Enhance Avisynth video workflow and metadata tools

tools/hdr10plus.py

@@ -0,0 +1,114 @@
+from __future__ import annotations
+
+import json
+import subprocess
+from pathlib import Path
+from typing import Any
+
+
+def extract_hdr10plus_metadata(
+    *,
+    ffmpeg: Path,
+    hdr10plus_tool: Path,
+    source: Path,
+    output: Path,
+) -> None:
+    """Extract HDR10+ JSON without creating a full intermediate video file."""
+    output.parent.mkdir(parents=True, exist_ok=True)
+    ffmpeg_process = subprocess.Popen(
+        [
+            str(ffmpeg),
+            "-v", "error", "-i", str(source), "-map", "0:v:0",
+            "-c", "copy", "-bsf:v", "hevc_mp4toannexb", "-f", "hevc", "-",
+        ],
+        stdout=subprocess.PIPE,
+        stderr=subprocess.PIPE,
+    )
+    assert ffmpeg_process.stdout is not None
+    try:
+        extracted = subprocess.run(
+            [str(hdr10plus_tool), "extract", "-o", str(output), "-"],
+            stdin=ffmpeg_process.stdout,
+            capture_output=True,
+            text=True,
+        )
+    finally:
+        ffmpeg_process.stdout.close()
+    ffmpeg_stderr = ffmpeg_process.stderr.read().decode(errors="replace")
+    ffmpeg_process.stderr.close()
+    ffmpeg_returncode = ffmpeg_process.wait()
+    if ffmpeg_returncode != 0:
+        raise RuntimeError(
+            f"FFmpeg could not extract the HEVC stream from {source}:\n{ffmpeg_stderr.strip()}"
+        )
+    if extracted.returncode != 0:
+        raise RuntimeError(
+            f"hdr10plus_tool could not extract metadata from {source}:\n"
+            f"{extracted.stderr.strip() or extracted.stdout.strip()}"
+        )
+    if not output.is_file():
+        raise RuntimeError(f"hdr10plus_tool did not create metadata JSON for {source}")
+
+
+def select_hdr10plus_frames(
+    *,
+    source_json: Path,
+    source_frames: list[int],
+    output_json: Path,
+) -> None:
+    """Keep metadata for the encoded source frames and rebuild scene indices."""
+    payload = json.loads(source_json.read_text(encoding="utf-8"))
+    entries = payload.get("SceneInfo")
+    if not isinstance(entries, list) or not entries:
+        raise RuntimeError(f"HDR10+ JSON has no SceneInfo entries: {source_json}")
+    if not source_frames:
+        raise RuntimeError("Cannot create HDR10+ metadata for an empty output")
+    if min(source_frames) < 0 or max(source_frames) >= len(entries):
+        raise RuntimeError(
+            "HDR10+ metadata frame count does not cover the validated source frames "
+            f"({len(entries)} metadata entries, highest source frame {max(source_frames)})"
+        )
+
+    selected = [dict(entries[index]) for index in source_frames]
+    _reindex_scenes(selected)
+    payload["SceneInfo"] = selected
+    payload["SceneInfoSummary"] = _scene_summary(selected)
+    output_json.write_text(json.dumps(payload, indent=2) + "\n", encoding="utf-8")
+
+
+def _reindex_scenes(entries: list[dict[str, Any]]) -> None:
+    scene_id = 0
+    scene_frame_index = 0
+    previous: dict[str, Any] | None = None
+    for sequence_frame_index, entry in enumerate(entries):
+        if previous is not None and _scene_payload(entry) != _scene_payload(previous):
+            scene_id += 1
+            scene_frame_index = 0
+        entry["SceneFrameIndex"] = scene_frame_index
+        entry["SceneId"] = scene_id
+        entry["SequenceFrameIndex"] = sequence_frame_index
+        scene_frame_index += 1
+        previous = entry
+
+
+def _scene_payload(entry: dict[str, Any]) -> dict[str, Any]:
+    return {
+        key: value
+        for key, value in entry.items()
+        if key not in {"SceneFrameIndex", "SceneId", "SequenceFrameIndex"}
+    }
+
+
+def _scene_summary(entries: list[dict[str, Any]]) -> dict[str, list[int]]:
+    starts = [
+        index
+        for index, entry in enumerate(entries)
+        if entry.get("SceneFrameIndex") == 0
+    ]
+    return {
+        "SceneFirstFrameIndex": starts,
+        "SceneFrameNumbers": [
+            next_start - start
+            for start, next_start in zip(starts, [*starts[1:], len(entries)])
+        ],
+    }