« MediaWiki:Common.js/Wikinews:Article principal/leadGenerator » : différence entre les versions
Traduction |
m a renommé Transwiki:Bawolff/sanbox/leadGenerator.js en MediaWiki:Common.js/Wikinews:Article principal/leadGenerator : requête de Sniff, comme précédemment |
||
(Aucune différence)
|
Version du 20 novembre 2009 à 17:26
importScriptURI('http://en.wikinews.org/w/index.php?title=user:Bawolff/mwapilib2.js&action=raw&ctype=text/javascript'); if (!window.Bawolff) {
window.Bawolff = {};
} // http://en.wikinews.org/w/api.php?action=query&prop=revisions&titles=template:Lead%20article%201%7Ctemplate:Lead%20article%202%7Ctemplate:Lead%20article%203%7Ctemplate:Lead%20article%204%7Ctemplate:Lead%20article%205&rvprop=timestamp%7Ccontent //Call as Bawolff.leadGen(pageName, 1, alert) //replacing 1 with which lead (1-5), alert with your callback //optionally takes summary method (as numeric) argument. choose 0-4 Bawolff.leadGen = function (title, leadNumb, callback, summaryMethod) {
api(title).getPage().lift(Bawolff.leadGen.extract, title, summaryMethod).lift(Bawolff.leadGen.create, leadNumb).lift(callback, title).exec();
}
//takes the source of a wikipage, extracts the first image name without the leading namespace Bawolff.leadGen.extractImg = function (page) {
var imgRegex = /\[\[(?:[iI][mM][aA][gG][eE]\:|[fF][Ii][lL][eE]\:)((?:[^\|\]])*?\.[pPsSjJgG][nNvVpPiI][gGeEfF][gG]?)[\|\]]?/; var img = imgRegex.exec(page); if (img && img.length >= 2) { return img[1]; } else { //return "Wikinews-logo.png"; //default return ""; }
} Bawolff.leadGen.extractType = function (page) {
//valid types are: breaking, special, original, exclusive, urgent or none. //this currently does not detect special or urget. if (page.match(/\{\{[eE]n(?: cours)?\}\}/)) { return "breaking"; } else if (page.match(/\{\{[iI]nterview(?:\|[^}]*)?\}\}/i)) { return "exclusive"; } else if (page.match(/\{\{[rR]eportage(?: original)?(?:\|[^}]*)?\}\}/i)) { return "original"; } else { //default return "none"; }
}
Bawolff.leadGen.takeIntro = function (pageText, method) {
//first test for redirects. var isRedirect = pageText.match(/^#redirect\s?\[\[([^\]]*)\]\]/i) if (isRedirect) { alert("Il semble que vous essayez d\'utiliser la création d\'articles principaux sur un page de redirection. Utiliser s\'il vous plaît le nom réel de la page au lieu de (" + isRedirect[1] + ")."); throw new Error("La page est une redirection. Merci de résoudre manuellement à : " + isRedirect[1] ); }
//doesn't handle links that make [1]. //intentionally doesn't strip ' chars (bold or italic) //as this often marks thigs with periods (E. coli) //method is a number (must be a number. no type conversion preformed). // 0: 1st sentence // 1: 1st two senetences // 2: 1st paragraph // 3: 1st 250 characters (+ a couple so we don't end in middle of word), or paragraph // 4: 1st 500 characters-ish, or paragraph var stripTemplates = /\{\{[^\}\{]*(?:\{\{[^\}\{]*(?:\{\{[^\}\{]*(?:\{\{[^\}\{]*\}\})?\}\})?\}\})?\}\}/g; //the img regex, looks for a start of the image, than checks for nested internal links, external links, and for ending ]] in caption //templates shold already be stripped at this point. var img = /\[\[[IifF][mMIi][aALl][GgEe][eE]?\:(?:\[\[(?:[^\]]*)\]\]|[^\]]|\](?!\]))*]]/g
