fix: honor audio_as_voice streaming + parse tests (#490) (thanks @jarvis-medmatic)

src/media/parse.test.ts

@@ -0,0 +1,19 @@
+import { describe, expect, it } from "vitest";
+
+import { splitMediaFromOutput } from "./parse.js";
+
+describe("splitMediaFromOutput", () => {
+  it("detects audio_as_voice tag and strips it", () => {
+    const result = splitMediaFromOutput("Hello [[audio_as_voice]] world");
+    expect(result.audioAsVoice).toBe(true);
+    expect(result.text).toBe("Hello world");
+  });
+
+  it("keeps audio_as_voice detection stable across calls", () => {
+    const input = "Hello [[audio_as_voice]]";
+    const first = splitMediaFromOutput(input);
+    const second = splitMediaFromOutput(input);
+    expect(first.audioAsVoice).toBe(true);
+    expect(second.audioAsVoice).toBe(true);
+  });
+});

src/media/parse.ts

@@ -34,6 +34,7 @@ function isInsideFence(
 
 // Regex to detect [[audio_as_voice]] tag
 const AUDIO_AS_VOICE_RE = /\[\[audio_as_voice\]\]/gi;
+const AUDIO_AS_VOICE_TEST_RE = /\[\[audio_as_voice\]\]/i;
 
 export function splitMediaFromOutput(raw: string): {
   text: string;
@@ -123,7 +124,7 @@ export function splitMediaFromOutput(raw: string): {
     .trim();
 
   // Detect and strip [[audio_as_voice]] tag
-  const hasAudioAsVoice = AUDIO_AS_VOICE_RE.test(cleanedText);
+  const hasAudioAsVoice = AUDIO_AS_VOICE_TEST_RE.test(cleanedText);
   if (hasAudioAsVoice) {
     cleanedText = cleanedText
       .replace(AUDIO_AS_VOICE_RE, "")