mcp-scraper 0.26.2 → 0.27.0

package/dist/bin/api-server.cjs

@@ -17330,7 +17330,7 @@ var init_schemas3 = __esm({
     "use strict";
     import_zod17 = require("zod");
     DEFAULT_PROXY_MODE = "none";
-    DEFAULT_MAPS_PROXY_MODE = "location";
+    DEFAULT_MAPS_PROXY_MODE = "none";
     HarvestOptionsSchema = import_zod17.z.object({
       query: import_zod17.z.string().min(1),
       location: import_zod17.z.string().optional(),
@@ -17372,6 +17372,7 @@ var init_schemas3 = __esm({
       gl: import_zod17.z.string().length(2).default("us"),
       hl: import_zod17.z.string().length(2).default("en"),
       maxResults: import_zod17.z.number().int().min(1).max(50).default(10),
+      includeServices: import_zod17.z.boolean().default(false),
       proxyMode: import_zod17.z.enum(["location", "configured", "none"]).default(DEFAULT_MAPS_PROXY_MODE),
       proxyZip: import_zod17.z.string().regex(/^\d{5}$/).optional(),
       debug: import_zod17.z.boolean().default(false),
@@ -19227,12 +19228,277 @@ var init_video_routes = __esm({
   }
 });
 
+// src/extractor/MapsSearchExtractor.ts
+function buildGoogleLocalResultsUrl(input) {
+  const query = [input.query, input.location].filter(Boolean).join(" ");
+  const params = new URLSearchParams({ q: query, udm: "1", gl: input.gl, hl: input.hl, pws: "0" });
+  if (input.location) params.set("uule", encodeUule(normalizeLocation(input.location)));
+  return `https://www.google.com/search?${params.toString()}`;
+}
+var LOCAL_RESULTS_PAGE_LIMIT, LOCAL_RESULTS_WAIT_MS, MapsSearchExtractor;
+var init_MapsSearchExtractor = __esm({
+  "src/extractor/MapsSearchExtractor.ts"() {
+    "use strict";
+    init_errors();
+    init_uule();
+    LOCAL_RESULTS_PAGE_LIMIT = 8;
+    LOCAL_RESULTS_WAIT_MS = 1e3;
+    MapsSearchExtractor = class {
+      constructor(driver) {
+        this.driver = driver;
+      }
+      driver;
+      async extract(options) {
+        const startMs = Date.now();
+        const searchQuery = [options.query, options.location].filter(Boolean).join(" ");
+        const searchUrl = buildGoogleLocalResultsUrl(options);
+        const config = {
+          headless: options.headless,
+          kernelApiKey: options.kernelApiKey,
+          kernelProxyId: options.kernelProxyId,
+          kernelProxyResolution: options.kernelProxyResolution,
+          proxyMode: options.proxyMode,
+          viewport: { width: 1280, height: 900 },
+          locale: `${options.hl}-${options.gl.toUpperCase()}`,
+          debug: options.debug
+        };
+        try {
+          await this.driver.launch(config);
+          const page = this.driver.getPage();
+          await page.goto(searchUrl, { waitUntil: "domcontentloaded", timeout: 6e4 });
+          await page.waitForTimeout(LOCAL_RESULTS_WAIT_MS);
+          if (await this.detectBlock(page)) throw new CaptchaError(RECAPTCHA_INSTRUCTIONS);
+          const results = await this.collectResults(page, options);
+          return {
+            query: options.query,
+            location: options.location ?? null,
+            searchQuery,
+            searchUrl,
+            extractedAt: (/* @__PURE__ */ new Date()).toISOString(),
+            requestedMaxResults: options.maxResults,
+            resultCount: results.length,
+            results,
+            durationMs: Date.now() - startMs
+          };
+        } finally {
+          await this.driver.close();
+        }
+      }
+      async detectBlock(page) {
+        return page.evaluate(() => {
+          const text2 = document.body.innerText.slice(0, 2e3);
+          return /unusual traffic|captcha|recaptcha|about this page/i.test(text2) || /\/sorry\//.test(location.href);
+        });
+      }
+      async collectResults(page, options) {
+        const seen = /* @__PURE__ */ new Set();
+        const results = [];
+        let pageCount = 0;
+        while (results.length < options.maxResults && pageCount < LOCAL_RESULTS_PAGE_LIMIT) {
+          if (await this.detectBlock(page)) throw new CaptchaError(RECAPTCHA_INSTRUCTIONS);
+          const cards = await this.extractVisibleCards(page, options.location);
+          for (const card of cards) {
+            if (results.length >= options.maxResults || seen.has(card.cardKey)) continue;
+            seen.add(card.cardKey);
+            const details = await this.openCardAndExtractDialog(page, card, options.includeServices);
+            results.push({
+              position: results.length + 1,
+              name: card.name,
+              placeUrl: details?.placeUrl ?? card.placeUrl,
+              cid: details?.cid ?? card.cid,
+              cidDecimal: details?.cidDecimal ?? card.cidDecimal,
+              rating: details?.rating ?? card.rating,
