Wikipedia

User:Polygnotus/DuplicateReferences.js: Difference between revisions

Browse history interactively
← Previous edit
Content deleted Content added
No edit summary
No edit summary
 
(96 intermediate revisions by the same user not shown)
Line 1:
//Testpage: https://en.wikipedia.org/wiki/User:Polygnotus/DuplicateReferencesTest
 
// <nowiki>
mw.loader.using(['mediawiki.util'], function () {
$(document).ready(function () {
 
if ((mw.config.get('wgNamespaceNumber') !== 0 && mw.config.get('wgPageName') !== 'User:Polygnotus/dupreftest') || mw.config.get('wgAction') !== 'view') {
const DEBUG = false;
 
function debug(...args) {
if (DEBUG) {
console.log('[DuplicateReferences]', ...args);
}
}
 
if (
mw.config.get('wgAction') !== 'view' ||
mw.config.get('wgDiffNewId') ||
mw.config.get('wgDiffOldId') ||
(mw.config.get('wgNamespaceNumber') !== 0 && mw.config.get('wgPageName') !== 'User:Polygnotus/DuplicateReferencesTest')
) {
debug("Not the correct page or action, script terminated");
return;
}
 
debug("Page title:", document.title);
debug("URL:", window.___location.href);
 
function findNextReflistDiv(element) {
let nextElement = element.nextElementSibling;
while (nextElement) {
if (nextElement.tagName.toLowerCase() === 'div' &&
(nextElement.classList.contains('reflist') || nextElement.classList.contains('mw-references-wrap'))) {
return nextElement;
}
nextElement = nextElement.nextElementSibling;
}
return null;
}
 
const referencesHeader = document.querySelector("h2#References");
if (!referencesHeader) {
debug("References heading not found, script terminated");
return;
}
 
letconst referencesHeadingcontainerDiv = documentreferencesHeader.getElementByIdclosest("Referencesdiv");
if (!referencesHeadingcontainerDiv) {
debug("Container div not found, script terminated");
return;
}
 
const reflistDiv = findNextReflistDiv(containerDiv);
if (!reflistDiv) {
debug("Reflist div not found, script terminated");
return;
}
 
const referencesList = reflistDiv.querySelector('ol.references');
if (!referencesList) {
debug("ol.references not found within reflist div");
return;
}
 
const style = document.createElement('style');
style.textContent = `
li:target { border: 4px1px dotted red; padding: 2px; background-color: #ffcccc !important;}
.duplicate-citation-highlight { background-color: #ffe6e6e1eeff; }
.duplicate-citation-hover { background-color: #ffcccccce0ff; border: 1px dotted blue; }
.duplicate-citation-clicked { border: 4px1px dotted red; padding: 2px; background-color: #ffe6e6; }
.mw-collapsible-toggle { font-weight: normal; float: right; }
.duplicate-references-table { width: 100%; }
@media only screen and (max-width: 768px) {
.duplicate-references-table { display: none; }
}
`;
document.head.appendChild(style);
 
function addDuplicateCitationsTemplate(linkElement) {
let parentDiv = referencesHeading.closest("div");
debug("Adding duplicate citations template");
let newParagraph = document.createElement("p");
showLoading(linkElement);
newParagraph.style.color = "red";
function addDuplicateCitationsTemplate() {
var api = new mw.Api();
var pageTitle = mw.config.get('wgPageName');
 
// First, get the duplicate information
let duplicateInfo = getDuplicateInfo();
 
// Get current date
const currentDate = new Date();
const monthNames = ["January", "February", "March", "April", "May", "June",
"July", "August", "September", "October", "November", "December"
];
const currentMonth = monthNames[currentDate.getMonth()];
const currentYear = currentDate.getFullYear();
const dateParam = `|date=${currentMonth} ${currentYear}`;
 
api.get({
Line 37 ⟶ 100:
rvslots: 'main',
formatversion: 2
}).then(function (data) {
var page = data.query.pages[0];
var content = page.revisions[0].slots.main.content;
 
// AddCreate the templatereason at the top of the pagestring
varlet newContentreason = '{{DuplicateDuplicateReferences citations}}script detected:\n\n' + content;
 
