README.md

About

Scripts to download audio from Jisho dictionary entries

Bookmarklet

The JavaScript can be stored in a browser bookmark, clicking the bookmark will initiate the download

Closure Compiled Bookmarklet

Functionally the same at the original bookmarklet, but optimized using Google's Closure Compiler

Userscript

Creates a "Download Audio" link below the "Play Audio" link on Jisho dictionary entries

Usage

Bookmarklet

Add a bookmark and set the URL as javascript: and then paste the closure compiled bookmarklet

Userscript

Add the userscript to your userscript manager

jisho-audio-downloader-bookmarklet-closure-compiled.js

(function(){function l(g,h){var m=window.URL||window.webkitURL,b=new XMLHttpRequest,c=document.createElement("a");b.open("GET","https://cors-anywhere.herokuapp.com/"+g,!0);b.responseType="blob";b.onerror=function(){c.href=g;c.download=h;c.click()};b.onload=function(){c.href=m.createObjectURL(new Blob([b.response],{type:"application/octet-stream"}));c.download=h;c.click()};b.send()}for(var k=document.getElementsByTagName("audio"),e=0;e<k.length;++e){var a=k[e],d=a.id,f=a.nextElementSibling;f&&f.dataset&&f.dataset.id==d&&(a=a.children[0].src,d=(!d||7>d.length?"unknown":d.match("(?<=_)(.*?)(?=:)")[0])+(a&&0!==a.length?"."+a.split(".").pop(-1):""),l(a,d))}})();

jisho-audio-downloader-bookmarklet.js

(function() {
    'use strict';
    function getWordNameFromAudioID(audioID) {
        if (!audioID || audioID.length < 7) return "unknown";

        // audioID is of the format: audio_元気:げんき
        // The regex below grabs the kanji between the underscore and the semicolon
        const regex = '(?<=_)(.*?)(?=:)';
        const match = audioID.match(regex);
        return match[0];
    }

    function getFileExtension(file) {
        if (!file || file.length === 0) return "";

        // Splits the file url and gets the extension
        const fileExtension = '.' + file.split('.').pop(-1);
        return fileExtension;
    }

    function downloadFileWithFileName(url, fileName) {
        const URL = window.URL || window.webkitURL;

        const xhr = new XMLHttpRequest()
        const a = document.createElement('a')

        // The server the audio files are hosted on does not have different origin policy
        // This proxy server will append an appropriate header to the response, allowing the browser to load it
        const proxy = 'https://cors-anywhere.herokuapp.com/';

        xhr.open('GET', proxy + url, true);
        xhr.responseType = 'blob';
        xhr.onerror = function() {
            a.href = url;
            a.download = fileName;
            a.click();
        };
        xhr.onload = function() {
            const file = new Blob([xhr.response], { type : 'application/octet-stream' });
            a.href = URL.createObjectURL(file);
            a.download = fileName;
            a.click();
        };
        xhr.send();
    }

    const audioElements = document.getElementsByTagName('audio');
    for (let i = 0; i < audioElements.length; ++i) {
        const audioElement = audioElements[i];
        const audioID = audioElement.id;
        const audioLinkElement = audioElement.nextElementSibling;
        if (!audioLinkElement || !audioLinkElement.dataset || audioLinkElement.dataset.id != audioID) {
            continue; // @TODO: Site layout has changed, skip this or show alert?
        }
        const url = audioElement.children[0].src;
        const fileName = getWordNameFromAudioID(audioID) + getFileExtension(url);
        downloadFileWithFileName(url, fileName);
    }
})();

jisho-audio-downloader-userscript.js

// ==UserScript==
// @name         Jisho Audio Downloader
// @namespace    http://miere.ru/
// @version      0.3
// @description  Adds download audio button to Jisho.org pages which downloads word audio clips.
// @author       Vladislav <miere> Vorobiev, Rob Weiss
// @match        http://jisho.org/search/*
// @match        http://jisho.org/word/*
// @grant        none
// ==/UserScript==

(function() {
    'use strict';
    function getWordNameFromAudioID(audioID) {
        if (!audioID || audioID.length < 7) return "unknown";
        
        // audioID is of the format: audio_元気:げんき
        // The regex below grabs the kanji between the underscore and the semicolon
        const regex = '(?<=_)(.*?)(?=:)';
        const match = audioID.match(regex);
        return match[0];
    }

    function getFileExtension(file) {
        if (!file || file.length === 0) return "";

        // Splits the file url and gets the extension
        const fileExtension = '.' + file.split('.').pop(-1);
        return fileExtension;
    }

    function insertNodeAfter(insertNode, afterNode) {
        afterNode.parentNode.insertBefore(insertNode, afterNode.nextSibling);
    }

    function downloadFileWithFileName(url, fileName) {
        const URL = window.URL || window.webkitURL;

        const xhr = new XMLHttpRequest()
        const a = document.createElement('a')

        // The server the audio files are hosted on does not have different origin policy
        // This proxy server will append an appropriate header to the response, allowing the browser to load it
        const proxy = 'https://cors-anywhere.herokuapp.com/';

        xhr.open('GET', proxy + url, true);
        xhr.responseType = 'blob';
        xhr.onerror = function() {
            a.href = url;
            a.download = fileName;
            a.click();
        };
        xhr.onload = function() {
            const file = new Blob([xhr.response], { type : 'application/octet-stream' });
            a.href = URL.createObjectURL(file);
            a.download = fileName;
            a.click();
        };
        xhr.send();
    }

    const audioElements = document.getElementsByTagName('audio');
    for (let i = 0; i < audioElements.length; ++i) {
        const audioElement = audioElements[i];
        const audioID = audioElement.id;
        const audioLinkElement = audioElement.nextElementSibling;
        if (!audioLinkElement || !audioLinkElement.dataset || audioLinkElement.dataset.id != audioID) {
            continue; // @TODO: Site layout has changed, skip this or show alert?
        }
        let downloadAudioElement = document.createElement('a');
        downloadAudioElement.className = 'concept_light-status_link';
        downloadAudioElement.dataset.href = audioElement.children[0].src;
        downloadAudioElement.dataset.download = getWordNameFromAudioID(audioID) + getFileExtension(downloadAudioElement.dataset.href);
        downloadAudioElement.innerText = 'Download audio';
        downloadAudioElement.addEventListener('click', function(ev) {
            downloadFileWithFileName(ev.target.dataset.href, ev.target.dataset.download);
        });
        insertNodeAfter(downloadAudioElement, audioLinkElement);
    }
})();