fix(compiler): don't scan media tags inside comments/scripts; DOM-check auto-start (#1938) (#1940)

The timing compiler scanned raw HTML with tag regexes that weren't
comment-aware, so a comment or script merely mentioning `<video>`/`<audio>`
was rewritten as a real element — injecting id/data-start/data-hf-auto-start
into the comment text. That phantom attribute then tripped the probe stage's
substring check (`html.includes("data-hf-auto-start")`), launching an
unnecessary browser probe on every render with an unexplained empty reasons list.

- Mask comments, <script>, and <style> regions before the tag scan, then
  restore them verbatim (compileTimingAttrs, extractResolvedMedia).
- Replace the probe's substring match with a DOM query
  (video[data-hf-auto-start]) and add "auto-start video(s)" to the reasons list.

packages/producer/src/services/render/stages/probeStage.test.ts

@@ -1,5 +1,5 @@
 import { describe, expect, it, mock } from "bun:test";
-import { hasScriptedAudioVolumeAutomation } from "./probeStage.js";
+import { hasAutoStartVideos, hasScriptedAudioVolumeAutomation } from "./probeStage.js";
 
 // ── Mocks for runProbeStage tests ────────────────────────────────────────────
 // Capture the cfg passed to createCaptureSession so we can assert it carries
@@ -224,6 +224,26 @@ describe("hasScriptedAudioVolumeAutomation", () => {
   });
 });
 
+describe("hasAutoStartVideos", () => {
+  it("detects a real auto-start video element", () => {
+    expect(hasAutoStartVideos(`<video src="a.mp4" data-hf-auto-start="">`)).toBe(true);
+  });
+
+  it("ignores the attribute mentioned in a comment (issue #1938)", () => {
+    expect(hasAutoStartVideos(`<!-- videos get data-hf-auto-start injected --><p>hi</p>`)).toBe(
+      false,
+    );
+  });
+
+  it("ignores the attribute in prose text", () => {
+    expect(hasAutoStartVideos(`<p>the data-hf-auto-start sentinel</p>`)).toBe(false);
+  });
+
+  it("returns false when there is no media", () => {
+    expect(hasAutoStartVideos(`<div class="clip"></div>`)).toBe(false);
+  });
+});
+
 describe("runProbeStage — forceScreenshot threading", () => {
   it("passes forceScreenshot:true to createCaptureSession when stage input carries it but cfg does not (low-memory mode fix #1236)", async () => {
     capturedCfgs.length = 0;

packages/producer/src/services/render/stages/probeStage.ts

@@ -114,6 +114,17 @@ export function hasScriptedAudioVolumeAutomation(html: string, audioCount: numbe
   );
 }
 
+/**
+ * True when the compiled HTML has at least one `<video>` carrying the
+ * auto-injected `data-hf-auto-start` sentinel. Uses a DOM query, not a
+ * substring scan — `html.includes("data-hf-auto-start")` false-fires on any
+ * comment or prose that merely mentions the attribute (issue #1938).
+ */
+export function hasAutoStartVideos(html: string): boolean {
+  const { document } = parseHTML(html);
+  return document.querySelector("video[data-hf-auto-start]") !== null;
+}
+
 export async function runProbeStage(input: ProbeStageInput): Promise<ProbeStageResult> {
   const {
     projectDir,
@@ -139,7 +150,7 @@ export async function runProbeStage(input: ProbeStageInput): Promise<ProbeStageR
   let lastBrowserConsole: string[] = [];
 
   const probeStart = Date.now();
-  const hasAutoStartVideos = compiled.html.includes("data-hf-auto-start");
+  const hasAutoStart = hasAutoStartVideos(compiled.html);
   const hasScriptedAudio = hasScriptedAudioVolumeAutomation(
     compiled.html,
     composition.audios.length,
@@ -147,7 +158,7 @@ export async function runProbeStage(input: ProbeStageInput): Promise<ProbeStageR
   const needsBrowser =
     composition.duration <= 0 ||
     compiled.unresolvedCompositions.length > 0 ||
-    hasAutoStartVideos ||
+    hasAutoStart ||
     hasScriptedAudio;
 
   if (needsBrowser) {
@@ -155,6 +166,7 @@ export async function runProbeStage(input: ProbeStageInput): Promise<ProbeStageR
     if (composition.duration <= 0) reasons.push("root duration unknown");
     if (compiled.unresolvedCompositions.length > 0)
       reasons.push(`${compiled.unresolvedCompositions.length} unresolved composition(s)`);
+    if (hasAutoStart) reasons.push("auto-start video(s)");
     if (hasScriptedAudio) reasons.push("scripted audio volume");
 
     log.info("Launching browser for composition probe...", {