Tries to recognize and add featured artists from selected text in group description
目前為
// ==UserScript==
// @name Gazelle extract featured artists from description
// @namespace https://greasyfork.org/cs/users/321857-anakunda
// @version 1.35
// @description Tries to recognize and add featured artists from selected text in group description
// @author Anakunda
// @copyright 2020, Anakunda (https://greasyfork.org/cs/users/321857-anakunda)
// @license GPL-3.0-or-later
// @match https://redacted.ch/torrents.php?*id=*
// @match https://orpheus.network/torrents.php?*id=*
// @match https://notwhat.cd/torrents.php?*id=*
// @grant RegExp
// @grant GM_xmlhttpRequest
// @grant GM_getValue
// @grant GM_setValue
// @grant GM_deleteValue
// @grant GM_log
// @require https://greasyfork.org/scripts/388280-xpathlib/code/XPathLib.js
// ==/UserScript==
const multiArtistParsers = [
/\s*[,;\u3001](?!\s*(?:[JjSs]r)\b)(?:\s*[Aa]nd\s+)?\s*/,
/\s+[\/\|\×]\s+/,
];
const pseudoArtistParsers = [
/^(?:#??N[\/\-]?A|[JS]r\.?)$/i,
/^(?:traditional|lidová)$/i,
/\b(?:traditional|lidová)$/,
/^(?:tradiční|lidová)\s+/,
/^(?:[Aa]nonym)/,
/^(?:[Ll]iturgical\b|[Ll]iturgick[áý])/,
/^(?:auditorium|[Oo]becenstvo|[Pp]ublikum)$/,
/^(?:Various\s+Composers)$/i,
/^(?:Guests|Friends)$/i,
];
const ampersandParsers = [
/\s+(?:vs\.?|X)\s+(?!\s*(?:[\&\/\+\,\;]|and))/i,
/\s*[;\/\|\×]\s*(?!\s*(?:\s*[\&\/\+\,\;]|and))/i,
/\s+(?:[\&\+]|and)\s+(?!his\b|her\b|Friends$|Strings$)/i, // /\s+(?:[\&\+]|and)\s+(?!(?:The|his|her|Friends)\b)/i,
/\s*\+\s*(?!(?:his\b|her\b|Friends$|Strings$))/i,
];
const featParsers = [
/\s+(?:[Ww]ith)\s+(?!his\b|her\b|Friends$|Strings$)(.*?)\s*$/,
/(?:\s+[\-\−\—\–\_])?\s+(?:[Ff](?:eaturing|t\.))\s+(.*?)\s*$/,
/(?:\s+[\-\−\—\–\_])?\s+(?:[Ff](?:ea)?t\.)\s+(.*?)\s*$/,
/\s+\[\s*f(?:eat(?:\.|uring)|t\.)\s+([^\[\]]+?)\s*\]/i,
/\s+\(\s*f(?:eat(?:\.|uring)|t\.)\s+([^\(\)]+?)\s*\)/i,
/\s+\[\s*(?:(?:en\s+)?duo\s+)?avec\s+([^\[\]]+?)\s*\]/i,
/\s+\(\s*(?:(?:en\s+)?duo\s+)?avec\s+([^\(\)]+?)\s*\)/i,
/\s+\[\s*(?:with|w\/)\s+(?![Hh]is\b|[Hh]er\b|Friends$|Strings$)([^\[\]]+?)\s*\]/,
/\s+\(\s*(?:with|w\/)\s+(?![Hh]is\b|[Hh]er\b|Friends$|Strings$)([^\(\)]+?)\s*\)/,
];
const remixParsers = [
/\s+\((?:The\s+)Remix(?:e[sd])?\)/i,
/\s+\[(?:The\s+)Remix(?:e[sd])?\]/i,
/\s+(?:The\s+)Remix(?:e[sd])?\s*$/i,
/^(Remixes)\b/,
/\s+\(([^\(\)]+?)(?:[\'\’\`]s)?\s+(?:(?:Extended|Enhanced)\s+)?Remix\)/i,
/\s+\[([^\[\]]+?)(?:[\'\’\`]s)?\s+(?:(?:Extended|Enhanced)\s+)?Remix\]/i,
/\s+\([^\(\)]*\b(?:(Extended|Enhanced)\s+)?Remix(?:ed)?\s+by\s+([^\(\)]+)\)/i,
/\s+\[[^\[\]]*\b(?:(Extended|Enhanced)\s+)?Remix(?:ed)?\s+by\s+([^\[\]]+)\]/i,
];
const otherArtistsParsers = [
[/^(.*?)\s+(?:under|(?:conducted)\s+by)\s+(.*)$/, 4],
[/^()(.*?)\s+\(conductor\)$/i, 4],
//[/^()(.*?)\s+\(.*\)$/i, 1],
];
const artistStrips = [
/\s+(?:aka|AKA)\.?\s+(.*)$/,
/\s+\(([^\(\)]+)\)$/,
/\s+\[([^\[\]]+)\]$/,
/\s+\{([^\{\}]+)\}$/,
];
const siteApiTimeframeStorageKey = document.location.hostname + ' API time frame';
const gazelleApiFrame = 10500;
var artist_index, siteArtistsCache = {}, notSiteArtistsCache = [], xhr = new XMLHttpRequest;
var modal = null, btnAdd = null, btnCustom = null, customCtrls = [], sel = null, ajaxRejects = 0;
var prefs = {
