User:Cobaltcigs/IllWill

From Wikipedia, the free encyclopedia
var IllWill = {};
IllWill.css = [
    "#ill-sandbox, #ill-row-sandbox { display: none !important; }", 
	".ill-sitelinks a { display: block; white-space: nowrap; }", 
	".illwill-ignore { text-decoration: line-through; color: gray; font-size: smaller; }", 
	".ill-x { width: 24px; }",
	".ill-input, .ill-output { font-family: monospace; white-space: pre-wrap; }", 
	"#ill-buttons { text-align: right; white-space: nowrap; }", 
	"#ill-buttons input { margin-left: 3px; margin-right: 3px; font-size: smaller; }",
	"#ill-table caption { background: lightgreen; }",
	"#ill-help { font-size: smaller; }",
	"#illwill { clear: both; }",
	".ill-na { font-style: italic; color: gray; font-size: smaller; }",
	".ill-input .ill-na, .ill-output .ill-na { display: block; }",
].join("\n");
IllWill.Version = "1.0.5";
IllWill.Attribution = 'created by <a target="_blank" href="/wiki/User:Cobaltcigs">cobaltcigs</a> (<a target="_blank" href="/wiki/User_talk:Cobaltcigs">talk</a>)';
IllWill.CaptionHTML = "IllWill.js (v. "+IllWill.Version+") \u2013 " + IllWill.Attribution;
IllWill.LocalApi = location.href.split(/\/w(?:iki)?\//)[0] + "/w/api.php";
IllWill.DataApi = "https://wikidata.org/w/api.php";
IllWill.LocalLang = mw.config.get("wgPageContentLanguage");

IllWill.UrlLengthLimit = 3600; // TODO: determine actual limit
IllWill.HelpText = 'For each row, determine the correct wikidata item corresponding to the red link found in wikitext. This will import a full list of interwiki prefixes and titles to use with template <code>\x7B\x7B<a href="/wiki/Template:Ill">ill</a>\x7D\x7D</code> when replacing the regular red link. If none of the listed wikidata items are correct, uncheck the box in the first column to clear selection for that row, and no replacement will occur. Click "OK" when finished.';
IllWill.Summary = "added interlanguage links using IllWill.js";

IllWill.SearchCb = [];
IllWill.SiteLinks = {};
IllWill.Disambiguator = / \(.*\)$/;
var textbox = document.getElementById("wpTextbox1");
var editform = document.getElementById("editform");

IllWill.MakeWikidataLink = function(q, t) {
	return '<a id="'+q+'-link" href="https://wikidata.org/wiki/' + q + '" target="_blank">'+t+'</a>';
	}
IllWill.EscapeHTML = function(s) {
    return s.replace(/&/g, "&amp;").replace(/</g, "&lt;").replace(/>/g, "&gt;")
		.replace(/"/g, "&quot;").replace(/'/g, "&#x27;"); // a.k.a. "&apos;"
	}
     
