feat: add ffprobe side_data parsing for HDR/SDR detection

data/tracker.csv

@@ -1,8 +1,4 @@
-ff,2
-simple_ffprobe_script,1
-show_path,2
-append,15
-deappend,17
-rename_to_first_n,3
-count_files,4
-remove_fname_last_n,2
+git-honney333-work-cred,794
+ff,53
+deappend,8
+rename_to_first_n,1

ff

@@ -264,6 +264,12 @@ class audio_lines:
         
         if self.language:
             string += f" [{self.language}]"
+        if len(self.language) >= 14:
+            pass
+        elif len(self.language) >= 8:
+            string += "\t"
+        else:
+            string += "\t\t"
             
         if self.duration:
             string += f" {self.duration}s"
@@ -318,6 +324,13 @@ class subtitles:
         if self.language:
             parts.append(f"[{self.language}]")
 
+        if len(self.language) >= 14:
+            pass
+        elif len(self.language) >= 7:
+            parts.append("\t")
+        else:
+            parts.append("\t\t")
+
         if self.duration:
             parts.append(f"{self.duration}s")
             
@@ -384,6 +397,212 @@ class subtitles:
 
 #     return duration, video_streams, audio_streams, subtitle_streams
 
+def _side_data_hints(side_data_list):
+    """Return set of hints from ffprobe side_data_list (case-insensitive)."""
+    hints = set()
+    for sd in side_data_list or []:
+        try:
+            t = str(sd.get("side_data_type", "")).lower()
+        except Exception:
+            continue
+        if "dovi" in t or "dolby" in t or "rpu" in t:
+            hints.add("DV")
+        if "2094-40" in t or "hdr dynamic" in t or "hdr10+" in t:
+            hints.add("HDR10+")
+        if "mastering" in t or "content light" in t or "content_light" in t or "cll" in t:
+            hints.add("HDR-static")
+    return hints
+
+
+def _classify_fast(video):
+    """Tier 0: pure-python, zero extra subprocess.
+
+    Returns (label, needs_deep). Final labels need no further probing.
+    UNKNOWN with needs_deep=True means Tier 1/2 should run.
+    Uses SDR? for highly-probable-but-untagged SDR, HDR? for probable HDR.
+    """
+    transfer = str(video.get("color_transfer") or "").lower()
+    primaries = str(video.get("color_primaries") or "").lower()
+    space = str(video.get("color_space") or "").lower()
+    pix_fmt = str(video.get("pix_fmt") or "").lower()
+    codec = str(video.get("codec_name") or "").lower()
+    profile = str(video.get("profile") or "").lower()
+    bits = video.get("bits_per_raw_sample")
+    try:
+        bits = int(bits) if bits is not None else 0
+    except Exception:
+        bits = 0
+
+    hints = _side_data_hints(video.get("side_data_list"))
+
+    # Cover art / thumbnails are not real video tracks — never deep-probe them.
+    dispo = video.get("disposition") or {}
+    try:
+        if dispo.get("attached_pic") == 1 or codec in ("mjpeg", "png", "gif", "bmp", "tiff"):
+            return ("SDR", False)
+    except Exception:
+        pass
+
+    # 1. Definitive transfer function
+    if transfer == "smpte2084":
+        if "DV" in hints:
+            return ("DV", False)
+        if "HDR10+" in hints:
+            return ("HDR10+", False)
+        return ("HDR10", False)
+    if transfer == "arib-std-b67":
+        return ("HLG", False)
+    sdr_transfers = {
+        "bt709", "bt470m", "bt470bg", "smpte170m", "smpte240m",
+        "gamma22", "gamma28", "iec61966-2-1", "iec61966-2-4", "srgb",
+    }
+    if transfer in sdr_transfers:
+        return ("SDR", False)
+
+    # 2. Primaries / matrix when transfer was stripped
+    is_bt2020 = primaries == "bt2020" or space in ("bt2020nc", "bt2020c", "bt2020_ncl", "bt2020_cl")
+    is_bt709 = primaries == "bt709" or space == "bt709"
+    if is_bt2020:
+        if "DV" in hints:
+            return ("DV", False)
+        if "HDR10+" in hints:
+            return ("HDR10+", False)
+        # Transfer missing so HDR10 vs HLG unknown — probable HDR.
+        return ("HDR?", False)
+    if is_bt709:
+        return ("SDR", False)
+    if "HDR-static" in hints or "DV" in hints or "HDR10+" in hints:
+        if "DV" in hints:
+            return ("DV", False)
+        if "HDR10+" in hints:
+            return ("HDR10+", False)
+        return ("HDR?", False)
+
+    # 3. All color tags empty — fall back to bit-depth / codec heuristics.
+    is_10bit_plus = ("p10" in pix_fmt or "p12" in pix_fmt or "p14" in pix_fmt
+                     or "p16" in pix_fmt or bits >= 10 or "10" in profile
+                     or "12" in profile)
+    if not transfer and not primaries and not space:
+        if is_10bit_plus:
+            # 10-bit SDR (anime etc.) and stripped HDR look identical here.
+            return ("UNKNOWN", True)
+        # 8-bit + common SDR codecs = highly probable untagged SDR.
+        return ("SDR?", False)
+
+    # Partial tags, unknown values — probable SDR, no expensive probe.
+    if not is_10bit_plus:
+        return ("SDR?", False)
+    return ("UNKNOWN", True)
+
+
+def _classify_colors(transfer, primaries, space, hints):