set: function(prop, def) { this[prop] = GM_getValue(prop, def) }
};
(function() {
'use strict';
const styleSheet = `
.modal {
position: fixed;
left: 0;
top: 0;
width: 100%;
height: 100%;
background-color: rgba(0, 0, 0, 0.5);
opacity: 0;
visibility: hidden;
transform: scale(1.1);
transition: visibility 0s linear 0.25s, opacity 0.25s 0s, transform 0.25s;
}
.modal-content {
position: absolute;
top: 50%;
left: 50%;
font-size: 17px;
transform: translate(-50%, -50%);
background-color: FloralWhite;
color: black;
width: 31rem;
border-radius: 0.5rem;
padding: 2rem 2rem 2rem 2rem;
font-family: monospace;
}
.show-modal {
opacity: 1;
visibility: visible;
transform: scale(1.0);
transition: visibility 0s linear 0s, opacity 0.25s 0s, transform 0.25s;
}
input[type="text"] { cursor: text; }
input[type="radio"] { cursor: pointer; }
.lbl { cursor: pointer; }
.tooltip {
position: relative;
}
.tooltip .tooltiptext {
visibility: hidden;
width: 120px;
background-color: #555;
color: #fff;
text-align: center;
border-radius: 6px;
padding: 5px 0;
position: absolute;
z-index: 1;
bottom: 125%;
left: 50%;
margin-left: -60px;
opacity: 0;
transition: opacity 0.3s;
}
.tooltip .tooltiptext::after {
position: absolute;
top: 100%;
left: 50%;
margin-left: -5px;
border-width: 5px;
border-style: solid;
border-color: #555 transparent transparent transparent;
}
.tooltip:hover .tooltiptext {
visibility: visible;
opacity: 1;
}
`;
var addBox = document.querySelector('form.add_form[name="artists"]');
if (addBox == null) return;
btnAdd = document.createElement('input');
btnAdd.id = 'add-artists-from-selection';
btnAdd.value = 'Extract from selection';
btnAdd.onclick = add_from_selection;
btnAdd.type = 'button';
btnAdd.style.marginLeft = '5px';
btnAdd.style.visibility = 'hidden';
addBox.appendChild(btnAdd);
var style = document.createElement('style');
document.head.appendChild(style);
style.id = 'artist-parser-form';
style.type = 'text/css';
style.innerHTML = styleSheet;
var el, elem = [];
elem.push(document.createElement('div'));
elem[elem.length - 1].className = 'modal';
elem[elem.length - 1].id = 'add-from-selection-form';
modal = elem[0];
elem.push(document.createElement('div'));
elem[elem.length - 1].className = 'modal-content';
elem.push(document.createElement('input'));
elem[elem.length - 1].id = 'btnFill';
elem[elem.length - 1].type = 'submit';
elem[elem.length - 1].value = 'Capture';
elem[elem.length - 1].style = "position: fixed; right: 30px; width: 80px; top: 30px;";
elem[elem.length - 1].onclick = do_parse;
elem.push(document.createElement('input'));
elem[elem.length - 1].id = 'btnCancel';
elem[elem.length - 1].type = 'button';
elem[elem.length - 1].value = 'Cancel';
elem[elem.length - 1].style = "position: fixed; right: 30px; width: 80px; top: 65px;";
elem[elem.length - 1].onclick = closeModal;
var presetIndex = 0;
function addPreset(val, label = 'Custom', rx = null, order = [1, 2]) {
elem.push(document.createElement('div'));
el = document.createElement('input');
elem[elem.length - 1].style.paddingBottom = '10px';
el.id = 'parse-preset-' + val;
el.name = 'parse-preset';
el.value = val;
if (val == 1) el.checked = true;
el.type = 'radio';
el.onchange = update_custom_ctrls;
if (rx) {
el.rx = rx;
el.order = order;
