mitus helper

interleave_cht_frames.py

@@ -0,0 +1,120 @@
+#!/usr/bin/env python3
+"""
+Build an enhanced transcript that interleaves audio segments (from a
+whisperx JSON) with screen frames sourced from a cht session's
+frames/index.json. Frames are placed at their real timestamps rather
+than appended at the end.
+
+Usage:
+    python interleave_cht_frames.py \\
+        <transcript.json> <cht_frames_index.json> [output.txt]
+"""
+import json
+import sys
+from pathlib import Path
+
+
+def fmt_ts(seconds: float) -> str:
+    seconds = int(seconds)
+    return f"{seconds // 60:02d}:{seconds % 60:02d}"
+
+
+def load_audio(transcript_json: Path):
+    data = json.loads(transcript_json.read_text())
+    out = []
+    for s in data.get("segments", []):
+        out.append({
+            "type": "audio",
+            "ts": float(s.get("start", 0)),
+            "speaker": s.get("speaker"),
+            "text": (s.get("text") or "").strip(),
+        })
+    return out
+
+
+def load_frames(index_json: Path):
+    data = json.loads(index_json.read_text())
+    out = []
+    for f in data:
+        out.append({
+            "type": "frame",
+            "ts": float(f["timestamp"]),
+            "path": f["path"],
+            "id": f.get("id", ""),
+        })
+    return out
+
+
+def merge(audio, frames):
+    events = sorted(audio + frames, key=lambda e: e["ts"])
+    grouped = []
+    cur = None
+    for e in events:
+        if e["type"] == "frame":
+            if cur:
+                grouped.append(cur)
+                cur = None
+            grouped.append(e)
+        else:
+            if cur and cur.get("speaker") == e["speaker"]:
+                cur["text"] += " " + e["text"]
+            else:
+                if cur:
+                    grouped.append(cur)
+                cur = {
+                    "type": "audio",
+                    "ts": e["ts"],
+                    "speaker": e["speaker"],
+                    "text": e["text"],
+                }
+    if cur:
+        grouped.append(cur)
+    return grouped
+
+
+def render(grouped):
+    lines = [
+        "=" * 80,
+        "ENHANCED MEETING TRANSCRIPT",
+        "Audio transcript + Screen frames (interleaved by timestamp)",
+        "=" * 80,
+        "",
+    ]
+    for e in grouped:
+        ts = fmt_ts(e["ts"])
+        if e["type"] == "frame":
+            lines.append(f"[{ts}] SCREEN CONTENT:")
+            lines.append(f"  Frame: {e['path']}")
+        else:
+            spk = e["speaker"] or "UNKNOWN"
+            lines.append(f"[{ts}] {spk}:")
+            lines.append(f"  {e['text']}")
+        lines.append("")
+    return "\n".join(lines)
+
+
+def main():
+    if len(sys.argv) < 3:
+        print(__doc__, file=sys.stderr)
+        sys.exit(2)
+    transcript = Path(sys.argv[1])
+    frames_index = Path(sys.argv[2])
+    output = Path(sys.argv[3]) if len(sys.argv) >= 4 else None
+
+    audio = load_audio(transcript)
+    frames = load_frames(frames_index)
+    grouped = merge(audio, frames)
+
+    text = render(grouped)
+    if output:
+        output.write_text(text)
+        print(f"Wrote {output}")
+        print(f"  audio segments in : {len(audio)}")
+        print(f"  frames in         : {len(frames)}")
+        print(f"  output blocks     : {len(grouped)}")
+    else:
+        sys.stdout.write(text)
+
+
+if __name__ == "__main__":
+    main()