Merge pull request #385 from toasted-nutbread/pitch-accents

Pitch accents

7 files changed, 125 insertions, 33 deletions

diff --git a/ext/bg/data/dictionary-term-meta-bank-v3-schema.json b/ext/bg/data/dictionary-term-meta-bank-v3-schema.json
index 1cc0557f..8475db81 100644
--- a/ext/bg/data/dictionary-term-meta-bank-v3-schema.json
+++ b/ext/bg/data/dictionary-term-meta-bank-v3-schema.json
@@ -13,13 +13,71 @@
             },
             {
                 "type": "string",
-                "enum": ["freq"],
-                "description": "Type of data. \"freq\" corresponds to frequency information."
+                "enum": ["freq", "pitch"],
+                "description": "Type of data. \"freq\" corresponds to frequency information; \"pitch\" corresponds to pitch information."
             },
             {
-                "type": ["string", "number"],
                 "description": "Data for the term/expression."
             }
+        ],
+        "oneOf": [
+            {
+                "items": [
+                    {},
+                    {"enum": ["freq"]},
+                    {
+                        "type": ["string", "number"],
+                        "description": "Frequency information for the term or expression."
+                    }
+                ]
+            },
+            {
+                "items": [
+                    {},
+                    {"enum": ["pitch"]},
+                    {
+                        "type": ["object"],
+                        "description": "Pitch accent information for the term or expression.",
+                        "required": [
+                            "reading",
+                            "pitches"
+                        ],
+                        "additionalProperties": false,
+                        "properties": {
+                            "reading": {
+                                "type": "string",
+                                "description": "Reading for the term or expression."
+                            },
+                            "pitches": {
+                                "type": "array",
+                                "description": "List of different pitch accent information for the term and reading combination.",
+                                "additionalItems": {
+                                    "type": "object",
+                                    "required": [
+                                        "position"
+                                    ],
+                                    "additionalProperties": false,
+                                    "properties": {
+                                        "position": {
+                                            "type": "integer",
+                                            "description": "Mora position of the pitch accent downstep. A value of 0 indicates that the word does not have a downstep (heiban).",
+                                            "minimum": 0
+                                        },
+                                        "tags": {
+                                            "type": "array",
+                                            "description": "List of tags for this pitch accent.",
+                                            "items": {
+                                                "type": "string",
+                                                "description": "Tag for this pitch accent. This typically corresponds to a certain type of part of speech."
+                                            }
+                                        }
+                                    }
+                                }
+                            }
+                        }
+                    }
+                ]
+            }
         ]
     }
 }
\ No newline at end of file
diff --git a/ext/bg/data/options-schema.json b/ext/bg/data/options-schema.json
index d6207952..cb759b72 100644
--- a/ext/bg/data/options-schema.json
+++ b/ext/bg/data/options-schema.json
@@ -105,7 +105,10 @@
                                     "customPopupCss",
                                     "customPopupOuterCss",
                                     "enableWanakana",
-                                    "enableClipboardMonitor"
+                                    "enableClipboardMonitor",
+                                    "showPitchAccentDownstepNotation",
+                                    "showPitchAccentPositionNotation",
+                                    "showPitchAccentGraph"
                                 ],
                                 "properties": {
                                     "enable": {
@@ -227,6 +230,18 @@
                                     "enableClipboardMonitor": {
                                         "type": "boolean",
                                         "default": false
+                                    },
+                                    "showPitchAccentDownstepNotation": {
+                                        "type": "boolean",
+                                        "default": true
+                                    },
+                                    "showPitchAccentPositionNotation": {
+                                        "type": "boolean",
+                                        "default": true
+                                    },
+                                    "showPitchAccentGraph": {
+                                        "type": "boolean",
+                                        "default": false
                                     }
                                 }
                             },
diff --git a/ext/bg/js/dictionary.js b/ext/bg/js/dictionary.js
index 3dd1d0c1..74bd5a64 100644
--- a/ext/bg/js/dictionary.js
+++ b/ext/bg/js/dictionary.js
@@ -137,30 +137,6 @@ function dictTermsGroup(definitions, dictionaries) {
     return dictTermsSort(results);
 }
 
-function dictAreSetsEqual(set1, set2) {
-    if (set1.size !== set2.size) {
-        return false;
-    }
-
-    for (const value of set1) {
-        if (!set2.has(value)) {
-            return false;
-        }
-    }
-
-    return true;
-}
-
-function dictGetSetIntersection(set1, set2) {
-    const result = [];
-    for (const value of set1) {
-        if (set2.has(value)) {
-            result.push(value);
-        }
-    }
-    return result;
-}
-
 function dictTermsMergeBySequence(definitions, mainDictionary) {
     const sequencedDefinitions = new Map();
     const nonSequencedDefinitions = [];
@@ -281,11 +257,11 @@ function dictTermsMergeByGloss(result, definitions, appendTo=null, mergedIndices
         const only = [];
         const expressionSet = definition.expression;
         const readingSet = definition.reading;
-        if (!dictAreSetsEqual(expressionSet, resultExpressionSet)) {
-            only.push(...dictGetSetIntersection(expressionSet, resultExpressionSet));
+        if (!areSetsEqual(expressionSet, resultExpressionSet)) {
+            only.push(...getSetIntersection(expressionSet, resultExpressionSet));
         }
-        if (!dictAreSetsEqual(readingSet, resultReadingSet)) {
-            only.push(...dictGetSetIntersection(readingSet, resultReadingSet));
+        if (!areSetsEqual(readingSet, resultReadingSet)) {
+            only.push(...getSetIntersection(readingSet, resultReadingSet));
         }
         definition.only = only;
     }
diff --git a/ext/bg/js/options.js b/ext/bg/js/options.js
index bd0bbe0e..b36fe812 100644
--- a/ext/bg/js/options.js
+++ b/ext/bg/js/options.js
@@ -124,7 +124,10 @@ function profileOptionsCreateDefaults() {
             customPopupCss: '',
             customPopupOuterCss: '',
             enableWanakana: true,
-            enableClipboardMonitor: false
+            enableClipboardMonitor: false,
+            showPitchAccentDownstepNotation: true,
+            showPitchAccentPositionNotation: true,
+            showPitchAccentGraph: false
         },
 
