Frequencies marker (#1074)

* Update japanese.js tests

* Simplify fallback/early exit

* Add overloads to furigana and furiganaPlain handlebars helper functions

* Expose unique expression/reading arrays (and subsequently counts)

* Add {frequencies} marker

README.md

@@ -163,6 +163,7 @@ Flashcard fields can be configured with the following steps:
     `{dictionary}` | Name of the dictionary from which the card is being created (unavailable in *grouped* mode).
     `{document-title}` | Title of the web page that the term appeared in.
     `{expression}` | Term expressed as kanji (will be displayed in kana if kanji is not available).
+    `{frequencies}` | Frequency information for the term.
     `{furigana}` | Term expressed as kanji with furigana displayed above it (e.g. <ruby>日本語<rt>にほんご</rt></ruby>).
     `{furigana-plain}` | Term expressed as kanji with furigana displayed next to it in brackets (e.g. 日本語[にほんご]).
     `{glossary}` | List of definitions for the term (output format depends on whether running in *grouped* mode).
@@ -188,6 +189,7 @@ Flashcard fields can be configured with the following steps:
     `{cloze-suffix}` | Fragment of the containing `{sentence}` starting at the end of `{cloze-body}` until the end of `{sentence}`.
     `{dictionary}` | Name of the dictionary from which the card is being created.
     `{document-title}` | Title of the web page that the kanji appeared in.
+    `{frequencies}` | Frequency information for the kanji.
     `{glossary}` | List of definitions for the kanji.
     `{kunyomi}` | Kunyomi (Japanese reading) for the kanji expressed as katakana.
     `{onyomi}` | Onyomi (Chinese reading) for the kanji expressed as hiragana.

docs/templates.md

@@ -44,21 +44,27 @@ This function can be helpful for debugging values when creating templates.
 
 ### `furigana`
 
-Converts a definition to its furigana representation.
+Converts a definition or expression/reading pair to its furigana representation.
 
 <details>
   <summary>Syntax:</summary>
 
-  <code>{{#furigana}}<i>&lt;definition&gt;</i>{{/furigana}}</code>
+  <code>{{#furigana}}<i>&lt;definition&gt;</i>{{/furigana}}</code><br>
+  <code>{{#furigana <i>expression</i> <i>reading</i>}}{{/furigana}}</code><br>
 
   * _`definition`_ <br>
     The definition to convert.
+  * _`expression`_ <br>
+    The expression to convert.
+  * _`reading`_ <br>
+    The reading to convert.
 </details>
 <details>
   <summary>Example:</summary>
 
   ```handlebars
   {{#furigana}}{{.}}{{/furigana}}
+  {{#furigana "読む" "よむ"}}{{/furigana}}
   ```
 
   Output:
@@ -73,21 +79,27 @@ Converts a definition to its furigana representation.
 
 ### `furiganaPlain`
 
-Converts a definition to its simplified furigana representation.
+Converts a definition or expression/reading pair to its simplified furigana representation.
 
 <details>
   <summary>Syntax:</summary>
 
   <code>{{#furiganaPlain}}<i>&lt;definition&gt;</i>{{/furigana}}</code>
+  <code>{{#furiganaPlain <i>expression</i> <i>reading</i>}}{{/furiganaPlain}}</code><br>
 
   * _`definition`_ <br>
     The definition to convert.
+  * _`expression`_ <br>
+    The expression to convert.
+  * _`reading`_ <br>
+    The reading to convert.
 </details>
 <details>
   <summary>Example:</summary>
 
   ```handlebars
   {{~#furiganaPlain~}}{{.}}{{~/furiganaPlain~}}
+  {{#furiganaPlain "読む" "よむ"}}{{/furiganaPlain}}
   ```
 
   Output:

ext/bg/data/anki-field-templates-upgrade-v6.handlebars

@@ -6,3 +6,20 @@
         {{~/each~}}
     {{~/if~}}
 {{/inline}}
+
+{{#*inline "frequencies"}}
+    {{~#if (op ">" definition.frequencies.length 0)~}}
+        <ul style="text-align: left;">
+        {{~#each definition.frequencies~}}
+            <li>
+            {{~#if (op "!==" ../definition.type "kanji")~}}
+                {{~#if (op "||" (op ">" ../uniqueExpressions.length 1) (op ">" ../uniqueReadings.length 1))~}}(
+                    {{~#furigana expression reading~}}{{~/furigana~}}
+                ) {{/if~}}
+            {{~/if~}}
+            {{~dictionary}}: {{frequency~}}
+            </li>
+        {{~/each~}}
+        </ul>
+    {{~/if~}}
+{{/inline}}