+              reviewCount: details?.reviewCount ?? card.reviewCount,
+              category: details?.category ?? card.category,
+              address: details?.address ?? card.address,
+              phone: details?.phone ?? card.phone,
+              hoursStatus: details?.hoursStatus ?? card.hoursStatus,
+              websiteUrl: details?.websiteUrl ?? card.websiteUrl,
+              directionsUrl: card.directionsUrl,
+              metadata: card.metadata,
+              services: details?.services ?? [],
+              areasServed: details?.areasServed ?? [],
+              profileDetailsStatus: details ? "collected" : "unavailable"
+            });
+          }
+          if (results.length >= options.maxResults) break;
+          const advanced = await this.advanceToMoreResults(page);
+          if (!advanced) break;
+          pageCount += 1;
+        }
+        return results;
+      }
+      async extractVisibleCards(page, location2) {
+        const fallbackLocation = location2 ?? "";
+        return page.evaluate((locationHint) => {
+          const normalize4 = (value) => {
+            const text2 = value?.replace(/\s+/g, " ").trim() ?? "";
+            return text2 || null;
+          };
+          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 = normalize4(button.querySelector("h3")?.textContent);
+            return Boolean(name) && !button.closest('[role="dialog"], dialog') && !/\bsponsored\b/i.test(button.innerText);
+          });
+          const seen = /* @__PURE__ */ new Set();
+          const out = [];
+          for (const [cardIndex, card] of cards.entries()) {
+            const name = normalize4(card.querySelector("h3")?.textContent);
+            if (!name) continue;
+            const text2 = normalize4(card.innerText) ?? name;
+            const key = `${name.toLowerCase()}|${text2.toLowerCase()}`;
+            if (seen.has(key)) continue;
+            seen.add(key);
+            const lines = text2.split(/\n+/).map((line) => normalize4(line)).filter((line) => Boolean(line));
+            const aria = normalize4(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;
+            const mapsSearchUrl = `https://www.google.com/maps/search/${encodeURIComponent([name, locationHint].filter(Boolean).join(" "))}`;
+            out.push({
+              cardIndex,
+              cardKey: key,
+              name,
+              placeUrl: mapsSearchUrl,
+              cid: null,
+              cidDecimal: null,
+              rating,
+              reviewCount,
+              category,
+              address,
+              phone,
+              hoursStatus,
+              websiteUrl: null,
+              directionsUrl: `https://www.google.com/maps/dir/?api=1&destination=${encodeURIComponent([name, address].filter(Boolean).join(", "))}`,
+              metadata: lines.slice(0, 20)
+            });
+          }
+          return out;
+        }, fallbackLocation);
+      }
+      async openCardAndExtractDialog(page, card, includeServices) {
+        const clicked = await page.evaluate((input) => {
+          const normalize4 = (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) => `${normalize4(button.querySelector("h3")?.textContent)}|${normalize4(button.innerText)}` === input.cardKey);
+          const target = exact ?? cards[input.cardIndex];
+          if (!target) return false;
+          target.click();
+          return true;
+        }, { cardIndex: card.cardIndex, 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(() => {
+        });
+        if (await dialog.count() === 0 || !await dialog.isVisible().catch(() => false)) return null;
+        const details = await dialog.evaluate((root, includeServicesValue) => {
+          const text2 = (value) => value?.replace(/\s+/g, " ").trim() ?? "";
+          const phonePattern = /(?:\+?1[\s.-]?)?\(?\d{3}\)?[\s.-]\d{3}[\s.-]\d{4}/;
+          const buttons = Array.from(root.querySelectorAll("button"));
+          const buttonTexts = buttons.map((button) => text2(button.innerText)).filter(Boolean);
+          const servicesText = includeServicesValue ? buttonTexts.find((value) => /^services\s*:/i.test(value)) : null;
+          const services = servicesText ? [...new Set(servicesText.replace(/^services\s*:\s*/i, "").split(/\s*[,·]\s*/).map((value) => value.trim()).filter(Boolean))] : [];
+          const areaText = includeServicesValue ? buttonTexts.find((value) => /^serves\b/i.test(value)) : null;
+          const areasServed = areaText ? [areaText.replace(/^serves\s*/i, "").trim()].filter(Boolean) : [];
+          const externalLinks = Array.from(root.querySelectorAll("a[href]"));
+          const website = externalLinks.find((link) => /^website$/i.test(text2(link.innerText)) || /^website$/i.test(link.getAttribute("aria-label") ?? ""));
