fixed bugs

sentence_api/processing.py

@@ -1,10 +1,11 @@
+import json
 import logging
 import shutil
 import subprocess
 import tempfile
 import uuid
 from pathlib import Path
-from typing import Any, Dict, List, Optional
+from typing import Any, Dict, List, Optional, Tuple
 
 from .audio_metrics import (
     AudioAnalysisError,
@@ -85,6 +86,126 @@ def _ensure_webm_cues_front(filepath: Path) -> None:
         tmp.unlink(missing_ok=True)
 
 
+_IOS_VIDEO_CODECS = {"h264", "hevc"}
+_IOS_AUDIO_CODECS = {"aac", "mp3", "ac3", "eac3", "alac"}
+
+
+def _probe_stream_codecs(filepath: Path) -> Tuple[Optional[str], Optional[str]]:
+    ffprobe = shutil.which("ffprobe")
+    if ffprobe is None:
+        return None, None
+    completed = subprocess.run(
+        [
+            ffprobe,
+            "-hide_banner",
+            "-loglevel",
+            "error",
+            "-show_streams",
+            "-of",
+            "json",
+            str(filepath),
+        ],
+        capture_output=True,
+        text=True,
+        timeout=60,
+    )
+    if completed.returncode != 0:
+        return None, None
+    try:
+        payload = json.loads(completed.stdout or "{}")
+    except json.JSONDecodeError:
+        return None, None
+    video_codec = None
+    audio_codec = None
+    for stream in payload.get("streams") or []:
+        codec_type = stream.get("codec_type")
+        codec_name = stream.get("codec_name")
+        if codec_type == "video" and video_codec is None and codec_name not in {"mjpeg", "png"}:
+            video_codec = codec_name
+        elif codec_type == "audio" and audio_codec is None:
+            audio_codec = codec_name
+    return video_codec, audio_codec
+
+
+def _needs_ios_compatible_transcode(video_codec: Optional[str], audio_codec: Optional[str]) -> bool:
+    if video_codec is None and audio_codec is None:
+        return False
+    video_ok = video_codec is None or video_codec in _IOS_VIDEO_CODECS
+    audio_ok = audio_codec is None or audio_codec in _IOS_AUDIO_CODECS
+    return not (video_ok and audio_ok)
+
+
+def _ensure_ios_compatible_media(filepath: Path) -> None:
+    """Transcode AV1/VP9/Opus etc. to H.264/AAC so iOS AVPlayer can show video."""
+    video_codec, audio_codec = _probe_stream_codecs(filepath)
+    if not _needs_ios_compatible_transcode(video_codec, audio_codec):
+        return
+    ffmpeg = shutil.which("ffmpeg")
+    if ffmpeg is None:
+        logger.warning(
+            "ffmpeg not found; skipping iOS-compatible transcode for %s (%s/%s)",
+            filepath.name,
+            video_codec,
+            audio_codec,
+        )
+        return
+    tmp = filepath.with_suffix(".ios_tmp.mp4")
+    command = [
+        ffmpeg,
+        "-hide_banner",
+        "-loglevel",
+        "error",
+        "-y",
+        "-i",
+        str(filepath),
+        "-map",
+        "0:v:0",
+        "-map",
+        "0:a:0?",
+        "-vf",
+        "scale='min(1920,iw)':-2",
+        "-c:v",
+        "libx264",
+        "-preset",
+        "veryfast",
+        "-crf",
+        "23",
+        "-pix_fmt",
+        "yuv420p",
+        "-c:a",
+        "aac",
+        "-b:a",
+        "128k",
+        "-ac",
+        "2",
+        "-movflags",
+        "+faststart",
+        str(tmp),
+    ]
+    logger.info(
+        "Transcoding %s for iOS playback (%s/%s -> h264/aac)",
+        filepath.name,
+        video_codec,
+        audio_codec,
+    )
+    try:
+        completed = subprocess.run(command, capture_output=True, text=True, timeout=7200)
+        if completed.returncode == 0 and tmp.is_file() and tmp.stat().st_size > 0:
+            tmp.replace(filepath)
+            logger.info("iOS-compatible transcode finished: %s", filepath.name)
+            return
+        logger.warning(
+            "iOS-compatible transcode failed for %s: rc=%d stderr=%s",
+            filepath.name,
+            completed.returncode,
+            (completed.stderr or "")[-400:],
+        )
+    except (subprocess.TimeoutExpired, OSError) as exc:
+        logger.warning("iOS-compatible transcode error for %s: %s", filepath.name, exc)
+    finally:
+        tmp.unlink(missing_ok=True)
+
+
 class VideoProcessor:
     def __init__(
         self,
@@ -110,6 +231,7 @@ class VideoProcessor:
         work_path: Optional[Path] = None
         try:
             _ensure_webm_cues_front(media_path)
+            _ensure_ios_compatible_media(media_path)
             if self.transcriber.available:
                 work_path = self.settings.work_dir / f"{video_hash}-{uuid.uuid4().hex}.wav"
                 extract_audio(media_path, work_path)