// Create the edit summary
let summary = '+{{Duplicate citations}}';
if (duplicateInfo.length > 0) {
summaryduplicateInfo.forEach((info) +=> ': ';{
duplicateInfo reason += `* ${info.forEach(url} (refs: ${info, index).refs.map(r => {r.number).join(', ')})\n\n`;
summary += `${info.url} (refs: ${info.refs.join(', ')})`;
if (index < duplicateInfo.length - 1) {
summary += '; ';
}
});
}
 
// MakeCreate the edittemplate to insert
const templateToInsert = `{{Duplicated citations|reason=${reason}${dateParam}}}\n`;
 
// Use Morebits to handle the template insertion
const wikitextPage = new Morebits.wikitext.page(content);
// Define templates that should come before the duplicated citations template
const precedingTemplates = [
'short description',
'displaytitle',
'lowercase title',
'italic title',
'about',
'redirect',
'distinguish',
'for',
'Featured list',
'Featured article',
'Good article',
'Other uses',
'Redirect2',
'Use mdy dates',
'Use dmy dates',
'Use American English',
'Use British English'
];
 
// Insert the template after the specified templates
// The third parameter is flags (default 'i' for case-insensitive)
// The fourth parameter can include pre-template content like HTML comments
wikitextPage.insertAfterTemplates(templateToInsert, precedingTemplates, 'i', ['<!--[\\s\\S]*?-->']);
var newContent = wikitextPage.getText();
 
let summary = `Tagged [[WP:DUPREF|duplicate citations]] using [[User:Polygnotus/DuplicateReferences|DuplicateReferences]]`;
 
return api.postWithToken('csrf', {
action: 'edit',
Line 63 ⟶ 154:
summary: summary
});
}).then(function () {
mw.notifyshowSuccess('Successfully added the Duplicate citations template!'linkElement);
//setTimeout(function Reload() the page to show the changes{
___location.reload();
}, 100); // Reload after 0.catch(function(error)1 {second
}).catch(function (error) {
console.error('Error:', error);
showError(linkElement);
mw.notify('Failed to add the template. See console for details.', {type: 'error'});
});
}
 
function showLoading(element) {
element.innerHTML = '<sup><small>[ Working... ]</small></sup>';
}
 
function showSuccess(element) {
element.innerHTML = '<sup><small>[ Done ]</small></sup>';
}
 
function showError(element) {
element.innerHTML = '<sup><small>[ Error ]</small></sup>';
}
 
function getVisibleText(element) {
// Recursively get the visible text content of an element
let text = '';
for (let node of element.childNodes) {
if (node.nodeType === Node.TEXT_NODE) {
text += node.textContent.trim() + ' ';
} else if (node.nodeType === Node.ELEMENT_NODE) {
// Skip hidden elements
const style = window.getComputedStyle(node);
if (style.display !== 'none' && style.visibility !== 'hidden') {
text += getVisibleText(node) + ' ';
}
}
}
return text.trim();
}
 
function calculateLevenshteinDistance(a, b) {
debug("Comparing:");
debug("Text 1:", a);
debug("Text 2:", b);
 
if (a.length === 0) return b.length;
if (b.length === 0) return a.length;
 
const matrix = [];
 
// Increment along the first column of each row
for (let i = 0; i <= b.length; i++) {
matrix[i] = [i];
}
 
// Increment each column in the first row
for (let j = 0; j <= a.length; j++) {
matrix[0][j] = j;
}
 
// Fill in the rest of the matrix
for (let i = 1; i <= b.length; i++) {
for (let j = 1; j <= a.length; j++) {
if (b.charAt(i - 1) === a.charAt(j - 1)) {
matrix[i][j] = matrix[i - 1][j - 1];
} else {
matrix[i][j] = Math.min(
matrix[i - 1][j - 1] + 1, // substitution
Math.min(
matrix[i][j - 1] + 1, // insertion
matrix[i - 1][j] + 1 // deletion
)
);
}
}
}
 
debug("Levenshtein distance:", matrix[b.length][a.length]);
return matrix[b.length][a.length];
}
 
