Improved API and added examples

dinhoabreu · Mar 16, 2012 · 3d5d2c7 · 3d5d2c7
1 parent 5ace01f
commit 3d5d2c7
Show file tree

Hide file tree

Showing 8 changed files with 163 additions and 85 deletions.
diff --git a/README.md b/README.md
@@ -0,0 +1,49 @@
+# codes
+
+Node.js iconv bindings - Streamable, fast and lightweight
+
+## Installing with [npm](http://npmjs.org/)
+
+	npm install codes
+
+## Usage
+
+Encode from one character encoding to another:
+
+	var assert = require('assert');
+	var codes = require('../index');
+
+	// convert from UTF-8 to ISO-8859-1
+	var str = 'É isso ai. Testando acentuação: áéíóú...';
+	var buffer = codes.encode(str, 'ISO-8859-1//TRANSLIT');
+	var str2 = codes.decode(buffer, 'ISO-8859-1');
+	assert.equal(str, str2);
+
+	// stream from LATIN1 to UTF-8 
+	var cs = codes.createStream('UTF-8', 'LATIN1');
+	var str4 = '';
+	cs.on('data', function (data) {
+		var str3 = data.toString();
+		console.log(str3);
+		str4 += str3;
+	});
+	cs.on('end', function () {
+		assert.equal(str4, str);
+		console.log(str4);
+	});
+	cs.on('error', function (error) {
+		console.log(error);
+	});
+	cs.write(buffer.slice(0,1));
+	cs.write(buffer.slice(1,8));
+	cs.write(buffer.slice(8,16));
+	cs.write(buffer.slice(16,24));
+	cs.write(buffer.slice(24, 28));
+	cs.write(buffer.slice(28, 32));
+	cs.write(buffer.slice(32));
+	cs.end();
+
+	// convert from UTF-8 to ASCII
+	var str = 'aéióuçÁEÍOÚC';
+	var ascii = codes.convert(str, 'ASCII//TRANSLIT', 'UTF-8');
+	console.log(str + ' => ' + ascii.toString());
diff --git a/deps.sh b/deps.sh
@@ -19,7 +19,7 @@ fi
 
 cd deps/libiconv-1.14/
 if [ ! -f Makefile ]; then
-	./configure --prefix=$PREFIX --enable-static
+	./configure --prefix=$PREFIX --enable-static --enable-extra-encodings
 fi;
 make -s && make -s install
 cd -

diff --git a/examples/ex01.js b/examples/ex01.js
@@ -0,0 +1,37 @@
+var assert = require('assert');
+var codes = require('../index');
+
+// convert from UTF-8 to ISO-8859-1
+var str = 'É isso ai. Testando acentuação: áéíóú...';
+var buffer = codes.encode(str, 'ISO-8859-1//TRANSLIT');
+var str2 = codes.decode(buffer, 'ISO-8859-1');
+assert.equal(str, str2);
+
+// stream from LATIN1 to UTF-8 
+var cs = codes.createStream('UTF-8', 'LATIN1');
+var str4 = '';
+cs.on('data', function (data) {
+	var str3 = data.toString();
+	console.log(str3);
+	str4 += str3;
+});
+cs.on('end', function () {
+	assert.equal(str4, str);
+	console.log(str4);
+});
+cs.on('error', function (error) {
+	console.log(error);
+});
+cs.write(buffer.slice(0,1));
+cs.write(buffer.slice(1,8));
+cs.write(buffer.slice(8,16));
+cs.write(buffer.slice(16,24));
+cs.write(buffer.slice(24, 28));
+cs.write(buffer.slice(28, 32));
+cs.write(buffer.slice(32));
+cs.end();
+
+// convert from UTF-8 to ASCII
+var str = 'aéióuçÁEÍOÚC';
+var ascii = codes.convert(str, 'ASCII//TRANSLIT', 'UTF-8');
+console.log(str + ' => ' + ascii.toString());
diff --git a/lib/index.js → index.js b/lib/index.js → index.js
@@ -1,23 +1,21 @@
-var bind = require('../build/Release/iconv_binding');
+var bind = require('./build/Release/iconv');
 var stream = require('stream');
 var util = require('util');
 
-function create(options) {
-	var from = options.from || 'UTF-8',
-			to = options.to,
-			size = options.size || 8192,
+function create(to, from, options) {
+	var from = from || 'UTF-8',
+			to = to.split('//'),
+			size = options && options.size || 8192,
 			iconv;
 	from = bind.canonicalize(from);
-	to = bind.canonicalize(to);
-	if (options.option)
-		to += options.option;
+	to[0] = bind.canonicalize(to[0]);
+	to = to.join('//');
 	iconv = new bind.Iconv(to, from);
 	return {from: from, to: to, size: size, iconv: iconv};
 }
 