IllWill.EscapeRegExp = function(s) { return s.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'); }
IllWill.Normalize = function(s) { return s.normalize().replace(/[\s_]+/g, " ").trim(); }
IllWill.NormEqual = function(a, b) {
	a = IllWill.Normalize(a); b = IllWill.Normalize(b);
	return (a[0].toUpperCase()+a.substring(1)) == (b[0].toUpperCase()+b.substring(1));
	}
IllWill.MakeRegex = function(t) {
	return new RegExp("\\[\\[\\s*" + IllWill.EscapeRegExp(t).replace(/\s/g, "[\\s_]+") + "\\s*(?:\\|[^\\[\\]]+)?\\]\\]", "ig");
	}
IllWill.Setup = function() {
	if(IllWill.SetupComplete) return;
	if(!["edit", "submit"].includes(mw.config.get('wgAction'))) return;
	var style = document.createElement("style");
	style.type = "text/css";
	style.innerText = IllWill.css;
	document.head.appendChild(style);
	mw.util.addPortletLink('p-cactions', `javascript:IllWill.MakePanel()`, 'Ill',
        'ca-illwill', "Add interlanguage links by searching wikidata.org");
	IllWill.SetupComplete = true;
	}
IllWill.GetRedLinks = async function() {
	var sandbox = document.getElementById("ill-sandbox");
	if(!textbox) { alert("can't find the textbox"); return; }
	var wl = textbox.value.match(/\[\[[^\[\]\n]+\]\]/g);
	if(!wl) return [];
	var b = IllWill.UrlLengthLimit, a = (b - 260);
	var r = new RegExp("([^\n]{"+a+","+b+"})(?=\\[\\[)", "g");
	var textParts = wl.join("").replace(/\|[^\]]+/g, "").replace(r, "$1\n").split("\n");
	sandbox.innerHTML = "";
	for(var i = 0; i < textParts.length; i++) {
		var url = IllWill.LocalApi + "?action=parse&format=json&text=" + encodeURIComponent(textParts[i]);
		var response = await fetch(url);
		var json = await response.json();
		sandbox.innerHTML += json.parse.text["*"];
		}
	var reds = sandbox.getElementsByClassName("new");
	return Array.from(reds).filter(x => x.tagName=="A")
		.map(x => decodeURIComponent(x.href.match(/(?<=[\?\&]title=)[^\&]+/)[0]).replace(/_/g, " "));
	}
IllWill.MakePanel = async function() {
	var old = document.getElementById("illwill");
	if(old) old.remove();
	var wrapper = document.createElement("div");
	wrapper.id = "illwill";
	wrapper.innerHTML = '<div id="ill-sandbox"></div>';
	editform.parentNode.insertBefore(wrapper, editform);
	var reds = await IllWill.GetRedLinks();
	var thead = document.createElement("thead");
	var caption = document.createElement("caption");
	caption.setAttribute("colspan", "5");
	caption.innerHTML = IllWill.CaptionHTML;
	var head_tr = document.createElement("tr");
	head_tr.innerHTML = '<th></th><th>Input</th><th>Wikidata item</th><th>Sitelinks</th><th>Output</th></tr>';
	thead.appendChild(head_tr);
	var wtLinks = [];
	var txt = textbox.value;
	for(var i = 0; i < reds.length; i++) {
		var re = IllWill.MakeRegex(reds[i]);
		var wl = txt.match(re)[0];
		if(!wl) continue;
		if(wtLinks.find(z => z.title == reds[i])) continue;
		var o = { link: wl, title: reds[i] };
		wtLinks.push(o);
		}
	var trs = [];
	for(var i = 0; i < wtLinks.length; i++) {
		var td0 = '<td class="ill-x"></td>';
		var td1 = '<td class="ill-input"></td>';
		var td2 = '<td class="ill-wd-item"></td>';
		var td3 = '<td class="ill-sitelinks"></td>';
		var td4 = '<td class="ill-output"></td>';
		trs.push('<tr id="ill-row-'+i+'">'+td0+td1+td2+td3+td4+'</tr>');
		}
	var tbody = document.createElement("tbody");
	tbody.innerHTML += trs.join("");
	var tfoot = document.createElement("tfoot");
	tfoot.innerHTML = '<tr id="ill-row-buttons"><td colspan="3" id="ill-help">'+IllWill.HelpText+'</td><td colspan="2" id="ill-buttons"></td></tr>';
	var table = document.createElement("table");
	table.setAttribute("id", "ill-table");
	table.className="wikitable";
	table.appendChild(caption);
	table.appendChild(thead);
	table.appendChild(tbody);
	table.appendChild(tfoot);
	wrapper.appendChild(table);
	editform.setAttribute("style", "display: none;");
	var pp = "action=query&list=search&format=json&srlimit=500&srnamespace=0&srprop=snippet";
	for(var i = 0; i < wtLinks.length; i++) {
		var rowId = 'ill-row-'+i;
		IllWill.SearchCb[i] = IllWill.CreateSearchCb(i);
		var enc = encodeURIComponent(wtLinks[i].title.replace(IllWill.Disambiguator, ""));
		wtLinks[i].searchUrl = IllWill.DataApi+'?'+pp+'&srsearch=%22'+enc+'%22&callback=IllWill.SearchCb['+i+']';
		mw.loader.getScript(wtLinks[i].searchUrl);
		var tds = document.getElementById(rowId).getElementsByTagName("td");
		tds[0].innerHTML = '<input type="checkbox" id="checkbox-'+i+'" />';
		tds[0].getElementsByTagName("input")[0].setAttribute("onclick", 'IllWill.OnCheckbox('+i+')');
		tds[1].innerHTML = wtLinks[i].link;
		}
	var buttonCell = document.getElementById("ill-buttons");
	buttonCell.innerHTML += '<input type="button" onclick="IllWill.OnCancelButton();" value="Cancel" />';
	buttonCell.innerHTML += '<input type="button" onclick="IllWill.OnCancelButton(); IllWill.MakePanel();" value="Refresh" />';
	buttonCell.innerHTML += '<input type="button" onclick="IllWill.OnOkButton();" value="OK" />';
	}

