add lingua libre audio source (#1129)

* add lingua libre audio source

* mvp

* run file requests in parallel

* remove redundant language var

* redundant api function

---------

Co-authored-by: Cashew <52880648+cashewnuttynuts@users.noreply.github.com>

10 files changed, 133 insertions, 48 deletions

diff --git a/ext/js/background/backend.js b/ext/js/background/backend.js
index a468a3f6..96e8206b 100644
--- a/ext/js/background/backend.js
+++ b/ext/js/background/backend.js
@@ -676,8 +676,8 @@ export class Backend {
     }
 
     /** @type {import('api').ApiHandler<'getTermAudioInfoList'>} */
-    async _onApiGetTermAudioInfoList({source, term, reading}) {
-        return await this._audioDownloader.getTermAudioInfoList(source, term, reading);
+    async _onApiGetTermAudioInfoList({source, term, reading, languageSummary}) {
+        return await this._audioDownloader.getTermAudioInfoList(source, term, reading, languageSummary);
     }
 
     /** @type {import('api').ApiHandler<'sendMessageToFrame'>} */
@@ -2171,7 +2171,7 @@ export class Backend {
         const {term, reading} = definitionDetails;
         if (term.length === 0 && reading.length === 0) { return null; }
 
-        const {sources, preferredAudioIndex, idleTimeout} = details;
+        const {sources, preferredAudioIndex, idleTimeout, languageSummary} = details;
         let data;
         let contentType;
         try {
@@ -2181,6 +2181,7 @@ export class Backend {
                 term,
                 reading,
                 idleTimeout,
+                languageSummary,
             ));
         } catch (e) {
             const error = this._getAudioDownloadError(e);
diff --git a/ext/js/comm/api.js b/ext/js/comm/api.js
index 7a98feb1..952b5bfa 100644
--- a/ext/js/comm/api.js
+++ b/ext/js/comm/api.js
@@ -147,10 +147,11 @@ export class API {
      * @param {import('api').ApiParam<'getTermAudioInfoList', 'source'>} source
      * @param {import('api').ApiParam<'getTermAudioInfoList', 'term'>} term
      * @param {import('api').ApiParam<'getTermAudioInfoList', 'reading'>} reading
+     * @param {import('api').ApiParam<'getTermAudioInfoList', 'languageSummary'>} languageSummary
      * @returns {Promise<import('api').ApiReturn<'getTermAudioInfoList'>>}
      */
-    getTermAudioInfoList(source, term, reading) {
-        return this._invoke('getTermAudioInfoList', {source, term, reading});
+    getTermAudioInfoList(source, term, reading, languageSummary) {
+        return this._invoke('getTermAudioInfoList', {source, term, reading, languageSummary});
     }
 
     /**
diff --git a/ext/js/data/anki-note-builder.js b/ext/js/data/anki-note-builder.js
index 38588439..7f7fe3a7 100644
--- a/ext/js/data/anki-note-builder.js
+++ b/ext/js/data/anki-note-builder.js
@@ -428,8 +428,8 @@ export class AnkiNoteBuilder {
         if (injectAudio && dictionaryEntryDetails.type !== 'kanji') {
             const audioOptions = mediaOptions.audio;
             if (typeof audioOptions === 'object' && audioOptions !== null) {
-                const {sources, preferredAudioIndex, idleTimeout} = audioOptions;
-                audioDetails = {sources, preferredAudioIndex, idleTimeout};
+                const {sources, preferredAudioIndex, idleTimeout, languageSummary} = audioOptions;
+                audioDetails = {sources, preferredAudioIndex, idleTimeout, languageSummary};
             }
         }
         if (injectScreenshot) {
diff --git a/ext/js/display/display-anki.js b/ext/js/display/display-anki.js
index fc242549..b3b05408 100644
--- a/ext/js/display/display-anki.js
+++ b/ext/js/display/display-anki.js
@@ -193,7 +193,11 @@ export class DisplayAnki {
      */
     _onOptionsUpdated({options}) {
         const {
-            general: {resultOutputMode, glossaryLayoutMode, compactTags},
+            general: {
+                resultOutputMode,
+                glossaryLayoutMode,
+                compactTags,
+            },
             dictionaries,
             anki: {
                 tags,
@@ -883,7 +887,13 @@ export class DisplayAnki {
     _getAnkiNoteMediaAudioDetails(details) {
         if (details.type !== 'term') { return null; }
         const {sources, preferredAudioIndex} = this._displayAudio.getAnkiNoteMediaAudioDetails(details.term, details.reading);
-        return {sources, preferredAudioIndex, idleTimeout: this._audioDownloadIdleTimeout};
+        const languageSummary = this._display.getLanguageSummary();
+        return {
+            sources,
+            preferredAudioIndex,
+            idleTimeout: this._audioDownloadIdleTimeout,
+            languageSummary,
+        };
     }
 
