Add part of speech info (#1561)

* Add part of speech info to headwords

* Expose parts of speech to Anki template rendering

* Expose parts of speech

* Update pitch accent categories

* Update docs

* Add part-of-speech

* Update options and tests

* Update markers

* Update test data

8 files changed, 103 insertions, 13 deletions

diff --git a/ext/data/templates/anki-field-templates-upgrade-v10.handlebars b/ext/data/templates/anki-field-templates-upgrade-v10.handlebars
new file mode 100644
index 00000000..8442a93c
--- /dev/null
+++ b/ext/data/templates/anki-field-templates-upgrade-v10.handlebars
@@ -0,0 +1,30 @@
+{{#*inline "part-of-speech-pretty"}}
+    {{~#if (op "===" . "v1")~}}Ichidan verb
+    {{~else if (op "===" . "v5")~}}Godan verb
+    {{~else if (op "===" . "vk")~}}Kuru verb
+    {{~else if (op "===" . "vs")~}}Suru verb
+    {{~else if (op "===" . "vz")~}}Zuru verb
+    {{~else if (op "===" . "adj-i")~}}I-adjective
+    {{~else if (op "===" . "n")~}}Noun
+    {{~else~}}{{.}}
+    {{~/if~}}
+{{/inline}}
+
+{{#*inline "part-of-speech"}}
+    {{~#scope~}}
+        {{~#if (op "!==" definition.type "kanji")~}}
+            {{~#set "first" true}}{{/set~}}
+            {{~#each definition.expressions~}}
+                {{~#each wordClasses~}}
+                    {{~#unless (get (concat "used_" .))~}}
+                        {{~> part-of-speech-pretty . ~}}
+                        {{~#unless (get "first")}}, {{/unless~}}
+                        {{~#set (concat "used_" .) true~}}{{~/set~}}
+                        {{~#set "first" false~}}{{~/set~}}
+                    {{~/unless~}}
+                {{~/each~}}
+            {{~/each~}}
+            {{~#if (get "first")~}}Unknown{{~/if~}}
+        {{~/if~}}
+    {{~/scope~}}
+{{/inline}}
diff --git a/ext/data/templates/default-anki-field-templates.handlebars b/ext/data/templates/default-anki-field-templates.handlebars
index 0f390e8d..ec57a826 100644
--- a/ext/data/templates/default-anki-field-templates.handlebars
+++ b/ext/data/templates/default-anki-field-templates.handlebars
@@ -341,4 +341,35 @@
     {{~/scope~}}
 {{/inline}}
 
+{{#*inline "part-of-speech-pretty"}}
+    {{~#if (op "===" . "v1")~}}Ichidan verb
+    {{~else if (op "===" . "v5")~}}Godan verb
+    {{~else if (op "===" . "vk")~}}Kuru verb
+    {{~else if (op "===" . "vs")~}}Suru verb
+    {{~else if (op "===" . "vz")~}}Zuru verb
+    {{~else if (op "===" . "adj-i")~}}I-adjective
+    {{~else if (op "===" . "n")~}}Noun
+    {{~else~}}{{.}}
+    {{~/if~}}
+{{/inline}}
+
+{{#*inline "part-of-speech"}}
+    {{~#scope~}}
+        {{~#if (op "!==" definition.type "kanji")~}}
+            {{~#set "first" true}}{{/set~}}
+            {{~#each definition.expressions~}}
+                {{~#each wordClasses~}}
+                    {{~#unless (get (concat "used_" .))~}}
+                        {{~> part-of-speech-pretty . ~}}
+                        {{~#unless (get "first")}}, {{/unless~}}
+                        {{~#set (concat "used_" .) true~}}{{~/set~}}
+                        {{~#set "first" false~}}{{~/set~}}
+                    {{~/unless~}}
+                {{~/each~}}
+            {{~/each~}}
+            {{~#if (get "first")~}}Unknown{{~/if~}}
+        {{~/if~}}
+    {{~/scope~}}
+{{/inline}}
+
 {{~> (lookup . "marker") ~}}
diff --git a/ext/js/data/anki-note-data-creator.js b/ext/js/data/anki-note-data-creator.js
index c7047633..fb9c8b8c 100644
--- a/ext/js/data/anki-note-data-creator.js
+++ b/ext/js/data/anki-note-data-creator.js
@@ -441,7 +441,7 @@ class AnkiNoteDataCreator {
         const results = [];
         const {headwords} = dictionaryEntry;
         for (let i = 0, ii = headwords.length; i < ii; ++i) {
-            const {term, reading, tags, sources: [{deinflectedText}]} = headwords[i];
+            const {term, reading, tags, sources: [{deinflectedText}], wordClasses} = headwords[i];
             const termTags = this.createCachedValue(this._convertTags.bind(this, tags));
             const frequencies = this.createCachedValue(this._getTermExpressionFrequencies.bind(this, dictionaryEntry, i));
             const pitches = this.createCachedValue(this._getTermExpressionPitches.bind(this, dictionaryEntry, i));
@@ -455,7 +455,8 @@ class AnkiNoteDataCreator {
                 get frequencies() { return self.getCachedValue(frequencies); },
                 get pitches() { return self.getCachedValue(pitches); },
                 get furiganaSegments() { return self.getCachedValue(furiganaSegments); },
-                get termFrequency() { return self.getCachedValue(termFrequency); }
+                get termFrequency() { return self.getCachedValue(termFrequency); },
+                wordClasses
             };
             results.push(item);
         }
diff --git a/ext/js/data/options-util.js b/ext/js/data/options-util.js
index 334445c6..fd62a558 100644
--- a/ext/js/data/options-util.js
+++ b/ext/js/data/options-util.js
@@ -459,7 +459,7 @@ class OptionsUtil {
             {async: false, update: this._updateVersion7.bind(this)},
             {async: true,  update: this._updateVersion8.bind(this)},
             {async: false, update: this._updateVersion9.bind(this)},
-            {async: false, update: this._updateVersion10.bind(this)}
+            {async: true,  update: this._updateVersion10.bind(this)}
         ];
     }
 