function calculateSimilarityPercentage(distance, maxLength) {
const similarity = ((maxLength - distance) / maxLength) * 100;
debug("Similarity percentage:", similarity.toFixed(2) + "%");
return Math.round(similarity) + '%';
}
 
function getDuplicateInfo() {
debug("Getting duplicate info");
const referenceSpans = document.querySelectorAll('span.reference-text');
 
const urlMap = new Map();
const duplicates = [];
const urlMap = new Map();
const referenceItems = Array.from(referencesList.children);
 
debug("Number of reference items:", referenceItems.length);
referenceSpans.forEach((span) => {
const links = span.querySelectorAll('a');
const refNumber = span.closest('li')?.id.split('-').pop() || 'Unknown';
 
referenceItems.forEach((item, let validLinkindex) => null;{
forif (let linkitem.tagName.toLowerCase() of=== links'li') {
const urlrefId = linkitem.hrefid;
const linkTextrefNumber = link.textContent.trim()index + 1;
debug(`Processing reference item ${refNumber} (${refId})`);
 
if (
// Get the visible linkTexttext !==of "Archived"the &&entire reference item
const refText = !url.includesgetVisibleText("wikipedia.org/wiki/"item) &&;
debug(` Reference text: !url.includes("_(identifier)"${refText}`);
 
) {
// Find the first validLink =valid link; in the reference
const links = breakitem.querySelectorAll('a');
let validLink = null;
for (let link of links) {
const url = link.href;
 
// Skip this reference if the URL doesn't contain 'http'
if (!url.includes('http')) {
debug(` Skipping reference ${refNumber} - URL does not contain 'http'`);
return; // This 'return' is equivalent to 'continue' in a regular for loop
}
const linkText = link.textContent.trim();
 
if (
// (!url.includes("wikipedia.org/wiki/") || url.includes("Special:BookSources")) &&
linkText !== "Archived" &&
!url.includes("wikipedia.org") &&
!url.includes("_(identifier)") && // Templates like ISBN and ISSN and OCLC and S2CID contain (identifier)
!url.startsWith("https://search.worldcat.org/") && // |issn= parameter in cite news
!url.startsWith("https://www.bbc.co.uk/news/live/") && // live articles get frequent updates
!url.startsWith("https://www.aljazeera.com/news/liveblog/") &&
!url.startsWith("https://www.nbcnews.com/news/world/live-blog/") &&
!url.startsWith("https://www.theguardian.com/world/live/") &&
!url.startsWith("https://www.nytimes.com/live/") &&
!url.startsWith("https://edition.cnn.com/world/live-news/") &&
!url.startsWith("https://www.timesofisrael.com/liveblog") &&
!url.startsWith("https://www.france24.com/en/live-news/") &&
!url.startsWith("https://books.google.com/") && //may be 2 different pages of the same book
!url.startsWith("https://archive.org/details/isbn_")
) {
validLink = link;
debug(` Valid link found: ${url}`);
break;
}
}
}
 
if (validLink) {
const url = validLink.href;
if (urlMap.has(url)) {
urlMap.get(url).push({id: refId, number: refNumber, text: refText});
debug(` Duplicate found for URL: ${url}`);
} else {
urlMap.set(url, [{id: refId, number: refNumber, text: refText}]);
debug(` New URL added to map: ${url}`);
}
} else {
urlMap.setdebug(url,` [refNumber] No valid link found in this item`);
}
}
Line 109 ⟶ 314:
urlMap.forEach((refs, url) => {
if (refs.length > 1) {
duplicates.push({// url,Calculate refsLevenshtein });distance for each pair of refs
for (let i = 0; i < refs.length - 1; i++) {
for (let j = i + 1; j < refs.length; j++) {
debug(`Comparing references ${refs[i].number} and ${refs[j].number}:`);
const distance = calculateLevenshteinDistance(refs[i].text, refs[j].text);
const maxLength = Math.max(refs[i].text.length, refs[j].text.length);
const similarity = calculateSimilarityPercentage(distance, maxLength);
refs[i].similarity = refs[i].similarity || {};
refs[i].similarity[refs[j].id] = similarity;
}
}
duplicates.push({url, refs});
}
});
 