     // View note functions
diff --git a/ext/js/display/display-audio.js b/ext/js/display/display-audio.js
index c7e08ffe..0d1ca029 100644
--- a/ext/js/display/display-audio.js
+++ b/ext/js/display/display-audio.js
@@ -57,6 +57,7 @@ export class DisplayAudio {
             ['jpod101', 'JapanesePod101'],
             ['jpod101-alternate', 'JapanesePod101 (Alternate)'],
             ['jisho', 'Jisho.org'],
+            ['lingua-libre', 'Lingua Libre'],
             ['text-to-speech', 'Text-to-speech'],
             ['text-to-speech-reading', 'Text-to-speech (Kana reading)'],
             ['custom', 'Custom URL'],
@@ -677,7 +678,8 @@ export class DisplayAudio {
      */
     async _getTermAudioInfoList(source, term, reading) {
         const sourceData = this._getSourceData(source);
-        const infoList = await this._display.application.api.getTermAudioInfoList(sourceData, term, reading);
+        const languageSummary = this._display.getLanguageSummary();
+        const infoList = await this._display.application.api.getTermAudioInfoList(sourceData, term, reading, languageSummary);
         return infoList.map((info) => ({info, audioPromise: null, audioResolved: false, audio: null}));
     }
 
diff --git a/ext/js/display/display.js b/ext/js/display/display.js
index f86d7b8c..3d18e416 100644
--- a/ext/js/display/display.js
+++ b/ext/js/display/display.js
@@ -192,6 +192,8 @@ export class Display extends EventDispatcher {
         this._onMenuButtonMenuCloseBind = this._onMenuButtonMenuClose.bind(this);
         /** @type {ThemeController} */
         this._themeController = new ThemeController(document.documentElement);
+        /** @type {import('language').LanguageSummary[]} */
+        this._languageSummaries = [];
 
         /* eslint-disable @stylistic/no-multi-spaces */
         this._hotkeyHandler.registerActions([
@@ -316,6 +318,8 @@ export class Display extends EventDispatcher {
             documentElement.dataset.browser = browser;
         }
 
+        this._languageSummaries = await this._application.api.getLanguageSummaries();
+
         // Prepare
         await this._hotkeyHelpController.prepare(this._application.api);
         await this._displayGenerator.prepare();
@@ -398,6 +402,16 @@ export class Display extends EventDispatcher {
     }
 
