add phonetic transcriptions term meta type (#434)

* move dictionary files to dictionary folder

* wip

* move dictionary files to dictionary folder

* add ipa term meta

* wip

* fixing comments wip

* fixing comments wip

* fixing comments wip

* fixing comments wip

* fixing comments wip

* fixing comments wip

* fix comments

* fix comments

* update test data

* fix gitignore

* engines

* add tests

* update database test

* fix test

5 files changed, 102 insertions, 20 deletions

diff --git a/types/ext/anki-templates.d.ts b/types/ext/anki-templates.d.ts
index 5c40f406..098873e6 100644
--- a/types/ext/anki-templates.d.ts
+++ b/types/ext/anki-templates.d.ts
@@ -76,6 +76,7 @@ export type NoteData = {
     readonly uniqueReadings: string[];
     readonly pitches: PitchGroup[];
     readonly pitchCount: number;
+    readonly phoneticTranscriptions: TranscriptionGroup[];
     readonly context: Context;
     media: Media;
     readonly dictionaryEntry: Dictionary.DictionaryEntry;
@@ -97,6 +98,20 @@ export type Pitch = {
     exclusiveReadings: string[];
 };
 
+export type TranscriptionGroup = {
+    dictionary: string;
+    phoneticTranscriptions: Transcription[];
+};
+
+export type Transcription = {
+    expressions: string[];
+    reading: string;
+    ipa: string;
+    tags: Dictionary.Tag[];
+    exclusiveExpressions: string[];
+    exclusiveReadings: string[];
+};
+
 /**
  * For legacy reasons, {@link Pitch} has a custom tag type that resembles {@link Dictionary.Tag}.
  */
@@ -175,7 +190,8 @@ export type TermDictionaryEntry = {
     readonly termTags?: Tag[];
     readonly definitions?: TermDefinition[];
     readonly frequencies: TermFrequency[];
-    readonly pitches: TermPronunciation[];
+    readonly pitches: TermPitchAccent[];
+    readonly phoneticTranscriptions: TermPhoneticTranscription[];
     sourceTermExactMatchCount: number;
     url: string;
     readonly cloze: Cloze;
@@ -225,7 +241,7 @@ export type TermFrequency = {
     frequency: number | string;
 };
 
-export type TermPronunciation = {
+export type TermPitchAccent = {
     index: number;
     expressionIndex: number;
     dictionary: string;
@@ -235,14 +251,32 @@ export type TermPronunciation = {
     };
     expression: string;
     reading: string;
-    readonly pitches: TermPitch[];
+    readonly pitches: PitchAccent[];
 };
 
-export type TermPitch = {
+export type PitchAccent = {
     position: number;
     tags: Tag[];
 };
 
+export type TermPhoneticTranscription = {
+    index: number;
+    expressionIndex: number;
+    dictionary: string;
+    dictionaryOrder: {
+        index: number;
+        priority: number;
+    };
+    expression: string;
+    reading: string;
+    readonly phoneticTranscriptions: PhoneticTranscription[];
+};
+
+export type PhoneticTranscription = {
+    ipa: string;
+    tags: Tag[];
+};
+
 export type TermFrequencyType = DictionaryDataUtil.TermFrequencyType;
 
 export type TermHeadword = {
@@ -251,7 +285,7 @@ export type TermHeadword = {
     reading: string;
     readonly termTags: Tag[];
     readonly frequencies: TermFrequency[];
-    readonly pitches: TermPronunciation[];
+    readonly pitches: TermPitchAccent[];
     readonly furiganaSegments: FuriganaSegment[];
     readonly termFrequency: TermFrequencyType;
     wordClasses: string[];
diff --git a/types/ext/dictionary-data-util.d.ts b/types/ext/dictionary-data-util.d.ts
index b78e6439..4ab06f11 100644
--- a/types/ext/dictionary-data-util.d.ts
+++ b/types/ext/dictionary-data-util.d.ts
@@ -64,21 +64,15 @@ export type KanjiFrequency = {
 export type TermFrequencyType = 'popular' | 'rare' | 'normal';
 
 export type GroupedPronunciationInternal = {
+    pronunciation: Dictionary.Pronunciation;
     terms: Set<string>;
     reading: string;
-    position: number;
-    nasalPositions: number[];
-    devoicePositions: number[];
-    tags: Dictionary.Tag[];
 };
 
 export type GroupedPronunciation = {
+    pronunciation: Dictionary.Pronunciation;
     terms: string[];
     reading: string;
-    position: number;
-    nasalPositions: number[];
-    devoicePositions: number[];
-    tags: Dictionary.Tag[];
     exclusiveTerms: string[];
     exclusiveReadings: string[];
 };
diff --git a/types/ext/dictionary-data.d.ts b/types/ext/dictionary-data.d.ts
index b194c190..0e0edd5c 100644
--- a/types/ext/dictionary-data.d.ts
+++ b/types/ext/dictionary-data.d.ts
@@ -125,7 +125,7 @@ export type GenericFrequencyData = string | number | {
 
 export type TermMetaArray = TermMeta[];
 