debug("Number of duplicate sets found:", duplicates.length);
debug("Duplicate sets:", duplicates);
return duplicates;
}
 
function calculateSimilaritycreateCollapsibleTable(str1, str2duplicateInfo) {
const longertable = str1document.length > str2.length ? str1 : str2createElement('table');
table.className = 'wikitable mw-collapsible duplicate-references-table';
const shorter = str1.length > str2.length ? str2 : str1;
consttable.setAttribute('role', longerLength = longer.length'presentation');
 
if (longerLength === 0) {
const tbody = return 1document.0createElement('tbody');
table.appendChild(tbody);
 
const headerRow = document.createElement('tr');
const headerCell = document.createElement('td');
headerCell.innerHTML = '<strong>Duplicate References</strong>';
 
const toggleSpan = document.createElement('span');
toggleSpan.className = 'mw-collapsible-toggle';
toggleSpan.innerHTML = '[<a href="#" class="mw-collapsible-text">hide</a>]';
headerCell.appendChild(toggleSpan);
 
 
// Check if the {{Duplicated citations}} template is already present
const duplicatedCitationsTemplate = document.querySelector('table.box-Duplicated_citations');
 
// Only add the link if the template is not present
if (!duplicatedCitationsTemplate) {
 
// Add the "add {{duplicated citations}}" link to the header
const addTemplateLink = document.createElement('a');
addTemplateLink.textContent = ' add {{duplicated citations}} ';
addTemplateLink.href = '#';
addTemplateLink.addEventListener('click', function (e) {
e.preventDefault();
addDuplicateCitationsTemplate(this);
});
//headerCell.appendChild(document.createTextNode(' ['));
headerCell.appendChild(addTemplateLink);
//headerCell.appendChild(document.createTextNode(']'));
}
headerRow.appendChild(headerCell);
return (longerLength - editDistance(longer, shorter)) / parseFloat(longerLength);
tbody.appendChild(headerRow);
}
 
const pageTitle = mw.config.get('wgPageName').replace(/_/g, ' ');
function editDistance(s1, s2) {
s1 = s1.toLowerCase();
s2 = s2.toLowerCase();
 
constduplicateInfo.forEach(({url, costsrefs}) => new Array();{
for (let i = 0;const irow <= s1document.lengthcreateElement('tr'); i++) {
letconst lastValuecell = idocument.createElement('td');
 
for (let j = 0; j <= s2.length; j++) {
// Create report if (i == 0)icon
const costs[j]reportIcon = jdocument.createElement('a');
reportIcon.href = `https://en.wikipedia.org/wiki/User_talk:Polygnotus?action=edit&section=new&preloadtitle=Reporting%20%5B%5BUser%3APolygnotus%2FDuplicateReferences%7CDuplicateReferences%5D%5D%20false-positive&preload=User:Polygnotus/$1&preloadparams%5b%5d=${encodeURIComponent(`[[${pageTitle}]] ${url}`)}%20~~~~`;
else {
reportIcon.innerHTML = '<img src="https://upload.wikimedia.org/wikipedia/commons/thumb/e/ef/Cross_CSS_Red.svg/15px-Cross_CSS_Red.svg.png" width="15" height="15" alt="Report false positive" title="Report false positive" />';
if (j > 0) {
let newValuereportIcon.style.marginRight = costs[j - 1]'5px';
if (s1cell.charAtappendChild(i - 1) != s2.charAt(j - 1)reportIcon);
 
newValue = Math.min(Math.min(newValue, lastValue),
let urlLink = costs[j]document.createElement('a') + 1;
costs[j - 1]urlLink.href = lastValueurl;
lastValueurlLink.textContent = newValueurl;
urlLink.target = }"_blank";
urlLink.rel = "noopener noreferrer";
 
cell.appendChild(urlLink);
cell.appendChild(document.createTextNode(' in refs: '));
 
const originalRef = refs[0];
refs.forEach((ref, index) => {
let link = document.createElement('a');
link.href = `#${ref.id}`;
link.textContent = ref.number;
cell.appendChild(link);
 