+          const mapLink = externalLinks.find((link) => /google\.[^/]+\/maps|\/maps\//i.test(link.href));
+          const addressLink = externalLinks.find((link) => /\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.test(text2(link.innerText)));
+          const allText = text2(root.innerText);
+          const rating = allText.match(/\b(?:rated\s+)?(\d(?:\.\d)?)\s+out of 5\b/i)?.[1] ?? allText.match(/\b(\d(?:\.\d)?)\s+rated\b/i)?.[1] ?? null;
+          const reviewCount = allText.match(/([\d,]+)\s+(?:user\s+)?reviews?\b/i)?.[1]?.replace(/,/g, "") ?? null;
+          const phoneButton = buttonTexts.find((value) => /^call\b/i.test(value) || phonePattern.test(value)) ?? "";
+          const phone = phoneButton.match(phonePattern)?.[0] ?? null;
+          const hoursStatus = buttonTexts.find((value) => /\b(?:open|closed|opens|closes)\b/i.test(value)) ?? null;
+          const placeUrl = mapLink?.href ?? addressLink?.href ?? location.href;
+          const fid = placeUrl.match(/!1s(0x[0-9a-f]+):(0x[0-9a-f]+)/i);
+          const cid = fid ? `${fid[1]}:${fid[2]}` : null;
+          let cidDecimal = null;
+          if (fid) {
+            try {
+              cidDecimal = BigInt(fid[2]).toString();
+            } catch {
+            }
+          }
+          if (!cidDecimal) {
+            const cidFromQuery = new URL(placeUrl).searchParams.get("cid");
+            cidDecimal = cidFromQuery && /^\d+$/.test(cidFromQuery) ? cidFromQuery : null;
+          }
+          return {
+            placeUrl,
+            cid,
+            cidDecimal,
+            rating,
+            reviewCount,
+            category: null,
+            address: text2(addressLink?.innerText) || null,
+            phone,
+            websiteUrl: website?.href ?? null,
+            hoursStatus,
+            services,
+            areasServed
+          };
+        }, includeServices);
+        await this.closeDialog(page);
+        return details;
+      }
+      async closeDialog(page) {
+        const closed = await page.evaluate(() => {
+          const dialogs = Array.from(document.querySelectorAll('dialog, [role="dialog"]'));
+          const dialog = dialogs.find((candidate) => candidate.offsetParent !== null);
+          const close = dialog && Array.from(dialog.querySelectorAll("button")).find(
+            (button) => /^close$/i.test(button.getAttribute("aria-label") ?? "") || /^close$/i.test(button.innerText.trim())
+          );
+          close?.click();
+          return Boolean(close);
+        });
+        if (closed) await page.waitForTimeout(250);
+      }
+      async advanceToMoreResults(page) {
+        const before = await page.evaluate(() => document.body.innerText.slice(-2e3));
+        const advanced = await page.evaluate(() => {
+          const visible = (element) => element.offsetParent !== null;
+          const mobile = Array.from(document.querySelectorAll("button, a")).find(
+            (element) => visible(element) && /^(more search results|more results)$/i.test(element.innerText.trim())
+          );
+          if (mobile) {
+            mobile.click();
+            return true;
+          }
+          const next = Array.from(document.querySelectorAll("a[href]")).find(
+            (link) => visible(link) && (/^next$/i.test(link.innerText.trim()) || /[?&]start=\d+/.test(link.href))
+          );
+          if (next) {
+            next.click();
+            return true;
+          }
+          return false;
+        });
+        if (!advanced) return false;
+        await page.waitForTimeout(LOCAL_RESULTS_WAIT_MS);
+        await page.waitForFunction((previous) => document.body.innerText.slice(-2e3) !== previous, before, { timeout: 5e3 }).catch(() => {
+        });
+        return true;
+      }
+    };
+  }
+});
+
 // src/extractor/MapsNavigator.ts
 var MapsNavigator;
 var init_MapsNavigator = __esm({
   "src/extractor/MapsNavigator.ts"() {
     "use strict";
     init_selectors();
+    init_MapsSearchExtractor();
     MapsNavigator = class {
       constructor(page) {
         this.page = page;
@@ -19290,32 +19556,26 @@ var init_MapsNavigator = __esm({
         await this.page.waitForSelector('[role="main"] h2', { timeout: 5e3 }).catch(() => {
         });
       }
-      async navigateToSerpBusinessPanel(businessName, location2) {
-        const query = `${businessName} ${location2}`;
-        const searchUrl = `https://www.google.com/search?q=${encodeURIComponent(query)}&udm=1`;
+      async navigateToSerpBusinessPanel(businessName, location2, gl, hl) {
+        const searchUrl = buildGoogleLocalResultsUrl({ query: businessName, location: location2, gl, hl });
         await this.page.goto(searchUrl, { waitUntil: "domcontentloaded", timeout: 45e3 }).catch(() => {
