Merge pull request #450 from toasted-nutbread/frequency-readings

Add support for filtering frequency metadata based on readings

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

@@ -25,9 +25,31 @@
                 "items": [
                     {},
                     {"enum": ["freq"]},
+                    {
+                        "oneOf": [
                             {
                                 "type": ["string", "number"],
                                 "description": "Frequency information for the term or expression."
+                            },
+                            {
+                                "type": ["object"],
+                                "required": [
+                                    "reading",
+                                    "frequency"
+                                ],
+                                "additionalProperties": false,
+                                "properties": {
+                                    "reading": {
+                                        "type": "string",
+                                        "description": "Reading for the term or expression."
+                                    },
+                                    "frequency": {
+                                        "type": ["string", "number"],
+                                        "description": "Frequency information for the term or expression."
+                                    }
+                                }
+                            }
+                        ]
                     }
                 ]
             },

ext/bg/js/translator.js

@@ -482,7 +482,9 @@ class Translator {
             switch (mode) {
                 case 'freq':
                     for (const term of termsUnique[index]) {
-                        term.frequencies.push({expression, frequency: data, dictionary});
+                        const frequencyData = this.getFrequencyData(expression, data, dictionary, term);
+                        if (frequencyData === null) { continue; }
+                        term.frequencies.push(frequencyData);
                     }
                     break;
                 case 'pitch':
@@ -575,6 +577,18 @@ class Translator {
         return tagMetaList;
     }
 
+    getFrequencyData(expression, data, dictionary, term) {
+        if (data !== null && typeof data === 'object') {
+            const {frequency, reading} = data;
+
+            const termReading = term.reading || expression;
+            if (reading !== termReading) { return null; }
+
+            return {expression, frequency, dictionary};
+        }
+        return {expression, frequency: data, dictionary};
+    }
+
     async getPitchData(expression, data, dictionary, term) {
         const reading = data.reading;
         const termReading = term.reading || expression;

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

@@ -2,6 +2,12 @@
     ["打", "freq", 1],
     ["打つ", "freq", 2],
     ["打ち込む", "freq", 3],
+    ["打", "freq", {"reading": "だ", "frequency": 4}],
+    ["打", "freq", {"reading": "ダース", "frequency": 5}],
+    ["打つ", "freq", {"reading": "うつ", "frequency": 6}],
+    ["打つ", "freq", {"reading": "ぶつ", "frequency": 7}],
+    ["打ち込む", "freq", {"reading": "うちこむ", "frequency": 8}],
+    ["打ち込む", "freq", {"reading": "ぶちこむ", "frequency": 9}],
     [
         "打ち込む",
         "pitch",

test/test-database.js

@@ -235,8 +235,8 @@ async function testDatabase1() {
             true
         );
         vm.assert.deepStrictEqual(counts, {
-            counts: [{kanji: 2, kanjiMeta: 2, terms: 32, termMeta: 6, tagMeta: 14}],
-            total: {kanji: 2, kanjiMeta: 2, terms: 32, termMeta: 6, tagMeta: 14}
+            counts: [{kanji: 2, kanjiMeta: 2, terms: 32, termMeta: 12, tagMeta: 14}],
+            total: {kanji: 2, kanjiMeta: 2, terms: 32, termMeta: 12, tagMeta: 14}
         });
 
         // Test find* functions
@@ -626,9 +626,9 @@ async function testFindTermMetaBulk1(database, titles) {
                 }
             ],
             expectedResults: {
-                total: 1,
+                total: 3,
                 modes: [
-                    ['freq', 1]
+                    ['freq', 3]
                 ]
             }
         },
@@ -639,9 +639,9 @@ async function testFindTermMetaBulk1(database, titles) {
                 }
             ],
             expectedResults: {
-                total: 1,
+                total: 3,
                 modes: [
-                    ['freq', 1]
+                    ['freq', 3]
                 ]
             }
         },
@@ -652,9 +652,9 @@ async function testFindTermMetaBulk1(database, titles) {
                 }
             ],
             expectedResults: {
-                total: 3,
+                total: 5,
                 modes: [
-                    ['freq', 1],
+                    ['freq', 3],
                     ['pitch', 2]
                 ]
             }