@@ -750,9 +750,11 @@ class OptionsUtil {
         return options;
     }
 
-    _updateVersion10(options) {
+    async _updateVersion10(options) {
         // Version 10 changes:
         //  Removed global option useSettingsV2.
+        //  Added part-of-speech field template.
+        await this._applyAnkiFieldTemplatesPatch(options, '/data/templates/anki-field-templates-upgrade-v10.handlebars');
         delete options.global.useSettingsV2;
         return options;
     }
diff --git a/ext/js/display/display-generator.js b/ext/js/display/display-generator.js
index 19e88b8a..724bec9c 100644
--- a/ext/js/display/display-generator.js
+++ b/ext/js/display/display-generator.js
@@ -250,10 +250,14 @@ class DisplayGenerator {
         node.dataset.readingIsSame = `${reading === expression}`;
         node.dataset.frequency = DictionaryDataUtil.getTermFrequency(termTags);
 
-        const pitchAccentCategories = this._getPitchAccentCategories(reading, pronunciations, headwordIndex);
+        const {wordClasses} = headword;
+        const pitchAccentCategories = this._getPitchAccentCategories(reading, pronunciations, wordClasses, headwordIndex);
         if (pitchAccentCategories !== null) {
             node.dataset.pitchAccentCategories = pitchAccentCategories;
         }
+        if (wordClasses.length > 0) {
+            node.dataset.wordClasses = wordClasses.join(' ');
+        }
 
         this._setTextContent(node.querySelector('.expression-reading'), reading);
 
@@ -762,13 +766,14 @@ class DisplayGenerator {
         }
     }
 
-    _getPitchAccentCategories(reading, pronunciations, headwordIndex) {
+    _getPitchAccentCategories(reading, pronunciations, wordClasses, headwordIndex) {
         if (pronunciations.length === 0) { return null; }
+        const isVerbOrAdjective = this._isVerbOrAdjective(wordClasses);
         const categories = new Set();
         for (const pronunciation of pronunciations) {
             if (pronunciation.headwordIndex !== headwordIndex) { continue; }
             for (const {position} of pronunciation.pitches) {
-                const category = this._japaneseUtil.getPitchCategory(reading, position, false);
+                const category = this._japaneseUtil.getPitchCategory(reading, position, isVerbOrAdjective);
                 if (category !== null) {
                     categories.add(category);
                 }
@@ -776,4 +781,19 @@ class DisplayGenerator {
         }
         return categories.size > 0 ? [...categories].join(' ') : null;
     }
+
+    _isVerbOrAdjective(wordClasses) {
+        for (const wordClass of wordClasses) {
+            switch (wordClass) {
+                case 'v1':
+                case 'v5':
+                case 'vs':
+                case 'vk':
+                case 'vz':
+                case 'adj-i':
+                    return true;
+            }
+        }
+        return false;
+    }
 }
diff --git a/ext/js/language/translator.js b/ext/js/language/translator.js
index ad9689ee..e99ac8db 100644
--- a/ext/js/language/translator.js
+++ b/ext/js/language/translator.js
@@ -917,8 +917,8 @@ class Translator {
         return {originalText, transformedText, deinflectedText, isPrimary};
     }
 