         });
-        await this.page.waitForTimeout(2e3);
+        await this.page.waitForTimeout(1e3);
         const clicked = await this.page.evaluate((name) => {
-          const trimmedName = name.trim();
-          const words = trimmedName.split(/\s+/);
-          const prefix = words.slice(0, Math.min(2, words.length)).join(" ");
-          const headings = Array.from(document.querySelectorAll('h3, div[role="heading"]'));
-          const heading = headings.find((el2) => el2.innerText?.trim() === trimmedName) ?? headings.find((el2) => el2.innerText?.trim().startsWith(prefix));
-          const clickable = heading?.closest('a, div[role="link"], div[jsaction]');
-          if (clickable) {
-            clickable.click();
-            return true;
-          }
-          if (heading) {
-            heading.click();
+          const normalized = name.replace(/\s+/g, " ").trim().toLowerCase();
+          const card = Array.from(document.querySelectorAll("button")).find(
+            (button) => button.querySelector("h3")?.textContent?.replace(/\s+/g, " ").trim().toLowerCase() === normalized
+          );
+          if (card) {
+            card.click();
             return true;
           }
           return false;
         }, businessName);
         if (!clicked) return false;
-        await this.page.waitForTimeout(2e3);
-        return true;
+        await this.page.waitForSelector('dialog, [role="dialog"]', { timeout: 5e3 }).catch(() => {
+        });
+        return this.page.locator('dialog, [role="dialog"]').count().then((count) => count > 0);
       }
       buildReviewsUrl(placeUrl) {
         const fid = placeUrl.match(/!1s(0x[0-9a-f]+:0x[0-9a-f]+)/i);
@@ -19508,7 +19768,7 @@ var init_MapsExtractor = __esm({
           let servicesStatus = "not_requested";
           if (options.includeServices) {
             try {
-              const opened = await navigator.navigateToSerpBusinessPanel(options.businessName, options.location);
+              const opened = await navigator.navigateToSerpBusinessPanel(options.businessName, options.location, options.gl, options.hl);
               if (opened) {
                 const result = await this.extractServicesAndAreasServed(page);
                 services = result.services;
@@ -19729,249 +19989,15 @@ var init_MapsExtractor = __esm({
         return result.data;
       }
       async extractServicesAndAreasServed(page) {
-        await page.waitForTimeout(1500);
-        let services = [];
-        if (await this.clickRowStartingWithRetry(page, "Services:")) {
-          await page.waitForSelector('[role="dialog"]', { timeout: 5e3 }).catch(() => {
-          });
-          await page.waitForTimeout(800);
-          services = await this.extractOpenDialogItems(page, "Services");
-          await this.closeOpenDialog(page, "Services");
-          await page.waitForTimeout(500);
-        }
-        let areasServed = [];
-        if (await this.clickRowStartingWithRetry(page, "Serves ")) {
-          await page.waitForSelector('[role="dialog"]', { timeout: 5e3 }).catch(() => {
-          });
-          await page.waitForTimeout(800);
-          areasServed = await this.extractOpenDialogItems(page, "Areas Served");
-          await this.closeOpenDialog(page, "Areas Served");
-        }
-        return { services, areasServed };
-      }
-      async clickRowStartingWithRetry(page, prefix) {
-        if (await this.clickRowStartingWith(page, prefix)) return true;
-        await page.waitForTimeout(2e3);
-        return this.clickRowStartingWith(page, prefix);
-      }
-      async clickRowStartingWith(page, prefix) {
-        return page.evaluate((p) => {
-          const el2 = Array.from(document.querySelectorAll('button, a, div[role="button"]')).find(
-            (e) => (e.innerText ?? "").trim().startsWith(p)
-          );
-          if (!el2) return false;
-          el2.click();
-          return true;
-        }, prefix);
-      }
-      async extractOpenDialogItems(page, expectedHeading) {
-        return page.evaluate((expected) => {
-          const dialogs = Array.from(document.querySelectorAll('[role="dialog"]'));
-          const withHeading = dialogs.find((d) => {
-            const h = d.querySelector('h1, h2, [role="heading"]');
-            const headingText = h?.innerText?.trim() ?? "";
-            return headingText.startsWith(expected);
-          });
-          if (!withHeading) return [];
-          const headingEl = withHeading.querySelector('h1, h2, [role="heading"]');
-          const headingLines = new Set(
-            (headingEl?.innerText ?? "").split("\n").map((s) => s.trim()).filter(Boolean)
-          );
-          const seen = /* @__PURE__ */ new Set();
