@agentproto/corpus-cli 0.1.1 → 0.3.0

package/dist/ports/index.d.ts

@@ -133,12 +133,22 @@ declare class OpenAiWhisperStt implements SttPort {
  * transcribes it with an injected SttPort (Whisper). Captions-independent
  * and robust: works whether or not the video has subtitles.
  *
- *   video URL → yt-dlp -f bestaudio -x (mp3) → SttPort.transcribe → text
+ *   video URL → yt-dlp -f bestaudio/best -x (mp3) → SttPort.transcribe → text
  *
- * yt-dlp handles the pot token / SABR negotiation / signature itself, so
- * no browser or cookies are needed for public videos. Non-video URLs
- * resolve to `null` so this fetcher composes ahead of a readability
- * fetcher (see CompositeFetcher).
+ * Modern YouTube requires three workarounds beyond a plain yt-dlp call:
+ *   • `-f bestaudio/best` — many videos no longer expose a standalone
+ *     bestaudio format; the /best fallback avoids "format not available".
+ *   • `--remote-components ejs:github` — solves the nsig "n challenge"
+ *     that otherwise yields "Only images are available".
+ *   • `--extractor-args youtube:player_client=<client>` — both the android
+ *     client (unauthenticated) and the web_creator client (authenticated)
+ *     fall back to progressive format 18, which requires no PO token and
+ *     avoids the HTTP 403 that DASH segment requests trigger. The client is
+ *     chosen based on whether cookies are provided: android does not support
+ *     cookie auth (yt-dlp skips it when cookies are present), so web_creator
+ *     is used instead for authenticated sessions.
+ * Cookies (--cookies-from-browser or --cookies) are still recommended to
+ * avoid bot-check rate-limits on burst downloads.
  *
  * The yt-dlp invocation is injected as `download` so the adapter is unit-
  * testable with a fake downloader + fake SttPort — no binary, no network.
@@ -178,6 +188,13 @@ interface YtDlpWhisperFetcherOptions {
     readonly cookiesFromBrowser?: string;
     /** Path to a Netscape cookies.txt — passed to yt-dlp as `--cookies`. */
     readonly cookiesFile?: string;
+    /**
+     * Directory that contains the `ffmpeg` / `ffprobe` binaries (e.g.
+     * `/opt/homebrew/bin`). Passed as `--ffmpeg-location` to yt-dlp so the
+     * postprocessor can find them even when PATH is not inherited by the
+     * subprocess. Omit to rely on yt-dlp's default PATH search.
+     */
+    readonly ffmpegLocation?: string;
 }
 declare class YtDlpWhisperFetcher implements FetcherPort {
     private readonly stt;
@@ -386,6 +403,8 @@ interface AnthropicDistillerOptions {
     readonly maxItems?: number;
     /** Optional sink for per-call token usage (cost + Langfuse export). */
     readonly onUsage?: (usage: DistillUsage) => void;
+    /** Output language code (e.g. "fr"). Absent → English (default). */
+    readonly lang?: string;
 }
 declare class AnthropicDistiller implements DistillPort {
     private readonly apiKey;
@@ -393,6 +412,7 @@ declare class AnthropicDistiller implements DistillPort {
     private readonly baseUrl;
     private readonly maxItems;
     private readonly onUsage;
+    private readonly lang;
     constructor(opts: AnthropicDistillerOptions);
     distill(input: DistillInput): Promise<readonly DistilledItem[]>;
 }
@@ -435,9 +455,14 @@ interface CliEngine {
     readonly subscriptionBilled: boolean;
     /** Executable to invoke — must be on PATH and logged in. */
     readonly command: string;
-    /** Build the argv for one completion. The prompt is fed over stdin, not argv. */
+    /**
+     * Build the argv for one completion.
+     * Most engines receive the prompt over stdin and ignore `opts.prompt`.
+     * Arg-based engines (hermes `-z`) embed the prompt in argv instead.
+     */
     buildArgs(opts: {
         readonly model?: string;
+        readonly prompt?: string;
     }): string[];
     /** Extract the model's text from captured stdout; null → fall back to raw stdout. */
     parseOutput(stdout: string): string | null;
@@ -473,6 +498,8 @@ interface CliAgentDistillerOptions {
     readonly timeoutMs?: number;
     /** Working dir. Default os.tmpdir() — neutral, no project CLAUDE.md / repo. */
     readonly cwd?: string;
+    /** Output language code (e.g. "fr"). Absent → English (default). */
+    readonly lang?: string;
 }
 declare class CliAgentDistiller implements DistillPort {
     private readonly engine;
@@ -480,6 +507,7 @@ declare class CliAgentDistiller implements DistillPort {
     private readonly maxItems;
     private readonly timeoutMs;
     private readonly cwd;
+    private readonly lang;
     constructor(opts: CliAgentDistillerOptions);
     distill(input: DistillInput): Promise<readonly DistilledItem[]>;
 }

