fix(core): root-cause id-less media wash in timingCompiler getAttr, drop the band-aid (#1792)

* fix(core): root-cause the id-less media wash in getAttr, drop the band-aid

The blank-wash/dropped-audio fix in #1790 added assignMissingMediaIds in the
producer to stamp ids onto id-less timed media. That was a band-aid: the real
cause is timingCompiler's getAttr, whose regex had no name boundary at all, so
getAttr(tag, "id") matched the trailing id="…" inside data-hf-id="…". compileTag
saw a phantom id and skipped its existing hf-video-N/hf-audio-N injection,
leaving the element with no real el.id — which the render pipeline keys off of.

Fix getAttr with the same (?<![\w-]) lookbehind used for the lint readAttr fix.
compileTag's auto-id injection now fires for data-hf-id-only media, in both the
main composition and sub-compositions (parseSubCompositions runs the same
compileTimingAttrs pass), so assignMissingMediaIds is removed entirely.

Extends the regression fixture with a standalone id-less <audio> (the dropped-
audio side, previously untested) and raises minAudioCorrelation to 0.9. Adds a
timingCompiler test for the data-hf-id/id boundary.

* test(producer): use seeded pink noise (not a pure sine) for fixture audio

A continuous sine anti-aligns under the audio cross-correlation (correlation
-1.0 from a sub-period offset). Broadband seeded noise correlates robustly.

* test: cover audio-side id injection via unit test; keep render fixture video-only

The audio render-baseline used synthetic sine/noise, which anti-aligns under
the harness audio cross-correlation (deterministic -1.0). Real audio fixtures
are unaffected. Cover the audio side of the boundary fix with a deterministic
timingCompiler unit test (id-less <audio> gets hf-audio-N) instead, and keep
the render fixture video-only.

* test(producer): regenerate baseline under the root fix (hf-video-N from compileTag)

packages/core/src/compiler/timingCompiler.test.ts

@@ -16,6 +16,29 @@ describe("compileTimingAttrs", () => {
     expect(unresolved).toHaveLength(0);
   });
 
+  it("injects a real id when the element has only data-hf-id (not a phantom match)", () => {
+    // Regression: getAttr(tag, "id") matched the trailing id="…" inside
+    // data-hf-id="…" and returned a phantom, so compileTag skipped its
+    // hf-video-N injection — leaving no real el.id and a blank-wash render.
+    const html = '<video data-hf-id="hf-bgvideo01" src="a.mp4" data-start="0" data-duration="2">';
+    const { html: compiled } = compileTimingAttrs(html);
+
+    expect(compiled).toContain('id="hf-video-0"');
+    expect(compiled).toContain('data-hf-id="hf-bgvideo01"');
+    expect(compiled).toContain('data-end="2"');
+  });
+
+  it("injects a real id on an audio element that has only data-hf-id", () => {
+    // Audio side of the same bug: the mixer selects `audio[id][src]`, so a
+    // phantom-id match meant the element was dropped (silent). compileTag must
+    // inject a real hf-audio-N so the mixer can find it.
+    const html = '<audio data-hf-id="hf-bgaudio01" src="a.mp3" data-start="0" data-duration="2">';
+    const { html: compiled } = compileTimingAttrs(html);
+
+    expect(compiled).toContain('id="hf-audio-0"');
+    expect(compiled).toContain('data-hf-id="hf-bgaudio01"');
+  });
+
   it("leaves data-end unchanged when already present", () => {
     const html = '<video id="v1" src="a.mp4" data-start="0" data-end="3">';
     const { html: compiled, unresolved } = compileTimingAttrs(html);

packages/core/src/compiler/timingCompiler.ts

@@ -60,7 +60,13 @@ export function shouldClampMediaDuration(declaredDuration: number, maxDuration:
 // ── Helpers ──────────────────────────────────────────────────────────────
 