-    _createTermHeadword(index, term, reading, sources, tags) {
-        return {index, term, reading, sources, tags};
+    _createTermHeadword(index, term, reading, sources, tags, wordClasses) {
+        return {index, term, reading, sources, tags, wordClasses};
     }
 
     _createTermDefinition(index, headwordIndices, dictionary, tags, entries) {
@@ -953,7 +953,7 @@ class Translator {
     }
 
     _createTermDictionaryEntryFromDatabaseEntry(databaseEntry, originalText, transformedText, deinflectedText, reasons, isPrimary, enabledDictionaryMap) {
-        const {expression, reading: rawReading, definitionTags, termTags, glossary, score, dictionary, id, sequence} = databaseEntry;
+        const {expression, reading: rawReading, definitionTags, termTags, glossary, score, dictionary, id, sequence, rules} = databaseEntry;
         const reading = (rawReading.length > 0 ? rawReading : expression);
         const {index: dictionaryIndex, priority: dictionaryPriority} = this._getDictionaryOrder(dictionary, enabledDictionaryMap);
         const sourceTermExactMatchCount = (isPrimary && deinflectedText === expression ? 1 : 0);
@@ -975,7 +975,7 @@ class Translator {
             dictionaryPriority,
             sourceTermExactMatchCount,
             maxTransformedTextLength,
-            [this._createTermHeadword(0, expression, reading, [source], headwordTagGroups)],
+            [this._createTermHeadword(0, expression, reading, [source], headwordTagGroups, rules)],
             [this._createTermDefinition(0, [0], dictionary, definitionTagGroups, glossary)]
         );
     }
@@ -1100,15 +1100,16 @@ class Translator {
 
     _addTermHeadwords(headwordsMap, headwords) {
         const headwordIndexMap = [];
-        for (const {term, reading, sources, tags} of headwords) {
+        for (const {term, reading, sources, tags, wordClasses} of headwords) {
             const key = this._createMapKey([term, reading]);
             let headword = headwordsMap.get(key);
             if (typeof headword === 'undefined') {
-                headword = this._createTermHeadword(headwordsMap.size, term, reading, [], []);
+                headword = this._createTermHeadword(headwordsMap.size, term, reading, [], [], []);
                 headwordsMap.set(key, headword);
             }
             this._addUniqueSources(headword.sources, sources);
             this._addUniqueTagGroups(headword.tags, tags);
+            this._addUniqueStrings(headword.wordClasses, wordClasses);
             headwordIndexMap.push(headword.index);
         }
         return headwordIndexMap;
diff --git a/ext/js/pages/settings/anki-controller.js b/ext/js/pages/settings/anki-controller.js
index 37ae2cf7..4e56594a 100644
--- a/ext/js/pages/settings/anki-controller.js
+++ b/ext/js/pages/settings/anki-controller.js
@@ -99,6 +99,7 @@ class AnkiController {
                     'glossary',
                     'glossary-brief',
                     'glossary-no-dictionary',
+                    'part-of-speech',
                     'pitch-accents',
                     'pitch-accent-graphs',
                     'pitch-accent-positions',
diff --git a/ext/settings.html b/ext/settings.html
index a9c29a73..2f8c012c 100644
--- a/ext/settings.html
+++ b/ext/settings.html
@@ -2652,6 +2652,10 @@
                     <td>List of definitions for the term, except the dictionary tag is omitted.</td>
                 </tr>
                 <tr>
+                    <td><code class="anki-field-marker">{part-of-speech}</code></td>
+                    <td>Part of speech information for the term.</td>
+                </tr>
+                <tr>
                     <td><code class="anki-field-marker">{pitch-accents}</code></td>
                     <td>List of pitch accent downstep notations for the term.</td>
                 </tr>