         audio: {
diff --git a/ext/bg/js/settings/main.js b/ext/bg/js/settings/main.js
index ebc443df..7caeaea0 100644
--- a/ext/bg/js/settings/main.js
+++ b/ext/bg/js/settings/main.js
@@ -84,6 +84,9 @@ async function formRead(options) {
     options.general.popupScalingFactor = parseFloat($('#popup-scaling-factor').val());
     options.general.popupScaleRelativeToPageZoom = $('#popup-scale-relative-to-page-zoom').prop('checked');
     options.general.popupScaleRelativeToVisualViewport = $('#popup-scale-relative-to-visual-viewport').prop('checked');
+    options.general.showPitchAccentDownstepNotation = $('#show-pitch-accent-downstep-notation').prop('checked');
+    options.general.showPitchAccentPositionNotation = $('#show-pitch-accent-position-notation').prop('checked');
+    options.general.showPitchAccentGraph = $('#show-pitch-accent-graph').prop('checked');
     options.general.popupTheme = $('#popup-theme').val();
     options.general.popupOuterTheme = $('#popup-outer-theme').val();
     options.general.customPopupCss = $('#custom-popup-css').val();
@@ -161,6 +164,9 @@ async function formWrite(options) {
     $('#popup-scaling-factor').val(options.general.popupScalingFactor);
     $('#popup-scale-relative-to-page-zoom').prop('checked', options.general.popupScaleRelativeToPageZoom);
     $('#popup-scale-relative-to-visual-viewport').prop('checked', options.general.popupScaleRelativeToVisualViewport);
+    $('#show-pitch-accent-downstep-notation').prop('checked', options.general.showPitchAccentDownstepNotation);
+    $('#show-pitch-accent-position-notation').prop('checked', options.general.showPitchAccentPositionNotation);
+    $('#show-pitch-accent-graph').prop('checked', options.general.showPitchAccentGraph);
     $('#popup-theme').val(options.general.popupTheme);
     $('#popup-outer-theme').val(options.general.popupOuterTheme);
     $('#custom-popup-css').val(options.general.customPopupCss);
diff --git a/ext/bg/js/translator.js b/ext/bg/js/translator.js
index 6f43f7b0..f16889ce 100644
--- a/ext/bg/js/translator.js
+++ b/ext/bg/js/translator.js
@@ -490,6 +490,7 @@ class Translator {
 
             // New data
             term.frequencies = [];
+            term.pitches = [];
         }
 
         const metas = await this.database.findTermMetaBulk(expressionsUnique, dictionaries);
@@ -500,6 +501,13 @@ class Translator {
                         term.frequencies.push({expression, frequency: data, dictionary});
                     }
                     break;
+                case 'pitch':
+                    for (const term of termsUnique[index]) {
+                        const pitchData = await this.getPitchData(expression, data, dictionary, term);
+                        if (pitchData === null) { continue; }
+                        term.pitches.push(pitchData);
+                    }
+                    break;
             }
         }
     }
@@ -583,6 +591,20 @@ class Translator {
         return tagMetaList;
     }
 
+    async getPitchData(expression, data, dictionary, term) {
+        const reading = data.reading;
+        const termReading = term.reading || expression;
+        if (reading !== termReading) { return null; }
+
+        const pitches = [];
+        for (let {position, tags} of data.pitches) {
+            tags = Array.isArray(tags) ? await this.getTagMetaList(tags, dictionary) : [];
+            pitches.push({position, tags});
+        }
+
+        return {reading, pitches, dictionary};
+    }
+
     static createExpression(expression, reading, termTags=null, termFrequency=null) {
         const furiganaSegments = jp.distributeFurigana(expression, reading);
         return {
diff --git a/ext/bg/settings.html b/ext/bg/settings.html
index cfe20be4..0b2e4f9c 100644
--- a/ext/bg/settings.html
+++ b/ext/bg/settings.html
@@ -163,6 +163,18 @@
                 </div>
 
                 <div class="checkbox options-advanced">
+                    <label><input type="checkbox" id="show-pitch-accent-downstep-notation"> Show downstep notation for pitch accents</label>
+                </div>
+
+                <div class="checkbox options-position">
+                    <label><input type="checkbox" id="show-pitch-accent-position-notation"> Show position notation for pitch accents</label>
+                </div>
+
+                <div class="checkbox options-advanced">
+                    <label><input type="checkbox" id="show-pitch-accent-graph"> Show graph for pitch accents</label>
+                </div>
+
+                <div class="checkbox options-advanced">
                     <label><input type="checkbox" id="show-debug-info"> Show debug information</label>
                 </div>