From 496b9e7ddbbe1367ba21bd19c98e9f439b5cb904 Mon Sep 17 00:00:00 2001
From: Mithgol the Webmaster <getgit@mithgol.ru>
Date: Thu, 23 Apr 2015 00:28:33 +0300
Subject: [PATCH] =?UTF-8?q?dirty=C2=A0hacks=20for=20ashtuchkin/iconv-lite#?=
 =?UTF-8?q?73=20and=20ashtuchkin/iconv-lite#96?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 README.md | 13 +++++++---
 fiunis.js | 74 ++++++++++++++++++++++++++++++++++++++++++++++++-------
 2 files changed, 74 insertions(+), 13 deletions(-)

diff --git a/README.md b/README.md
index 18a04e2..796f4ba 100644
--- a/README.md
+++ b/README.md
@@ -34,13 +34,18 @@ Replaces all occurrences of Fidonet Unicode substrings in the given text by 
 
 For example, `.decode('The video “&+mAJcFlwNbHpOS3p/iTJbUHvH-;” is interesting.')` returns `'The video “頂尖對決之穿褲子篇” is interesting.'`.
 
-### encode(text)
+### encode(text, encoding)
 
-Returns the Fidonet Unicode substring equivalent of the given text.
+This method makes one or more Fidonet Unicode substrings from the given `text`. This method has the two different possible behaviours:
 
-For example, `.encode('頂尖對決之穿褲子篇')` returns `'&+mAJcFlwNbHpOS3p/iTJbUHvH-;'`.
+* If `encoding` is `undefined`, this method returns a JavaScript **string** with the Fidonet Unicode substring equivalent of the given `text`.
+   * For example, `.encode('頂尖對決之穿褲子篇')` returns `'&+mAJcFlwNbHpOS3p/iTJbUHvH-;'`.
+   * The whole given `text` is converted (including ASCII characters). You should either detect (beforehand) which substrings should be given to this encoder (and what other substrings would be left to a traditional 8-bit encoding) or use the other encoder that decides that automatically (i.e. also give some `encoding` to the method, see below).
 
-Any given text is converted (including ASCII characters). You should detect (beforehand) which substrings should be given to this encoder and what other substrings would be left to a traditional 8-bit encoding.
+* If `encoding` is not `undefined`, this method returns a Node.js **Buffer** with the given `text` converted to the given `encoding`.
+   * If some fragments of the given `text` consist of characters that cannot be represented in the given `encoding`, then Fidonet Unicode substrings are used to encode such fragments.
+   * The given `encoding` must be an encoding understood by the [`iconv-lite`](https://github.com/ashtuchkin/iconv-lite) module (otherwise an error is thrown).
+   * The given `encoding` should also be a single-byte encoding (that generates exactly one byte for each of the characters it can represent). Otherwise the behaviour of this method is not reliable. (An error is thrown only if a multi-byte encoding can be detected, i.e. if the Buffer's length is not equal to the text's length.)
 
 ## Testing Fiunis
 
diff --git a/fiunis.js b/fiunis.js
index 98b5ba4..ca29617 100644
--- a/fiunis.js
+++ b/fiunis.js
@@ -1,4 +1,5 @@
-require('iconv-lite').extendNodeEncodings();
+var iconvLite = require('iconv-lite');
+iconvLite.extendNodeEncodings();
 
 var Fiunis = function(){
    if (!(this instanceof Fiunis)) return new Fiunis();
@@ -14,15 +15,70 @@ Fiunis.prototype.decode = function(text){
    }).join('');
 };
 
-Fiunis.prototype.encode = function(text){
-   var base64string = Buffer(text, 'utf16be').toString('base64');
-   while(
-      base64string.length > 0 &&
-      base64string.charAt(base64string.length - 1) === '='
-   ){
-      base64string = base64string.slice(0, -1);
+Fiunis.prototype.encode = function(text, targetEncoding){
+   if( typeof targetEncoding === 'undefined' ){ // encode whole string
+      var base64string = Buffer(text, 'utf16be').toString('base64');
+      while(
+         base64string.length > 0 &&
+         base64string.charAt(base64string.length - 1) === '='
+      ){
+         base64string = base64string.slice(0, -1);
+      }
+      return '&+' + base64string + '-;';
+   }
+
+   // otherwise detect and render Fidonet Unicode substrings
+   if( !iconvLite.encodingExists(targetEncoding) ){
+      throw new Error(this.errors.UNKNOWN_ENCODING);
+   }
+   if( iconvLite.defaultCharSingleByte !== '?' ){
+      throw new Error(this.errors.ICONVLITE_TAINTED);
+   }
+
+   var primalBuffer = Buffer(text, targetEncoding);
+   if( primalBuffer.length !== text.length ){
+      throw new Error(this.errors.MULTIBYTE_ENCODING);
    }
-   return '&+' + base64string + '-;';
+   var zebra = primalBuffer.toString(targetEncoding).split( /(\?+)/ );
+   if( zebra.length < 2 ){ // zero non-encodable substrings
+      return primalBuffer;
+   } else primalBuffer = void 0;
+
+   var _here = this;
+   var remainingStr = text;
+   var collected = [];
+   zebra.forEach(function(zebraLine, IDX){
+      if( IDX % 2 === 0 ){ // encodable substring's index: 0, 2, 4...
+         collected = collected.concat([zebraLine]);
+         remainingStr = remainingStr.slice(zebraLine.length);
+      } else { // non-encodable substring's index: 1, 3, 5...
+         var srcZebra = remainingStr.slice(0, zebraLine.length).split(
+            /(\?+)/
+         ).map(function(srcLine, srcIDX){
+            if( srcIDX % 2 === 0 ){
+               // actual non-encodable substring's index: 0, 2, 4...
+               if( srcLine.length < 1 ) return srcLine;
+               return _here.encode(srcLine);
+            } else {
+               // encodable source defaultCharSingleByte index: 1, 3, 5...
+               return srcLine;
+            }
+         });
+         collected = collected.concat(srcZebra);
+         remainingStr = remainingStr.slice(zebraLine.length);
+      }
+   });
+   return Buffer(collected.join(''), targetEncoding);
+};
+
+Fiunis.prototype.errors = {
+   UNKNOWN_ENCODING: 'The given encoding is unknown!',
+   MULTIBYTE_ENCODING: 'The given encoding is not a single-byte encoding!',
+   ICONVLITE_TAINTED: [
+      'The issue ',
+      'https://github.com/ashtuchkin/iconv-lite/issues/96',
+      ' is possible!'
+   ].join('')
 };
 
 module.exports = new Fiunis();
\ No newline at end of file