fix wrong pairing of kanji and phonetic notation (reported in #10)

hexenq · Jun 5, 2017 · 9d8ed24 · 9d8ed24
1 parent 972492d
commit 9d8ed24
Show file tree

Hide file tree

Showing 7 changed files with 67 additions and 27 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,3 +1,10 @@
+<a name="0.1.5"></a>
+## [0.1.5](https://github.com/hexenq/kuroshiro.js/compare/0.1.4...0.1.5) (2017-06-05)
+
+### Bug Fixes
+
+* fix wrong pairing of kanji and phonetic notation (reported in [#10](https://github.com/hexenq/kuroshiro.js/issues/10))
+
 <a name="0.1.4"></a>
 ## [0.1.4](https://github.com/hexenq/kuroshiro.js/compare/0.1.3...0.1.4) (2017-05-25)
 

diff --git a/dist/browser/kuroshiro.js b/dist/browser/kuroshiro.js
@@ -9682,23 +9682,32 @@ var convert = function(str, options){
                     break;
                 case 1:
                     var pattern = '';
+                    var isLastTokenKanji = false;
+                    var subs = []; // recognize kanjis and group them
                     for(var c=0;c<tokens[i].surface_form.length;c++){
                         if(isKanji(tokens[i].surface_form[c])){
-                            pattern += '(.*)';
+                            if(!isLastTokenKanji){   // ignore successive kanji tokens (#10)
+                                isLastTokenKanji = true;
+                                pattern += '(.*?)';
+                                subs.push(tokens[i].surface_form[c]);
+                            }else{
+                                subs[subs.length-1] += tokens[i].surface_form[c];
+                            }
                         }else{
+                            isLastTokenKanji = false;
+                            subs.push(tokens[i].surface_form[c]);
                             pattern += wanakana.isKatakana(tokens[i].surface_form[c]) ? wanakana.toHiragana(tokens[i].surface_form[c]):tokens[i].surface_form[c];
                         }
                     }
-                    var reg = new RegExp(pattern);
+                    var reg = new RegExp('^' + pattern + '$');
                     var matches = reg.exec(tokens[i].reading);
                     if(matches){
-                        var pickKanji = 0;
-                        for(var c1=0;c1<tokens[i].surface_form.length;c1++){
-                            if(isKanji(tokens[i].surface_form[c1])){
-                                notations.push([tokens[i].surface_form[c1],1,matches[pickKanji+1]]);
-                                pickKanji++;
+                        var pickKanji = 1;
+                        for(var c1=0;c1<subs.length;c1++){
+                            if(isKanji(subs[c1][0])){
+                                notations.push([subs[c1],1,matches[pickKanji++]]);
                             }else{
-                                notations.push([tokens[i].surface_form[c1],2,wanakana.toHiragana(tokens[i].surface_form[c1])]);
+                                notations.push([subs[c1],2,wanakana.toHiragana(subs[c1])]);
                             }
                         }
                     }else{

diff --git a/dist/browser/kuroshiro.min.js b/dist/browser/kuroshiro.min.js
diff --git a/dist/node/kuroshiro.js b/dist/node/kuroshiro.js
diff --git a/package.json b/package.json
@@ -1,6 +1,6 @@
 {
   "name": "kuroshiro",
-  "version": "0.1.4",
+  "version": "0.1.5",
   "description": "kuroshiro.js is a japanese language utility mainly for converting Kanji-mixed sentence to Hiragana, Katakana or Romaji with furigana and okurigana modes supported.",
   "main": "./dist/node/kuroshiro.js",
   "scripts": {

diff --git a/src/kuroshiro.js b/src/kuroshiro.js
@@ -193,23 +193,32 @@ var convert = function(str, options){
                     break;
                 case 1:
                     var pattern = '';
+                    var isLastTokenKanji = false;
+                    var subs = []; // recognize kanjis and group them
                     for(var c=0;c<tokens[i].surface_form.length;c++){
                         if(isKanji(tokens[i].surface_form[c])){
-                            pattern += '(.*)';
+                            if(!isLastTokenKanji){   // ignore successive kanji tokens (#10)
+                                isLastTokenKanji = true;
+                                pattern += '(.*?)';
+                                subs.push(tokens[i].surface_form[c]);
+                            }else{
+                                subs[subs.length-1] += tokens[i].surface_form[c];
+                            }
                         }else{
+                            isLastTokenKanji = false;
+                            subs.push(tokens[i].surface_form[c]);
                             pattern += wanakana.isKatakana(tokens[i].surface_form[c]) ? wanakana.toHiragana(tokens[i].surface_form[c]):tokens[i].surface_form[c];
                         }
                     }
-                    var reg = new RegExp(pattern);
+                    var reg = new RegExp('^' + pattern + '$');
                     var matches = reg.exec(tokens[i].reading);
                     if(matches){
-                        var pickKanji = 0;
-                        for(var c1=0;c1<tokens[i].surface_form.length;c1++){
-                            if(isKanji(tokens[i].surface_form[c1])){
-                                notations.push([tokens[i].surface_form[c1],1,matches[pickKanji+1]]);
-                                pickKanji++;
+                        var pickKanji = 1;
+                        for(var c1=0;c1<subs.length;c1++){
+                            if(isKanji(subs[c1][0])){
+                                notations.push([subs[c1],1,matches[pickKanji++]]);
                             }else{
-                                notations.push([tokens[i].surface_form[c1],2,wanakana.toHiragana(tokens[i].surface_form[c1])]);
+                                notations.push([subs[c1],2,wanakana.toHiragana(subs[c1])]);
                             }
                         }
                     }else{

diff --git a/test/kuroshiroTest.js b/test/kuroshiroTest.js
@@ -9,6 +9,7 @@ describe("kuroshiro.js Test", function () {
     const EXAMPLE_TEXT = "感じ取れたら手を繋ごう、重なるのは人生のライン and レミリア最高！";
     const EXAMPLE_TEXT2 = "ブラウン管への愛が足りねぇな";
     const EXAMPLE_TEXT3 = "関ヶ原の戦い";
+    const EXAMPLE_TEXT4 = "綺麗な花。面白い映画。面白かったです。";
 
     before(function(done){
         kuroshiro.init(done);
@@ -78,6 +79,11 @@ describe("kuroshiro.js Test", function () {
         var result = kuroshiro.convert(ori,{mode:'okurigana', to:'hiragana'});
         expect(result).to.eql('関ヶ原(せきがはら)の戦(たたか)い');
     });
+    it("Kanji to Hiragana with okurigana(4)", function () {
+        var ori = EXAMPLE_TEXT4;
+        var result = kuroshiro.convert(ori,{mode:'okurigana', to:'hiragana'});
+        expect(result).to.eql('綺麗(きれい)な花(はな)。面白(おもしろ)い映画(えいが)。面白(おもしろ)かったです。');
+    });
     it("Kanji to Katakana with okurigana", function () {
         var ori = EXAMPLE_TEXT;
         var result = kuroshiro.convert(ori,{mode:'okurigana', to:'katakana'});