     /**
+     * @returns {import('language').LanguageSummary}
+     * @throws {Error}
+     */
+    getLanguageSummary() {
+        if (this._options === null) { throw new Error('Options is null'); }
+        const language = this._options.general.language;
+        return /** @type {import('language').LanguageSummary} */ (this._languageSummaries.find(({iso}) => iso === language));
+    }
+
+    /**
      * @returns {import('settings').OptionsContext}
      */
     getOptionsContext() {
diff --git a/ext/js/language/language-descriptors.js b/ext/js/language/language-descriptors.js
index f9fb4f09..2e8ece55 100644
--- a/ext/js/language/language-descriptors.js
+++ b/ext/js/language/language-descriptors.js
@@ -50,6 +50,7 @@ const capitalizationPreprocessors = {
 const languageDescriptors = [
     {
         iso: 'ar',
+        iso639_3: 'ara',
         name: 'Arabic',
         exampleText: 'قَرَأَ',
         textPreprocessors: {
@@ -58,6 +59,7 @@ const languageDescriptors = [
     },
     {
         iso: 'de',
+        iso639_3: 'deu',
         name: 'German',
         exampleText: 'gelesen',
         textPreprocessors: {
@@ -68,12 +70,14 @@ const languageDescriptors = [
     },
     {
         iso: 'el',
+        iso639_3: 'ell',
         name: 'Greek',
         exampleText: 'διαβάζω',
         textPreprocessors: capitalizationPreprocessors,
     },
     {
         iso: 'en',
+        iso639_3: 'eng',
         name: 'English',
         exampleText: 'read',
         textPreprocessors: capitalizationPreprocessors,
@@ -81,6 +85,7 @@ const languageDescriptors = [
     },
     {
         iso: 'es',
+        iso639_3: 'spa',
         name: 'Spanish',
         exampleText: 'leer',
         textPreprocessors: capitalizationPreprocessors,
@@ -88,6 +93,7 @@ const languageDescriptors = [
     },
     {
         iso: 'fa',
+        iso639_3: 'fas',
         name: 'Persian',
         exampleText: 'خواندن',
         textPreprocessors: {
@@ -96,18 +102,21 @@ const languageDescriptors = [
     },
     {
         iso: 'fi',
+        iso639_3: 'fin',
         name: 'Finnish',
         exampleText: 'lukea',
         textPreprocessors: capitalizationPreprocessors,
     },
     {
         iso: 'fr',
+        iso639_3: 'fra',
         name: 'French',
         exampleText: 'lire',
         textPreprocessors: capitalizationPreprocessors,
     },
     {
         iso: 'grc',
+        iso639_3: 'grc',
         name: 'Ancient Greek',
         exampleText: 'γράφω',
         textPreprocessors: {
@@ -117,24 +126,28 @@ const languageDescriptors = [
     },
     {
         iso: 'hu',
+        iso639_3: 'hun',
         name: 'Hungarian',
         exampleText: 'olvasni',
         textPreprocessors: capitalizationPreprocessors,
     },
     {
         iso: 'id',
+        iso639_3: 'ind',
         name: 'Indonesian',
         exampleText: 'membaca',
         textPreprocessors: capitalizationPreprocessors,
     },
     {
         iso: 'it',
+        iso639_3: 'ita',
         name: 'Italian',
         exampleText: 'leggere',
         textPreprocessors: capitalizationPreprocessors,
     },
     {
         iso: 'la',
+        iso639_3: 'lat',
         name: 'Latin',
         exampleText: 'legere',
         textPreprocessors: {
@@ -145,11 +158,13 @@ const languageDescriptors = [
     },
     {
         iso: 'lo',
+        iso639_3: 'lao',
         name: 'Lao',
         exampleText: 'ອ່ານ',
     },
     {
         iso: 'ja',
+        iso639_3: 'jpn',
         name: 'Japanese',
         exampleText: '読め',
         isTextLookupWorthy: isStringPartiallyJapanese,
@@ -165,11 +180,13 @@ const languageDescriptors = [
     },
     {
         iso: 'km',
+        iso639_3: 'khm',
         name: 'Khmer',
         exampleText: 'អាន',
     },
