Jump to content

User:Polygnotus/Scripts/DetectPromo-v2.js

From Wikipedia, the free encyclopedia
This is an old revision of this page, as edited by Polygnotus (talk | contribs) at 22:00, 6 April 2025. The present address (URL) is a permanent link to this revision, which may differ significantly from the current revision.
Note: After saving, you have to bypass your browser's cache to see the changes. Google Chrome, Firefox, Microsoft Edge and Safari: Hold down the ⇧ Shift key and click the Reload toolbar button. For details and instructions about other browsers, see Wikipedia:Bypass your cache.
//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

// <nowiki>

/*
- 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.
*/



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', 'leverag', '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'
    ];

    /** 
     * @type {Object} 
     * Dictionary of words and their exception patterns
     * Each entry is a word and regex pattern that should NOT be flagged
     */
      exceptions = {
        'leading': /leading\s+to\b/i,
        'outstanding': /outstanding\s+warrant/i,
        'numerous': /numerous\s+witnesses\b/i,
        'significant': /significant\s+blow\b/i,
        'historic': /National\s+Register\s+of\s+Historic\s+Places/i
    };


    /**
     * @param {Object} mw
     * @param {jQuery} $
     */
    constructor(mw, $) {
        this.mw = mw;
        this.$ = $;
    }

    async execute() {
        if (!this.shouldRunOnThisPage()) {
            return;
        }

        const title = this.mw.config.get('wgPageName');
        const wikicode = await this.getWikicode(title);
        if (!wikicode) return;

        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 = `
            <div id="DetectPromo" style="background-color: #ccc; padding: 10px; margin-bottom: 10px; display: flex; justify-content: space-between; align-items: center;">
                <div>
                    <span style="font-weight: bold;">Potentially promotional words detected:</span> 
        `;

        html += displayedResults.map(word => 
            `<a href="#" class="promo-word" data-word="${word}" style="color: blue; text-decoration: underline; cursor: pointer;">${word}</a>`
        ).join(', ');

        if (searchResults.length > MAX_DISPLAYED_RESULTS) {
            html += ', ...... and more.';
        }

        html += `</div>
                <div>
                    <a href="#" id="report-false-positive" style="color: #d33; text-decoration: underline; font-size: 0.9em;">Report false positive</a>
                </div>
            </div>`;

        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${this.escapeRegEx(word)}\\b`, 'gi');
            content.find('*').contents().filter(function() {
                return this.nodeType === Node.TEXT_NODE;
            }).each((_, textNode) => {
                const text = textNode.textContent;
                if (regex.test(text)) {
                    const newHtml = text.replace(regex, `<span class="promo-highlight" data-word="${word}" style="${highlightStyle}">$&</span>`);
                    const newElement = document.createElement('span');
                    newElement.innerHTML = newHtml;
                    textNode.parentNode.replaceChild(newElement, textNode);
                }
            });
        });
    }

    /**
     * @param {string} wikicode
     * @return {string[]} searchResults
     */
    getSearchResults(wikicode) {
        return this.wordsToSearch.filter(word => {
            // Check if the word has exceptions
            if (this.exceptions[word.toLowerCase()]) {
                // Regular expression to find the word with word boundaries
                const wordRegex = new RegExp(`\\b${this.escapeRegEx(word)}\\b`, 'gi');
                const text = wikicode;
                let match;
                let foundPromoUse = false;
                
                // Check each occurrence of the word
                while ((match = wordRegex.exec(text)) !== null) {
                    // Get context around this match (40 chars before and after)
                    const startPos = Math.max(0, match.index - 40);
                    const endPos = Math.min(text.length, match.index + word.length + 40);
                    const context = text.substring(startPos, endPos);
                    
                    // If this occurrence doesn't match the exception pattern, it's promotional
                    if (!this.exceptions[word.toLowerCase()].test(context)) {
                        foundPromoUse = true;
                        break;
                    }
                }
                
                return foundPromoUse;
            } else {
                // Regular case, no exceptions
                const regEx = new RegExp(`\\b${this.escapeRegEx(word)}\\b`, 'i');
                return regEx.test(wikicode);
            }
        });
    }