-function IconvStream(options) {
-	var h = create(options);
-	this.options = options;
+function CodesStream(to, from, options) {
+	var h = create(to, from, options);
 	this._iconv = h.iconv;
 	this._input = new Buffer(h.size);
 	this._input.start = 0;
@@ -28,15 +26,15 @@ function IconvStream(options) {
 	this.readable = true;
 	this.writable = true;
 }
-util.inherits(IconvStream, stream.Stream);
+util.inherits(CodesStream, stream.Stream);
 
-IconvStream.prototype.write = function (data) {
+CodesStream.prototype.write = function (data) {
 	if (!this.writable) {
 		this.emit('error', new Error('stream not writable'))	
 		return false;
 	}
   if (!Buffer.isBuffer(data)) {
-    var encoding = 'utf8';
+    var encoding = 'UTF-8';
     if (typeof(arguments[1]) == 'string') encoding = arguments[1];
     data = new Buffer('' + data, encoding);
   }
@@ -54,7 +52,7 @@ IconvStream.prototype.write = function (data) {
 	return true;
 }
 
-IconvStream.prototype.end = function (data) {
+CodesStream.prototype.end = function (data) {
 	if (data)
 		this.write.apply(this, arguments);
 	if (this.writable) {
@@ -65,7 +63,7 @@ IconvStream.prototype.end = function (data) {
 	this.writable = false;
 }
 
-IconvStream.prototype._write = function (chunk) {
+CodesStream.prototype._write = function (chunk) {
 	var input = this._input;
 	chunk.stop = false;
 	while (chunk.length > 0 && !chunk.stop)
@@ -78,14 +76,13 @@ IconvStream.prototype._write = function (chunk) {
 	}
 }
 
-IconvStream.prototype._convert = function (chunk) {
+CodesStream.prototype._convert = function (chunk) {
 	var iconv = this._iconv;
 	var output = this._output;
 	var r = iconv.convert(chunk, output);
-	var chunkEnd = chunk.length - r.leftBytesIn;
-	var outputEnd = output.length - r.leftBytesOut;
-	var chunk = chunk.slice(chunkEnd);
-	var data = output.slice(0, outputEnd);
+	var offsetIn = r.offsetIn;
+	var chunk = chunk.slice(offsetIn);
+	var data = output.slice(0, r.offsetOut);
 	chunk.stop = true;
 	if (data.length)
 		this.emit('data', data);
@@ -98,7 +95,7 @@ IconvStream.prototype._convert = function (chunk) {
 		case 'EILSEQ':
 			var error = new Error(r.error);
 			error.code = r.errno;
-			chunk = chunk.slice(chunkEnd + 1);
+			chunk = chunk.slice(offsetIn + 1);
 			this.emit('error', error);
 			return chunk.slice(0, 0);
 			break;
@@ -112,18 +109,25 @@ IconvStream.prototype._convert = function (chunk) {
 	return chunk;
 }
 
-exports.createStream = function (options) {
-	return new IconvStream(options);
+exports.CodesStream = CodesStream;
+exports.createStream = function (to, from, options) {
+	return new CodesStream(to, from, options);
 }
-exports.convert = function (input, options) {
+exports.convert = function (input, to, from, options) {
 	if (typeof input == 'string')
-		input = new Buffer(input, 'utf8');
-	var h = create(options);
-	var output = new Buffer(input.length * 2);
+		input = new Buffer(input, 'UTF-8');
+	var h = create(to, from, options);
+	var output = new Buffer(input.length * 4);
 	var r = h.iconv.convert(input, output);
-	var outputEnd = output.length - r.leftBytesOut;
-	return output.slice(0, outputEnd);
+	return output.slice(0, r.offsetOut);
+}
+exports.encode = function (input, to, options) {
+	return this.convert(input, to, 'UTF-8', options);
+}
+exports.decode = function (input, from, options) {
+	return this.convert(input, 'UTF-8', from, options).toString();
 }
 exports.encodings = bind.encodings;
+exports.canonicalize = bind.canonicalize;
 exports.TRANSLIT = '//TRANSLIT';
 exports.IGNORE = '//IGNORE';
diff --git a/package.json b/package.json
@@ -2,12 +2,14 @@
   "author": "Edison E. Abreu <dinho.abreu@gmail.com>",
   "name": "codes",
   "description": "Node.js iconv bindings - Streamable, fast and lightweight",
-  "version": "0.0.2",
+	"keywords": "iconv charset encoding encode decode convert code codes stream fast ligthweight",
+	"homepage": "https://github.com/dinhoabreu/codes",
+  "version": "0.1.0",
   "repository": {
     "type": "git",
     "url": "git@github.com:dinhoabreu/codes.git"
   },
-  "main": "lib/index.js",
+  "main": "index.js",
 	"scripts": {
 		"prepublish": "node-waf distclean && rm -rf libiconv",
 		"preinstall": "./deps.sh",
@@ -20,6 +22,6 @@
 	},
   "optionalDependencies": {},
   "engines": {
-    "node": "0.6.x"
+    "node": ">=0.6.x"
   }
 }
diff --git a/src/iconv_binding.cc → src/iconv.cc b/src/iconv_binding.cc → src/iconv.cc
@@ -57,15 +57,15 @@ class Iconv {
 		static void Initialize(Handle<Object> target) {
 			HandleScope scope;
 