// Add similarity information
if (index > 0) {
const similarity = calculateSimilarityPercentage(
calculateLevenshteinDistance(originalRef.text, ref.text),
Math.max(originalRef.text.length, ref.text.length)
);
let similarityInfo = document.createElement('span');
similarityInfo.textContent = ` (${similarity})`;
cell.appendChild(similarityInfo);
}
} link.addEventListener('mouseover', () => {
if refs.forEach(ir => 0){
costs[s2.length] const citationElement = lastValuedocument.getElementById(r.id);
} if (citationElement) {
return costs[s2 if (r.length];id === ref.id) {
citationElement.classList.add('duplicate-citation-hover');
}
} else {
citationElement.classList.add('duplicate-citation-highlight');
}
}
});
});
link.addEventListener('mouseout', () => {
refs.forEach(r => {
const citationElement = document.getElementById(r.id);
if (citationElement) {
citationElement.classList.remove('duplicate-citation-hover');
citationElement.classList.remove('duplicate-citation-highlight');
}
});
});
 
link.addEventListener('click', () => {
function getAllVisibleText(element) {
document.querySelectorAll('.duplicate-citation-clicked').forEach(el => {
if (element.nodeType === Node.TEXT_NODE) {
return element el.textContentclassList.trimremove('duplicate-citation-clicked');
});
refs.forEach(r => {
const citationElement = document.getElementById(r.id);
if (element.nodeType === Node.ELEMENT_NODE) {
if (window.getComputedStyle(element).display === 'none' if (citationElement) {
return citationElement.classList.add('duplicate-citation-clicked');
}
});
return Array.from(element.childNodes });
.map(child => getAllVisibleText(child))
.join(' ')
.replace(/\s+/g, ' ')
.trim();
}
return '';
}
 
function extractVisibleText if (htmlStringindex < refs.length - 1) {
const parser = new DOMParser cell.appendChild(document.createTextNode(', '));
const doc = parser.parseFromString(htmlString, 'text/html'); }
return getAllVisibleText(doc.body });
 
row.appendChild(cell);
tbody.appendChild(row);
});
 
return table;
}
 
function checkDuplicateReferenceLinks() {
constdebug("Checking referenceSpansfor =duplicate document.querySelectorAll('span.reference-text' links");
const urlMapduplicateInfo = new MapgetDuplicateInfo();
 
const duplicates = new Map();
letif totalLinks(duplicateInfo.length => 0;) {
debug("Duplicates found, creating collapsible table");
 
// Create a map of cite_note ids to their correct reference numbers
const citeNoteMaptable = new MapcreateCollapsibleTable(duplicateInfo);
containerDiv.after(table);
document.querySelectorAll('.reference').forEach(ref => {
 
const link = ref.querySelector('a');
if// (link)Set {up collapsible functionality
const linkHreftoggleLink = linktable.getAttributequerySelector('href.mw-collapsible-toggle a');
const tableBody = if $(linkHreftable) {.find('tr:not(:first-child)');
const citeNoteIdstorageKey = linkHref.substring(1); // Remove the leading '#duplicateReferencesTableState';
 
