fix(media-use): clean up temporary TTS text directories (#5477)

* fix(media-use): clean up temporary TTS text files

* chore(skills): refresh manifest for TTS cleanup

skills-manifest.json

@@ -54,8 +54,8 @@
       "files": 1
     },
     "media-use": {
-      "hash": "1a841e4288d862af",
-      "files": 110
+      "hash": "f9ff3d9b23853032",
+      "files": 111
     },
     "motion-graphics": {
       "hash": "6db77031ad839f6a",

skills/media-use/audio/scripts/lib/tts.cleanup.test.mjs

@@ -0,0 +1,175 @@
+import { test } from "node:test";
+import assert from "node:assert/strict";
+import { spawnSync } from "node:child_process";
+import { chmodSync, mkdirSync, mkdtempSync, rmSync, writeFileSync } from "node:fs";
+import { tmpdir } from "node:os";
+import { join } from "node:path";
+
+const ttsUrl = new URL("./tts.mjs", import.meta.url).href;
+const stub = `#!${process.execPath}
+const fs = require("node:fs");
+const path = require("node:path");
+const args = process.argv.slice(2);
+if (args[0] === "--version") {
+  console.log("Python 3.12.0");
+  process.exit(0);
+}
+const input = args[2];
+const output = args[0] === "-c" ? args[4] : args[args.indexOf("--output") + 1];
+const text = fs.readFileSync(input, "utf8");
+fs.writeFileSync(output + ".started", input);
+const finish = () => {
+  if (text === "slow" && !fs.existsSync(process.env.RELEASE_FILE)) {
+    setTimeout(finish, 5);
+    return;
+  }
+  const consumed = fs.readFileSync(input, "utf8");
+  fs.writeFileSync(output + ".json", JSON.stringify({ input, text, consumed, args }));
+  if (process.env.CASE_MODE === "nonzero") process.exit(7);
+  if (process.env.CASE_MODE !== "no-output") fs.writeFileSync(output, consumed);
+};
+setTimeout(finish, 10);
+`;
+
+const exercise = `
+import assert from "node:assert/strict";
+import fs from "node:fs";
+import childProcess from "node:child_process";
+import { syncBuiltinESMExports } from "node:module";
+import { join, dirname } from "node:path";
+const { CASE_ROOT: root, CASE_MODE: mode, CASE_PROVIDER: provider } = process.env;
+const temp = join(root, "tmp");
+const preexisting = join(temp, "hf-txt-preexisting");
+fs.mkdirSync(preexisting);
+fs.writeFileSync(join(preexisting, "marker"), "keep");
+const injected = Object.assign(new Error("synthetic " + mode), { code: "EACCES" });
+const originalMkdtemp = fs.mkdtempSync;
+const originalWrite = fs.writeFileSync;
+const originalRm = fs.rmSync;
+if (mode === "allocation") fs.mkdtempSync = (prefix, ...args) => {
+  if (prefix.startsWith(join(temp, "hf-txt-"))) throw injected;
+  return originalMkdtemp(prefix, ...args);
+};
+if (mode === "write") fs.writeFileSync = (file, ...args) => {
+  if (String(file).endsWith("line.txt")) {
+    originalWrite(file, "partial");
+    throw injected;
+  }
+  return originalWrite(file, ...args);
+};
+if (mode === "spawn-throw") childProcess.spawn = () => { throw injected; };
+if (mode === "cleanup") fs.rmSync = (dir, ...args) => {
+  if (String(dir).startsWith(join(temp, "hf-txt-"))) throw injected;
+  return originalRm(dir, ...args);
+};
+syncBuiltinESMExports();
+const { synthesizeOne } = await import(${JSON.stringify(ttsUrl)});
+const project = join(root, "project");
+const wavAbs = join(project, "line.wav");
+const request = (chosen = provider, text = "synthetic café", output = wavAbs) =>
+  synthesizeOne({ provider: chosen, text, voiceId: "synthetic-voice", lang: "es",
+    speed: 0.8, wavAbs: output, hyperframesDir: project });
+const owned = () => fs.readdirSync(temp).filter(name => name !== "hf-txt-preexisting");
+if (["allocation", "write", "spawn-throw"].includes(mode)) {
+  await assert.rejects(request(), error => error === injected);
+  assert.equal(fs.existsSync(wavAbs), false);
+} else if (mode === "concurrent") {
+  const slowWav = join(project, "slow.wav");
+  const slow = request("elevenlabs", "slow", slowWav);
+  try {
+    const deadline = Date.now() + 5000;
+    while (!fs.existsSync(slowWav + ".started")) {
+      assert.ok(Date.now() < deadline, "slow subprocess must start");
+      await new Promise(resolve => setTimeout(resolve, 5));
+    }
+    const slowInput = fs.readFileSync(slowWav + ".started", "utf8");
+    assert.deepEqual(await request("kokoro"), { ok: true, words: null });
+    assert.deepEqual(owned(), [dirname(slowInput).split(/[\\\\/]/).pop()]);
+    assert.equal(fs.readFileSync(slowInput, "utf8"), "slow");
+  } finally {
+    fs.writeFileSync(process.env.RELEASE_FILE, "release");
+    await slow;
+  }
+  assert.equal(fs.readFileSync(slowWav, "utf8"), "slow");
+} else {
+  const result = await request();
+  const label = provider === "elevenlabs" ? "elevenlabs (python)" : "kokoro (npx hyperframes tts)";
+  if (["nonzero", "enoent", "no-output"].includes(mode)) {
+    const error = mode === "no-output" ? label + " produced no wav file" :
+      label + " exited with status " + (mode === "enoent" ? -1 : 7);
+    assert.deepEqual(result, { ok: false, words: null, error });
+    assert.equal(fs.existsSync(wavAbs), false);
+  } else {
+    assert.deepEqual(result, { ok: true, words: null });
+    assert.equal(fs.readFileSync(wavAbs, "utf8"), "synthetic café");
+  }
+  if (mode !== "enoent") {
+    const log = JSON.parse(fs.readFileSync(wavAbs + ".json", "utf8"));
+    assert.equal(log.text, "synthetic café");
+    assert.equal(log.consumed, log.text);
+    if (provider === "kokoro") assert.deepEqual(log.args.slice(-4), ["--lang", "es", "--speed", "0.8"]);
+  }
+}
+if (mode === "cleanup") assert.equal(owned().length, 1);
+else assert.deepEqual(owned(), [], "owned text directories must be removed");
+assert.equal(fs.readFileSync(join(preexisting, "marker"), "utf8"), "keep");
+`;
+
+function runCase(provider, mode) {
+  const root = mkdtempSync(join(tmpdir(), "tts-cleanup-test-"));
+  try {
+    for (const dir of ["home", "tmp", "bin", "project"]) mkdirSync(join(root, dir));
+    if (mode !== "enoent") {
+      for (const name of ["npx", "python3"]) {
+        const executable = join(root, "bin", name);
+        writeFileSync(executable, stub);
+        chmodSync(executable, 0o755);
+      }
+    }
+    const result = spawnSync(process.execPath, ["--input-type=module", "-e", exercise], {
+      encoding: "utf8",
+      timeout: 15000,
+      env: {
+        HOME: join(root, "home"),
+        USERPROFILE: join(root, "home"),
+        TMPDIR: join(root, "tmp"),
+        TMP: join(root, "tmp"),
+        TEMP: join(root, "tmp"),
+        PATH: join(root, "bin"),
+        CASE_ROOT: root,
+        CASE_PROVIDER: provider,
+        CASE_MODE: mode,
+        RELEASE_FILE: join(root, "release"),
+      },
+    });
+    assert.equal(result.error, undefined);
+    assert.equal(result.status, 0, result.stdout + result.stderr);
+  } finally {
+    rmSync(root, { recursive: true, force: true });
+  }
+}
+
+for (const provider of ["elevenlabs", "kokoro"]) {
+  for (const mode of [
+    "success",
+    "nonzero",
+    "no-output",
+    "enoent",
+    "spawn-throw",
+    "allocation",
+    "write",
+    "cleanup",
+  ]) {
+    test(
+      `synthesizeOne(${provider}) preserves ${mode} behavior and cleans only its text directory`,
+      { skip: process.platform === "win32" },
+      () => runCase(provider, mode),
+    );
+  }
+}
+
+test(
+  "concurrent providers retain each other's text until their subprocess finishes",
+  { skip: process.platform === "win32" },
+  () => runCase("kokoro", "concurrent"),
+);

