Devoice and nasal pronunciation info (#1832)

* Update schema to support information about nasal and devoiced mora

* Expose nasalPositions and devoicePositions in dictionary entry data

* Expose nasalPositions, devoicePositions in grouped pitch info

* Update display generator

* Update test dictionary data

* Update test data

docs/interfaces/dictionary-entry.ts

@@ -354,6 +354,14 @@ namespace Translation {
          * Position of the downstep, as a number of mora.
          */
         position: number;
+        /**
+         * Positions of moras that have a .
+         */
+        nasalPositions: number[];
+        /**
+         * Position of the downstep, as a number of mora.
+         */
+        devoicePositions: [];
         /**
          * Tags for the pitch accent.
          */

ext/css/display.css

@@ -1512,6 +1512,31 @@ button.definition-item-expansion-button:focus:focus-visible+.definition-item-con
     padding-right: 0.1em;
     margin-right: 0.1em;
 }
+.pitch-accent-character-devoice-indicator {
+    display: block;
+    position: absolute;
+    left: 50%;
+    top: 50%;
+    width: 1.125em;
+    height: 1.125em;
+    border: calc(1.5em / var(--font-size-no-units)) dotted var(--danger-color);
+    border-radius: 50%;
+    box-sizing: border-box;
+    z-index: 1;
+    transform: translate(-50%, -50%);
+}
+.pitch-accent-character-nasal-indicator {
+    display: block;
+    position: absolute;
+    right: -0.125em;
+    top: 0.125em;
+    width: 0.375em;
+    height: 0.375em;
+    border: calc(1.5em / var(--font-size-no-units)) solid var(--danger-color);
+    border-radius: 50%;
+    box-sizing: border-box;
+    z-index: 1;
+}
 .pitch-accent-position::before {
     content: ' [';
 }

ext/data/schemas/dictionary-term-meta-bank-v3-schema.json

@@ -85,6 +85,40 @@
                                             "description": "Mora position of the pitch accent downstep. A value of 0 indicates that the word does not have a downstep (heiban).",
                                             "minimum": 0
                                         },
+                                        "nasal": {
+                                            "oneOf": [
+                                                {
+                                                    "type": "integer",
+                                                    "description": "Position of a mora with nasal sound.",
+                                                    "minimum": 0
+                                                },
+                                                {
+                                                    "type": "array",
+                                                    "description": "Positions of morae with nasal sound.",
+                                                    "items": {
+                                                        "type": "integer",
+                                                        "minimum": 0
+                                                    }
+                                                }
+                                            ]
+                                        },
+                                        "devoice": {
+                                            "oneOf": [
+                                                {
+                                                    "type": "integer",
+                                                    "description": "Position of a mora with devoiced sound.",
+                                                    "minimum": 0
+                                                },
+                                                {
+                                                    "type": "array",
+                                                    "description": "Positions of morae with devoiced sound.",
+                                                    "items": {
+                                                        "type": "integer",
+                                                        "minimum": 0
+                                                    }
+                                                }
+                                            ]
+                                        },
                                         "tags": {
                                             "type": "array",
                                             "description": "List of tags for this pitch accent.",

ext/js/display/display-generator.js

@@ -468,12 +468,17 @@ class DisplayGenerator {
 
     _createPitch(details) {
         const jp = this._japaneseUtil;
-        const {reading, position, tags, exclusiveTerms, exclusiveReadings} = details;
+        const {reading, position, nasalPositions, devoicePositions, tags, exclusiveTerms, exclusiveReadings} = details;
         const morae = jp.getKanaMorae(reading);
 
