doocus first ver

doocus/extractors/media.py

@@ -0,0 +1,79 @@
+"""
+Media-family extractor: mp4.
+
+doocus does not transcribe here. It probes container metadata via ffprobe, grabs
+a thumbnail via ffmpeg, and records a pointer delegating full transcription to
+the meetus pipeline (process_meeting.py). Meeting videos are thus discoverable
+and previewable in doocus while their transcript lives in meetus.
+"""
+import json
+import shutil
+import subprocess
+import tempfile
+from pathlib import Path
+
+from .base import ExtractionResult
+
+EXTRACTOR_ID = "media/ffprobe@1"
+
+
+def _ffprobe(source: Path) -> dict:
+    if not shutil.which("ffprobe"):
+        return {}
+    cmd = ["ffprobe", "-v", "quiet", "-print_format", "json",
+           "-show_format", "-show_streams", str(source)]
+    try:
+        out = subprocess.run(cmd, capture_output=True, text=True, check=True).stdout
+        return json.loads(out)
+    except (subprocess.CalledProcessError, json.JSONDecodeError):
+        return {}
+
+
+def _thumbnail(source: Path, at_seconds: float = 1.0) -> bytes:
+    if not shutil.which("ffmpeg"):
+        return b""
+    with tempfile.NamedTemporaryFile(suffix=".jpg", delete=False) as tmp:
+        tmp_path = Path(tmp.name)
+    try:
+        cmd = ["ffmpeg", "-v", "quiet", "-y", "-ss", str(at_seconds),
+               "-i", str(source), "-frames:v", "1", "-vf", "scale=640:-1",
+               str(tmp_path)]
+        subprocess.run(cmd, check=True)
+        return tmp_path.read_bytes() if tmp_path.exists() else b""
+    except subprocess.CalledProcessError:
+        return b""
+    finally:
+        tmp_path.unlink(missing_ok=True)
+
+
+def extract(source: Path, options: dict) -> ExtractionResult:
+    warnings = []
+    probe = _ffprobe(source)
+    metadata = {"delegate": "meetus", "delegate_hint": "run process_meeting.py for transcript"}
+
+    fmt = probe.get("format", {})
+    if fmt:
+        if "duration" in fmt:
+            metadata["duration_seconds"] = float(fmt["duration"])
+        tags = fmt.get("tags", {})
+        if "creation_time" in tags:
+            metadata["created"] = tags["creation_time"]
+    video_streams = [s for s in probe.get("streams", []) if s.get("codec_type") == "video"]
+    if video_streams:
+        v = video_streams[0]
+        metadata["width"], metadata["height"] = v.get("width"), v.get("height")
+    if not probe:
+        warnings.append("ffprobe unavailable or failed; metadata limited")
+
+    thumb = _thumbnail(source) or None
+    if thumb is None:
+        warnings.append("thumbnail not generated (ffmpeg unavailable or failed)")
+
+    dur = metadata.get("duration_seconds")
+    content = (f"_Video {source.name}"
+               + (f", {dur:.0f}s" if dur else "")
+               + ". Transcript is produced by the meetus pipeline "
+                 "(`process_meeting.py`), not doocus._\n")
+
+    return ExtractionResult(content=content, metadata=metadata, thumb=thumb,
+                            warnings=warnings, extractor=EXTRACTOR_ID)