **Be careful:** any word that resembles the structure of pinyin (has a number at the end of it, for example) may be given tone marks. Therefore, it is recommended that you avoid this option if possible.
 
-If you pass a string that meets one of the requirements above, the function will return a string with the tone numbers converted to tone marks. Otherwise, it will return an unchanged string.
+If you pass a string that meets one of the requirements above, the function will return a string with the tone numbers converted to tone marks. Otherwise, it will return an unchanged string.
+
+## Limitations
+<!-- TODO lowercase only, incomplete validation, etc.-->

pinyinify.js

@@ -1,3 +1,11 @@
+var toneMarks = {
+	a: ["a", "\u0101", "\u00e1", "\u01ce", "\u00e0", "a"],
+	e: ["e", "\u0113", "\u00e9", "\u011b", "\u00e8", "e"],
+	i: ["i", "\u012b", "\u00ed", "\u01d0", "\u00ec", "i"],
+	o: ["o", "\u014d", "\u00f3", "\u01d2", "\u00f2", "o"],
+	u: ["u", "\u016b", "\u00fa", "\u01d4", "\u00f9", "u"],
+	v: ["\u00fc", "\u01d6", "\u01d8", "\u01da", "\u01dc", "\u00fc"]
+};
 
 String.prototype.isAlpha = function() {
 
@@ -9,18 +17,64 @@ String.prototype.isPinyinVowel = function() {
 	return /^[aeiouv\u00fc]$/.test(this);
 }
 
+String.prototype.lastIndexOfRegex = function(regExp) {
+
+	var lastIndex = -1;
+	for (var i = 0; i < this.length; i++) {
+
+		if (regExp.test(this.charAt(i))) {
+
+			lastIndex = i;
+		}
+	}
+
+	return lastIndex;
+}
+
+String.prototype.replaceAt = function(index, replacement) {
+
+	//TODO
+}
+
 /**
- * Unicode tone marks (in order)
- * a: u0101 u00e1 u01ce u00e0
- * e: u0113 u00e9 u011b u00e8
- * i: u012b u00ed u01d0 u00ec
- * o: u014d u00f3 u01d2 u00f2
- * u: u016b u00fa u01d4 u00f9
- * v: u01d6 u01d8 u01da u01dc
+ * Takes a single pinyin word using tone numbers and converts to tone symbols.
  */
 String.prototype.convertPinyin = function() {
 
-	//TODO
+	var str = this.toLocaleLowerCase();
+
+	var toneNumIndex = str.search(/[1-5]/);
+	var firstVowelIndex = str.search(/[aeiouv\u00fc]/);
+	if (str.length > 7 || toneNumIndex < 1 || 
+	    toneNumIndex !== str.length - 1 ||
+	    firstVowelIndex < 0) {
+
+		console.log("String.prototype.convertPinyin:" + this + 
+		            " is not a valid pinyin word.")
+		return this;
+	}
+
+	var toneNum = parseInt(str[toneNumIndex]);
+	if (/[ae]/.test(str)) {
+
+		var index = str.search(/[ae]/);
+		str = str.replaceAt(index, toneMarks[str.charAt(index)][toneNum]);
+	} else if (/ou/.test(str)) {
+
+		var index = str.search(/ou/);
+		str = str.replaceAt(index, toneMarks[str.charAt(index)][toneNum]);
+	} else {
+
+		var index = str.lastIndexOfRegex(/[aeiouv\u00fc]/);
+		var vowel = str.charAt(index);
+		if (vowel == "\u00fc") {
+
+			vowel = "v";
+		}
+		str = str.replaceAt(index, toneMarks[vowel][toneNum]);
+	}
+
+	return str;
 }
 
 var pinyinify = function(str) {
@@ -37,14 +91,17 @@ var pinyinify = function(str) {
 		var char = str.charAt(i);
 		if (char.isAlpha()) {
 
-			res += str.substring(0, i);
-			str = str.substring(i);
-			i = 0;
+			if (i !== 0) {
+
+				res += str.substring(0, i);
+				str = str.substring(i);
+				i = 0;
+			}
 			var toneNumIndex = str.search(/[1-5]/);
 			if (toneNumIndex > 0 && toneNumIndex < 7) {
+
 				res += str.substring(0, toneNumIndex + 1).convertPinyin();
 				str = str.substring(toneNumIndex + 1);
-
 			} else {
 
 				var whitespaceIndex = str.search(/\s/);

test.html

@@ -4,7 +4,7 @@
 	<script>
 	var submit = function(str) {
 		// document.getElementById('result').innerHTML = pinyinify(str);
-		document.getElementById('result').innerHTML = isAlpha(str);
+		document.getElementById('result').innerHTML = /ou/.test(str);
 	}
 	</script>
 	<script src="pinyinify.js"></script>