IllWill.OnCancelButton = function() {
	editform.removeAttribute("style");
	document.getElementById("illwill").remove();
	}
IllWill.OnOkButton = function() {
	var table = document.getElementById("ill-table");
	var xboxes = Array.from(table.getElementsByClassName("ill-x")).map(x => x.getElementsByTagName("input")[0]);
	var inputs = table.getElementsByClassName("ill-input");
	var outputs = table.getElementsByClassName("ill-output");
	var textbox = document.getElementById("wpTextbox1");
	var t = textbox.value;
	for(var i = 0; i < inputs.length; i++) {
		if(!xboxes[i].checked) continue;
		var re = new RegExp(IllWill.EscapeRegExp(inputs[i].innerText), "gi");
		t = t.replace(re, outputs[i].innerText);
		}
	if(t != textbox.value) {
		document.getElementById("wpSummary").value = IllWill.Summary;
		textbox.value = t;
		document.getElementById("wpDiff").click();
		}
	}
IllWill.OnCheckbox = function(i) {
	var row = document.getElementById('ill-row-'+i);
	var inputs = row.getElementsByTagName("input");
	if(inputs[0].checked) {
		var selected = false;
		for(var i = 1; i < inputs.length; i++) if(inputs[i].checked) selected = true;
		if(!selected) inputs[0].checked = false;
		return;
		}
	for(var i = 1; i < inputs.length; i++) inputs[i].checked = false;
	var tds = row.getElementsByTagName("td");
	tds[3].innerHTML = tds[4].innerHTML = "";
	}

IllWill.CreateSearchCb = function(i) {
	var rowId = 'ill-row-'+i;
	return function(obj) {
		var li = [];
		for(var j = 0; j < obj.query.search.length; j++) {
			var z = obj.query.search[j];
			var qNum = z.title;
			var div = document.createElement("div");
			div.setAttribute("id", z.title+'-li');
			var input = document.createElement("input");
			input.setAttribute("type", "radio");
			input.setAttribute("name", "selection-"+i);
			input.setAttribute("value", qNum);
			input.setAttribute("id", qNum+'-radio');
			var label = document.createElement("label");
			label.setAttribute("for", qNum);
			label.innerHTML = IllWill.MakeWikidataLink(qNum, qNum) + ': <span id="'+qNum+'-desc"></span>';
			div.appendChild(input);
			div.appendChild(label);
			li[j] = div.outerHTML;
			}
		var cell = document.getElementById(rowId).getElementsByTagName("td")[2];
		cell.innerHTML = (li.length == 0)? `<span class="ill-na">(no search result!]</span>` : li.join("");
		for(var j = 0; j < obj.query.search.length; j++) {
			var qNum = obj.query.search[j].title;
			var radio = document.getElementById(qNum+'-radio');
			radio.setAttribute("onclick", 'IllWill.ShowSiteLinks("'+qNum+'")');
			var qNumUrl = IllWill.DataApi+"?action=query&prop=revisions&titles="+qNum+"&rvslots=*&rvprop=content&format=json&callback=IllWill.qNumCb";
			mw.loader.getScript(qNumUrl);
			}
		};
	}

IllWill.ShowSiteLinks = function(qNum) {
	var div = document.getElementById(qNum+"-li");
	var tr = div.parentNode.parentNode;
	var tds = tr.getElementsByTagName("td");
	var checkbox = tds[0].getElementsByTagName("input")[0]
	var q = IllWill.SiteLinks[qNum];
	var langs = Object.keys(q);
	langs.sort();
	var aa = '';
	for(var i = 0; i < langs.length; i++) {
		var lang = langs[i];
		var a = document.createElement("a");
		var url = "https://" + lang + ".wikipedia.org/wiki/" + encodeURIComponent(q[lang]);
		a.setAttribute("href", url);
		a.setAttribute("target", "_blank");
		a.innerText = lang + ":" + q[lang];
		aa += a.outerHTML;
		}
	tds[3].innerHTML = aa;
	var wl = tds[1].innerText.match(/(?<=^\[\[).*(?=\]\]$)/)[0].split('|');
	var pp = IllWill.Normalize(wl[0]);
	if(!langs.length) {
		tds[3].innerHTML = '<span class="ill-na">(no sitelinks found)</span>';
		tds[4].innerHTML = '<span class="ill-na">(no replacement)</span>' + tds[1].innerText;
		checkbox.checked = false;
		checkbox.disabled = true;
		return;
		}
	checkbox.checked = true;
	checkbox.disabled = false;
	if(langs.includes(IllWill.LocalLang)) {
		var qq = IllWill.Normalize(q[IllWill.LocalLang]);
		if(wl[1] && !IllWill.NormEqual(wl[1], qq)) qq += "|" + IllWill.Normalize(wl[1]);
		else if(IllWill.Disambiguator.test(qq)) qq += "|" + qq.replace(IllWill.Disambiguator, "");
		tds[4].innerText = "[[" + qq + "]]";
		return;
		}
	if(wl[1]) pp += "|lt=" + IllWill.Normalize(wl[1]);
	else if(IllWill.Disambiguator.test(pp)) pp += "|lt=" + pp.replace(IllWill.Disambiguator, "");
	for(var i = 0; i < langs.length; i++) {
		var lang = IllWill.Normalize(langs[i]);
		pp += "|" + lang;
		pp += "|" + (IllWill.NormEqual(q[lang], wl[0])? "" : IllWill.Normalize(q[lang]));
		}
	pp = pp.replace(/\|+$/, "");
	var ill = "\x7B\x7Bill|" + IllWill.Normalize(pp) + "\x7D\x7D";
	tds[4].innerText = ill;
	}