const refNumber = citeNoteId.split('-').pop();
function citeNoteMap.setsetTableState(citeNoteId,isCollapsed) refNumber);{
}if (isCollapsed) {
} tableBody.hide();
}) toggleLink.textContent = 'show';
referenceSpans.forEach((span) => {
const links = span.querySelectorAll('a');
const citeNote = span.closest('li')?.id;
const refNumber = citeNoteMap.get(citeNote) || 'Unknown';
const refText = extractVisibleText(span.outerHTML); // Extract visible text content
let validLink = null;
for (let link of links) {
const url = link.href;
const linkText = link.textContent.trim();
if (
linkText !== "Archived" &&
!url.includes("wikipedia.org/wiki/") &&
!url.includes("_(identifier)")
) {
validLink = link;
break;
}
}
if (validLink) {
const url = validLink.href;
totalLinks++;
if (urlMap.has(url)) {
if (duplicates.has(url)) {
duplicates.get(url).push({ refNumber, citeNote, refText });
} else {
duplicates.set(url, [urlMap.get(url), { refNumber, citeNote, refText }]);
}
} else {
urlMaptableBody.setshow(url, { refNumber, citeNote, refText });
toggleLink.textContent = 'hide';
}
localStorage.setItem(storageKey, isCollapsed);
}
});
if (duplicates.size > 0) {
if (document.querySelector('table.box-Duplicated_citations') === null) {
const editSections = parentDiv.querySelectorAll('span.mw-editsection');
editSections.forEach(editSection => {
let spanBefore = document.createElement('span');
spanBefore.className = 'mw-editsection-bracket';
spanBefore.textContent = '[';
let addTemplateLink = document.createElement('a');
addTemplateLink.textContent = ' add {{duplicated citations}} ';
addTemplateLink.href = '#';
addTemplateLink.addEventListener('click', function(e) {
e.preventDefault();
addDuplicateCitationsTemplate();
});
let spanAfter = document.createElement('span');
spanAfter.className = 'mw-editsection-bracket';
spanAfter.textContent = ']';
editSection.appendChild(spanBefore);
editSection.appendChild(addTemplateLink);
editSection.appendChild(spanAfter);
});
}
duplicates.forEach((refInfo, url) => {
let paragraphInfo = document.createElement('span');
let urlLink = document.createElement('a');
urlLink.href = url;
urlLink.textContent = url;
urlLink.target = "_blank";
urlLink.rel = "noopener noreferrer";
paragraphInfo.appendChild(document.createTextNode('Duplicate URL: '));
paragraphInfo.appendChild(urlLink);
paragraphInfo.appendChild(document.createTextNode(' in refs: '));
refInfo.forEach((ref, index) => {
if (ref.citeNote) {
let link = document.createElement('a');
link.href = `#${ref.citeNote}`;
link.textContent = ref.refNumber;
paragraphInfo.appendChild(link);
 
// Highlight only the specificInitialize duplicatesstate onfrom hoverlocalStorage
const initialState = linklocalStorage.addEventListener('mouseover', getItem(storageKey) =>== {'true';
refInfo.forEachsetTableState(duplicate => {initialState);
const citationElement = document.getElementById(duplicate.citeNote);
if (citationElement) {
if (duplicate.citeNote === ref.citeNote) {
citationElement.classList.add('duplicate-citation-hover');
} else {
citationElement.classList.add('duplicate-citation-highlight');
}
}
});
});
link.addEventListener('mouseout', () => {
refInfo.forEach(duplicate => {
const citationElement = document.getElementById(duplicate.citeNote);
if (citationElement) {
citationElement.classList.remove('duplicate-citation-hover');
citationElement.classList.remove('duplicate-citation-highlight');
}
});
});
 
// Highlight duplicates on toggleLink.addEventListener('click', andfunction allow(e) navigation{
linke.addEventListener('click', preventDefault() => {;
const isCurrentlyCollapsed = // Remove previous click highlightstableBody.is(':hidden');
document.querySelectorAllsetTableState('.duplicate-citation-clicked'!isCurrentlyCollapsed).forEach(el => {;
el.classList.remove('duplicate-citation-clicked');
});
// Add new click highlights
refInfo.forEach(duplicate => {
const citationElement = document.getElementById(duplicate.citeNote);
if (citationElement) {
citationElement.classList.add('duplicate-citation-clicked');
}
});
// The default behavior (navigation) will now occur
});
} else {
paragraphInfo.appendChild(document.createTextNode(ref.refNumber));
}
// Calculate similarity with the next reference
if (index < refInfo.length - 1) {
const similarity = calculateSimilarity(ref.refText, refInfo[index + 1].refText);
const similarityPercentage = Math.round(similarity * 100);
paragraphInfo.appendChild(document.createTextNode(` (${similarityPercentage}%)`));
}
if (index < refInfo.length - 1) {
paragraphInfo.appendChild(document.createTextNode(', '));
}
});
paragraphInfo.appendChild(document.createElement('br'));
newParagraph.appendChild(paragraphInfo);
});
} else {
parentDiv.afterdebug(newParagraph"No duplicates found");
}
}
 
checkDuplicateReferenceLinks();
debug("Script execution completed");
});
});
// </nowiki>
Retrieved from "https://en.wikipedia.org/wiki/User:Polygnotus/DuplicateReferences.js"