-export type TermMeta = TermMetaFrequency | TermMetaPitch;
+export type TermMeta = TermMetaFrequency | TermMetaPitch | TermMetaPhonetic;
 
 export type TermMetaFrequencyDataWithReading = {
     reading: string;
@@ -154,6 +154,20 @@ export type TermMetaPitch = [
     data: TermMetaPitchData,
 ];
 
+export type TermMetaPhonetic = [
+    expression: string,
+    mode: 'ipa',
+    data: TermMetaPhoneticData,
+];
+
+export type TermMetaPhoneticData = {
+    reading: string;
+    transcriptions: {
+        ipa: string;
+        tags?: string[];
+    }[];
+};
+
 export type KanjiMetaArray = KanjiMeta[];
 
 export type KanjiMeta = KanjiMetaFrequency;
diff --git a/types/ext/dictionary-database.d.ts b/types/ext/dictionary-database.d.ts
index 3202ef60..3cf68543 100644
--- a/types/ext/dictionary-database.d.ts
+++ b/types/ext/dictionary-database.d.ts
@@ -102,7 +102,7 @@ export type Tag = {
     dictionary: string;
 };
 
-export type DatabaseTermMeta = DatabaseTermMetaFrequency | DatabaseTermMetaPitch;
+export type DatabaseTermMeta = DatabaseTermMetaFrequency | DatabaseTermMetaPitch | DatabaseTermMetaPhoneticData;
 
 export type DatabaseTermMetaFrequency = {
     expression: string;
@@ -118,12 +118,19 @@ export type DatabaseTermMetaPitch = {
     dictionary: string;
 };
 
+export type DatabaseTermMetaPhoneticData = {
+    expression: string;
+    mode: 'ipa';
+    data: DictionaryData.TermMetaPhoneticData;
+    dictionary: string;
+};
+
 export type TermMetaFrequencyDataWithReading = {
     reading: string;
     frequency: DictionaryData.GenericFrequencyData;
 };
 
-export type TermMeta = TermMetaFrequency | TermMetaPitch;
+export type TermMeta = TermMetaFrequency | TermMetaPitch | TermMetaPhoneticData;
 
 export type TermMetaType = TermMeta['mode'];
 
@@ -136,13 +143,21 @@ export type TermMetaFrequency = {
 };
 
 export type TermMetaPitch = {
+    mode: 'pitch';
     index: number;
     term: string;
-    mode: 'pitch';
     data: DictionaryData.TermMetaPitchData;
     dictionary: string;
 };
 
+export type TermMetaPhoneticData = {
+    mode: 'ipa';
+    index: number;
+    term: string;
+    data: DictionaryData.TermMetaPhoneticData;
+    dictionary: string;
+};
+
 export type DatabaseKanjiMeta = DatabaseKanjiMetaFrequency;
 
 export type DatabaseKanjiMetaFrequency = {
diff --git a/types/ext/dictionary.d.ts b/types/ext/dictionary.d.ts
index 3e90dec0..7c348e7f 100644
--- a/types/ext/dictionary.d.ts
+++ b/types/ext/dictionary.d.ts
@@ -365,15 +365,21 @@ export type TermPronunciation = {
      */
     dictionaryPriority: number;
     /**
-     * The pitch accent representations for the term.
+     * The pronunciations for the term.
      */
-    pitches: TermPitch[];
+    pronunciations: Pronunciation[];
 };
 
+export type Pronunciation = PitchAccent | PhoneticTranscription;
+
 /**
  * Pitch accent information for a term, represented as the position of the downstep.
  */
-export type TermPitch = {
+export type PitchAccent = {
+    /**
+     * Type of the pronunciation, for disambiguation between union type members.
+     */
+    type: 'pitch-accent';
     /**
      * Position of the downstep, as a number of mora.
      */
@@ -392,6 +398,25 @@ export type TermPitch = {
     tags: Tag[];
 };
 
+export type PhoneticTranscription = {
+    /**
+     * Type of the pronunciation, for disambiguation between union type members.
+     */
+    type: 'phonetic-transcription';
+    /**
+     * An IPA transcription.
+     */
+    ipa: string;
+    /**
+     * Tags for the IPA transcription.
+     */
+    tags: Tag[];
+};
+
+export type PronunciationType = Pronunciation['type'];
+
+export type PronunciationGeneric<T extends PronunciationType> = Extract<Pronunciation, {type: T}>;
+
 /**
  * Frequency information corresponds to how frequently a term appears in a corpus,
  * which can be a number of occurrences or an overall rank.