+    """Shared helper for Tier 1/2 dicts. Returns label or None."""
+    transfer = str(transfer or "").lower()
+    primaries = str(primaries or "").lower()
+    space = str(space or "").lower()
+    if transfer == "smpte2084":
+        if "DV" in hints:
+            return "DV"
+        if "HDR10+" in hints:
+            return "HDR10+"
+        return "HDR10"
+    if transfer == "arib-std-b67":
+        return "HLG"
+    if transfer in ("bt709", "smpte170m", "smpte240m", "gamma22", "gamma28",
+                    "iec61966-2-1", "iec61966-2-4", "srgb", "bt470m", "bt470bg"):
+        return "SDR"
+    if primaries == "bt2020" or space in ("bt2020nc", "bt2020c", "bt2020_ncl", "bt2020_cl"):
+        if "DV" in hints:
+            return "DV"
+        if "HDR10+" in hints:
+            return "HDR10+"
+        return "HDR?"
+    if primaries == "bt709" or space == "bt709":
+        return "SDR"
+    if "DV" in hints:
+        return "DV"
+    if "HDR10+" in hints:
+        return "HDR10+"
+    if "HDR-static" in hints:
+        return "HDR?"
+    return None
+
+
+def _probe_frame_tier1(path, v_idx):
+    """Tier 1 (cheap): 1-frame + full stream dump for one v stream. No decode."""
+    try:
+        cmd = [
+            "ffprobe", "-v", "error",
+            "-select_streams", f"v:{v_idx}",
+            "-show_streams",
+            "-show_frames", "-read_intervals", "%+#1",
+            "-show_entries", "frame=color_space,color_transfer,color_primaries,pix_fmt,side_data_list",
+            "-of", "json",
+            path,
+        ]
+        r = subprocess.run(cmd, capture_output=True, text=True, timeout=10)
+        if r.returncode != 0:
+            return None
+        info = json.loads(r.stdout or "{}")
+        streams = info.get("streams") or []
+        frames = info.get("frames") or []
+        s0 = streams[0] if streams else {}
+        f0 = frames[0] if frames else {}
+        hints = _side_data_hints(s0.get("side_data_list")) | _side_data_hints(f0.get("side_data_list"))
+        # Prefer stream tags, fall back to frame tags per field.
+        label = _classify_colors(
+            s0.get("color_transfer") or f0.get("color_transfer"),
+            s0.get("color_primaries") or f0.get("color_primaries"),
+            s0.get("color_space") or f0.get("color_space"),
+            hints,
+        )
+        return label
+    except Exception:
+        return None
+
+
+def _probe_trace_tier2(path):
+    """Tier 2 (expensive): bitstream VUI/SEI via trace_headers, 1 frame, no decode."""
+    try:
+        import shutil
+        if shutil.which("ffmpeg") is None:
+            return None
+        cmd = [
+            "ffmpeg", "-hide_banner", "-v", "info",
+            "-i", path,
+            "-c:v", "copy", "-bsf:v", "trace_headers",
+            "-frames:v", "1", "-f", "null", "-",
+        ]
+        r = subprocess.run(cmd, capture_output=True, text=True, timeout=15)
+        out = ((r.stderr or "") + "\n" + (r.stdout or "")).lower()
+        if not out.strip():
+            return None
+        has_dv = "dovi" in out or "rpu" in out
+        has_hdr10plus = "2094-40" in out or "hdr10+" in out or "dhdr10" in out
+        has_static = "mastering" in out or "content_light" in out or "max_cll" in out
+        # VUI transfer_characteristics: 16=PQ(ST2084), 18=HLG; primaries 9=BT.2020
+        import re
+        m = re.search(r"transfer_characteristics\s*[:=]\s*(\d+)", out)
+        trc = m.group(1) if m else ""
+        m2 = re.search(r"colour_primaries\s*[:=]\s*(\d+)|color_primaries\s*[:=]\s*(\d+)", out)
+        prim = (m2.group(1) or m2.group(2)) if m2 else ""
+        if has_dv:
+            return "DV"
+        if has_hdr10plus:
+            return "HDR10+"
+        if trc == "16":
+            return "HDR10"
+        if trc == "18":
+            return "HLG"
+        if prim == "9" or has_static:
+            return "HDR?"
+        if trc in ("1", "6", "7"):
+            return "SDR"
+        return None
+    except Exception:
+        return None
+
+
 def get_media_info(file):
     cmd = [
         "ffprobe",
@@ -391,12 +610,12 @@ def get_media_info(file):
         "-show_entries",
         (
             "format=duration:format_tags=title:"
-            "stream=index,codec_type,codec_name,"
+            "stream=index,codec_type,codec_name,profile,"
             "width,height,r_frame_rate,bit_rate,duration,nb_frames,"
             "pix_fmt,field_order,time_base,display_aspect_ratio,"
             "color_space,color_transfer,color_primaries,bits_per_raw_sample,"
             "sample_rate,channels,bits_per_sample,"
-            "stream_disposition=forced,default:"
+            "stream_disposition=forced,default,attached_pic:"
             "stream_tags=language,title"
         ),
         "-of", "json",
@@ -408,14 +627,14 @@ def get_media_info(file):
         info = json.loads(result.stdout)
     except subprocess.CalledProcessError as e:
         print(f"Error running ffprobe: {e.stderr}")
-        return None, [], [], [], None, "UNKNOWN"
+        return None, [], [], [], None, []
 