    /**
     * @param {string} wikicode
     * @return {string} cleanedWikicode
     */
    cleanWikicode(wikicode) {
        return wikicode
            .replace(/\[\[|\]\]/g, '') // Remove [[ ]]
            .replace(/<ref[^<]*<\/ref>|<ref[^>]*\/>/gm, ''); // Remove <ref></ref> and <ref />
    }

    /**
     * @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<string|null>} 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 = $('<div>')
                .addClass('detected-words-list')
                .css({
                    'margin-bottom': '10px',
                    'background-color': '#f8f9fa',
                    'padding': '8px',
                    'border-radius': '2px',
                    'border': '1px solid #eaecf0'
                })
                .append($('<strong>').text('Detected promotional words: '))
                .append(document.createTextNode(detectedWords.join(', ')));

            // Add fields to layout
            layout.addItems([
                new OO.ui.FieldLayout(wordCombinationField, {
                    label: 'Enter a word combination that is NOT promotional',
                    align: 'top'
                })
            ]);
            
            // Add the detected words info before the layout
            layout.$element.prepend(detectedWordsHtml);

            // Add instructions
            const instructionsLayout = new OO.ui.PanelLayout({
                padded: true,
                expanded: false
            });
            
            instructionsLayout.$element.append(
                $('<div>')
                    .css('margin-bottom', '1em')
                    .append($('<p>').text('Please provide a multi-word combination that includes one of the promotional words but is NOT promotional.'))
                    .append($('<p>').html('<strong>The combination must:</strong>'))
                    .append($('<ul>')
                        .append($('<li>').text('Include at least 2 words'))
                        .append($('<li>').text('Include one of the detected promotional words'))
                        .append($('<li>').text('Form a phrase that is NOT promotional'))
                    )
                    .append($('<p>').text('For example:'))
                    .append($('<ul>')
                        .append($('<li>').text('"outstanding" can be promotional, but "outstanding warrant" is not'))
                        .append($('<li>').text('"leading" can be promotional, but "leading to" is not'))
                        .append($('<li>').text('"award-winning" can be promotional, but "award-winning ceremony" is not'))
                    )
            );

            // Define a proper dialog class with a name
            function FalsePositiveDialog(config) {
                FalsePositiveDialog.super.call(this, config);
            }
            OO.inheritClass(FalsePositiveDialog, OO.ui.ProcessDialog);
            
            // Define static properties
            FalsePositiveDialog.static.name = 'falsePositiveDialog';
            FalsePositiveDialog.static.title = 'Report False Positive';
            FalsePositiveDialog.static.actions = [
                {
                    action: 'cancel',
                    label: 'Cancel',
                    flags: ['safe', 'close']
                },
                {
                    action: 'submit',
                    label: 'Submit Report',
                    flags: ['primary', 'progressive']
                }
            ];
            FalsePositiveDialog.static.size = 'medium';
            
            // Create message dialog
            const messageDialog = new FalsePositiveDialog();

            // Define dialog process
            FalsePositiveDialog.prototype.getActionProcess = function(action) {
                const self = this;
                
                if (action === 'submit') {
                    const wordCombination = wordCombinationField.getValue().trim();
                    
                    // Check if the word combination has at least 2 words
                    const wordCount = wordCombination.split(/\s+/).filter(w => w.length > 0).length;
                    
                    if (wordCount < 2) {
                        // Show error message for insufficient words
                        wordCombinationField.setValidityFlag(false);
                        return new OO.ui.Process(function() {
                            self.showErrors(new OO.ui.Error(
                                'Please enter a multi-word combination (at least 2 words). We only want to exclude word combinations that are not promotional in context.',
                                { recoverable: true }
                            ));
                        });
                    }
                    