}
if (val == 999) btnCustom = el;
elem[elem.length - 1].appendChild(el);
el = document.createElement('label');
el.style.marginLeft = '10px';
el.style.marginRight = '10px';
el.htmlFor = 'parse-preset-' + val;
el.className = 'lbl';
el.innerHTML = label;
elem[elem.length - 1].appendChild(el);
if (val != 999) return;
el = document.createElement('input');
el.type = 'text';
el.id = 'custom-pattern';
el.style.width = '20rem';
el.style.fontFamily = 'monospace';
el.autoComplete = "on";
addTooltip(el, 'RegExp to parse lines, first two captured groups are used');
customCtrls.push(elem[elem.length - 1].appendChild(el));
el = document.createElement('input');
el.type = 'radio';
el.name = 'parse-order';
el.id = 'parse-order-1';
el.value = 1;
el.checked = true;
el.style.marginLeft = '1rem';
addTooltip(el, 'Captured regex groups assigned in order $1: artist(s), $2: assignment');
customCtrls.push(elem[elem.length - 1].appendChild(el));
el = document.createElement('label');
el.htmlFor = 'parse-order-1';
el.textContent = '→';
el.style.marginLeft = '5px';
elem[elem.length - 1].appendChild(el);
el = document.createElement('input');
el.type = 'radio';
el.name = 'parse-order';
el.id = 'parse-order-2';
el.value = 2;
el.style.marginLeft = '10px';
addTooltip(el, 'Captured regex groups assigned in order $1: assignment, $2: artist(s)');
customCtrls.push(elem[elem.length - 1].appendChild(el));
el = document.createElement('label');
el.htmlFor = 'parse-order-2';
el.textContent = '←';
el.style.marginLeft = '5px';
elem[elem.length - 1].appendChild(el);
}
addPreset(++presetIndex, escapeHTML('<artist(s)> - <assignment>]'), /^\s*(.+?)(?:\:|\s+[\-\−\—\~\–]+\s+(.*?))?\s*$/);
addPreset(++presetIndex, escapeHTML('<artist>[, <assignment>]') +
'<span style="font-family: initial;"> <i>(HRA style)</i></span>', /^\s*(.+?)(?:\:|\s*,\s*(.*?))?\s*$/);
addPreset(++presetIndex, escapeHTML('<artist(s)>[: <assignment>]'), /^\s*(.+?)(?:\:|\s*:+\s*(.*?))?(?:\s*,)?\s*$/);
addPreset(++presetIndex, escapeHTML('<artist(s)>[ (<assignment>)]'), /^\s*(.+?)(?:\:|\s+(?:\([^\(\)]+\)|\[[^\[\]]+\]|\{[^\{\}]+\}))?(?:\s*,)?\s*$/);
addPreset(++presetIndex, escapeHTML('[<assignment> - ]<artist(s)>'), /^\s*(?:(.*?)\s+[\-\−\—\~\–]+\s+)?(.+?)\:?(?:\s*,)?\s*$/, [2, 1]);
addPreset(++presetIndex, escapeHTML('[<assignment>: ]<artist(s)>'), /^\s*(?:(.*?)\s*:+\s*)?(.+?)\:?(?:\s*,)?\s*$/, [2, 1]);
addPreset(++presetIndex, escapeHTML('<artist>[ / <assignment>]'), /^\s*(.+?)(?:\:|\s*\/+\s*(.*?))?(?:\s*,)?\s*$/);
addPreset(++presetIndex, escapeHTML('<artist>[; <assignment>]'), /^\s*(.+?)(?:\:|\s*;\s*(.*?))?(?:\s*,)?\s*$/);
addPreset(++presetIndex, escapeHTML('[<assignment> / ]<artist(s)>'), /^\s*(?:(.*?)\s*\/+\s*)?(.+?)\:?(?:\s*,)?\s*$/, [2, 1]);
addPreset(++presetIndex, '<span style="font-family: initial;">From tracklist</span>', /^\s*\d+(?:\s*[\-\−\—\~\–\.\:]\s*|\s+)(.+?)\s*$/, []);
addPreset(999);
elem.slice(2).forEach(k => { elem[1].appendChild(k) });
elem[0].appendChild(elem[1]);
document.body.appendChild(elem[0]);
window.addEventListener("click", windowOnClick);
document.addEventListener('selectionchange', () => {
var cs = window.getComputedStyle(modal);
if (!btnAdd || window.getComputedStyle(modal).visibility != 'hidden') return;
