Store Scraper MCP

Instructions

Implementation Reference

• src/server.js:656-696 (handler)

  The main handler function for the 'gp_reviews' tool. It validates input, builds the reviews URL, fetches the HTML, parses reviews using parseGPReviews, and returns the result as JSON.
  async function handleGPReviews(args) { try { const { appId, country = 'us', lang = 'en', page = 0, sort = 0, } = args; if (!appId) { throw new Error('appId is required'); } const url = buildGPReviewsUrl({ appId, country, lang, page, sort }); const html = await fetchText(url); const result = parseGPReviews(html); return { content: [ { type: 'text', text: JSON.stringify({ page, ...result, }, null, 2), }, ], }; } catch (error) { return { content: [ { type: 'text', text: JSON.stringify({ error: error.message }, null, 2), }, ], isError: true, }; } }
• src/server.js:1269-1302 (registration)

  Registration of the 'gp_reviews' tool in the MCP tools list, including name, description, and input schema definition.
  { name: 'gp_reviews', description: '[Google Play] Get app reviews with pagination', inputSchema: { type: 'object', properties: { appId: { type: 'string', description: 'Google Play app ID', }, country: { type: 'string', description: 'Two-letter country code (default: us)', default: 'us', }, lang: { type: 'string', description: 'Language code (default: en)', default: 'en', }, page: { type: 'number', description: 'Page number (default: 0)', default: 0, }, sort: { type: 'number', description: 'Sort order: 0 = most recent, 2 = most helpful (default: 0)', default: 0, }, }, required: ['appId'], }, },
• src/endpoints/googlePlay.js:80-104 (helper)

  Helper function buildReviewsUrl (imported as buildGPReviewsUrl) that constructs the Google Play reviews page URL based on appId, country, lang, page, and sort parameters.
  export function buildReviewsUrl(params) { const { appId, lang = 'en', country = 'us', page = 0, sort = 0, // 0 = most recent, 2 = most helpful } = params; if (!appId) { throw new Error('appId is required'); } // Google Play uses pagination tokens, but we can use page numbers as approximation const queryParams = new URLSearchParams({ id: appId, gl: country, hl: lang, reviewSortOrder: sort.toString(), reviewType: '0', // All reviews pageNum: page.toString(), }); return `${GOOGLE_PLAY_BASE}/store/apps/details?id=${appId}&gl=${country}&hl=${lang}#Reviews`; }
• src/parsers/googlePlay/reviews.js:11-224 (helper)

