mcp-scraper 0.32.0 → 0.32.1

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

@@ -7,7 +7,7 @@ import {
   registerMemoryMcpTools,
   registerPaaExtractorMcpTools,
   registerSerpIntelligenceCaptureTools
-} from "../chunk-YOOEAXWP.js";
+} from "../chunk-2FY4VBAY.js";
 import "../chunk-R7EETU7Z.js";
 import "../chunk-MTSBI7ZH.js";
 import {
@@ -16,7 +16,7 @@ import {
 import "../chunk-PTHY7NYD.js";
 import {
   PACKAGE_VERSION
-} from "../chunk-CPQF3PYW.js";
+} from "../chunk-VJBS6RPB.js";
 import "../chunk-4767BA2O.js";
 import "../chunk-M2S27J6Z.js";
 import "../chunk-ICT7DDHL.js";

package/dist/{chunk-YOOEAXWP.js → chunk-2FY4VBAY.js}

@@ -19,7 +19,7 @@ import {
 } from "./chunk-PTHY7NYD.js";
 import {
   PACKAGE_VERSION
-} from "./chunk-CPQF3PYW.js";
+} from "./chunk-VJBS6RPB.js";
 import {
   MC_PER_CREDIT
 } from "./chunk-4767BA2O.js";
@@ -10110,4 +10110,4 @@ export {
   MEMORY_TOOL_SCHEMAS,
   registerMemoryMcpTools
 };
-//# sourceMappingURL=chunk-YOOEAXWP.js.map
+//# sourceMappingURL=chunk-2FY4VBAY.js.map

package/dist/chunk-VJBS6RPB.js

@@ -0,0 +1,7 @@
+// src/version.ts
+var PACKAGE_VERSION = "0.32.1";
+
+export {
+  PACKAGE_VERSION
+};
+//# sourceMappingURL=chunk-VJBS6RPB.js.map

package/dist/chunk-VJBS6RPB.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/version.ts"],"sourcesContent":["export const PACKAGE_VERSION = '0.32.1'\n"],"mappings":";AAAO,IAAM,kBAAkB;","names":[]}