                    // Check if any of the detected promotional words are part of the combination
                    const containsPromoWord = detectedWords.some(word => 
                        wordCombination.toLowerCase().includes(word.toLowerCase())
                    );
                    
                    if (!containsPromoWord) {
                        // Show error message for missing any promotional word
                        wordCombinationField.setValidityFlag(false);
                        return new OO.ui.Process(function() {
                            self.showErrors(new OO.ui.Error(
                                'Your word combination must include one of the detected promotional words.',
                                { recoverable: true }
                            ));
                        });
                    }
                    
                    // Find which promotional word is included in the combination
                    let includedWord = '';
                    for (const word of detectedWords) {
                        if (wordCombination.toLowerCase().includes(word.toLowerCase())) {
                            includedWord = word;
                            break;
                        }
                    }
                    
                    // All validation passed, proceed with submission
                    return new OO.ui.Process(function() {
                        self.close({
                            action: action,
                            selectedWord: includedWord,
                            wordCombination: wordCombination
                        });
                    });
                }
                
                // Handle cancel and close actions explicitly
                if (action === 'cancel' || action === 'close') {
                    return new OO.ui.Process(function() {
                        self.close({ action: action });
                    });
                }
                
                // Fallback for other actions
                return FalsePositiveDialog.super.prototype.getActionProcess.call(this, action);
            };

            // Define dialog setup process - this is the proper way to add content
            FalsePositiveDialog.prototype.getSetupProcess = function() {
                return FalsePositiveDialog.super.prototype.getSetupProcess.call(this).next(function() {
                    // Add content to the body
                    this.$body.append(instructionsLayout.$element, layout.$element);
                }, this);
            };

            // This section is now handled in the static properties

            // Add dialog to window manager
            const windowManager = new OO.ui.WindowManager();
            this.$('body').append(windowManager.$element);
            windowManager.addWindows([messageDialog]);
            
            // Open dialog
            windowManager.openWindow(messageDialog).closed.then(data => {
                if (data && data.action === 'submit') {
                    this.reportFalsePositive(data.selectedWord, data.wordCombination);
                }
            });
        });
    }

    /**
     * Report false positive by redirecting to the talk page with the word combination
     * @param {string} selectedWord - The promotional word selected
     * @param {string} wordCombination - The non-promotional word combination
     */
    reportFalsePositive(selectedWord, wordCombination) {
        const currentPage = this.mw.config.get('wgPageName');
        
        // Use exactly the same URL structure as DuplicateReferences
        const baseUrl = 'https://en.wikipedia.org/wiki/User_talk:Polygnotus';
        const action = 'edit';
        const section = 'new';
        const preloadtitle = 'Reporting%20%5B%5BUser%3APolygnotus%2FDetectPromo%7CDetectPromo%5D%5D%20false-positive';
        
        // Note the $1 placeholder in the preload parameter - this is key!
        const preload = 'User:Polygnotus/$1';
        
        // Format the preloadparams with the word combination
        const preloadparams = encodeURIComponent(`[[${currentPage}]] "${selectedWord}" in phrase "${wordCombination}" is not promotional [[User:Polygnotus|Polygnotus]] ([[User talk:Polygnotus|talk]]) 09:37, 5 April 2025 (UTC)`);
        
        // Construct the final URL to match the example format exactly
        const reportURL = `${baseUrl}?action=${action}&section=${section}&preloadtitle=${preloadtitle}&preload=${preload}&preloadparams%5b%5d=${preloadparams}`;
        
        // Redirect to the report page
        window.location.href = reportURL;
    }

    /**
     * @param {string} string
     * @return {string} escapedString
     */
    escapeRegEx(string) {
        return string.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
    }
}

$(() => {
    mw.loader.using(['mediawiki.api', 'oojs-ui-core', 'oojs-ui-windows', 'oojs-ui-widgets']).then(() => {
        new DetectPromo(mw, $).execute();
    });
});