package/dist/ports/index.mjs

@@ -339,7 +339,8 @@ var YtDlpWhisperFetcher = class {
     this.download = opts.download ?? defaultYtDlpDownloader(opts.ytDlpBin ?? "yt-dlp", {
       maxDurationSec: opts.maxDurationSec,
       cookiesFromBrowser: opts.cookiesFromBrowser,
-      cookiesFile: opts.cookiesFile
+      cookiesFile: opts.cookiesFile,
+      ffmpegLocation: opts.ffmpegLocation
     });
     this.extraHosts = opts.extraVideoHosts ? new Set(opts.extraVideoHosts) : void 0;
   }
@@ -383,32 +384,44 @@ function isAuthError(e) {
   const m = msg(e).toLowerCase();
   return m.includes(" 401") || m.includes(" 403") || m.includes("unauthorized") || m.includes("forbidden") || m.includes("api key") || m.includes("api_key");
 }
+function buildYtDlpArgs(url, dir, opts) {
+  const durationGuard = opts?.maxDurationSec && opts.maxDurationSec > 0 ? ["--match-filter", `duration <= ${Math.floor(opts.maxDurationSec)}`] : [];
+  const cookieArgs = opts?.cookiesFromBrowser ? ["--cookies-from-browser", opts.cookiesFromBrowser] : opts?.cookiesFile ? ["--cookies", opts.cookiesFile] : [];
+  const ffmpegArgs = opts?.ffmpegLocation ? ["--ffmpeg-location", opts.ffmpegLocation] : [];
+  const client = cookieArgs.length === 0 ? "android" : "web_creator";
+  const extractorArgs = ["--extractor-args", `youtube:player_client=${client}`];
+  return [
+    "-f",
+    "bestaudio/best",
+    // /best fallback: many videos lack a standalone bestaudio
+    "-x",
+    "--audio-format",
+    "mp3",
+    "--audio-quality",
+    "64K",
+    "--no-playlist",
+    "--no-progress",
+    "--remote-components",
+    "ejs:github",
+    // solves YouTube nsig "n challenge"
+    ...extractorArgs,
+    ...durationGuard,
+    ...cookieArgs,
+    ...ffmpegArgs,
+    "--write-info-json",
+    "-o",
+    join(dir, "track.%(ext)s"),
+    url
+  ];
+}
 function defaultYtDlpDownloader(bin, opts) {
   return async (url) => {
     const dir = await mkdtemp(join(tmpdir(), "corpus-ytdlp-"));
     const cleanup = () => rm(dir, { recursive: true, force: true });
     try {
-      const durationGuard = opts?.maxDurationSec && opts.maxDurationSec > 0 ? ["--match-filter", `duration <= ${Math.floor(opts.maxDurationSec)}`] : [];
-      const cookieArgs = opts?.cookiesFromBrowser ? ["--cookies-from-browser", opts.cookiesFromBrowser] : opts?.cookiesFile ? ["--cookies", opts.cookiesFile] : [];
       await execFileAsync(
         bin,
-        [
-          "-f",
-          "bestaudio",
-          "-x",
-          "--audio-format",
-          "mp3",
-          "--audio-quality",
-          "64K",
-          "--no-playlist",
-          "--no-progress",
-          ...durationGuard,
-          ...cookieArgs,
-          "--write-info-json",
-          "-o",
-          join(dir, "track.%(ext)s"),
-          url
-        ],
+        buildYtDlpArgs(url, dir, opts),
         { maxBuffer: 16 * 1024 * 1024 }
       );
       const files = await readdir(dir);
@@ -761,15 +774,17 @@ var AnthropicDistiller = class {
   baseUrl;
   maxItems;
   onUsage;
+  lang;
   constructor(opts) {
     this.apiKey = opts.apiKey;
     this.model = opts.model ?? "claude-sonnet-4-6";
     this.baseUrl = (opts.baseUrl ?? "https://api.anthropic.com/v1").replace(/\/+$/, "");
     this.maxItems = opts.maxItems ?? 8;
     this.onUsage = opts.onUsage;
+    this.lang = opts.lang;
   }
   async distill(input) {
-    const prompt = buildDistillPrompt(input, this.maxItems);
+    const prompt = buildDistillPrompt(input, this.maxItems, { lang: this.lang });
     const startedAt = (/* @__PURE__ */ new Date()).toISOString();
     const res = await fetch(`${this.baseUrl}/messages`, {
       method: "POST",
@@ -808,18 +823,20 @@ var CliAgentDistiller = class {
   maxItems;
   timeoutMs;
   cwd;
+  lang;
   constructor(opts) {
     this.engine = opts.engine;
     this.model = opts.model;
     this.maxItems = opts.maxItems ?? 8;
     this.timeoutMs = opts.timeoutMs ?? 12e4;
     this.cwd = opts.cwd ?? tmpdir();
+    this.lang = opts.lang;
   }
   async distill(input) {
-    const prompt = buildDistillPrompt(input, this.maxItems);
+    const prompt = buildDistillPrompt(input, this.maxItems, { lang: this.lang });
     const stdout = await spawnWithStdin({
       command: this.engine.command,
-      args: this.engine.buildArgs({ ...this.model ? { model: this.model } : {} }),
+      args: this.engine.buildArgs({ ...this.model ? { model: this.model } : {}, prompt }),
       stdin: prompt,
       cwd: this.cwd,
       timeoutMs: this.timeoutMs
@@ -894,11 +911,23 @@ var OPENCODE = {
   buildArgs: ({ model }) => ["run", ...model ? ["-m", model] : []],
   parseOutput: plainTextOutput
 };
+var HERMES = {
+  id: "hermes",
+  subscriptionBilled: true,
+  command: "hermes",
+  buildArgs: ({ model, prompt }) => [
+    "--oneshot",
+    prompt ?? "",
+    ...model ? ["-m", model] : []
+  ],
+  parseOutput: plainTextOutput
+};
 var CLI_ENGINES = {
   [CLAUDE_CODE.id]: CLAUDE_CODE,
   [GEMINI.id]: GEMINI,
   [CODEX.id]: CODEX,
-  [OPENCODE.id]: OPENCODE
+  [OPENCODE.id]: OPENCODE,
+  [HERMES.id]: HERMES
 };
 var RPC_RESPONSE = z.object({ result: z.unknown().optional(), error: z.unknown().optional() }).loose();
 var RPC_MESSAGE = z.object({