package/dist/{server-OBOTNFD6.js → server-GXTYKIAX.js}

@@ -36,7 +36,7 @@ import {
   sanitizeAttempts,
   sanitizeHarvestResult,
   transcribeMediaUrl
-} from "./chunk-YOOEAXWP.js";
+} from "./chunk-2FY4VBAY.js";
 import {
   auditImages,
   buildLinkReport,
@@ -77,7 +77,7 @@ import {
   RawMapsOverviewSchema,
   RawMapsReviewStatsSchema
 } from "./chunk-PTHY7NYD.js";
-import "./chunk-CPQF3PYW.js";
+import "./chunk-VJBS6RPB.js";
 import {
   completeExtractJob,
   countSuccessfulPages,
@@ -13205,29 +13205,38 @@ var MapsSearchExtractor = class {
       };
       const phonePattern = /(?:\+?1[\s.-]?)?\(?\d{3}\)?[\s.-]\d{3}[\s.-]\d{4}/;
       const addressPattern = /\b[A-Z]{2}\s+\d{5}(?:-\d{4})?\b|\b(?:St|Street|Ave|Avenue|Rd|Road|Blvd|Boulevard|Dr|Drive|Ln|Lane|Way|Pkwy|Parkway)\b/i;
-      const cards = Array.from(document.querySelectorAll("button")).filter((button) => {
-        const name = normalize(button.querySelector("h3")?.textContent);
-        return Boolean(name) && !button.closest('[role="dialog"], dialog') && !/\bsponsored\b/i.test(button.innerText);
+      const headings = Array.from(document.querySelectorAll('div[role="heading"]')).filter((heading) => {
+        const text2 = normalize(heading.textContent);
+        return Boolean(text2) && !/^(?:businesses|places|sponsored|people also ask|discussions and forums|more businesses|more places|custom date range|choose what you)/i.test(text2 ?? "");
       });
       const seen = /* @__PURE__ */ new Set();
       const out = [];
-      for (const [cardIndex, card] of cards.entries()) {
-        const name = normalize(card.querySelector("h3")?.textContent);
+      let cardIndex = -1;
+      for (const heading of headings) {
+        let card = heading;
+        for (let hop = 0; hop < 8; hop++) {
+          const parent = card.parentElement;
+          if (!parent || parent.querySelectorAll('div[role="heading"]').length > 1) break;
+          card = parent;
+        }
+        if (card.closest('[role="dialog"], dialog')) continue;
+        const name = normalize(heading.textContent);
         if (!name) continue;
-        const text2 = normalize(card.innerText) ?? name;
-        const key = `${name.toLowerCase()}|${text2.toLowerCase()}`;
+        const lines = (card.innerText ?? "").split("\n").map((line) => normalize(line)).filter((line) => Boolean(line));
+        if (/^sponsored$/i.test(lines[0] ?? "")) continue;
+        const ratingLine = lines.find((line) => /\d(?:\.\d)?\s*\(/.test(line)) ?? "";
+        const rating = ratingLine.match(/(\d(?:\.\d)?)\s*\(/)?.[1] ?? null;
+        const reviewCount = ratingLine.match(/\((\d[\d,]*)\)/)?.[1]?.replace(/,/g, "") ?? null;
+        const category = normalize(ratingLine.split("\xB7").slice(1).join("\xB7"));
+        const phone = lines.map((line) => line.match(phonePattern)?.[0]).find((value) => Boolean(value)) ?? null;
+        const hoursStatus = normalize((lines.find((line) => /\b(?:open|closed|opens|closes)\b/i.test(line)) ?? "").split("\xB7")[0] ?? "");
+        const address = lines.find((line) => addressPattern.test(line)) ?? null;
+        const websiteUrl = Array.from(card.querySelectorAll("a[href]")).map((anchor) => anchor.href).find((href) => /^https?:/i.test(href) && !/google\.|gstatic|googleusercontent/.test(href)) ?? null;
+        if (!rating && !(phone && websiteUrl)) continue;
+        const key = `${name.toLowerCase()}|${lines.join(" ").toLowerCase()}`;
         if (seen.has(key)) continue;
         seen.add(key);
-        const lines = text2.split(/\n+/).map((line) => normalize(line)).filter((line) => Boolean(line));
-        const aria = normalize(card.getAttribute("aria-label"));
-        const rating = (aria ?? text2).match(/(?:rated\s+)?(\d(?:\.\d)?)\s*(?:out of 5)?/i)?.[1] ?? null;
-        const reviewCount = (aria ?? text2).match(/(?:\(|·\s*)([\d,]+)\s+(?:user\s+)?reviews?\b/i)?.[1]?.replace(/,/g, "") ?? null;
-        const phone = lines.find((line) => phonePattern.test(line)) ?? null;
-        const address = lines.find((line) => addressPattern.test(line)) ?? null;
-        const hoursStatus = lines.find((line) => /\b(?:open|closed|opens|closes)\b/i.test(line)) ?? null;
-        const category = lines.find(
-          (line) => line !== name && line !== phone && line !== address && line !== hoursStatus && !/^(?:rated|\d[\d.,]*\s*(?:out of 5|reviews?)?)\b/i.test(line) && !/sponsored|website|directions|call|save|share|more/i.test(line)
-        ) ?? null;
+        cardIndex += 1;
         const mapsSearchUrl = `https://www.google.com/maps/search/${encodeURIComponent([name, locationHint].filter(Boolean).join(" "))}`;
         out.push({
           cardIndex,
@@ -13242,7 +13251,7 @@ var MapsSearchExtractor = class {
           address,
           phone,
           hoursStatus,
-          websiteUrl: null,
+          websiteUrl,
           directionsUrl: `https://www.google.com/maps/dir/?api=1&destination=${encodeURIComponent([name, address].filter(Boolean).join(", "))}`,
           metadata: lines.slice(0, 20)
         });
@@ -13251,17 +13260,27 @@ var MapsSearchExtractor = class {
     }, fallbackLocation);
   }
   async openCardAndExtractDialog(page, card, includeServices) {
+    if (!includeServices) return null;
     const clicked = await page.evaluate((input) => {
       const normalize = (value) => (value ?? "").replace(/\s+/g, " ").trim().toLowerCase();
-      const cards = Array.from(document.querySelectorAll("button")).filter(
-        (button) => Boolean(button.querySelector("h3")) && !button.closest('[role="dialog"], dialog') && !/\bsponsored\b/i.test(button.innerText)
-      );
-      const exact = cards.find((button) => `${normalize(button.querySelector("h3")?.textContent)}|${normalize(button.innerText)}` === input.cardKey);
-      const target = exact ?? cards[input.cardIndex];
-      if (!target) return false;
-      target.click();
-      return true;
-    }, { cardIndex: card.cardIndex, cardKey: card.cardKey });
+      const headings = Array.from(document.querySelectorAll('div[role="heading"]'));
+      for (const heading of headings) {
+        let target = heading;
+        for (let hop = 0; hop < 8; hop++) {
+          const parent = target.parentElement;
+          if (!parent || parent.querySelectorAll('div[role="heading"]').length > 1) break;
+          target = parent;
+        }
+        if (target.closest('[role="dialog"], dialog')) continue;
+        const name = normalize(heading.textContent);
+        if (!name) continue;
+        const lines = (target.innerText ?? "").split("\n").map((line) => normalize(line)).filter(Boolean);
+        if (`${name}|${lines.join(" ")}` !== input.cardKey) continue;
+        target.click();
+        return true;
+      }
+      return false;
+    }, { cardKey: card.cardKey });
     if (!clicked) return null;
     const dialog = page.locator('dialog, [role="dialog"]').filter({ has: page.locator('h1, h2, [role="heading"]') }).last();
     await dialog.waitFor({ state: "visible", timeout: 6e3 }).catch(() => {
@@ -27608,4 +27627,4 @@ app.get("/blog/:slug/", (c) => {
 export {
   app
 };
-//# sourceMappingURL=server-OBOTNFD6.js.map
+//# sourceMappingURL=server-GXTYKIAX.js.map