mcp-scraper 0.2.14 → 0.2.16

package/dist/bin/mcp-scraper-cli.js

@@ -9,7 +9,7 @@ import {
 } from "../chunk-F44RBOJ5.js";
 import {
   PACKAGE_VERSION
-} from "../chunk-ROCXJOVY.js";
+} from "../chunk-KE4JZDLV.js";
 
 // src/cli/human-cli.ts
 import { Command } from "commander";

package/dist/bin/mcp-scraper-combined-stdio-server.cjs

@@ -108,6 +108,9 @@ var HttpMcpToolExecutor = class {
   facebookAdTranscribe(input) {
     return this.call("/facebook/transcribe", input);
   }
+  facebookVideoTranscribe(input) {
+    return this.call("/facebook/video-transcribe", input, this.httpTimeoutOverrideMs ?? 24e4);
+  }
   mapsPlaceIntel(input) {
     return this.call("/maps/place", input);
   }
@@ -138,7 +141,7 @@ var import_node_os2 = require("os");
 var import_node_path2 = require("path");
 
 // src/version.ts
-var PACKAGE_VERSION = "0.2.14";
+var PACKAGE_VERSION = "0.2.16";
 
 // src/mcp/browser-agent-tool-schemas.ts
 var import_zod = require("zod");
@@ -1468,7 +1471,8 @@ ${adBlocks}`,
     `
 ---
 \u{1F4A1} **Tips**
-- Transcribe video ads: use \`facebook_ad_transcribe\` with the \`videoUrl\` above
+- Transcribe video ads: use \`facebook_ad_transcribe\` with the direct \`videoUrl\` above
+- Transcribe organic Facebook reels/posts: use \`facebook_video_transcribe\` with the public Facebook URL
 - Find other advertisers: use \`facebook_ad_search\``
   ].filter(Boolean).join("\n");
   return {
@@ -1914,10 +1918,66 @@ ${text}`,
 ${chunkRows}` : "",
     `
 ---
-\u{1F4A1} Get more ads from this advertiser: use \`facebook_page_intel\``
+\u{1F4A1} Get more ads from this advertiser: use \`facebook_page_intel\`. For public Facebook reel/post URLs, use \`facebook_video_transcribe\`.`
   ].filter(Boolean).join("\n");
   return oneBlock(full);
 }
+function formatFacebookVideoTranscribe(raw, input) {
+  const parsed = parseData(raw);
+  if ("error" in parsed) return { content: [{ type: "text", text: parsed.error }], isError: true };
+  const d = parsed.data;
+  const text = d.text ?? "";
+  const chunks = d.chunks ?? [];
+  const wordCount = text.trim() ? text.trim().split(/\s+/).length : 0;
+  const durSec = d.durationMs ? (d.durationMs / 1e3).toFixed(0) : "\u2014";
+  const videoDuration = typeof d.videoDurationSec === "number" ? `${Math.round(d.videoDurationSec)}s` : "\u2014";
+  const label = d.videoId ? `Facebook Organic Video \`${d.videoId}\`` : "Facebook Organic Video";
+  const chunkRows = chunks.slice(0, 50).map((c) => {
+    const sec = Number.isFinite(c.timestamp[0]) ? Math.floor(c.timestamp[0]) : 0;
+    const mm = String(Math.floor(sec / 60)).padStart(2, "0");
+    const ss = String(sec % 60).padStart(2, "0");
+    return `| ${mm}:${ss} | ${cell(truncate(c.text, 120))} |`;
+  }).join("\n");
+  const full = [
+    `# ${label} Transcript`,
+    d.ownerName ? `**Owner:** ${d.ownerName}` : "",
+    `**Video duration:** ${videoDuration} \xB7 **Transcribed in:** ${durSec}s \xB7 **${wordCount} words**`,
+    d.pageUrl ? `**Page URL:** ${d.pageUrl}` : `**Page URL:** ${input.url}`,
+    d.videoUrl ? `**Extracted MP4:** \`${d.videoUrl}\`` : "",
+    `
+## Full Transcript
+${text}`,
+    chunks.length ? `
+## Timestamped Chunks
+| Time | Text |
+|------|------|
+${chunkRows}` : "",
+    `
+---
+\u{1F4A1} Use \`videoUrl\` as the extracted MP4 for download or follow-up processing. Use \`facebook_ad_transcribe\` only for Ad Library videoUrl values.`
+  ].filter(Boolean).join("\n");
+  return {
+    ...oneBlock(full),
+    structuredContent: {
+      sourceUrl: d.sourceUrl ?? input.url,
+      pageUrl: d.pageUrl ?? input.url,
+      videoId: d.videoId ?? null,
+      ownerName: d.ownerName ?? null,
+      selectedQuality: d.selectedQuality ?? input.quality ?? "best",
+      bitrate: typeof d.bitrate === "number" ? d.bitrate : null,
+      videoDurationSec: typeof d.videoDurationSec === "number" ? d.videoDurationSec : null,
+      videoUrl: d.videoUrl ?? "",
+      wordCount,
+      chunkCount: chunks.length,
+      transcriptText: text,
+      chunks: chunks.map((c) => ({
+        startSec: Number.isFinite(c.timestamp[0]) ? c.timestamp[0] : 0,
+        endSec: Number.isFinite(c.timestamp[1]) ? c.timestamp[1] : 0,
+        text: c.text
+      }))
+    }
+  };
+}
 