skills/media-use/audio/scripts/lib/tts.mjs

@@ -274,23 +274,21 @@ export async function synthesizeOne({
     } catch {
       return { ok: false, words: null };
     }
-    const { cmd, args } = pythonInvocation([
-      "-c",
-      ELEVENLABS_PY,
-      writeTmpText(text),
-      voiceId,
-      wavAbs,
-    ]);
-    const r = await spawnP(cmd, args, {});
-    return synthResult(r, wavAbs, "elevenlabs (python)");
+    return withTmpText(text, async (textPath) => {
+      const { cmd, args } = pythonInvocation(["-c", ELEVENLABS_PY, textPath, voiceId, wavAbs]);
+      const r = await spawnP(cmd, args, {});
+      return synthResult(r, wavAbs, "elevenlabs (python)");
+    });
   }
   // kokoro — via the published CLI; --output is relative to the project dir.
   const wavRel = relTo(hyperframesDir, wavAbs);
-  const args = ["hyperframes", "tts", writeTmpText(text), "--voice", voiceId, "--output", wavRel];
-  if (lang !== "en") args.push("--lang", lang);
-  if (speed !== 1) args.push("--speed", String(speed));
-  const r = await spawnP("npx", args, { cwd: hyperframesDir });
-  return synthResult(r, wavAbs, "kokoro (npx hyperframes tts)");
+  return withTmpText(text, async (textPath) => {
+    const args = ["hyperframes", "tts", textPath, "--voice", voiceId, "--output", wavRel];
+    if (lang !== "en") args.push("--lang", lang);
+    if (speed !== 1) args.push("--speed", String(speed));
+    const r = await spawnP("npx", args, { cwd: hyperframesDir });
+    return synthResult(r, wavAbs, "kokoro (npx hyperframes tts)");
+  });
 }
 
 // Shape a spawn result into { ok, words, error }, naming why on failure so the
@@ -378,11 +376,18 @@ export async function transcribeWav({ wavRel, lang = "en", hyperframesDir }) {
 }
 
 // ── tiny local utils ──────────────────────────────────────────────────────────
-function writeTmpText(text) {
+async function withTmpText(text, run) {
   const td = mkdtempSync(join(tmpdir(), "hf-txt-"));
-  const p = join(td, "line.txt");
-  writeFileSync(p, text);
-  return p;
+  try {
+    const p = join(td, "line.txt");
+    writeFileSync(p, text);
+    return await run(p);
+  } finally {
+    // Cleanup must preserve the synthesis result or the original write/spawn error.
+    try {
+      rmSync(td, { recursive: true, force: true });
+    } catch {}
+  }
 }
 function relTo(base, abs) {
   return abs.startsWith(base + "/") ? abs.slice(base.length + 1) : abs;