fix(engine): preserve bounded ffprobe diagnostics (#2772)

* fix(engine): preserve bounded ffprobe diagnostics

* test(engine): harden ffprobe diagnostic redaction

packages/engine/src/utils/ffprobe.test.ts

@@ -183,6 +183,7 @@ describe("ffprobe missing-binary fallback", () => {
 
   it("spawns the configured absolute FFprobe path when HYPERFRAMES_FFPROBE_PATH is set", async () => {
     process.env.HYPERFRAMES_FFPROBE_PATH = "/tools/ffprobe.exe";
+    const successfulStderr = "recoverable diagnostic on a successful probe";
     const { spawn, calls } = createSpawnSpy([
       {
         kind: "exit",
@@ -191,6 +192,7 @@ describe("ffprobe missing-binary fallback", () => {
           streams: [{ codec_type: "audio", codec_name: "aac", sample_rate: "48000", channels: 2 }],
           format: { duration: "1.25", bit_rate: "128000" },
         }),
+        stderr: successfulStderr,
       },
     ]);
     vi.resetModules();
@@ -200,7 +202,9 @@ describe("ffprobe missing-binary fallback", () => {
     const meta = await extractAudioMetadata("/tmp/uses-configured-ffprobe.wav");
 
     expect(meta.durationSeconds).toBe(1.25);
+    expect(JSON.stringify(meta)).not.toContain(successfulStderr);
     expect(calls[0]?.command).toBe(resolve("/tools/ffprobe.exe"));
+    expect(calls[0]?.args.slice(0, 2)).toEqual(["-v", "error"]);
   });
 
   it.each([
@@ -363,6 +367,65 @@ describe("ffprobe missing-binary fallback", () => {
     await expect(extractMediaMetadataMocked("/tmp/no-such-video.mp4")).rejects.toThrow(/ffprobe/);
   });
 
+  it("surfaces bounded ffprobe stderr for invalid media", async () => {
+    const leadingNoise = "x".repeat(10_000);
+    const diagnostic = "Invalid data found when processing input";
+    const inputPath = "/tmp/render/My Secret Video.mp4";
+    const { spawn, calls } = createSpawnSpy([
+      {
+        kind: "exit",
+        code: 1,
+        stderr: `${leadingNoise}${inputPath}: ${diagnostic}`,
+      },
+    ]);
+    vi.resetModules();
+    vi.doMock("child_process", () => ({ spawn }));
+
+    const { extractAudioMetadata } = await import("./ffprobe.js");
+
+    let thrown: unknown;
+    try {
+      await extractAudioMetadata(inputPath);
+    } catch (error) {
+      thrown = error;
+    }
+    expect(String(thrown)).toContain(diagnostic);
+    expect(String(thrown)).toContain("[input]");
+    expect(String(thrown)).not.toContain(inputPath);
+    expect(String(thrown)).not.toContain("My Secret Video.mp4");
+    expect(String(thrown).length).toBeLessThan(4_500);
+    expect(calls[0]?.args.slice(0, 2)).toEqual(["-v", "error"]);
+    expect(calls[0]?.args.at(-1)).toBe(inputPath);
+  });
+
+  it("redacts an input path fragment when the stderr tail starts inside its basename", async () => {
+    const diagnostic = "Invalid data found when processing input";
+    const inputPath = "/tmp/render/Confidential Client Preview.mp4";
+    const retainedPathFragment = "Client Preview.mp4";
+    const diagnosticPrefix = `: ${diagnostic} `;
+    const remainingBytes =
+      8 * 1024 - Buffer.byteLength(retainedPathFragment) - Buffer.byteLength(diagnosticPrefix);
+    const multibyteCount = Math.floor(remainingBytes / Buffer.byteLength("€"));
+    const trailingAscii = "x".repeat(remainingBytes - multibyteCount * Buffer.byteLength("€"));
+    const stderr = `${inputPath}${diagnosticPrefix}${"€".repeat(multibyteCount)}${trailingAscii}`;
+    const { spawn } = createSpawnSpy([{ kind: "exit", code: 1, stderr }]);
+    vi.resetModules();
+    vi.doMock("child_process", () => ({ spawn }));
+
+    const { extractAudioMetadata } = await import("./ffprobe.js");
+
+    let thrown: unknown;
+    try {
+      await extractAudioMetadata(inputPath);
+    } catch (error) {
+      thrown = error;
+    }
+    expect(String(thrown)).toContain(diagnostic);
+    expect(String(thrown)).toContain("[input]");
+    expect(String(thrown)).not.toContain(retainedPathFragment);
+    expect(String(thrown)).not.toContain("Client Preview.mp4");
+  });
+
   it("extractAudioMetadata surfaces a ffprobe-missing error verbatim", async () => {
     const { spawn, calls } = createSpawnSpy([{ kind: "missing" }]);
     hidePathBinaries();

packages/engine/src/utils/ffprobe.ts

@@ -1,15 +1,54 @@
 // fallow-ignore-file code-duplication complexity
 import { spawn } from "child_process";
 import { readFileSync } from "fs";
-import { extname } from "path";
+import { basename, extname } from "path";
+import { redactTelemetryString } from "@hyperframes/core";
 import { FFPROBE_PATH_ENV, getFfprobeBinary } from "./ffmpegBinaries.js";
 import { ManagedChildProcess } from "./managedChildProcess.js";
 import { trackChildProcess } from "./processTracker.js";
 