var sel = document.getSelection();
ShowHideAddbutton();
});
})();
function add_from_selection() {
sel = document.getSelection();
if (sel.isCollapsed || modal == null) return;
prefs.set('preset', 1);
prefs.set('custom_pattern', '^\\s*(.+?)(?:\\s*:+\\s*(.*?)|\\:)?\\s*$');
prefs.set('custom_pattern_order', 1);
setRadiosValue('parse-preset', prefs.preset);
customCtrls[0].value = prefs.custom_pattern;
setRadiosValue('parse-order', prefs.custom_pattern_order);
sel = sel.toString();
update_custom_ctrls();
modal.classList.add("show-modal");
}
function do_parse(expr, flags = '') {
closeModal();
if (!sel) return;
var preset = getSelectedRadio('parse-preset');
if (preset == null) return;
prefs.preset = preset.value;
var order = preset.order;
var custom_parse_order = getSelectedRadio('parse-order');
var rx = preset.rx;
if (!rx && preset.value == 999 && custom_parse_order != null) {
rx = [new RegExp(customCtrls[0].value)];
order = custom_parse_order != null ?
custom_parse_order.value == 1 ? [1, 2] : custom_parse_order.value == 2 ? [2, 1] : null : [1, 2];
}
const guest_parser = /^(.*?)(?:\s+(?:feat(?:\.|uring)|with|meets)\s+(.*))?$/;
function extr_artists(kind) { return document.querySelectorAll('ul#artist_list > li.' + kind + ' > a') }
var artists = [
extr_artists('artist_main'),
extr_artists('artist_guest'),
extr_artists('artists_remix'),
extr_artists('artists_composers'),
extr_artists('artists_conductors'),
extr_artists('artists_dj'),
extr_artists('artists_producer'),
];
cleanupArtistsForm();
var addedartists = [];
for (var i = 0; i < artists.length; ++i) addedartists[i] = [];
artist_index = 0;
sel.split(/(?:\r?\n)+/).forEach(function(line) {
if (!line || !line.trim()) return;
if (line.search(/^\s*(?:Recorded|Mastered)\b/i) >= 0) return;
line = line.replace(/\s+\(tracks?\b[^\(\)]+\)/, '').replace(/\s+\[tracks?\b[^\[\]]+\]/, '')
let matches = /^\s*(?:Produced)[ \-\−\—\~\–]by (.+?)\s*$/.exec(line);
if (matches != null) splitAmpersands(matches[1]).forEach(producer => { add_artist(producer, 7) });
else if (rx instanceof RegExp && (matches = rx.exec(line)) != null) {
if (!Array.isArray(order) || order.length < 2) {
let title = matches[1];
if (/^(.+?) - /.test(title)) {
let artist = RegExp.$1;
if ((matches = featParsers.slice(0, 3).reduce((m, rx) => m || rx.exec(artist), null)) != null) {
splitAmpersands(artist.slice(0, matches.index)).forEach(artist => { add_artist(artist, 1) });
splitAmpersands(matches[1]).forEach(artist => { add_artist(artist, 2) });
} else splitAmpersands(artist).forEach(artist => { add_artist(artist, 1) });
}
if ((matches = featParsers.slice(1).reduce((m, rx) => m || rx.exec(title), null)) != null)
splitAmpersands(matches[1]).forEach(artist => { add_artist(artist, 2) });
} else if (matches[order[0]]) {
let role = deduce_artist(matches[order[1]]);
splitAmpersands(matches[order[0]]).forEach(artist => { add_artist(artist, role) });
} else splitAmpersands(matches[order[1]]).forEach(artist => { add_artist(artist, 2) });
}
});
prefs.custom_pattern = customCtrls[0].value;
prefs.custom_pattern_order = custom_parse_order != null ? custom_parse_order.value : 1;
for (i in prefs) { if (typeof prefs[i] != 'function') GM_setValue(i, prefs[i]) }
return;
function deduce_artist(str) {