     {
         iso: 'ko',
+        iso639_3: 'kor',
         name: 'Korean',
         exampleText: '읽어',
         textPreprocessors: {
@@ -182,24 +199,28 @@ const languageDescriptors = [
     },
     {
         iso: 'nl',
+        iso639_3: 'nld',
         name: 'Dutch',
         exampleText: 'lezen',
         textPreprocessors: capitalizationPreprocessors,
     },
     {
         iso: 'pl',
+        iso639_3: 'pol',
         name: 'Polish',
         exampleText: 'czytacie',
         textPreprocessors: capitalizationPreprocessors,
     },
     {
         iso: 'pt',
+        iso639_3: 'por',
         name: 'Portuguese',
         exampleText: 'ler',
         textPreprocessors: capitalizationPreprocessors,
     },
     {
         iso: 'ro',
+        iso639_3: 'ron',
         name: 'Romanian',
         exampleText: 'citit',
         textPreprocessors: {
@@ -209,6 +230,7 @@ const languageDescriptors = [
     },
     {
         iso: 'ru',
+        iso639_3: 'rus',
         name: 'Russian',
         exampleText: 'читать',
         textPreprocessors: {
@@ -219,6 +241,7 @@ const languageDescriptors = [
     },
     {
         iso: 'sga',
+        iso639_3: 'sga',
         name: 'Old Irish',
         exampleText: 'légaid',
         textPreprocessors: {
@@ -229,6 +252,7 @@ const languageDescriptors = [
     },
     {
         iso: 'sh',
+        iso639_3: 'hbs',
         name: 'Serbo-Croatian',
         exampleText: 'čitaše',
         textPreprocessors: {
@@ -238,6 +262,7 @@ const languageDescriptors = [
     },
     {
         iso: 'sq',
+        iso639_3: 'sqi',
         name: 'Albanian',
         exampleText: 'ndihmojme',
         textPreprocessors: capitalizationPreprocessors,
@@ -245,23 +270,27 @@ const languageDescriptors = [
     },
     {
         iso: 'sv',
+        iso639_3: 'swe',
         name: 'Swedish',
         exampleText: 'läsa',
         textPreprocessors: capitalizationPreprocessors,
     },
     {
         iso: 'th',
+        iso639_3: 'tha',
         name: 'Thai',
         exampleText: 'อ่าน',
     },
     {
         iso: 'tr',
+        iso639_3: 'tur',
         name: 'Turkish',
         exampleText: 'okuyor',
         textPreprocessors: capitalizationPreprocessors,
     },
     {
         iso: 'vi',
+        iso639_3: 'vie',
         name: 'Vietnamese',
         exampleText: 'đọc',
         textPreprocessors: {
@@ -271,11 +300,13 @@ const languageDescriptors = [
     },
     {
         iso: 'yue',
+        iso639_3: 'yue',
         name: 'Cantonese',
         exampleText: '讀',
     },
     {
         iso: 'zh',
+        iso639_3: 'zho',
         name: 'Chinese',
         exampleText: '读',
         isTextLookupWorthy: isStringPartiallyChinese,
diff --git a/ext/js/language/languages.js b/ext/js/language/languages.js
index 7759fda5..96f7a080 100755
--- a/ext/js/language/languages.js
+++ b/ext/js/language/languages.js
@@ -22,8 +22,8 @@ import {languageDescriptorMap} from './language-descriptors.js';
  */
 export function getLanguageSummaries() {
     const results = [];
-    for (const {name, iso, exampleText} of languageDescriptorMap.values()) {
-        results.push({name, iso, exampleText});
+    for (const {name, iso, iso639_3, exampleText} of languageDescriptorMap.values()) {
+        results.push({name, iso, iso639_3, exampleText});
     }
     return results;
 }
diff --git a/ext/js/media/audio-downloader.js b/ext/js/media/audio-downloader.js
index 2d1bc4ec..d378d043 100644
--- a/ext/js/media/audio-downloader.js
+++ b/ext/js/media/audio-downloader.js
@@ -25,6 +25,16 @@ import {NativeSimpleDOMParser} from '../dom/native-simple-dom-parser.js';
 import {SimpleDOMParser} from '../dom/simple-dom-parser.js';
 import {isStringEntirelyKana} from '../language/ja/japanese.js';
 