  Core parsing helper parseReviews (imported as parseGPReviews) that extracts review data from Google Play HTML using multiple fallback strategies including JSON-LD, script JSON, and DOM patterns.
  export function parseReviews(html) { if (!html || typeof html !== 'string') { return { data: [], nextPaginationToken: null, }; } const reviews = []; const seenReviewIds = new Set(); try { // Strategy 1: Look for JSON-LD structured data const jsonLdMatches = html.matchAll(/<script[^>]*type=["']application\/ld\+json["'][^>]*>([\s\S]*?)<\/script>/gi); for (const match of jsonLdMatches) { try { const jsonLd = JSON.parse(match[1]); if (jsonLd['@type'] === 'Review' || (jsonLd['@type'] === 'ItemList' && jsonLd.itemListElement)) { const reviewList = jsonLd['@type'] === 'Review' ? [jsonLd] : jsonLd.itemListElement; reviewList.forEach(item => { if (item['@type'] === 'Review' || item.reviewBody) { const review = extractFromJsonLd(item); if (review && !seenReviewIds.has(review.reviewId || review.text)) { seenReviewIds.add(review.reviewId || review.text); reviews.push(review); } } }); } } catch (e) { // Not valid JSON-LD, continue } } // Strategy 2: Look for embedded JSON data in script tags (Google Play uses _df_ prefix) const scriptMatches = html.matchAll(/<script[^>]*>([\s\S]*?)<\/script>/gi); for (const match of scriptMatches) { const scriptContent = match[1]; // Look for review data in various JSON formats if (scriptContent.includes('review') || scriptContent.includes('rating') || scriptContent.includes('_df_')) { // Try to find JSON objects/arrays const jsonPatterns = [ /\[([\s\S]{100,}?)\]/g, // Array of reviews /\{[\s\S]*?"reviews"[\s\S]*?:[\s\S]*?\[([\s\S]*?)\][\s\S]*?\}/g, // Object with reviews array /\{[\s\S]*?"data"[\s\S]*?:[\s\S]*?\[([\s\S]*?)\][\s\S]*?\}/g, // Object with data array ]; for (const pattern of jsonPatterns) { const matches = scriptContent.matchAll(pattern); for (const jsonMatch of matches) { try { const jsonStr = '[' + jsonMatch[1] + ']'; const jsonData = JSON.parse(jsonStr); if (Array.isArray(jsonData)) { jsonData.forEach(item => { if (item && (item.reviewId || item.text || item.rating || item.comment)) { const review = normalizeReview(item); if (review && !seenReviewIds.has(review.reviewId || review.text)) { seenReviewIds.add(review.reviewId || review.text); reviews.push(review); } } }); } } catch (e) { // Not valid JSON, try next pattern } } } // Try direct JSON parsing if content looks like JSON if (scriptContent.trim().startsWith('{') || scriptContent.trim().startsWith('[')) { try { const jsonData = JSON.parse(scriptContent); if (Array.isArray(jsonData)) { jsonData.forEach(item => { if (item && (item.reviewId || item.text || item.rating)) { const review = normalizeReview(item); if (review && !seenReviewIds.has(review.reviewId || review.text)) { seenReviewIds.add(review.reviewId || review.text); reviews.push(review); } } }); } else if (jsonData.reviews || jsonData.data || jsonData[0]) { const reviewList = jsonData.reviews || jsonData.data || (Array.isArray(jsonData) ? jsonData : []); reviewList.forEach(item => { const review = normalizeReview(item); if (review && !seenReviewIds.has(review.reviewId || review.text)) { seenReviewIds.add(review.reviewId || review.text); reviews.push(review); } }); } } catch (e) { // Not JSON, continue } } } } // Strategy 3: Extract from visible HTML structure with improved patterns const reviewPatterns = [ /<div[^>]*class=["'][^"']*review["'][^>]*>([\s\S]*?)<\/div>/gi, /<div[^>]*itemprop=["']review["'][^>]*>([\s\S]*?)<\/div>/gi, /<article[^>]*class=["'][^"']*review["'][^>]*>([\s\S]*?)<\/article>/gi, /<div[^>]*data-review-id=["'][^"']*["'][^>]*>([\s\S]*?)<\/div>/gi, ]; for (const pattern of reviewPatterns) { const reviewBlockMatches = html.matchAll(pattern); for (const blockMatch of reviewBlockMatches) { const reviewHtml = blockMatch[1]; // Extract rating with multiple patterns const ratingMatch = reviewHtml.match(/aria-label=["'](\d+)\s*stars?["']/i) || reviewHtml.match(/<div[^>]*class=["'][^"']*rating["'][^>]*>(\d+)[^<]*<\/div>/i) || reviewHtml.match(/ratingValue["']:\s*["']?(\d+)/i) || reviewHtml.match(/<meta[^>]*itemprop=["']ratingValue["'][^>]*content=["'](\d+)["']/i) || reviewHtml.match(/<span[^>]*class=["'][^"']*star-rating["'][^>]*>(\d+)/i); const rating = ratingMatch ? parseInt(ratingMatch[1], 10) : null; // Extract text with multiple patterns const textMatch = reviewHtml.match(/<span[^>]*class=["'][^"']*review-body["'][^>]*>([\s\S]*?)<\/span>/i) || reviewHtml.match(/<div[^>]*class=["'][^"']*review-text["'][^>]*>([\s\S]*?)<\/div>/i) || reviewHtml.match(/<p[^>]*class=["'][^"']*review-text["'][^>]*>([\s\S]*?)<\/p>/i) || reviewHtml.match(/reviewBody["']:\s*["']([^"']+)["']/i) || reviewHtml.match(/<span[^>]*itemprop=["']reviewBody["'][^>]*>([\s\S]*?)<\/span>/i); let text = textMatch ? textMatch[1].replace(/<[^>]+>/g, '').trim() : null; if (text) { text = text.replace(/&quot;/g, '"').replace(/&amp;/g, '&').replace(/&lt;/g, '<').replace(/&gt;/g, '>'); } // Extract author with multiple patterns const authorMatch = reviewHtml.match(/<span[^>]*class=["'][^"']*author-name["'][^>]*>([^<]+)<\/span>/i) || reviewHtml.match(/<a[^>]*class=["'][^"']*author["'][^>]*>([^<]+)<\/a>/i) || reviewHtml.match(/<span[^>]*itemprop=["']author["'][^>]*>([^<]+)<\/span>/i) || reviewHtml.match(/author["']:\s*["']([^"']+)["']/i); const author = authorMatch ? authorMatch[1].trim() : null; // Extract date with multiple patterns const dateMatch = reviewHtml.match(/<span[^>]*class=["'][^"']*review-date["'][^>]*>([^<]+)<\/span>/i) || reviewHtml.match(/<time[^>]*datetime=["']([^"']+)["']/i) || reviewHtml.match(/<span[^>]*itemprop=["']datePublished["'][^>]*>([^<]+)<\/span>/i) || reviewHtml.match(/datePublished["']:\s*["']([^"']+)["']/i); const date = dateMatch ? dateMatch[1].trim() : null; // Extract review ID const reviewIdMatch = reviewHtml.match(/data-review-id=["']([^"']+)["']/i) || reviewHtml.match(/reviewId["']:\s*["']([^"']+)["']/i); const reviewId = reviewIdMatch ? reviewIdMatch[1] : null; // Extract thumbs up const thumbsUpMatch = reviewHtml.match(/(\d+)\s*(?:thumbs?|helpful|útil)/i) || reviewHtml.match(/thumbsUp["']:\s*["']?(\d+)/i); const thumbsUp = thumbsUpMatch ? parseInt(thumbsUpMatch[1], 10) : 0; // Only add if we have meaningful data if (rating || text || author) { const reviewKey = reviewId || text || `${author}-${date}`; if (!seenReviewIds.has(reviewKey)) { seenReviewIds.add(reviewKey); reviews.push({ reviewId: reviewId, userName: author || 'Anonymous', userImage: null, date: date, dateText: date, score: rating || 0, scoreText: rating ? rating.toString() : '0', title: null, text: text, replyDate: null, replyText: null, version: null, thumbsUp: thumbsUp, criterias: [], }); } } } } // Strategy 4: Look for pagination token const paginationPatterns = [ /nextPaginationToken["']:\s*["']([^"']+)["']/i, /"paginationToken"["']:\s*["']([^"']+)["']/i, /data-pagination-token=["']([^"']+)["']/i, ]; let nextToken = null; for (const pattern of paginationPatterns) { const match = html.match(pattern); if (match) { nextToken = match[1]; break; } } return { data: reviews, nextPaginationToken: nextToken, }; } catch (error) { console.error('Error parsing Google Play reviews:', error); return { data: [], nextPaginationToken: null, }; } }
• src/server.js:1472-1473 (registration)

  Dispatcher switch case that routes 'gp_reviews' tool calls to the handleGPReviews handler.
  case 'gp_reviews': return await handleGPReviews(args);