Merge pull request #440 from toasted-nutbread/collapse-emphatic-sequences

Add support for collapsing emphatic character sequences

ext/bg/data/options-schema.json

@@ -388,7 +388,8 @@
                                     "convertNumericCharacters",
                                     "convertAlphabeticCharacters",
                                     "convertHiraganaToKatakana",
-                                    "convertKatakanaToHiragana"
+                                    "convertKatakanaToHiragana",
+                                    "collapseEmphaticSequences"
                                 ],
                                 "properties": {
                                     "convertHalfWidthCharacters": {
@@ -415,6 +416,11 @@
                                         "type": "string",
                                         "enum": ["false", "true", "variant"],
                                         "default": "variant"
+                                    },
+                                    "collapseEmphaticSequences": {
+                                        "type": "string",
+                                        "enum": ["false", "true", "full"],
+                                        "default": "false"
                                     }
                                 }
                             },

ext/bg/js/japanese.js

@@ -82,6 +82,9 @@
 
     const ITERATION_MARK_CODE_POINT = 0x3005;
 
+    const HIRAGANA_SMALL_TSU_CODE_POINT = 0x3063;
+    const KATAKANA_SMALL_TSU_CODE_POINT = 0x30c3;
+    const KANA_PROLONGED_SOUND_MARK_CODE_POINT = 0x30fc;
 
     // Existing functions
 
@@ -372,6 +375,40 @@
     }
 
 
+    // Miscellaneous
+
+    function collapseEmphaticSequences(text, fullCollapse, sourceMap=null) {
+        let result = '';
+        let collapseCodePoint = -1;
+        const hasSourceMap = (sourceMap !== null);
+        for (const char of text) {
+            const c = char.codePointAt(0);
+            if (
+                c === HIRAGANA_SMALL_TSU_CODE_POINT ||
+                c === KATAKANA_SMALL_TSU_CODE_POINT ||
+                c === KANA_PROLONGED_SOUND_MARK_CODE_POINT
+            ) {
+                if (collapseCodePoint !== c) {
+                    collapseCodePoint = c;
+                    if (!fullCollapse) {
+                        result += char;
+                        continue;
+                    }
+                }
+            } else {
+                collapseCodePoint = -1;
+                result += char;
+                continue;
+            }
+
+            if (hasSourceMap) {
+                sourceMap.combine(Math.max(0, result.length - 1), 1);
+            }
+        }
+        return result;
+    }
+
+
     // Exports
 
     Object.assign(jp, {
@@ -383,6 +420,7 @@
         convertHalfWidthKanaToFullWidth,
         convertAlphabeticToKana,
         distributeFurigana,
-        distributeFuriganaInflected
+        distributeFuriganaInflected,
+        collapseEmphaticSequences
     });
 })();

ext/bg/js/options.js

@@ -170,7 +170,8 @@ function profileOptionsCreateDefaults() {
             convertNumericCharacters: 'false',
             convertAlphabeticCharacters: 'false',
             convertHiraganaToKatakana: 'false',
-            convertKatakanaToHiragana: 'variant'
+            convertKatakanaToHiragana: 'variant',
+            collapseEmphaticSequences: 'false'
         },
 
         dictionaries: {},

ext/bg/js/settings/main.js

@@ -118,6 +118,7 @@ async function formRead(options) {
     options.translation.convertAlphabeticCharacters = $('#translation-convert-alphabetic-characters').val();
     options.translation.convertHiraganaToKatakana = $('#translation-convert-hiragana-to-katakana').val();
     options.translation.convertKatakanaToHiragana = $('#translation-convert-katakana-to-hiragana').val();
+    options.translation.collapseEmphaticSequences = $('#translation-collapse-emphatic-sequences').val();
 
     options.parsing.enableScanningParser = $('#parsing-scan-enable').prop('checked');
     options.parsing.enableMecabParser = $('#parsing-mecab-enable').prop('checked');
@@ -199,6 +200,7 @@ async function formWrite(options) {
     $('#translation-convert-alphabetic-characters').val(options.translation.convertAlphabeticCharacters);
     $('#translation-convert-hiragana-to-katakana').val(options.translation.convertHiraganaToKatakana);
     $('#translation-convert-katakana-to-hiragana').val(options.translation.convertKatakanaToHiragana);
+    $('#translation-collapse-emphatic-sequences').val(options.translation.collapseEmphaticSequences);
 
     $('#parsing-scan-enable').prop('checked', options.parsing.enableScanningParser);
     $('#parsing-mecab-enable').prop('checked', options.parsing.enableMecabParser);

ext/bg/js/translator.js

@@ -347,17 +347,27 @@ class Translator {
 
     getAllDeinflections(text, options) {
         const translationOptions = options.translation;
+        const collapseEmphaticOptions = [[false, false]];
+        switch (translationOptions.collapseEmphaticSequences) {
+            case 'true':
+                collapseEmphaticOptions.push([true, false]);
+                break;
+            case 'full':
+                collapseEmphaticOptions.push([true, false], [true, true]);
+                break;
+        }
         const textOptionVariantArray = [
             Translator.getTextOptionEntryVariants(translationOptions.convertHalfWidthCharacters),
             Translator.getTextOptionEntryVariants(translationOptions.convertNumericCharacters),
             Translator.getTextOptionEntryVariants(translationOptions.convertAlphabeticCharacters),
             Translator.getTextOptionEntryVariants(translationOptions.convertHiraganaToKatakana),
-            Translator.getTextOptionEntryVariants(translationOptions.convertKatakanaToHiragana)
+            Translator.getTextOptionEntryVariants(translationOptions.convertKatakanaToHiragana),
+            collapseEmphaticOptions
         ];
 