if (/\b(?:remix)/i.test(str)) return 3; // remixer
if (/\b(?:composer|libretto|lyric\w*|written[ \-\−\—\~\–]by)\b/i.test(str)) return 4; // composer
if (/\b(?:conduct|rirector\b)/i.test(str)) return 5; // conductor
if (/\b(?:compiler\b)/i.test(str)) return 6; // compiler
if (/\b(?:producer\b|produced[ \-\−\—\~\–]by\b)/i.test(str)) return 7; // producer
return 2;
}
function add_artist(name, type = 1) {
if (!name || !type) return false;
if (/^(?:(?:Special\s+)?Guests?):?$/i.test(name)) return false;
// avoid dupes
var n = name.toLowerCase();
for (var i of artists[0]) { if (n == i.textContent.toLowerCase()) return false }
if (type >= 2) for (i of artists[type - 1]) { if (n == i.textContent.toLowerCase()) return false }
for (i of addedartists[0]) { if (n == i.toLowerCase()) return false }
if (type >= 2) for (i of addedartists[type - 1]) { if (n == i.toLowerCase()) return false }
var id = get_artist_field(artist_index);
if (id == null) {
AddArtistField();
id = get_artist_field(artist_index);
if (id == null) return false;
}
id.value = name;
id.nextElementSibling.value = type;
addedartists[type - 1].push(name);
++artist_index;
return true;
}
}
function get_artist_field(index) {
var id = document.getElementById('artist');
if (index <= 0) return id;
for (var i = 0; i < index; ++i) {
do { id = id.nextElementSibling } while (id != null && (id.localName != 'input' || id.name != 'aliasname[]'));
if (id == null) break;
}
return id;
}
function closeModal() {
if (modal == null) return;
ShowHideAddbutton();
modal.classList.remove("show-modal");
}
function windowOnClick(event) {
if (modal != null && event.target === modal) closeModal();
}
function update_custom_ctrls() {
function en(elem) {
if (elem == null || btnCustom == null) return;
elem.disabled = !btnCustom.checked;
elem.style.opacity = btnCustom.checked ? 1 : 0.5;
}
customCtrls.forEach(k => { en(k) });
}
function getSelectedRadio(name) {
for (var i of document.getElementsByName(name)) { if (i.checked) return i }
return null;
}
function setRadiosValue(name, val) {
for (var i of document.getElementsByName(name)) { if (i.value == val) i.checked = true }
}
function ShowHideAddbutton() {
//btnAdd.style.visibility = document.getSelection().type == 'Range' ? 'visible' : 'hidden';
btnAdd.style.visibility = document.getSelection().isCollapsed ? 'hidden' : 'visible';
}
function escapeHTML(string) {
var pre = document.createElement('pre');
var text = document.createTextNode(string);
pre.appendChild(text);
return pre.innerHTML;
}
function cleanupArtistsForm() {
var id = get_artist_field(0);
do {
id.value = null;
id = id.nextElementSibling;
if (id == null) break;
id.value = 1;
do { id = id.nextElementSibling } while (id != null && (id.localName != 'input' || id.name != 'aliasname[]'));
} while (id != null);
}
function addTooltip(elem, text) {
if (elem == null) return;
elem.classList.add('tooltip');
var tt = document.createElement('span');
tt.className = 'tooltiptext';
tt.textContent = text;
elem.appendChild(tt);
}
function twoOrMore(artist) { return artist.length >= 2 && !pseudoArtistParsers.some(rx => rx.test(artist)) };
function looksLikeTrueName(artist, index = 0) {
return twoOrMore(artist)
&& (index == 0 || !/^(?:his\b|her\b|Friends$|Strings$)/i.test(artist))
