@revclamp/google-ads-transparency-mcp-actor 1.0.1 → 1.0.2

package/actor_config.json

@@ -21,7 +21,7 @@
   ],
   "manifest": {
     "name": "google-ads-transparency-mcp-actor",
-    "version": "1.0.1",
+    "version": "1.0.2",
     "title": "RevClamp AdPulse - For Google Ads Transparency Center (MCP & Apify Actor)",
     "websiteUrl": "https://revclamp.com",
     "description": "Find verified Google Ads advertiser entities and extract active or historical ad creatives (Text, Image, Video), preview links, and run dates without browser or proxy overhead. Independent intelligence tool by RevClamp; not affiliated with or endorsed by Google LLC."

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@revclamp/google-ads-transparency-mcp-actor",
   "mcpName": "com.revclamp/google-ads-transparency-mcp-actor",
-  "version": "1.0.1",
+  "version": "1.0.2",
   "type": "module",
   "description": "RevClamp AdPulse - For Google Ads Transparency Center (MCP & Apify Actor). Independent intelligence tool by RevClamp; not affiliated with or endorsed by Google LLC.",
   "homepage": "https://revclamp.com",

package/server.json

@@ -3,7 +3,7 @@
   "name": "com.revclamp/google-ads-transparency-mcp-actor",
   "title": "RevClamp AdPulse (Google Ads Transparency MCP)",
   "description": "Google Ads Transparency Center advertiser & creative intelligence MCP server by RevClamp.",
-  "version": "1.0.1",
+  "version": "1.0.2",
   "websiteUrl": "https://revclamp.com",
   "repository": {
     "url": "https://github.com/lhphong09t2/revclamp-mcp",
@@ -27,7 +27,7 @@
       "registryType": "npm",
       "registryBaseUrl": "https://registry.npmjs.org",
       "identifier": "@revclamp/google-ads-transparency-mcp-actor",
-      "version": "1.0.1",
+      "version": "1.0.2",
       "transport": {
         "type": "stdio"
       }

package/src/apify_actor.mjs

@@ -1255,6 +1255,7 @@ export async function runGoogleAdsTransparencyActor(input = {}, hooks = {}) {
         advertiserId: targetItem.advertiserId,
         creativeId: targetItem.creativeId,
         includeAdCopy: Boolean(input?.includeAdCopy),
+        fallbackPreviewUrl: targetItem.previewUrl,
         fetchImpl: hooks.fetchImpl,
         nowFn: nowMs,
         sleepFn,
@@ -1282,6 +1283,14 @@ export async function runGoogleAdsTransparencyActor(input = {}, hooks = {}) {
             rateLimitedInEnrich = true;
           }
         }
+      } else if (detailRes.ok && detailRes.untrustedAdContent) {
+        targetItem.untrustedAdContent = detailRes.untrustedAdContent;
+        if (
+          detailRes.untrustedAdContent.extractionStatus === 'UPSTREAM_RATE_LIMITED' ||
+          detailRes.untrustedAdContent.contentJsFetchStatus === 'UPSTREAM_RATE_LIMITED'
+        ) {
+          rateLimitedInEnrich = true;
+        }
       }
     }
   }

package/src/mcp_server.mjs

@@ -43,7 +43,7 @@ export const ENCODED_BRAND_ICON_DATA_URI = `data:image/svg+xml;utf8,${encodeURIC
 export const MCP_SERVER_INFO = Object.freeze({
   name: 'google-ads-transparency-mcp-actor',
   title: 'RevClamp AdPulse - For Google Ads Transparency Center (MCP & Apify Actor)',
-  version: '1.0.1',
+  version: '1.0.2',
   description:
     'Find verified Google Ads advertiser entities and extract active or historical ad creatives (Text, Image, Video), preview links, and run dates. Independent intelligence tool by RevClamp; not affiliated with or endorsed by Google LLC.',
   // Forward-compatible MCP Registry & Glama metadata extensions (ignored by older stdio clients)

package/src/rpc_client.mjs

@@ -1928,7 +1928,7 @@ export async function resolveAdvertiserEntities(options = {}) {
   const endpointUrl = options.endpointUrl || GOOGLE_ADS_RPC_ENDPOINTS.SearchSuggestions;
   const rpcRes = await executeGoogleAdsRpcPost(
     endpointUrl,
-    { '1': searchQueryStr, '2': 10, '3': 10 },
+    { '1': searchQueryStr, '2': 50, '3': 10 },
     options
   );
 
@@ -2291,8 +2291,20 @@ function isAllowedLandingHostname(hostname) {
   return true;
 }
 