+const FFPROBE_STDERR_MAX_BYTES = 8 * 1024;
+const FFPROBE_ERROR_MAX_CHARS = 4 * 1024;
+
+function redactFfprobeInput(stderr: string, filePath: string): string {
+  if (!filePath) return stderr;
+
+  let redacted = stderr.split(filePath).join("[input]");
+  const inputBasename = basename(filePath);
+  if (inputBasename) redacted = redacted.split(inputBasename).join("[input]");
+
+  // ManagedChildProcess retains an 8 KiB byte tail. When that boundary lands
+  // inside the input path, stderr starts with only a suffix of the path, so
+  // neither exact-path nor generic absolute-path redaction can recognize it.
+  if (!redacted.startsWith("[input]")) {
+    for (let offset = 1; offset < filePath.length; offset += 1) {
+      const suffix = filePath.slice(offset);
+      if (suffix.length < 4 || !redacted.startsWith(suffix)) continue;
+      const boundary = redacted[suffix.length];
+      if (boundary !== undefined && !/[\s:'")\]}]/.test(boundary)) continue;
+      redacted = `[input]${redacted.slice(suffix.length)}`;
+      break;
+    }
+  }
+
+  return redacted;
+}
+
+function sanitizeFfprobeDiagnostic(stderr: string, filePath: string): string {
+  const stderrWithoutInput = redactFfprobeInput(stderr, filePath);
+  const redacted = redactTelemetryString(stderrWithoutInput, FFPROBE_STDERR_MAX_BYTES);
+  if (redacted.length <= FFPROBE_ERROR_MAX_CHARS) return redacted;
+  return `…${redacted.slice(-(FFPROBE_ERROR_MAX_CHARS - 1))}`;
+}
+
 /** Spawn ffprobe with given args, return stdout. Throws on non-zero exit or missing binary. */
-async function runFfprobe(args: string[], signal?: AbortSignal): Promise<string> {
+async function runFfprobe(
+  filePath: string,
+  argsWithoutInput: string[],
+  signal?: AbortSignal,
+): Promise<string> {
   const command = getFfprobeBinary();
-  const proc = spawn(command, args);
+  const proc = spawn(command, ["-v", "error", ...argsWithoutInput, filePath]);
   trackChildProcess(proc);
   let stdout = "";
   proc.stdout.on("data", (data) => {
@@ -18,6 +57,7 @@ async function runFfprobe(args: string[], signal?: AbortSignal): Promise<string>
   const managed = new ManagedChildProcess(proc, {
     signal,
     deadlineAtMs: Date.now() + 30_000,
+    stderrMaxBytes: FFPROBE_STDERR_MAX_BYTES,
   });
   const outcome = await managed.wait();
   if (outcome.reason === "spawn_error") {
@@ -32,8 +72,9 @@ async function runFfprobe(args: string[], signal?: AbortSignal): Promise<string>
     throw outcome.error ?? new Error(outcome.stderr);
   }
   if (outcome.reason !== "exit" || outcome.exitCode !== 0) {
+    const diagnostic = sanitizeFfprobeDiagnostic(outcome.stderr, filePath);
     throw new Error(
-      `[FFmpeg] ffprobe ${outcome.reason} with code ${outcome.exitCode}: ${outcome.stderr}`,
+      `[FFmpeg] ffprobe ${outcome.reason} with code ${outcome.exitCode}: ${diagnostic}`,
     );
   }
   return stdout;
@@ -263,14 +304,11 @@ export async function extractMediaMetadata(filePath: string): Promise<VideoMetad
 
     let output: FFProbeOutput | null = null;
     try {
-      const stdout = await runFfprobe([
+      const stdout = await runFfprobe(filePath, [
-        "-v",
-        "quiet",
         "-print_format",
         "json",
         "-show_format",
         "-show_streams",
-        filePath,
       ]);
       output = parseProbeJson(stdout);
     } catch (error) {
@@ -361,7 +399,8 @@ export async function extractAudioMetadata(
 
   const probePromise = (async (): Promise<AudioMetadata> => {
     const stdout = await runFfprobe(
-      ["-v", "quiet", "-print_format", "json", "-show_format", "-show_streams", filePath],
+      filePath,
+      ["-print_format", "json", "-show_format", "-show_streams"],
       options?.signal,
     );
     const output = parseProbeJson(stdout);
@@ -373,9 +412,7 @@ export async function extractAudioMetadata(
     const sampleRate = audioStream.sample_rate ? parseInt(audioStream.sample_rate) : 44100;
     const audioCodec = audioStream.codec_name || "unknown";
     if (audioCodec === "aac" && sampleRate > 0) {
-      const packetStdout = await runFfprobe([
+      const packetStdout = await runFfprobe(filePath, [
-        "-v",
-        "quiet",
         "-select_streams",
         "a:0",
         "-count_packets",
@@ -383,7 +420,6 @@ export async function extractAudioMetadata(
         "stream=nb_read_packets",
         "-print_format",
         "json",
-        filePath,
       ]);
       const packetOutput = parseProbeJson(packetStdout);
       const packetCount = Number(packetOutput.streams[0]?.nb_read_packets);
@@ -441,9 +477,7 @@ export async function analyzeKeyframeIntervals(filePath: string): Promise<Keyfra
 }
 
 async function analyzeKeyframeIntervalsUncached(filePath: string): Promise<KeyframeAnalysis> {
-  const stdout = await runFfprobe([
+  const stdout = await runFfprobe(filePath, [
-    "-v",
-    "quiet",
     "-select_streams",
     "v:0",
     "-skip_frame",
@@ -452,7 +486,6 @@ async function analyzeKeyframeIntervalsUncached(filePath: string): Promise<Keyfr
     "frame=pts_time",
     "-of",
     "csv=p=0",
-    filePath,
   ]);
 
   const timestamps = stdout