User:Polygnotus/Scripts/DetectPromo-v2.js

//Forked from User:Novem Linguae/Scripts/DetectPromo.js

//now highlights the words in the article text and those in the top bar are now clickable

//

/*

- Let reviewer know when certain promotional and POV keywords are detected.

- Displays a bar at the top of the article, listing the detected keywords.

- Highlights detected promotional words with a yellow background and red border within the article text.

- Makes the words in the top bar clickable, scrolling to their first occurrence in the article.

- Added "Report false positive" link to allow users to report incorrect detections.

- Added confirmation dialog explaining false positive criteria before reporting.

- Ignores promotional words that appear in the page title.

- Ignores promotional words that appear in wiki link titles.

• /

class DetectPromo {

/** @type {string[]} */

wordsToSearch = [

'% growth', '6-figure', '7-figure', '8-figure', '9-figure',

'B2B', 'B2C', 'a record', 'acclaimed', 'accomplished',

'are a necessity', 'around the world', 'award winning', 'award-winning',

'beloved', 'best available', 'bestselling', 'boasts', 'comprehensive',

'countless hours', 'create a revolution', 'critical acclaim',

'disrupt', 'drastically', 'dynamic', 'elevate', 'eminent', 'engaging',

'entrepreneur', 'evangelist', 'excelled', 'exceptional', 'exemplified',

'exemplify', 'expert', 'expertise', 'extensive', 'famous', 'fascinating',

'fast growing', 'fast-growing', 'fastest growing', 'fastest-growing',

'finest', 'fully integrated', 'fully-integrated', 'globally',

'globally recognized', 'growing popularity', 'highlights',

'highly accomplished', 'highly praised', 'highly specialized',

'historic', 'honored with', 'hypnotic', 'illustrious', 'impressive',

'indelible', 'inexhaustible', 'influential', 'innovation', 'innovative',

'insights', 'inspired by', 'integrate', 'invaluable', 'leader in',

'leading', 'legendary', 'leverage', 'massive', 'mastermind', 'more than',

'most highly', 'most important', 'most impressive', 'most notable',

'mystical', 'natural charm', 'noteworthy', 'numerous', 'organically',

'outstanding', 'perfect', 'philanthropist', 'picturesque', 'pioneer',

'pioneering', 'popular destination', 'popularity', 'premiere',

'prestigious', 'prominence', 'prominent', 'promising', 'promulgator',

'ranked', 'reinvent', 'remarkable', 'renowed', 'renowned', 'resonating',

'respected', 'revolutionary', 'rising star', 'save millions', 'savvy',

'seamless', 'sensual', 'several offers', 'showcased', 'signature',

'significant', 'soulful', 'spanning', 'state of art', 'state of the art',

'state-of-art', 'state-of-the-art', 'striking', 'super famous',

'tailored', 'tranquility', 'transcend', 'transform', 'underpin',

'ventured into', 'very first', 'visionary', 'wide selection',

'widely used', 'world class', 'world-class', 'worldwide', 'zero to hero'

];

/**

* Lists of allowed exception phrases - these will NOT be flagged even if they contain a listed word

* Key is the base word, value is an array of exception phrases

* @type {Object}

*/

exceptionPhrases = {

'outstanding': ['outstanding debts'],

'drastically': ['vary drastically'],

'revolutionary': ['revolutionary war'],

'revolutionary': ['American revolutionary'],

'Revolutionary': ['Revolutionary Committee'],

'dynamic': ['dynamic list'],

'inspired': ['inspired by'],

'leading': ['leading to'],

'outstanding': ['outstanding warrant'],

'numerous': ['numerous witnesses'],

'significant': ['significant blow'],

'historic': ['National Register of Historic Places']

};

/**

* @param {Object} mw

* @param {jQuery} $

*/

constructor(mw, $) {

this.mw = mw;

this.$ = $;

// Store page title for later use

this.pageTitle = this.mw.config.get('wgTitle');

this.pageName = this.mw.config.get('wgPageName');