+        const nasalPositionsSet = nasalPositions.length > 0 ? new Set(nasalPositions) : null;
+        const devoicePositionsSet = devoicePositions.length > 0 ? new Set(devoicePositions) : null;
+
         const node = this._templates.instantiate('pitch-accent');
 
         node.dataset.pitchAccentPosition = `${position}`;
+        if (nasalPositions.length > 0) { node.dataset.nasalMoraPosition = nasalPositions.join(' '); }
+        if (devoicePositions.length > 0) { node.dataset.devoiceMoraPosition = devoicePositions.join(' '); }
         node.dataset.tagCount = `${tags.length}`;
 
         let n = node.querySelector('.pitch-accent-position');
@@ -487,18 +492,39 @@ class DisplayGenerator {
 
         n = node.querySelector('.pitch-accent-characters');
         for (let i = 0, ii = morae.length; i < ii; ++i) {
+            const i1 = i + 1;
             const mora = morae[i];
             const highPitch = jp.isMoraPitchHigh(i, position);
-            const highPitchNext = jp.isMoraPitchHigh(i + 1, position);
+            const highPitchNext = jp.isMoraPitchHigh(i1, position);
+            const nasal = nasalPositionsSet !== null && nasalPositionsSet.has(i1);
+            const devoice = devoicePositionsSet !== null && devoicePositionsSet.has(i1);
 
-            const n1 = this._templates.instantiate('pitch-accent-character');
+            const n1 = document.createElement('span');
-            const n2 = n1.querySelector('.pitch-accent-character-inner');
+            n1.className = 'pitch-accent-character';
+
+            const n2 = document.createElement('span');
+            n2.className = 'pitch-accent-character-inner';
+
+            n1.appendChild(n2);
 
             n1.dataset.position = `${i}`;
             n1.dataset.pitch = highPitch ? 'high' : 'low';
             n1.dataset.pitchNext = highPitchNext ? 'high' : 'low';
             this._setTextContent(n2, mora, 'ja');
 
+            if (devoice) {
+                n1.dataset.devoice = 'true';
+                const n3 = document.createElement('span');
+                n3.className = 'pitch-accent-character-devoice-indicator';
+                n1.appendChild(n3);
+            }
+            if (nasal) {
+                n1.dataset.nasal = 'true';
+                const n3 = document.createElement('span');
+                n3.className = 'pitch-accent-character-nasal-indicator';
+                n1.appendChild(n3);
+            }
+
             n.appendChild(n1);
         }
 

ext/js/language/dictionary-data-util.js

@@ -108,13 +108,15 @@ class DictionaryDataUtil {
                 dictionaryPitchAccentInfoList = [];
                 pitchAccentInfoMap.set(dictionary, dictionaryPitchAccentInfoList);
             }
-            for (const {position, tags} of pitches) {
+            for (const {position, nasalPositions, devoicePositions, tags} of pitches) {
-                let pitchAccentInfo = this._findExistingPitchAccentInfo(reading, position, tags, dictionaryPitchAccentInfoList);
+                let pitchAccentInfo = this._findExistingPitchAccentInfo(reading, position, nasalPositions, devoicePositions, tags, dictionaryPitchAccentInfoList);
                 if (pitchAccentInfo === null) {
                     pitchAccentInfo = {
                         terms: new Set(),
                         reading,
                         position,
+                        nasalPositions,
+                        devoicePositions,
                         tags,
                         exclusiveTerms: [],
                         exclusiveReadings: []
@@ -228,11 +230,13 @@ class DictionaryDataUtil {
         return results;
     }
 
-    static _findExistingPitchAccentInfo(reading, position, tags, pitchAccentInfoList) {
+    static _findExistingPitchAccentInfo(reading, position, nasalPositions, devoicePositions, tags, pitchAccentInfoList) {
         for (const pitchInfo of pitchAccentInfoList) {
             if (
                 pitchInfo.reading === reading &&
                 pitchInfo.position === position &&
+                this._areArraysEqual(pitchInfo.nasalPositions, nasalPositions) &&
+                this._areArraysEqual(pitchInfo.devoicePositions, devoicePositions) &&
                 this._areTagListsEqual(pitchInfo.tags, tags)
             ) {
                 return pitchInfo;
@@ -241,6 +245,15 @@ class DictionaryDataUtil {
         return null;
     }
 