&& artist.split(/\s+/).length >= 2
&& !pseudoArtistParsers.some(rx => rx.test(artist)) || getSiteArtist(artist);
}
function strip(art) {
return artistStrips.reduce(function(acc, rx, ndx) {
return ndx != 1 || rx.test(acc)/* && !notMonospaced(RegExp.$1)*/ ? acc.replace(rx, '') : acc;
}, art);
}
function getSiteArtist(artist) {
//if (isOPS) return undefined;
if (!artist || notSiteArtistsCache.includesCaseless(artist)) return null;
var key = Object.keys(siteArtistsCache).find(it => it.toLowerCase() == artist.toLowerCase());
if (key) return siteArtistsCache[key];
var now = Date.now();
try { var apiTimeFrame = JSON.parse(window.localStorage[siteApiTimeframeStorageKey]) } catch(e) { apiTimeFrame = {} }
if (!apiTimeFrame.timeStamp || now > apiTimeFrame.timeStamp + gazelleApiFrame) {
apiTimeFrame.timeStamp = now;
apiTimeFrame.requestCounter = 1;
} else ++apiTimeFrame.requestCounter;
window.localStorage[siteApiTimeframeStorageKey] = JSON.stringify(apiTimeFrame);
if (apiTimeFrame.requestCounter > 5) {
console.debug('getSiteArtist() request exceeding AJAX API time frame: /ajax.php?action=artist&artistname="' +
artist + '" (' + apiTimeFrame.requestCounter + ')');
++ajaxRejects;
return undefined;
}
try {
var requestUrl = '/ajax.php?action=artist&artistname=' + encodeURIComponent(artist);
xhr.open('GET', requestUrl, false);
//if (isRED && prefs.redacted_api_key) xhr.setRequestHeader('Authorization', prefs.redacted_api_key);
xhr.send();
if (xhr.status == 404) {
notSiteArtistsCache.pushUniqueCaseless(artist);
return null;
}
if (xhr.readyState != XMLHttpRequest.DONE || xhr.status < 200 || xhr.status >= 400) {
console.warn('getSiteArtist("' + artist + '") error:', xhr, 'url:', document.location.origin + requestUrl);
return undefined; // error
}
let response = JSON.parse(xhr.responseText);
if (response.status != 'success') {
notSiteArtistsCache.pushUniqueCaseless(artist);
return null;
}
siteArtistsCache[artist] = response.response;
if (prefs.diag_mode) console.log('getSiteArtist("' + artist + '") success:', siteArtistsCache[artist]);
return (siteArtistsCache[artist]);
} catch(e) {
console.error('UA::getSiteArtist("' + artist + '"):', e, xhr);
return undefined;
}
}
function splitArtists(str, parsers = multiArtistParsers) {
var result = [str];
parsers.forEach(function(parser) {
for (var i = result.length; i > 0; --i) {
var j = result[i - 1].split(parser).map(strip);
if (j.length > 1 && j.every(twoOrMore) && !j.some(artist => pseudoArtistParsers.some(rx => rx.test(artist)))
&& !getSiteArtist(result[i - 1])) result.splice(i - 1, 1, ...j);
}
});
return result;
}
function splitAmpersands(artists) {
if (typeof artists == 'string') var result = splitArtists(artists);
else if (Array.isArray(artists)) result = Array.from(artists); else return [];
ampersandParsers.forEach(function(ampersandParser) {
for (let i = result.length; i > 0; --i) {
let j = result[i - 1].split(ampersandParser).map(strip);
if (j.length <= 1 || !j.every(looksLikeTrueName) || getSiteArtist(result[i - 1])) continue;
result.splice(i - 1, 1, ...j.filter(function(artist) {
return !result.includesCaseless(artist) && !pseudoArtistParsers.some(rx => rx.test(artist));
}));
}
});
return result;
}