// Convert to lowercase for case-insensitive comparison

this.pageTitleLower = this.pageTitle.toLowerCase();

// Pre-process the words list for more efficient matching

this.processWordsList();

}

/**

* Process words list to identify base words for exception checking

*/

processWordsList() {

// Create a mapping of base words to their searchable variations

this.baseWordMap = {};

// For each searchable word, find if it has a base word with exceptions

for (const word of this.wordsToSearch) {

for (const baseWord in this.exceptionPhrases) {

// If the word contains the base word, map it back

if (word.toLowerCase().includes(baseWord.toLowerCase())) {

if (!this.baseWordMap[word.toLowerCase()]) {

this.baseWordMap[word.toLowerCase()] = [];

}

this.baseWordMap[word.toLowerCase()].push(baseWord.toLowerCase());

}

}

}

}

async execute() {

if (!this.shouldRunOnThisPage()) {

return;

}

const title = this.mw.config.get('wgPageName');

const wikicode = await this.getWikicode(title);

if (!wikicode) return;

// Clean wikicode by removing links and references

const cleanedWikicode = this.cleanWikicode(wikicode);

const searchResults = this.getSearchResults(cleanedWikicode);

if (searchResults.length > 0) {

this.displayResults(searchResults);

this.highlightPromoWords(searchResults);

}

}

/**

* @param {string[]} searchResults

*/

displayResults(searchResults) {

const MAX_DISPLAYED_RESULTS = 20;

const displayedResults = searchResults.slice(0, MAX_DISPLAYED_RESULTS);

let html = `

`;

this.$('#contentSub').after(html);

// Add click event listeners for promo words

this.$('.promo-word').on('click', (e) => {

e.preventDefault();

const word = this.$(e.target).data('word');

this.scrollToWord(word);

});

// Add click event listener for false positive reporting

this.$('#report-false-positive').on('click', (e) => {

e.preventDefault();

this.showFalsePositiveDialog(searchResults);

});

}

/**

* Scroll to the first occurrence of a word in the article

* @param {string} word

*/

scrollToWord(word) {

const content = this.$('#mw-content-text');

const regex = new RegExp(`\\b${this.escapeRegEx(word)}\\b`, 'i');

const elements = content.find('*').contents().filter(function() {

return this.nodeType === Node.TEXT_NODE && regex.test(this.textContent);

});

if (elements.length > 0) {

const firstOccurrence = elements[0];

firstOccurrence.parentElement.scrollIntoView({ behavior: 'smooth', block: 'center' });

}

}

/**

* Highlight promotional words with yellow background and red border within the article text

* @param {string[]} searchResults

*/

highlightPromoWords(searchResults) {

const content = this.$('#mw-content-text');

const highlightStyle = 'background-color: yellow; border: 1px solid red; padding: 2px; margin: -2px;';

searchResults.forEach(word => {

const regex = new RegExp(`\\b${this.escapeRegEx(word)}\\b`, 'gi');

content.find('*').contents().filter(function() {

return this.nodeType === Node.TEXT_NODE;

}).each((_, textNode) => {

const text = textNode.textContent;

if (regex.test(text)) {

const newHtml = text.replace(regex, `$&`);

const newElement = document.createElement('span');

newElement.innerHTML = newHtml;

textNode.parentNode.replaceChild(newElement, textNode);

}

});

});

}

/**

* Check if a word appears in the page title

* @param {string} word - The word to check

* @return {boolean} - True if the word is in the page title

*/

isWordInPageTitle(word) {

const wordLower = word.toLowerCase();

// Simple check for exact word in title

if (this.pageTitleLower.includes(wordLower)) {

// Check with word boundaries to ensure it's a complete word

const wordRegex = new RegExp(`\\b${this.escapeRegEx(wordLower)}\\b`, 'i');

return wordRegex.test(this.pageTitleLower);

}

return false;