+const VAST_CDATA_WRAPPER_SUFFIX_RE =
+  /(?:\]\]>|\]\]%3E|%5D%5D%3E|%5D%5D%253E|<\/[A-Za-z0-9_:-]+>|%3C\/[A-Za-z0-9_:-]+%3E|%3C%2F[A-Za-z0-9_:-]+%3E|%253C%2F[A-Za-z0-9_:-]+%253E).*$/i;
+
+function stripVastXmlCdataArtifacts(rawStr) {
+  if (typeof rawStr !== 'string' || !rawStr) return '';
+  return rawStr
+    .replace(/^(?:<!\[CDATA\[|%3C!%5BCDATA%5B)+/i, '')
+    .replace(VAST_CDATA_WRAPPER_SUFFIX_RE, '')
+    .replace(/[\]>]+$/, '')
+    .trim();
+}
+
 function normalizeCandidateLandingUrl(rawUrl) {
-  const cleaned = sanitizeUntrustedAdText(rawUrl, 2048);
+  const cleaned = stripVastXmlCdataArtifacts(sanitizeUntrustedAdText(rawUrl, 2048));
   if (!cleaned) return null;
   try {
     const parsed = new URL(cleaned);
@@ -2363,7 +2375,7 @@ function unwrapCandidateUrlOrAdUrlParam(rawUrl) {
   const adUrlMatch = decodedRaw.match(/[?&]adurl=([^"'\s&\\]+)/i);
   if (adUrlMatch && adUrlMatch[1]) {
     try {
-      const target = decodeURIComponent(adUrlMatch[1]);
+      const target = decodeURIComponent(stripVastXmlCdataArtifacts(adUrlMatch[1]));
       const normalizedTarget = normalizeCandidateLandingUrl(target);
       if (normalizedTarget) return normalizedTarget;
     } catch {
@@ -2583,7 +2595,9 @@ export function extractStaticAdCopyFromContentJs(rawJsText) {
       const adUrlParamMatch = surface.match(/[?&]adurl=(https?%3A%2F%2F[^"'\s&\\]+)/i);
       if (adUrlParamMatch && adUrlParamMatch[1]) {
         try {
-          landingPageUrl = normalizeCandidateLandingUrl(decodeURIComponent(adUrlParamMatch[1]));
+          landingPageUrl = normalizeCandidateLandingUrl(
+            decodeURIComponent(stripVastXmlCdataArtifacts(adUrlParamMatch[1]))
+          );
           if (landingPageUrl) break;
         } catch {
           landingPageUrl = null;
@@ -3558,17 +3572,33 @@ export async function getCreativeDetailsById(options = {}) {
     };
   }
 
+  const nowFn = typeof options.nowFn === 'function' ? options.nowFn : Date.now;
+  const fallbackPreviewUrl =
+    typeof options.fallbackPreviewUrl === 'string' && options.fallbackPreviewUrl.trim()
+      ? options.fallbackPreviewUrl.trim()
+      : null;
+
   if (canary.status === 'EMPTY_RESULT') {
+    let fallbackUntrustedAdContent = null;
+    if (
+      options.includeAdCopy === true &&
+      fallbackPreviewUrl &&
+      (!options.deadlineAtMs || nowFn() + 200 < options.deadlineAtMs)
+    ) {
+      fallbackUntrustedAdContent = await fetchAndExtractStaticAdCopy(fallbackPreviewUrl, options);
+    }
     return {
       ok: true,
       status: 'EMPTY_RESULT',
       detail: null,
+      ...(fallbackUntrustedAdContent ? { untrustedAdContent: fallbackUntrustedAdContent } : {}),
       rpcAttempts: rpcRes.attempts,
     };
   }
 
   const detailObj = rpcRes.data['1'];
   const { previewUrl, archiveImageUrls } = extractCreativeAssetUrls(detailObj?.['5'] ?? detailObj);
+  const effectivePreviewUrl = previewUrl || fallbackPreviewUrl;
   const rawRegions = Array.isArray(detailObj?.['17']) ? detailObj['17'] : [];
   const regionalPresence = rawRegions
     .filter((r) => r && typeof r === 'object')
@@ -3588,15 +3618,14 @@ export async function getCreativeDetailsById(options = {}) {
   let untrustedAdContent = extractStaticAdCopyFromContentJs(
     JSON.stringify(detailObj?.['5'] ?? detailObj)
   );
-  const nowFn = typeof options.nowFn === 'function' ? options.nowFn : Date.now;
   if (
     !untrustedAdContent.headline &&
     !untrustedAdContent.bodyText &&
     options.includeAdCopy === true &&
-    previewUrl &&
+    effectivePreviewUrl &&
     (!options.deadlineAtMs || nowFn() + 200 < options.deadlineAtMs)
   ) {
-    const fetchedCopy = await fetchAndExtractStaticAdCopy(previewUrl, options);
+    const fetchedCopy = await fetchAndExtractStaticAdCopy(effectivePreviewUrl, options);
     if (fetchedCopy.extractionStatus === 'EXTRACTED') {
       untrustedAdContent = {
         ...fetchedCopy,