mcp-scraper 0.4.2 → 0.4.3

package/dist/bin/mcp-stdio-server.js

@@ -7,7 +7,7 @@ import {
   registerMemoryMcpTools,
   registerPaaExtractorMcpTools,
   registerSerpIntelligenceCaptureTools
-} from "../chunk-GFY6XHZS.js";
+} from "../chunk-BAMVP57P.js";
 import "../chunk-GWRIO6JT.js";
 import "../chunk-HPV4VOQX.js";
 import {
@@ -16,7 +16,7 @@ import {
 import "../chunk-XGIPATLV.js";
 import {
   PACKAGE_VERSION
-} from "../chunk-7HICT4GY.js";
+} from "../chunk-SRMGSCC3.js";
 import "../chunk-M2S27J6Z.js";
 
 // bin/mcp-stdio-server.ts

package/dist/{chunk-GFY6XHZS.js → chunk-BAMVP57P.js}

@@ -17,7 +17,7 @@ import {
 } from "./chunk-XGIPATLV.js";
 import {
   PACKAGE_VERSION
-} from "./chunk-7HICT4GY.js";
+} from "./chunk-SRMGSCC3.js";
 import {
   sanitizeVendorName
 } from "./chunk-M2S27J6Z.js";
@@ -2037,6 +2037,51 @@ ${rows}`,
     }
   };
 }
+function formatTrustpilotReviews(raw, input) {
+  const parsed = parseData(raw);
+  if ("error" in parsed) return { content: [{ type: "text", text: parsed.error }], isError: true };
+  const d = parsed.data;
+  const domain = d.domain ?? input.domain;
+  const reviewUrl = d.reviewUrl;
+  const extractedAt = d.extractedAt;
+  const requestedMaxPages = d.requestedMaxPages ?? input.maxPages ?? 5;
+  const pagesFetched = d.pagesFetched;
+  const reviews = d.reviews ?? [];
+  const durationMs = d.durationMs;
+  const rows = reviews.map((r) => {
+    const stars = r.rating != null ? "\u2605".repeat(r.rating) + "\u2606".repeat(r.ratingScale - r.rating) : "\u2014";
+    const answer = r.body[0]?.answer ?? "";
+    return `| ${cell(r.reviewer.name ?? "Anonymous")} | ${stars} | ${cell(r.title)} | ${cell(r.date ? r.date.slice(0, 10) : null)} | ${r.flags.origin ?? "\u2014"} | ${r.flags.companyReplied ? "yes" : "no"} | ${cell(truncate(answer, 100))} |`;
+  }).join("\n");
+  const full = [
+    `# Trustpilot Reviews: ${domain}`,
+    `**Collected:** ${reviews.length} review${reviews.length === 1 ? "" : "s"} across ${pagesFetched ?? 0}/${requestedMaxPages} page${requestedMaxPages === 1 ? "" : "s"}`,
+    reviewUrl ? `**Source:** ${reviewUrl}` : null,
+    `
+## Reviews
+| Reviewer | Rating | Title | Date | Origin | Company Replied | Excerpt |
+|----------|--------|-------|------|--------|------------------|---------|
+${rows}`,
+    `
+---
+\u{1F4A1} **Note:** sampling tool, default 5 pages (~100 reviews), max 50. For full-corpus export use Trustpilot's official Business API.`,
+    durationMs != null ? `
+*Extracted in ${(durationMs / 1e3).toFixed(1)}s*` : null
+  ].filter(Boolean).join("\n");
+  return {
+    ...oneBlock(full),
+    structuredContent: {
+      domain,
+      reviewUrl: reviewUrl ?? null,
+      extractedAt: extractedAt ?? null,
+      requestedMaxPages,
+      pagesFetched: pagesFetched ?? 0,
+      reviewCount: reviews.length,
+      reviews,
+      durationMs: durationMs ?? 0
+    }
+  };
+}
 async function formatDirectoryWorkflow(raw, input, ctx) {
   const parsed = parseData(raw);
   if ("error" in parsed) return { content: [{ type: "text", text: parsed.error }], isError: true };
@@ -2788,6 +2833,34 @@ var MapsPlaceIntelInputSchema = {
   maxReviews: z.number().int().min(1).max(500).default(50).describe("Max review cards when includeReviews is true. Default 50, maximum 500."),
   includeServices: z.boolean().default(false).describe("Fetch the business's configured services list and areas-served list, when the profile has them. Adds one extra page visit; not present for every business.")
 };
+var TrustpilotReviewsInputSchema = {
+  domain: z.string().min(1).describe(`The business's domain as it appears in its Trustpilot URL, e.g. "www.bhphotovideo.com" (include the www. if the site uses it \u2014 pass the domain as-is, do not guess).`),
+  maxPages: z.number().int().min(1).max(50).default(5).describe("Review pages to fetch (~20 reviews per page). Default 5 (~100 reviews). Maximum 50 \u2014 large companies can have 1,000+ pages; this tool is for sampling, not full-corpus export.")