+/** @type {RequestInit} */
+const DEFAULT_REQUEST_INIT_PARAMS = {
+    method: 'GET',
+    mode: 'cors',
+    cache: 'default',
+    credentials: 'omit',
+    redirect: 'follow',
+    referrerPolicy: 'no-referrer',
+};
+
 export class AudioDownloader {
     /**
      * @param {RequestBuilder} requestBuilder
@@ -39,6 +49,7 @@ export class AudioDownloader {
             ['jpod101', this._getInfoJpod101.bind(this)],
             ['jpod101-alternate', this._getInfoJpod101Alternate.bind(this)],
             ['jisho', this._getInfoJisho.bind(this)],
+            ['lingua-libre', this._getInfoLinguaLibre.bind(this)],
             ['text-to-speech', this._getInfoTextToSpeech.bind(this)],
             ['text-to-speech-reading', this._getInfoTextToSpeechReading.bind(this)],
             ['custom', this._getInfoCustom.bind(this)],
@@ -50,13 +61,14 @@ export class AudioDownloader {
      * @param {import('audio').AudioSourceInfo} source
      * @param {string} term
      * @param {string} reading
+     * @param {import('language').LanguageSummary} languageSummary
      * @returns {Promise<import('audio-downloader').Info[]>}
      */
-    async getTermAudioInfoList(source, term, reading) {
+    async getTermAudioInfoList(source, term, reading, languageSummary) {
         const handler = this._getInfoHandlers.get(source.type);
         if (typeof handler === 'function') {
             try {
-                return await handler(term, reading, source);
+                return await handler(term, reading, source, languageSummary);
             } catch (e) {
                 // NOP
             }
@@ -70,12 +82,13 @@ export class AudioDownloader {
      * @param {string} term
      * @param {string} reading
      * @param {?number} idleTimeout
+     * @param {import('language').LanguageSummary} languageSummary
      * @returns {Promise<import('audio-downloader').AudioBinaryBase64>}
      */
-    async downloadTermAudio(sources, preferredAudioIndex, term, reading, idleTimeout) {
+    async downloadTermAudio(sources, preferredAudioIndex, term, reading, idleTimeout, languageSummary) {
         const errors = [];
         for (const source of sources) {
-            let infoList = await this.getTermAudioInfoList(source, term, reading);
+            let infoList = await this.getTermAudioInfoList(source, term, reading, languageSummary);
             if (typeof preferredAudioIndex === 'number') {
                 infoList = (preferredAudioIndex >= 0 && preferredAudioIndex < infoList.length ? [infoList[preferredAudioIndex]] : []);
             }
@@ -137,12 +150,8 @@ export class AudioDownloader {
             vulgar: 'true',
         });
         const response = await this._requestBuilder.fetchAnonymous(fetchUrl, {
+            ...DEFAULT_REQUEST_INIT_PARAMS,
             method: 'POST',
-            mode: 'cors',
-            cache: 'default',
-            credentials: 'omit',
-            redirect: 'follow',
-            referrerPolicy: 'no-referrer',
             headers: {
                 'Content-Type': 'application/x-www-form-urlencoded',
             },
@@ -181,14 +190,7 @@ export class AudioDownloader {
     /** @type {import('audio-downloader').GetInfoHandler} */
     async _getInfoJisho(term, reading) {
         const fetchUrl = `https://jisho.org/search/${term}`;
-        const response = await this._requestBuilder.fetchAnonymous(fetchUrl, {
-            method: 'GET',
-            mode: 'cors',
-            cache: 'default',
-            credentials: 'omit',
-            redirect: 'follow',
-            referrerPolicy: 'no-referrer',
-        });
+        const response = await this._requestBuilder.fetchAnonymous(fetchUrl, DEFAULT_REQUEST_INIT_PARAMS);
         const responseText = await response.text();
 