 // src/mcp/mcp-tool-schemas.ts
 var import_zod2 = require("zod");
@@ -1971,7 +2031,11 @@ var FacebookAdSearchInputSchema = {
   maxResults: import_zod2.z.number().int().min(1).max(20).default(10)
 };
 var FacebookAdTranscribeInputSchema = {
-  videoUrl: import_zod2.z.string().url().describe("Facebook CDN video URL from a facebook_page_intel result")
+  videoUrl: import_zod2.z.string().url().describe("Direct Facebook CDN video URL from a facebook_page_intel ad result. Do not pass a public Facebook reel/post/share URL here; use facebook_video_transcribe for organic Facebook URLs.")
+};
+var FacebookVideoTranscribeInputSchema = {
+  url: import_zod2.z.string().url().describe("Organic Facebook reel, video, watch, post, or share URL from facebook.com, m.facebook.com, or fb.watch. The tool renders the page, extracts the best matching public Facebook CDN MP4 URL, then transcribes it. Use this when the user pastes a normal Facebook video page URL and asks for the transcript or downloadable MP4."),
+  quality: import_zod2.z.enum(["best", "hd", "sd"]).default("best").describe("Preferred progressive MP4 quality. Use best by default; hd prefers the highest HD progressive URL; sd forces the SD URL.")
 };
 var MapsPlaceIntelInputSchema = {
   businessName: import_zod2.z.string().min(1).describe('Business name only. If user says "Elite Roofing Denver CO", use businessName="Elite Roofing" and location="Denver, CO".'),
@@ -2338,6 +2402,24 @@ var FacebookPageIntelOutputSchema = {
     variations: import_zod2.z.number().int().nullable()
   }))
 };
+var FacebookVideoTranscribeOutputSchema = {
+  sourceUrl: import_zod2.z.string().url(),
+  pageUrl: import_zod2.z.string().url(),
+  videoId: NullableString,
+  ownerName: NullableString,
+  selectedQuality: import_zod2.z.string(),
+  bitrate: import_zod2.z.number().int().nullable(),
+  videoDurationSec: import_zod2.z.number().nullable(),
+  videoUrl: import_zod2.z.string().url(),
+  wordCount: import_zod2.z.number().int().min(0),
+  chunkCount: import_zod2.z.number().int().min(0),
+  transcriptText: import_zod2.z.string(),
+  chunks: import_zod2.z.array(import_zod2.z.object({
+    startSec: import_zod2.z.number(),
+    endSec: import_zod2.z.number(),
+    text: import_zod2.z.string()
+  }))
+};
 var CreditsInfoInputSchema = {
   item: import_zod2.z.string().optional().describe('Optional tool, action, or feature to look up, e.g. "maps reviews", "extract_url", "YouTube transcription", or "concurrency"'),
   includeLedger: import_zod2.z.boolean().default(false).describe("Whether to include recent credit ledger entries")
@@ -2814,7 +2896,7 @@ function registerPaaExtractorMcpTools(server2, executor, options = {}) {
   }, async (input) => formatYoutubeTranscribe(await executor.youtubeTranscribe(input), input));
   server2.registerTool("facebook_page_intel", {
     title: "Facebook Advertiser Ad Intel",
-    description: withReportNote("Harvest ads from a Facebook advertiser. Returns ad copy, headlines, CTAs, creative type, status, landing URLs, and video URLs ready for transcription. Accepts pageId, libraryId, or a brand/advertiser name as query. Use after facebook_ad_search when possible."),