-          const items = [];
-          withHeading.querySelectorAll("*").forEach((el2) => {
-            if (el2.children.length > 0) return;
-            const text2 = el2.innerText?.trim() ?? el2.textContent?.trim() ?? "";
-            if (!text2 || headingLines.has(text2)) return;
-            if (/^(back|close|done|thanks)$/i.test(text2)) return;
-            if (/^serves\s.*following/i.test(text2)) return;
-            if (seen.has(text2)) return;
-            seen.add(text2);
-            items.push(text2);
-          });
-          return items;
-        }, expectedHeading);
-      }
-      async closeOpenDialog(page, expectedHeading) {
-        await page.evaluate((expected) => {
-          const dialog = Array.from(document.querySelectorAll('[role="dialog"]')).find((d) => {
-            const h = d.querySelector('h1, h2, [role="heading"]');
-            const headingText = h?.innerText?.trim() ?? "";
-            return headingText.startsWith(expected);
-          });
-          const closeBtn = dialog?.querySelector('[aria-label*="Close" i], button[aria-label*="close" i]');
-          closeBtn?.click();
-        }, expectedHeading);
-      }
-    };
-  }
-});
-
-// src/extractor/MapsSearchExtractor.ts
-var MAPS_SEARCH_SCROLL_BUDGET_MS, MAPS_SEARCH_SCROLL_STEP_MS, MAPS_SEARCH_MAX_NO_GROWTH_ROUNDS, MapsSearchExtractor;
-var init_MapsSearchExtractor = __esm({
-  "src/extractor/MapsSearchExtractor.ts"() {
-    "use strict";
-    init_errors();
-    MAPS_SEARCH_SCROLL_BUDGET_MS = 6e4;
-    MAPS_SEARCH_SCROLL_STEP_MS = 1200;
-    MAPS_SEARCH_MAX_NO_GROWTH_ROUNDS = 4;
-    MapsSearchExtractor = class {
-      constructor(driver) {
-        this.driver = driver;
-      }
-      driver;
-      async extract(options) {
-        const startMs = Date.now();
-        const searchQuery = [options.query, options.location].filter(Boolean).join(" ");
-        const searchUrl = `https://www.google.com/maps/search/${encodeURIComponent(searchQuery)}?hl=${encodeURIComponent(options.hl)}`;
-        const config = {
-          headless: options.headless,
-          kernelApiKey: options.kernelApiKey,
-          kernelProxyId: options.kernelProxyId,
-          kernelProxyResolution: options.kernelProxyResolution,
-          proxyMode: options.proxyMode,
-          viewport: { width: 1280, height: 900 },
-          locale: `${options.hl}-${options.gl.toUpperCase()}`,
-          debug: options.debug
-        };
-        try {
-          await this.driver.launch(config);
-          const page = this.driver.getPage();
-          await page.goto(searchUrl, { waitUntil: "domcontentloaded", timeout: 6e4 });
-          await page.waitForTimeout(3e3);
-          const blocked = await this.detectBlock(page);
-          if (blocked) throw new CaptchaError(RECAPTCHA_INSTRUCTIONS);
-          const results = await this.collectResults(page, options.maxResults);
-          return {
-            query: options.query,
-            location: options.location ?? null,
-            searchQuery,
-            searchUrl,
-            extractedAt: (/* @__PURE__ */ new Date()).toISOString(),
-            requestedMaxResults: options.maxResults,
-            resultCount: results.length,
-            results,
-            durationMs: Date.now() - startMs
-          };
-        } finally {
-          await this.driver.close();
-        }
-      }
-      async detectBlock(page) {
         return page.evaluate(() => {
-          const text2 = document.body.innerText.slice(0, 2e3);
-          return /unusual traffic|captcha|recaptcha|about this page/i.test(text2) || /\/sorry\//.test(location.href);
-        });
-      }
-      async collectResults(page, maxResults) {
-        const seen = /* @__PURE__ */ new Map();
-        const started = Date.now();
-        let noGrowthRounds = 0;
-        while (Date.now() - started < MAPS_SEARCH_SCROLL_BUDGET_MS) {
-          const before = seen.size;
-          const batch = await this.extractVisibleResults(page);
-          for (const result of batch) {
-            const key = this.resultKey(result);
-            if (!seen.has(key)) seen.set(key, { ...result, position: seen.size + 1 });
-            if (seen.size >= maxResults) break;
-          }
-          if (seen.size >= maxResults) break;
-          if (seen.size === before) noGrowthRounds += 1;
-          else noGrowthRounds = 0;
-          if (noGrowthRounds >= MAPS_SEARCH_MAX_NO_GROWTH_ROUNDS) break;
-          await page.evaluate(() => {
-            const feed = document.querySelector('[role="feed"]');
-            if (feed) {
-              feed.scrollTop = feed.scrollHeight;
-            } else {
-              window.scrollTo(0, document.body.scrollHeight);
-            }
-          });