pageText = pageText.replace(stripTemplates, ) pageText = pageText.replace(img, ); var pipedLink = /\[\[[^\]\|]*\|([^\]\|]*)\]\]/g; pageText = pageText.replace(pipedLink, '$1'); var normLink = /\[\[([^\]\|]*)\]\]/g; pageText = pageText.replace(normLink, '$1'); var extLink = /\[(?:http|ftp|gopher|irc|https)\:[^\]\s]*\s?([^\]]*)]/g pageText = pageText.replace(extLink, '$1'); var firstPar; switch (method) { case 1: //1st 2 sentence or 1st paragraph firstPar = /[^\n]+?\s[\w"]*\.(?=\s)(?:[^\n]+?\s\w*\.(?=\s)|(?=\s))/; pageText = pageText.match(firstPar)[0]; break; case 2: //1st paragraph firstPar = /[^\n]+?(?=\n)/; pageText = pageText.match(firstPar)[0]; break; case 3: // 1st 250 characters-ish firstPar = /[^\n]{2,250}.*?\b/; pageText = pageText.match(firstPar)[0] + "..."; break; case 5: //1st 500 characters-ish firstPar = /[^\n]{2,500}.*?\b/; pageText = pageText.match(firstPar)[0] + "..."; break; default: //aka case 0. 1st sentence firstPar = /[^\n]+?\s[\w"]*\.(?=\s)/; pageText = pageText.match(firstPar)[0]; break; } return pageText;
}
Bawolff.leadGen.extract = function(pageText, pageName, summaryMethod) { //editlink is overriden later. return { width:'100x100px',
image: Bawolff.leadGen.extractImg(pageText), title: pageName, synopsis: Bawolff.leadGen.takeIntro(pageText, summaryMethod), order: 'right', edit_this: 'Wikinews:Sandbox', type: Bawolff.leadGen.extractType(pageText) };
}
Bawolff.leadGen.create = function (leadObj, leadNumb) {
if (leadNumb === 1) { leadObj.width = "150x150px"; //ugly i know, but whatever. } var res = '{' + '{Article principal'; res += "\n |modif=" + leadNumb; res += "\n |Image=" + leadObj.image; res += "\n |Image width=" + leadObj.width; res += "\n |type=" + leadObj.type; res += "\n |Titre=" + leadObj.title; res += "\n |Synopsis=" + leadObj.synopsis; res += "\n}}\n{{" res += "Article principal/Documentation}}";
return res;
} //calls its argument giving it an object with meta info about the current leads. //probably want to use Bawolff.leadGen.makeLeadTable instead.
Bawolff.leadGen.makeLeadMetaObject = function (callback) {
var leadObj = function (doc) { var exTitle = function (text) { text = text.replace(/[\s\S]*?\|title=([^\|]*)[\s\S]*/, '$1'); text = text.replace(/\s*$/, ); return text; } var pages = doc.getElementsByTagName('page');
var obj = {}, time, time2 = Infinity; for (var i = 0; i < pages.length; i++) { pages[i].getElementsByTagName('rev')[0].normalize(); time = Bawolff.mwapi.parseAPIDate(pages[i].getElementsByTagName('rev')[0].getAttribute('timestamp')); //This is really ugly... //put the oldest lead in a global variable. if (time < time2) { //this compares miliseconds after epoch Bawolff.leadGen.oldestLead = pages[i].getAttribute('title'); time2 = time; } obj[pages[i].getAttribute('title')] = {timestamp: time, title: exTitle(pages[i].getElementsByTagName('rev')[0].firstChild.data)}; } return obj; } api().makeRequest({action: 'query', prop: 'revisions', titles: 'Modèle:Une 1|Modèle:Une 2', prop: 'revisions', rvprop: 'timestamp|content'}, leadObj).lift(callback).exec();
}
Bawolff.leadGen.oldestLead = null;
Bawolff.leadGen.makeLeadTable = function (callback) {
var wrapper = function(leadObj) {
var html = '
'; html += '<thead></thead><tbody>';for (var i in leadObj) { if (leadObj.hasOwnProperty(i)) { var numb = i.charAt(i.length - 1);html += '"; } } html += '</tbody>
# | Position | Article | Âge |
---|---|---|---|
';
html += numb; //last characterhtml += ' | ';
html += Bawolff.leadGen.leadToPosition(numb);html += ' | ';
html += leadObj[i].title;html += ' | ';
var time = leadObj[i].timestamp.getTime(); //convert to miliseconds var d = (new Date).getTime(); var delta = Math.round((d - time) / (1000*60*60)); var timeStr; if (delta === 1) { timeStr = delta + ' heure'; } else { timeStr = delta + ' heures'; } html += timeStr;html += " |
';
callback(html); } Bawolff.leadGen.makeLeadMetaObject(wrapper);
}
Bawolff.leadGen.leadToPosition = function (numb) { //double equal sign intentional to convert from string.
if (numb == 1) { return "gauche"; } if (numb == 2) { return "droite"; } else { throw new Error("Chiffre de l\'article principal invalide (" + numb +") trasmis à Bawolff.leadGen.leadToPosition"); }
}