}

/**

* Scans text for promotional words while respecting exceptions

* @param {string} text - The wikicode text to search

* @return {string[]} - List of found promotional words

*/

getSearchResults(text) {

const results = [];

const lowerText = text.toLowerCase();

// For each word to search

for (const word of this.wordsToSearch) {

const lowerWord = word.toLowerCase();

// Check if this word is in the text (quick filter)

if (!lowerText.includes(lowerWord)) {

continue;

}

// Test with proper word boundaries

const wordRegex = new RegExp(`\\b${this.escapeRegEx(word)}\\b`, 'i');

if (!wordRegex.test(text)) {

continue;

}

// Check if word is in page title

if (this.isWordInPageTitle(word)) {

continue; // Skip this word if it's in the page title

}

// See if this word has any base words with exceptions

const baseWords = this.baseWordMap[lowerWord] || [];

// Check if this is an exact match for an exception phrase

let isException = false;

for (const baseWord of baseWords) {

const exceptionList = this.exceptionPhrases[baseWord] || [];

for (const exceptionPhrase of exceptionList) {

// If this word is part of an exception phrase that exists in the text, flag it

if (exceptionPhrase.toLowerCase() === lowerWord ||

(lowerText.includes(exceptionPhrase.toLowerCase()) &&

new RegExp(`\\b${this.escapeRegEx(exceptionPhrase)}\\b`, 'i').test(text))) {

isException = true;

break;

}

}

if (isException) break;

}

// Add it to results if it's not an exception

if (!isException) {

results.push(word);

}

}

return results;

}

/**

* Clean wikicode by removing links and references

* This focuses on removing wikilinks to avoid detecting promotional words

* within link targets/titles

*

* @param {string} wikicode

* @return {string} cleanedWikicode

*/

cleanWikicode(wikicode) {

// First, remove all wikilinks while keeping display text

let cleanedCode = wikicode.replace(/\[\[([^\]\|]+)\|([^\]]+)\]\]/g, '$2');

// Then, remove all simple wikilinks

cleanedCode = cleanedCode.replace(/\[\[([^\]]+)\]\]/g, '$1');

// Remove tags

cleanedCode = cleanedCode.replace(/|]*\/>/gm, '');

return cleanedCode;

}

/**

* @return {boolean}

*/

shouldRunOnThisPage() {

const action = this.mw.config.get('wgAction');

const isDiff = this.mw.config.get('wgDiffNewId');

const isDeletedPage = !this.mw.config.get('wgCurRevisionId');

const namespace = this.mw.config.get('wgNamespaceNumber');

const title = this.mw.config.get('wgPageName');

return (

action === 'view' &&

!isDiff &&

!isDeletedPage &&

([0, 118].includes(namespace) ||

title === 'User:Polygnotus')

);

}

/**

* @param {string} title

* @return {Promise} wikicode

*/

async getWikicode(title) {

try {

const api = new this.mw.Api();

const response = await api.get({

action: 'parse',

page: title,

prop: 'wikitext',

formatversion: '2',

format: 'json'

});

return response.parse.wikitext;

} catch (error) {

console.error('Error fetching wikicode:', error);

return null;

}

}

/**

* Show enhanced dialog with text field for reporting false positives

* @param {string[]} detectedWords

*/

showFalsePositiveDialog(detectedWords) {

// Use MediaWiki's OOjs UI dialog framework

mw.loader.using(['oojs-ui-core', 'oojs-ui-windows', 'oojs-ui-widgets'], () => {

// Create layout for dialog content

const layout = new OO.ui.FieldsetLayout({

label: 'Report False Positive'

});

// Create a more direct text field for entering word combination

const wordCombinationField = new OO.ui.TextInputWidget({

placeholder: 'Enter word combination (e.g., "outstanding warrant")',

title: 'Enter the exact multi-word combination that should not be flagged as promotional'

});

// Display the detected promotional words for reference

const detectedWordsHtml = $('