-          await page.waitForTimeout(MAPS_SEARCH_SCROLL_STEP_MS);
-        }
-        return [...seen.values()].slice(0, maxResults);
-      }
-      resultKey(result) {
-        return result.cidDecimal ?? result.placeUrl.replace(/[?&].*$/, "") ?? result.name;
-      }
-      async extractVisibleResults(page) {
-        return page.evaluate(() => {
-          function normalizeText(value) {
-            const text2 = value?.replace(/\s+/g, " ").trim() ?? "";
-            return text2 || null;
-          }
-          function cidFromUrl(url) {
-            const fid = url.match(/!1s(0x[0-9a-f]+):(0x[0-9a-f]+)/i);
-            if (!fid) return { cid: null, cidDecimal: null };
-            let cidDecimal = null;
-            try {
-              cidDecimal = BigInt(fid[2]).toString();
-            } catch {
-            }
-            return { cid: `${fid[1]}:${fid[2]}`, cidDecimal };
-          }
-          function textParts(card) {
-            if (!card) return [];
-            const parts = [];
-            card.querySelectorAll("div, span").forEach((el2) => {
-              const text2 = Array.from(el2.childNodes).filter((node) => node.nodeType === 3).map((node) => node.textContent?.trim() ?? "").filter((text3) => text3.length > 1 && text3.length < 140).join(" ");
-              if (text2 && !parts.includes(text2)) parts.push(text2);
-            });
-            return parts;
-          }
-          function firstMatching(parts, pattern) {
-            const value = parts.find((part) => pattern.test(part));
-            return value ?? null;
-          }
-          function normalizedSet(values) {
-            return new Set(values.filter(Boolean).map((value) => value.toLowerCase()));
-          }
-          const out = [];
-          const seen = /* @__PURE__ */ new Set();
-          const anchors = Array.from(document.querySelectorAll('a[href*="/maps/place/"]'));
-          for (const anchor of anchors) {
-            const placeUrl = anchor.href;
-            const stableUrl = placeUrl.replace(/[?&].*$/, "");
-            if (seen.has(stableUrl)) continue;
-            seen.add(stableUrl);
-            const card = anchor.closest('.Nv2PK, [role="article"], .bfdHYd') ?? anchor.parentElement;
-            const parts = textParts(card);
-            const aria = normalizeText(anchor.getAttribute("aria-label"));
-            const heading = normalizeText(card?.querySelector('.qBF1Pd, .fontHeadlineSmall, [role="heading"]')?.textContent);
-            const name = aria ?? heading ?? parts[0] ?? stableUrl;
-            const links = Array.from(card?.querySelectorAll("a[href]") ?? []);
-            const websiteUrl = links.find((link) => link.href.startsWith("http") && !link.href.includes("google."))?.href ?? null;
-            const rating = firstMatching(parts, /^\d(?:\.\d)?$/);
-            const reviewCountRaw = firstMatching(parts, /^\(?[\d,]+\)?(?:\s+reviews?)?$/i);
-            const phone = firstMatching(parts, /(?:\+?1[\s.-]?)?\(?\d{3}\)?[\s.-]\d{3}[\s.-]\d{4}/);
-            const hoursStatus = parts.find((part) => /^(open|closed|closes|opens)\b|^·\s*(opens|closes)\b/i.test(part)) ?? null;
-            const address = parts.find((part) => /\b[A-Z]{2}\s+\d{5}\b|\b(?:St|Street|Ave|Avenue|Rd|Road|Blvd|Drive|Dr)\b/i.test(part)) ?? null;
-            const directionsUrl = links.find((link) => /google\.[^/]+\/maps\/dir|\/dir\//i.test(link.href))?.href ?? `https://www.google.com/maps/dir/?api=1&destination=${encodeURIComponent([name, address].filter(Boolean).join(", ") || name)}`;
-            const excluded = normalizedSet([name, rating, reviewCountRaw, phone, hoursStatus, address, "Website", "Directions"]);
-            const category = parts.find((part) => {
-              const normalized = part.toLowerCase();
-              return !excluded.has(normalized) && !/^\d(?:\.\d)?$|^\(?[\d,]+\)?(?:\s+reviews?)?$/i.test(part) && !/(?:\+?1[\s.-]?)?\(?\d{3}\)?[\s.-]\d{3}[\s.-]\d{4}/.test(part) && !/\b[A-Z]{2}\s+\d{5}\b|\b(?:St|Street|Ave|Avenue|Rd|Road|Blvd|Drive|Dr)\b/i.test(part) && !/^(open|closed|closes|opens)\b|^·\s*(opens|closes)\b|directions|website|book online|sponsored|visit site|financing/i.test(part);
-            }) ?? null;
-            const { cid, cidDecimal } = cidFromUrl(placeUrl);
-            out.push({
-              position: out.length + 1,
-              name,
-              placeUrl,
-              cid,
-              cidDecimal,
-              rating,
-              reviewCount: reviewCountRaw ? reviewCountRaw.replace(/[()]/g, "") : null,
-              category,
-              address,
-              phone,
-              hoursStatus,
-              websiteUrl,
-              directionsUrl,
-              metadata: parts.slice(0, 20)
-            });