ext/bg/data/default-anki-field-templates.handlebars

@@ -306,4 +306,21 @@
     {{~/if~}}
 {{/inline}}
 
+{{#*inline "frequencies"}}
+    {{~#if (op ">" definition.frequencies.length 0)~}}
+        <ul style="text-align: left;">
+        {{~#each definition.frequencies~}}
+            <li>
+            {{~#if (op "!==" ../definition.type "kanji")~}}
+                {{~#if (op "||" (op ">" ../uniqueExpressions.length 1) (op ">" ../uniqueReadings.length 1))~}}(
+                    {{~#furigana expression reading~}}{{~/furigana~}}
+                ) {{/if~}}
+            {{~/if~}}
+            {{~dictionary}}: {{frequency~}}
+            </li>
+        {{~/each~}}
+        </ul>
+    {{~/if~}}
+{{/inline}}
+
 {{~> (lookup . "marker") ~}}

ext/bg/js/anki-note-builder.js

@@ -108,9 +108,19 @@ class AnkiNoteBuilder {
     _createNoteData(definition, mode, context, resultOutputMode, glossaryLayoutMode, compactTags) {
         const pitches = DictionaryDataUtil.getPitchAccentInfos(definition);
         const pitchCount = pitches.reduce((i, v) => i + v.pitches.length, 0);
+        const uniqueExpressions = new Set();
+        const uniqueReadings = new Set();
+        if (definition.type !== 'kanji') {
+            for (const {expression, reading} of definition.expressions) {
+                uniqueExpressions.add(expression);
+                uniqueReadings.add(reading);
+            }
+        }
         return {
             marker: null,
             definition,
+            uniqueExpressions: [...uniqueExpressions],
+            uniqueReadings: [...uniqueReadings],
             pitches,
             pitchCount,
             group: resultOutputMode === 'group',

ext/bg/js/settings/anki-controller.js

@@ -83,6 +83,7 @@ class AnkiController {
                     'dictionary',
                     'document-title',
                     'expression',
+                    'frequencies',
                     'furigana',
                     'furigana-plain',
                     'glossary',

ext/bg/js/template-renderer.js

@@ -117,13 +117,13 @@ class TemplateRenderer {
         return this._escape(dump);
     }
 
-    _furigana(context, options) {
+    _furigana(context, ...args) {
-        const definition = options.fn(context);
+        const {expression, reading} = this._getFuriganaExpressionAndReading(context, ...args);
-        const segs = jp.distributeFurigana(definition.expression, definition.reading);
+        const segs = jp.distributeFurigana(expression, reading);
 
         let result = '';
         for (const seg of segs) {
-            if (seg.furigana) {
+            if (seg.furigana.length > 0) {
                 result += `<ruby>${seg.text}<rt>${seg.furigana}</rt></ruby>`;
             } else {
                 result += seg.text;
@@ -133,20 +133,31 @@ class TemplateRenderer {
         return result;
     }
 
-    _furiganaPlain(context, options) {
+    _furiganaPlain(context, ...args) {
-        const definition = options.fn(context);
+        const {expression, reading} = this._getFuriganaExpressionAndReading(context, ...args);
-        const segs = jp.distributeFurigana(definition.expression, definition.reading);
+        const segs = jp.distributeFurigana(expression, reading);
 
         let result = '';
         for (const seg of segs) {
-            if (seg.furigana) {
+            if (seg.furigana.length > 0) {
+                if (result.length > 0) { result += ' '; }
                 result += `${seg.text}[${seg.furigana}]`;
             } else {
                 result += seg.text;
             }
         }
 
-        return result.trimLeft();
+        return result;
+    }
+
+    _getFuriganaExpressionAndReading(context, ...args) {
+        const options = args[args.length - 1];
+        if (args.length >= 3) {
+            return {expression: args[0], reading: args[1]};
+        } else {
+            const {expression, reading} = options.fn(context);
+            return {expression, reading};
+        }
     }
 
     _kanjiLinks(context, options) {

ext/mixed/js/japanese.js

@@ -411,9 +411,9 @@ const jp = (() => {
         // Furigana distribution
 
         distributeFurigana(expression, reading) {
-            const fallback = [{furigana: reading, text: expression}];
+            if (!reading || reading === expression) {
-            if (!reading) {
+                // Same
-                return fallback;
+                return [{furigana: '', text: expression}];
             }
 
             let isAmbiguous = false;
@@ -471,7 +471,9 @@ const jp = (() => {
             if (segments && !isAmbiguous) {
                 return segments;
             }
-            return fallback;
+
+            // Fallback
+            return [{furigana: reading, text: expression}];
         }
 
         distributeFuriganaInflected(expression, reading, source) {

test/test-japanese.js

@@ -334,6 +334,25 @@ function testDistributeFurigana() {
             [
                 {text: '長い間', furigana: 'ながいあいだ'}
             ]
+        ],
+        // Same/empty reading
+        [
+            ['飼い犬', ''],
+            [
+                {text: '飼い犬', furigana: ''}
+            ]
+        ],
+        [
+            ['かいいぬ', 'かいいぬ'],
+            [
+                {text: 'かいいぬ', furigana: ''}
+            ]
+        ],
+        [
+            ['かいぬ', 'かいぬ'],
+            [
+                {text: 'かいぬ', furigana: ''}
+            ]
         ]
     ];