+};
+var ReviewCardSchema = z.object({
+  source: z.enum(["trustpilot", "g2"]),
+  sourceReviewId: z.string().nullable(),
+  reviewUrl: z.string().nullable(),
+  reviewer: z.object({
+    name: z.string().nullable(),
+    profileId: z.string().nullable(),
+    title: z.string().nullable(),
+    companySegment: z.string().nullable()
+  }),
+  rating: z.number().nullable(),
+  ratingScale: z.number(),
+  title: z.string().nullable(),
+  date: z.string().nullable(),
+  body: z.array(z.object({ question: z.string().nullable(), answer: z.string() })),
+  truncated: z.boolean(),
+  flags: z.object({
+    origin: z.string().nullable(),
+    incentivized: z.boolean().nullable(),
+    validated: z.boolean().nullable(),
+    currentUser: z.boolean().nullable(),
+    companyReplied: z.boolean().nullable()
+  })
+});
 var MapsSearchInputSchema = {
   query: z.string().min(1).describe('Business category, niche, or search term, e.g. "roofers". Do not include location here \u2014 use location instead.'),
   location: z.string().optional().describe('City, region, country, or service area, e.g. "Denver, CO".'),
@@ -3104,6 +3177,16 @@ var MapsPlaceIntelOutputSchema = {
   areasServed: z.array(z.string()),
   servicesStatus: z.string()
 };
+var TrustpilotReviewsOutputSchema = {
+  domain: z.string(),
+  reviewUrl: z.string(),
+  extractedAt: z.string(),
+  requestedMaxPages: z.number().int(),
+  pagesFetched: z.number().int(),
+  reviewCount: z.number().int(),
+  reviews: z.array(ReviewCardSchema),
+  durationMs: z.number()
+};
 var CreditsInfoOutputSchema = {
   balanceCredits: z.number().nullable(),
   matchedCost: z.object({
@@ -4163,6 +4246,13 @@ function registerPaaExtractorMcpTools(server, executor, options = {}) {
     outputSchema: recordOutputSchema("maps_search", MapsSearchOutputSchema),
     annotations: liveWebToolAnnotations("Google Maps Business Search")
   }, async (input) => formatMapsSearch(await executor.mapsSearch(input), input));
+  server.registerTool("trustpilot_reviews", {
+    title: "Trustpilot Review Harvest",
+    description: "Extract customer reviews for a business from Trustpilot \u2014 reviewer, rating, title, body, date, invited/organic origin, company-reply flag. Sampling tool, not a full-corpus export: default 5 pages (~100 reviews), max 50 pages. For bulk/complete extraction across thousands of pages, use Trustpilot's official Business API instead.",
+    inputSchema: TrustpilotReviewsInputSchema,
+    outputSchema: recordOutputSchema("trustpilot_reviews", TrustpilotReviewsOutputSchema),
+    annotations: liveWebToolAnnotations("Trustpilot Review Harvest")
+  }, async (input) => formatTrustpilotReviews(await executor.trustpilotReviews(input), input));
   server.registerTool("directory_workflow", {
     title: "Directory Workflow: Markets + Maps",
     description: `Build directory/prospecting datasets: selects US city markets from Census population data, optionally joins configured ZIP groups, then runs Google Maps business searches per city in parallel. Use for "all cities over 100k population in a state" or market+Maps workflows. ${fileBehavior("Saves a CSV of results per city.", "Large results are stored as a retrievable artifact \u2014 you get an inline summary plus an artifactId for report_artifact_read.")}`,
@@ -4494,6 +4584,9 @@ var HttpMcpToolExecutor = class {
   mapsSearch(input) {
     return this.call("/maps/search", input);
   }
+  trustpilotReviews(input) {
+    return this.call("/trustpilot/reviews", input, this.httpTimeoutOverrideMs ?? 3e5);
+  }
   directoryWorkflow(input) {
     const cityCount = typeof input.maxCities === "number" ? input.maxCities : 25;
     const concurrency = typeof input.concurrency === "number" && input.concurrency > 0 ? input.concurrency : 5;
@@ -8096,4 +8189,4 @@ export {
   registerMemoryMcpTools,
   MemoryMcpToolExecutor
 };
-//# sourceMappingURL=chunk-GFY6XHZS.js.map
+//# sourceMappingURL=chunk-BAMVP57P.js.map