-          }
-          return out;
+          const visible = Array.from(document.querySelectorAll('dialog, [role="dialog"]')).find((dialog) => dialog.offsetParent !== null);
+          if (!visible) return { services: [], areasServed: [] };
+          const values = Array.from(visible.querySelectorAll("button")).map((button) => button.innerText.replace(/\s+/g, " ").trim()).filter(Boolean);
+          const servicesText = values.find((value) => /^services\s*:/i.test(value));
+          const services = servicesText ? [...new Set(servicesText.replace(/^services\s*:\s*/i, "").split(/\s*[,·]\s*/).map((value) => value.trim()).filter(Boolean))] : [];
+          const served = values.find((value) => /^serves\b/i.test(value));
+          const areasServed = served ? [served.replace(/^serves\s*/i, "").trim()].filter(Boolean) : [];
+          return { services, areasServed };
         });
       }
     };
@@ -22721,7 +22747,10 @@ function formatMapsSearch(raw, input) {
   const normalizedResults = results.map((result) => ({
     ...result,
     phone: result.phone ?? null,
-    hoursStatus: result.hoursStatus ?? null
+    hoursStatus: result.hoursStatus ?? null,
+    services: result.services ?? [],
+    areasServed: result.areasServed ?? [],
+    profileDetailsStatus: result.profileDetailsStatus ?? "unavailable"
   }));
   const searchQuery = d.searchQuery ?? [input.query, input.location].filter(Boolean).join(" ");
   const requestedMax = d.requestedMaxResults ?? input.maxResults ?? 10;
@@ -22749,9 +22778,14 @@ ${meta}`;
 |---|------|----------|--------|---------|-----|---------|------|
 ${rows}`,
     metadataSection,
+    results.some((result) => result.services?.length || result.areasServed?.length) ? `
+## Services and Areas Served
+${results.filter((result) => result.services?.length || result.areasServed?.length).map((result) => `### ${result.position}. ${result.name}
+${result.services?.length ? `- **Services:** ${result.services.join(", ")}` : ""}${result.areasServed?.length ? `
+- **Serves:** ${result.areasServed.join("; ")}` : ""}`).join("\n\n")}` : null,
     `
 ---
-\u{1F4A1} **Next step:** use \`maps_place_intel\` with a selected business name and location to hydrate full hours, phone, review topics, and optional review cards.`,
+\u{1F4A1} **Next step:** use \`maps_place_intel\` with a selected business name and location for full hours, review topics, and optional review cards. Set \`includeServices: true\` on this search to enrich the ranked list in one local-results session.`,
     durationMs != null ? `
 *Extracted in ${(durationMs / 1e3).toFixed(1)}s*` : null
   ].filter(Boolean).join("\n");
@@ -23802,6 +23836,7 @@ async function searchCity(options, market) {
       gl: options.gl,
       hl: options.hl,
       maxResults: options.maxResultsPerCity,
+      includeServices: false,
       headless: options.headless,
       kernelApiKey: options.kernelApiKey,
       proxyMode: options.proxyMode,
@@ -29688,7 +29723,7 @@ var PACKAGE_VERSION;
 var init_version = __esm({
   "src/version.ts"() {
     "use strict";
-    PACKAGE_VERSION = "0.26.2";
+    PACKAGE_VERSION = "0.27.0";
   }
 });
 
@@ -30279,8 +30314,9 @@ var init_mcp_tool_schemas = __esm({
       gl: import_zod35.z.string().length(2).default("us").describe("Google country code inferred from location."),
       hl: import_zod35.z.string().length(2).default("en").describe("Language inferred from user request."),
       maxResults: import_zod35.z.number().int().min(1).max(50).default(10).describe("Number of candidates to return. Default 10, maximum 50."),
-      proxyMode: import_zod35.z.enum(["location", "configured", "none"]).default(DEFAULT_MAPS_PROXY_MODE).describe("Defaults to location (city/state residential proxy targeting). configured forces the service proxy without city/ZIP targeting; none is local debugging only."),
-      proxyZip: import_zod35.z.string().regex(/^\d{5}$/).optional().describe("Optional US ZIP override for residential proxy targeting."),
+      includeServices: import_zod35.z.boolean().default(false).describe("Open each returned Google Business Profile and include configured services and areas served when available. Does not collect review cards."),
+      proxyMode: import_zod35.z.enum(["location", "configured", "none"]).default(DEFAULT_MAPS_PROXY_MODE).describe("Leave unset for the default direct browser route. Google localization comes from the city in the query plus UULE, gl, and hl. location is an explicit residential-proxy override."),
+      proxyZip: import_zod35.z.string().regex(/^\d{5}$/).optional().describe("Optional US ZIP override when proxyMode is location."),
       debug: import_zod35.z.boolean().default(false).describe("Include sanitized browser/proxy diagnostics.")