IllWill.GetBest = function(d, x) {
	// prefer local language, otherwise prefer latin script, otherwise take whatever
	// to-do: make step 2 prefer same script as user's local language would be better.
	// e.g. user on ru: with de: and uk: data available should prefer the latter, because cyrillic. 
	if(!d) return null;
	var dx = d[x];
	if(!dx) return null;
	var dxll = dx[IllWill.LocalLang];
	if(dxll) return dxll.value.trim();
	var dxv = Object.values(dx);
	if(!dxv.length) return null;
	var dxlatin = dxv.find(zz => /[a-z]/i.test(zz.value));
	if(dxlatin) return dxlatin.value.trim();
	return dxv[0].value.trim();
	}
IllWill.ShouldIgnore = function(data) {
	var s = JSON.stringify(data);
	if(/"Q13442814"/.test(s)) return true; // "scientific article"
	return false;
	}

IllWill.qNumCb = function(obj) {
	var pages = obj.query.pages;
	var keys = Object.keys(pages);
	if(keys.length != 1) return;
	var page = pages[keys[0]];
	if(page.revisions.length != 1) return;
	var qNum = page.title;
	var data = JSON.parse(page.revisions[0].slots.main["*"]);
	var failMsg = '<span class="ill-na">no description found</span>';
	var desc = IllWill.GetBest(data, "descriptions");
	var ignoreMe = IllWill.ShouldIgnore(data);
	var span = document.getElementById(qNum + "-desc");
	span.innerHTML = desc? IllWill.EscapeHTML(desc) : failMsg;
	var a = document.getElementById(qNum + "-link");
	var label = IllWill.GetBest(data, "labels");
	if(label) {
		if(ignoreMe && label.length > 40) label = label.substring(0,36) + "\u2026";
		a.innerText = label;
		}
	if(ignoreMe) {
		span.parentNode.className = "illwill-ignore";
		document.getElementById(qNum + "-radio").disabled = true;
		}
	if(!data.sitelinks) return;
	var slKeys = Object.keys(data.sitelinks);
	IllWill.SiteLinks[qNum] = {};
	for(var i = 0; i < slKeys.length; i++) {
		var wikiName = slKeys[i];
		if(!/wiki$/i.test(wikiName) || /commons/i.test(wikiName)) continue;
		var lang = wikiName.replace(/wiki$/, "");
		var langTitle = data.sitelinks[wikiName].title;
		IllWill.SiteLinks[qNum][lang] = langTitle;
		}
	}

IllWill.Setup();

IllWill.js does the following:

  1. Collects a list of all red links on a page,
  2. Searches for them on wikidata,
  3. Presents possible matches to the user with Q-number and short description,
  4. Fetches site links for each selected wikidata item,
  5. Formats each language code + title using the "interlanguage link" template (a.k.a. {{ill}}).
  6. Replaces the red link with a transclusion of this {{ill}}
Example (fictitious)
Input Wikidata item Site links Output
[[John Johnsonson (actor)|John Johnsonson]]

John Johnsonson

Q99999999: Asgardian film and stage actor
{{ill|John Johnsonson (actor)|lt=John Johnsonson|de|John Johnsonson (Schauspieler)|fr|John Johnsonson (acteur)}}

John Johnsonson [de; fr]

Real example diff: Special:Diff/981601881.