+    description: withReportNote("Harvest ads from a Facebook advertiser. Returns ad copy, headlines, CTAs, creative type, status, landing URLs, and direct ad video URLs ready for facebook_ad_transcribe. Accepts pageId, libraryId, or a brand/advertiser name as query. Use after facebook_ad_search when possible. For normal public Facebook reels/posts/watch/share URLs, use facebook_video_transcribe instead."),
     inputSchema: FacebookPageIntelInputSchema,
     outputSchema: FacebookPageIntelOutputSchema,
     annotations: liveWebToolAnnotations("Facebook Advertiser Ad Intel")
@@ -2828,10 +2910,17 @@ function registerPaaExtractorMcpTools(server2, executor, options = {}) {
   }, async (input) => formatFacebookAdSearch(await executor.facebookAdSearch(input), input));
   server2.registerTool("facebook_ad_transcribe", {
     title: "Facebook Ad Transcription",
-    description: "Transcribe audio from a Facebook ad video. Returns full transcript and timestamped chunks. Use the videoUrl value from facebook_page_intel results.",
+    description: "Transcribe audio from a Facebook ad video CDN URL. Returns full transcript and timestamped chunks. Use only with the direct videoUrl value from facebook_page_intel results, not public Facebook post/reel/share URLs.",
     inputSchema: FacebookAdTranscribeInputSchema,
     annotations: liveWebToolAnnotations("Facebook Ad Transcription")
   }, async (input) => formatFacebookAdTranscribe(await executor.facebookAdTranscribe(input), input));
+  server2.registerTool("facebook_video_transcribe", {
+    title: "Facebook Organic Video Transcription",
+    description: withReportNote("Transcribe audio from an organic Facebook reel, video, watch, post, or share URL, including fb.watch links. Use this when the user pastes a normal Facebook video page URL and wants the transcript or downloadable MP4. Renders the Facebook page in a browser, selects the best matching public Facebook CDN MP4 URL from page state, then returns sourceUrl, resolved pageUrl, videoId, ownerName, selectedQuality, bitrate, videoDurationSec, extracted MP4 URL, full transcript, and timestamped chunks."),
+    inputSchema: FacebookVideoTranscribeInputSchema,
+    outputSchema: FacebookVideoTranscribeOutputSchema,
+    annotations: liveWebToolAnnotations("Facebook Organic Video Transcription")
+  }, async (input) => formatFacebookVideoTranscribe(await executor.facebookVideoTranscribe(input), input));
   server2.registerTool("maps_place_intel", {
     title: "Google Maps Business Profile Details",
     description: withReportNote('Extract Google Maps business intelligence for one known/named business: rating, review count, category, address, phone, website, hours, booking URL, review histogram, review topics, about attributes, entity IDs, and optional review cards. Do not use this for category searches, local market prospect lists, or requests for multiple GMB/GBP profiles; use maps_search first for those. Split business name from location (e.g. "Elite Roofing Denver CO" => businessName "Elite Roofing", location "Denver, CO"). Pass includeReviews true when the user asks for reviews/customer pain.'),