         const dom = this._createSimpleDOMParser(responseText);
@@ -212,6 +214,44 @@ export class AudioDownloader {
     }
 
     /** @type {import('audio-downloader').GetInfoHandler} */
+    async _getInfoLinguaLibre(term, _reading, _details, languageSummary) {
+        if (typeof languageSummary !== 'object' || languageSummary === null) {
+            throw new Error('Invalid arguments');
+        }
+        const {iso639_3} = languageSummary;
+        const searchCategory = `incategory:"Lingua_Libre_pronunciation-${iso639_3}"`;
+        const searchString = `-${term}.wav`;
+        const fetchUrl = `https://commons.wikimedia.org/w/api.php?action=query&format=json&list=search&srsearch=intitle:/${searchString}/i+${searchCategory}&srnamespace=6&origin=*`;
+
+        const response = await this._requestBuilder.fetchAnonymous(fetchUrl, DEFAULT_REQUEST_INIT_PARAMS);
+
+        /** @type {import('audio-downloader').LinguaLibreLookupResponse} */
+        const lookupResponse = await readResponseJson(response);
+
+        const lookupResults = lookupResponse.query.search;
+
+        const fetchFileInfos = lookupResults.map(async ({title}) => {
+            const fileInfoURL = `https://commons.wikimedia.org/w/api.php?action=query&format=json&titles=${title}&prop=imageinfo&iiprop=user|url&origin=*`;
+            const response2 = await this._requestBuilder.fetchAnonymous(fileInfoURL, DEFAULT_REQUEST_INIT_PARAMS);
+            /** @type {import('audio-downloader').LinguaLibreFileResponse} */
+            const fileResponse = await readResponseJson(response2);
+            const fileResults = fileResponse.query.pages;
+            const results = [];
+            for (const page of Object.values(fileResults)) {
+                const fileUrl = page.imageinfo[0].url;
+                const fileUser = page.imageinfo[0].user;
+                const validFilenameTest = new RegExp(`^File:LL-Q\\d+\\s+\\(${iso639_3}\\)-(\\w+ \\()?${fileUser}\\)?-${term}\\.wav$`, 'i');
+                if (validFilenameTest.test(title)) {
+                    results.push({type: 'url', url: fileUrl, name: fileUser});
+                }
+            }
+            return /** @type {import('audio-downloader').Info1[]} */ (results);
+        });
+
+        return (await Promise.all(fetchFileInfos)).flat();
+    }
+
+    /** @type {import('audio-downloader').GetInfoHandler} */
     async _getInfoTextToSpeech(term, reading, details) {
         if (typeof details !== 'object' || details === null) {
             throw new Error('Invalid arguments');
@@ -259,14 +299,7 @@ export class AudioDownloader {
         }
         url = this._getCustomUrl(term, reading, url);
 
-        const response = await this._requestBuilder.fetchAnonymous(url, {
-            method: 'GET',
-            mode: 'cors',
-            cache: 'default',
-            credentials: 'omit',
-            redirect: 'follow',
-            referrerPolicy: 'no-referrer',
-        });
+        const response = await this._requestBuilder.fetchAnonymous(url, DEFAULT_REQUEST_INIT_PARAMS);
 
         if (!response.ok) {
             throw new Error(`Invalid response: ${response.status}`);
@@ -345,12 +378,7 @@ export class AudioDownloader {
         }
 
         const response = await this._requestBuilder.fetchAnonymous(url, {
-            method: 'GET',
-            mode: 'cors',
-            cache: 'default',
-            credentials: 'omit',
-            redirect: 'follow',
-            referrerPolicy: 'no-referrer',
+            ...DEFAULT_REQUEST_INIT_PARAMS,
             signal,
         });
 
@@ -429,12 +457,8 @@ export class AudioDownloader {
     async _getCustomAudioListSchema() {
         const url = chrome.runtime.getURL('/data/schemas/custom-audio-list-schema.json');
         const response = await fetch(url, {
-            method: 'GET',
+            ...DEFAULT_REQUEST_INIT_PARAMS,
             mode: 'no-cors',
-            cache: 'default',
-            credentials: 'omit',
-            redirect: 'follow',
-            referrerPolicy: 'no-referrer',
         });
         return await readResponseJson(response);
     }
diff --git a/ext/js/pages/settings/audio-controller.js b/ext/js/pages/settings/audio-controller.js
index 34d7adaa..ac79b51a 100644
--- a/ext/js/pages/settings/audio-controller.js
+++ b/ext/js/pages/settings/audio-controller.js
@@ -221,6 +221,7 @@ export class AudioController extends EventDispatcher {
             'jpod101',
             'jpod101-alternate',
             'jisho',
+            'lingua-libre',
             'custom',
         ];
         for (const type of typesAvailable) {
@@ -488,6 +489,7 @@ class AudioSourceEntry {
             case 'jpod101':
             case 'jpod101-alternate':
             case 'jisho':
+            case 'lingua-libre':
             case 'text-to-speech':
             case 'text-to-speech-reading':
             case 'custom':