         const deinflections = [];
         const used = new Set();
-        for (const [halfWidth, numeric, alphabetic, katakana, hiragana] of Translator.getArrayVariants(textOptionVariantArray)) {
+        for (const [halfWidth, numeric, alphabetic, katakana, hiragana, [collapseEmphatic, collapseEmphaticFull]] of Translator.getArrayVariants(textOptionVariantArray)) {
             let text2 = text;
             const sourceMap = new TextSourceMap(text2);
             if (halfWidth) {
@@ -375,6 +385,9 @@ class Translator {
             if (hiragana) {
                 text2 = jp.convertKatakanaToHiragana(text2);
             }
+            if (collapseEmphatic) {
+                text2 = jp.collapseEmphaticSequences(text2, collapseEmphaticFull, sourceMap);
+            }
 
             for (let i = text2.length; i > 0; --i) {
                 const text2Substring = text2.substring(0, i);

ext/bg/settings.html

@@ -427,7 +427,7 @@
 
                 <p class="help-block">
                     The conversion options below are listed in the order that the conversions are applied to the input text.
-                    Each conversion has three possible values:
+                    Conversions commonly have three possible values:
                 </p>
 
                 <ul class="help-block">
@@ -490,6 +490,15 @@
                         <option value="variant">Use both variants</option>
                     </select>
                 </div>
+
+                <div class="form-group">
+                    <label for="translation-collapse-emphatic-sequences">Collapse emphatic character sequences <span class="label-light">(すっっごーーい &rarr; すっごーい / すごい)</span></label>
+                    <select class="form-control" id="translation-collapse-emphatic-sequences">
+                        <option value="false">Disabled</option>
+                        <option value="true">Collapse into single character</option>
+                        <option value="full">Remove all characters</option>
+                    </select>
+                </div>
             </div>
 
             <div id="popup-content-scanning">

test/test-japanese.js

@@ -393,6 +393,59 @@ function testDistributeFuriganaInflected() {
     }
 }
 
+function testCollapseEmphaticSequences() {
+    const data = [
+        [['かこい', false], ['かこい', [1, 1, 1]]],
+        [['かこい', true], ['かこい', [1, 1, 1]]],
+        [['かっこい', false], ['かっこい', [1, 1, 1, 1]]],
+        [['かっこい', true], ['かこい', [2, 1, 1]]],
+        [['かっっこい', false], ['かっこい', [1, 2, 1, 1]]],
+        [['かっっこい', true], ['かこい', [3, 1, 1]]],
+        [['かっっっこい', false], ['かっこい', [1, 3, 1, 1]]],
+        [['かっっっこい', true], ['かこい', [4, 1, 1]]],
+
+        [['こい', false], ['こい', [1, 1]]],
+        [['こい', true], ['こい', [1, 1]]],
+        [['っこい', false], ['っこい', [1, 1, 1]]],
+        [['っこい', true], ['こい', [2, 1]]],
+        [['っっこい', false], ['っこい', [2, 1, 1]]],
+        [['っっこい', true], ['こい', [3, 1]]],
+        [['っっっこい', false], ['っこい', [3, 1, 1]]],
+        [['っっっこい', true], ['こい', [4, 1]]],
+
+        [['すごい', false], ['すごい', [1, 1, 1]]],
+        [['すごい', true], ['すごい', [1, 1, 1]]],
+        [['すごーい', false], ['すごーい', [1, 1, 1, 1]]],
+        [['すごーい', true], ['すごい', [1, 2, 1]]],
+        [['すごーーい', false], ['すごーい', [1, 1, 2, 1]]],
+        [['すごーーい', true], ['すごい', [1, 3, 1]]],
+        [['すっごーい', false], ['すっごーい', [1, 1, 1, 1, 1]]],
+        [['すっごーい', true], ['すごい', [2, 2, 1]]],
+        [['すっっごーーい', false], ['すっごーい', [1, 2, 1, 2, 1]]],
+        [['すっっごーーい', true], ['すごい', [3, 3, 1]]],
+
+        [['', false], ['', []]],
+        [['', true], ['', []]],
+        [['っ', false], ['っ', [1]]],
+        [['っ', true], ['', [1]]],
+        [['っっ', false], ['っ', [2]]],
+        [['っっ', true], ['', [2]]],
+        [['っっっ', false], ['っ', [3]]],
+        [['っっっ', true], ['', [3]]]
+    ];
+
+    for (const [[text, fullCollapse], [expected, expectedSourceMapping]] of data) {
+        const sourceMap = new TextSourceMap(text);
+        const actual1 = jp.collapseEmphaticSequences(text, fullCollapse, null);
+        const actual2 = jp.collapseEmphaticSequences(text, fullCollapse, sourceMap);
+        assert.strictEqual(actual1, expected);
+        assert.strictEqual(actual2, expected);
+        if (typeof expectedSourceMapping !== 'undefined') {
+            assert.ok(sourceMap.equals(new TextSourceMap(text, expectedSourceMapping)));
+        }
+    }
+}
+
 function testIsMoraPitchHigh() {
     const data = [
         [[0, 0], false],
@@ -462,6 +515,7 @@ function main() {
     testConvertAlphabeticToKana();
     testDistributeFurigana();
     testDistributeFuriganaInflected();
+    testCollapseEmphaticSequences();
     testIsMoraPitchHigh();
     testGetKanaMorae();
 }