     };
     DirectoryWorkflowInputSchema = {
@@ -30294,7 +30330,7 @@ var init_mcp_tool_schemas = __esm({
       includeZipGroups: import_zod35.z.boolean().default(true).describe("Attach ZIP groups from a configured US ZIPS CSV when available (MCP_SCRAPER_USZIPS_CSV_PATH or usZipsCsvPath)."),
       usZipsCsvPath: import_zod35.z.string().optional().describe("Local/test-only path to a US ZIPS CSV (state_abbr, zipcode, county, city columns). Deployed APIs should use MCP_SCRAPER_USZIPS_CSV_PATH instead. For ZIP enrichment, set MCP_SCRAPER_USZIPS_CSV_PATH on the server, or pass this in local/test mode."),
       saveCsv: import_zod35.z.boolean().default(true).describe("Save a directory-ready CSV of results to the MCP Scraper output directory and return its path."),
-      proxyMode: import_zod35.z.enum(["location", "configured", "none"]).default(DEFAULT_MAPS_PROXY_MODE).describe("Proxy targeting per city search. Defaults to location (city/ZIP-group residential targeting); configured forces the service proxy; none is local debugging only."),
+      proxyMode: import_zod35.z.enum(["location", "configured", "none"]).default(DEFAULT_MAPS_PROXY_MODE).describe("Proxy behavior per city search. Leave unset for the direct localized Google route; location is an explicit residential-proxy override."),
       proxyZip: import_zod35.z.string().regex(/^\d{5}$/).optional().describe("Optional ZIP override for proxy targeting; normally omitted."),
       debug: import_zod35.z.boolean().default(false).describe("Include sanitized browser/proxy diagnostics.")
     };
@@ -30365,7 +30401,10 @@ var init_mcp_tool_schemas = __esm({
         hoursStatus: NullableString,
         websiteUrl: NullableString,
         directionsUrl: NullableString,
-        metadata: import_zod35.z.array(import_zod35.z.string())
+        metadata: import_zod35.z.array(import_zod35.z.string()),
+        services: import_zod35.z.array(import_zod35.z.string()).default([]),
+        areasServed: import_zod35.z.array(import_zod35.z.string()).default([]),
+        profileDetailsStatus: import_zod35.z.enum(["collected", "none_exist", "unavailable", "not_requested"]).default("unavailable")
       })),
       attempts: import_zod35.z.array(MapsSearchAttemptOutput),
       durationMs: import_zod35.z.number().int().min(0)
@@ -32111,7 +32150,7 @@ function registerPaaExtractorMcpTools(server, executor, options = {}) {
   }, async (input) => formatMapsPlaceIntel(await executor.mapsPlaceIntel(input), input));
   server.registerTool("maps_search", {
     title: "Google Maps Business Search",
-    description: "Search Google Maps for multiple businesses by category, niche, or local market \u2014 leads, prospects, competitors, or beyond the 3-pack. Leave proxyMode unset. Returns up to 50 candidates (default 10) with names, place URLs, CIDs, ratings.",
+    description: "Search Google local results for multiple businesses by category, niche, or local market \u2014 leads, prospects, competitors, or beyond the 3-pack. It paginates the rendered local-results list, opens each exact business card, then closes the profile dialog before continuing. Set includeServices for services and areas served; review cards are never collected by this tool.",
     inputSchema: MapsSearchInputSchema,
     outputSchema: recordOutputSchema("maps_search", MapsSearchOutputSchema),
     annotations: liveWebToolAnnotations("Google Maps Business Search")