-  			push_sym = Persistent<String>::New(String::NewSymbol("push"));
-  			touppercase_sym = Persistent<String>::New(String::NewSymbol("toUpperCase"));
-  			convert_sym = Persistent<String>::New(String::NewSymbol("convert"));
-  			iconv_sym = Persistent<String>::New(String::NewSymbol("Iconv"));
-  			canonicalize_sym = Persistent<String>::New(String::NewSymbol("canonicalize"));
-  			encodings_sym = Persistent<String>::New(String::NewSymbol("encodings"));
-
-			leftBytesIn_sym = Persistent<String>::New(String::NewSymbol("leftBytesIn"));
-			leftBytesOut_sym = Persistent<String>::New(String::NewSymbol("leftBytesOut"));
+  		push_sym = Persistent<String>::New(String::NewSymbol("push"));
+  		touppercase_sym = Persistent<String>::New(String::NewSymbol("toUpperCase"));
+  		convert_sym = Persistent<String>::New(String::NewSymbol("convert"));
+  		iconv_sym = Persistent<String>::New(String::NewSymbol("Iconv"));
+  		canonicalize_sym = Persistent<String>::New(String::NewSymbol("canonicalize"));
+  		encodings_sym = Persistent<String>::New(String::NewSymbol("encodings"));
+
+			offsetIn_sym = Persistent<String>::New(String::NewSymbol("offsetIn"));
+			offsetOut_sym = Persistent<String>::New(String::NewSymbol("offsetOut"));
 			code_sym = Persistent<String>::New(String::NewSymbol("code"));
 			errno_sym = Persistent<String>::New(String::NewSymbol("errno"));
 			error_sym = Persistent<String>::New(String::NewSymbol("error"));
@@ -160,13 +160,15 @@ class Iconv {
 			Local<Object> target_obj = args[1]->ToObject();
 			char *source_data = Buffer::Data(source_obj);
 			size_t source_length = Buffer::Length(source_obj);
+			size_t source_left = source_length;
 			char *target_data = Buffer::Data(target_obj);
 			size_t target_length = Buffer::Length(target_obj);
-			size_t r = ic->convert(&source_data, &source_length, &target_data, &target_length);
+			size_t target_left = target_length;
+			size_t r = ic->convert(&source_data, &source_left, &target_data, &target_left);
 
 			Local<Object> obj = Object::New();
-			obj->Set(leftBytesIn_sym, Integer::NewFromUnsigned(source_length));
-			obj->Set(leftBytesOut_sym, Integer::NewFromUnsigned(target_length));
+			obj->Set(offsetIn_sym, Integer::NewFromUnsigned(source_length - source_left));
+			obj->Set(offsetOut_sym, Integer::NewFromUnsigned(target_length - target_left));
 			obj->Set(code_sym, Integer::New(r));
 			if (r == -1) {
 				obj->Set(error_sym, String::New(strerror(errno)));
@@ -184,8 +186,8 @@ class Iconv {
 			return scope.Close(obj);
 		}
 		static Persistent<FunctionTemplate> constructor_template;
-		static Persistent<String> leftBytesIn_sym;
-		static Persistent<String> leftBytesOut_sym;
+		static Persistent<String> offsetIn_sym;
+		static Persistent<String> offsetOut_sym;
 		static Persistent<String> code_sym;
 		static Persistent<String> errno_sym;
 		static Persistent<String> error_sym;
@@ -194,13 +196,13 @@ class Iconv {
 };
 
 Persistent<FunctionTemplate> Iconv::constructor_template;
-Persistent<String> Iconv::leftBytesIn_sym;
-Persistent<String> Iconv::leftBytesOut_sym;
+Persistent<String> Iconv::offsetIn_sym;
+Persistent<String> Iconv::offsetOut_sym;
 Persistent<String> Iconv::code_sym;
 Persistent<String> Iconv::errno_sym;
 Persistent<String> Iconv::error_sym;
 
 void init(Handle<Object> target) {
 	Iconv::Initialize(target);
 }
-NODE_MODULE(iconv_binding, init)
+NODE_MODULE(iconv, init)