 function getAttr(tag: string, attr: string): string | null {
-  const match = tag.match(new RegExp(`${attr}=["']([^"']+)["']`));
+  // `(?<![\w-])` anchors the attribute name to a fresh start. Without it,
+  // `getAttr(tag, "id")` matches the trailing `id="…"` inside `data-hf-id="…"`
+  // (and "src" inside `data-src`, etc.) and returns a phantom value. That bug
+  // made compileTag believe a Studio-stamped `data-hf-id`-only element already
+  // had an `id`, so it skipped its `hf-video-N` injection — leaving the element
+  // with no real `el.id`, which the render pipeline keys off of (blank wash).
+  const match = tag.match(new RegExp(`(?<![\\w-])${attr}=["']([^"']+)["']`));
   return match ? (match[1] ?? null) : null;
 }
 

packages/producer/src/services/htmlCompiler.ts

@@ -1360,32 +1360,6 @@ function rewriteUnresolvableGsapToCdn(html: string, projectDir: string): string
  * with all media metadata resolved.
  */
 // fallow-ignore-next-line complexity
-/**
- * Every render stage identifies `<video>`/`<audio>` by their real `id`: frame
- * extraction keys injected stills as `__render_frame_<id>__`, the runtime
- * frame-swap matches on `el.id` (runtime/media.ts), and the audio mixer selects
- * `audio[id][src]`. A timed media element with no `id` — e.g. one carrying only
- * `data-hf-id`, which is what Studio stamps — has an empty `el.id`, so its
- * injected frames never match: the video renders as a blank wash and any
- * separate `<audio>` is silently dropped. Assign a stable positional `id` to
- * every timed media element missing one, on the same HTML that is parsed for
- * media and served to the renderer, so the whole pipeline shares one identity.
- * `data-hf-id` is intentionally NOT reused as the id — it is a Studio edit
- * handle, not the render identity.
- */
-function assignMissingMediaIds(html: string): string {
-  const { document } = parseHTML(html);
-  const media = document.querySelectorAll("video[data-start], audio[data-start]");
-  let seq = 0;
-  let changed = false;
-  for (const el of Array.from(media)) {
-    if (el.getAttribute("id")) continue;
-    el.setAttribute("id", `hf-media-${seq++}`);
-    changed = true;
-  }
-  return changed ? document.toString() : html;
-}
-
 export async function compileForRender(
   projectDir: string,
   htmlPath: string,
@@ -1510,15 +1484,7 @@ export async function compileForRender(
   // Collect assets that resolve outside projectDir (e.g. ../shared-assets/hero.png).
   // These can't be served by the file server, so we map them to paths the
   // orchestrator will copy into the compiled output directory.
-  const { html: htmlBeforeMediaIds, externalAssets } = collectExternalAssets(
-    embeddedHtml,
-    projectDir,
-  );
-
-  // Give every timed <video>/<audio> a real `id` before any stage parses or
-  // serves this HTML — id-less media (e.g. carrying only `data-hf-id`) would
-  // otherwise render as a blank wash with dropped audio. See assignMissingMediaIds.
-  const html = assignMissingMediaIds(htmlBeforeMediaIds);
+  const { html, externalAssets } = collectExternalAssets(embeddedHtml, projectDir);
 
   for (const [relPath, absPath] of remoteMediaAssets) {
     externalAssets.set(relPath, absPath);

packages/producer/tests/video-hfid-no-id/output/compiled.html

@@ -44,7 +44,7 @@
            pipeline could not match its injected frames (keyed on the empty
            el.id), so the footage rendered as a blank white/grey wash. The
            baseline must show the testsrc2 footage, not a flat fill. -->
-      <video id="hf-media-0" data-hf-id="hf-bgvideo01" class="clip" src="clip.mp4" muted playsinline data-start="0" data-duration="2" data-track-index="0" data-end="2" data-has-audio="false"></video>
+      <video data-hf-id="hf-bgvideo01" class="clip" src="clip.mp4" muted playsinline data-start="0" data-duration="2" data-track-index="0" id="hf-video-0" data-end="2" data-has-audio="false"></video>
 
       <div id="caption" class="label clip" data-start="0" data-duration="2" data-track-index="10">
         FOOTAGE