+    static _areArraysEqual(array1, array2) {
+        const ii = array1.length;
+        if (ii !== array2.length) { return false; }
+        for (let i = 0; i < ii; ++i) {
+            if (array1[i] !== array2[i]) { return false; }
+        }
+        return true;
+    }
+
     static _areTagListsEqual(tagList1, tagList2) {
         const ii = tagList1.length;
         if (tagList2.length !== ii) { return false; }

ext/js/language/translator.js

@@ -828,12 +828,14 @@ class Translator {
                         {
                             if (data.reading !== reading) { continue; }
                             const pitches = [];
-                            for (const {position, tags} of data.pitches) {
+                            for (const {position, tags, nasal, devoice} of data.pitches) {
                                 const tags2 = [];
                                 if (Array.isArray(tags) && tags.length > 0) {
                                     tags2.push(this._createTagGroup(dictionary, tags));
                                 }
-                                pitches.push({position, tags: tags2});
+                                const nasalPositions = this._toNumberArray(nasal);
+                                const devoicePositions = this._toNumberArray(devoice);
+                                pitches.push({position, nasalPositions, devoicePositions, tags: tags2});
                             }
                             for (const {pronunciations, headwordIndex} of targets) {
                                 pronunciations.push(this._createTermPronunciation(
@@ -968,6 +970,10 @@ class Translator {
         return JSON.stringify(array);
     }
 
+    _toNumberArray(value) {
+        return Array.isArray(value) ? value : (typeof value === 'number' ? [value] : []);
+    }
+
     // Kanji data
 
     _createKanjiStat(name, value, databaseInfo, dictionary) {

test/data/dictionaries/valid-dictionary1/term_meta_bank_1.json

@@ -41,5 +41,35 @@
                 {"position": 0, "tags": ["P2"]}
             ]
         }
+    ],
+    [
+        "番号",
+        "pitch",
+        {
+            "reading": "ばんごう",
+            "pitches": [
+                {"position": 3, "nasal": 3}
+            ]
+        }
+    ],
+    [
+        "中腰",
+        "pitch",
+        {
+            "reading": "ちゅうごし",
+            "pitches": [
+                {"position": 0, "nasal": 3}
+            ]
+        }
+    ],
+    [
+        "土木工事",
+        "pitch",
+        {
+            "reading": "どぼくこうじ",
+            "pitches": [
+                {"position": 4, "devoice": 3}
+            ]
+        }
     ]
 ]

test/data/translator-test-results.json

@@ -1235,10 +1235,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -1366,10 +1370,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -1497,10 +1505,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -1628,10 +1640,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -3236,10 +3252,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -3367,10 +3387,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -3734,10 +3758,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -3865,10 +3893,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -4942,10 +4974,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -5120,10 +5156,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -5966,10 +6006,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -5983,10 +6027,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -6660,10 +6708,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -6795,10 +6847,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -6930,10 +6986,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -7065,10 +7125,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -7875,10 +7939,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -8006,10 +8074,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -8137,10 +8209,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -8268,10 +8344,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -9078,10 +9158,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -9209,10 +9293,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -9340,10 +9428,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -9471,10 +9563,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -10754,10 +10850,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]
@@ -10771,10 +10871,14 @@
             "pitches": [
               {
                 "position": 0,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               },
               {
                 "position": 3,
+                "nasalPositions": [],
+                "devoicePositions": [],
                 "tags": []
               }
             ]

test/test-database.js

@@ -162,8 +162,8 @@ async function testDatabase1() {
             true
         );
         vm.assert.deepStrictEqual(counts, {
-            counts: [{kanji: 2, kanjiMeta: 2, terms: 15, termMeta: 12, tagMeta: 15, media: 2}],
+            counts: [{kanji: 2, kanjiMeta: 2, terms: 15, termMeta: 15, tagMeta: 15, media: 2}],
-            total: {kanji: 2, kanjiMeta: 2, terms: 15, termMeta: 12, tagMeta: 15, media: 2}
+            total: {kanji: 2, kanjiMeta: 2, terms: 15, termMeta: 15, tagMeta: 15, media: 2}
         });
 
         // Test find* functions