     # Duration
     duration = None
     if "format" in info:
         duration = info["format"].get("duration")
 
-    # ✅ Extract container title
+    # ✅  Extract container title
     title = None
     if "format" in info:
         title = info["format"].get("tags", {}).get("title")
@@ -433,27 +652,28 @@ def get_media_info(file):
         elif stream_type == "subtitle":
             subtitle_streams.append(stream)
 
-    # Detect HDR type from the first video stream
+    # Tiered HDR/SDR detection: Tier 0 fast path first, expensive only on UNKNOWN.
     hdr_types = []
 
     if video_streams:
-        for video in video_streams:
-            color_transfer = video.get("color_transfer")
-            color_primaries = video.get("color_primaries")
-
-            if color_transfer == "smpte2084":
-                hdr_types.append("HDR10")
-
-            elif color_transfer == "arib-std-b67":
-                hdr_types.append("HLG")
-
-            elif (
-                color_transfer == "bt709"
-                and color_primaries == "bt709"
-            ):
-                hdr_types.append("SDR")
-            else:
-                hdr_types.append("Unknown HDR")
+        trace_cache = {}  # file-level Tier 2 result, run at most once per file
+        for v_pos, video in enumerate(video_streams):
+            label, needs_deep = _classify_fast(video)
+            if not needs_deep:
+                hdr_types.append(label)
+                continue
+            # Tier 1: 1-frame + stream dump (cheap, no decode)
+            deep = _probe_frame_tier1(file, v_pos)
+            if deep:
+                hdr_types.append(deep)
+                continue
+            # Tier 2: bitstream trace_headers (expensive, no decode)
+            if "trace" not in trace_cache:
+                trace_cache["trace"] = _probe_trace_tier2(file)
+            if trace_cache["trace"]:
+                hdr_types.append(trace_cache["trace"])
+                continue
+            hdr_types.append("UNKNOWN")
 
     if duration is not None:
         duration = float(duration)
@@ -531,10 +751,16 @@ class video_file:
                 np(f"\t{self.name} ({self.size}, {self.duration}, {self.bitrate} MB/s):", NORMAL_STYLE)
             if self.title:
                 np(f"\t\tTitle: \"{self.title}\"", NORMAL_STYLE)
+            if self.videos:
+                np(f"\t    Videos:", INFO_STYLE)
             for video in self.videos:
                 np(f"\t\t{video}", NORMAL_STYLE)
+            if self.audios:
+                np(f"\t    Audios:", INFO_STYLE)
             for audio in self.audios:
                 np(f"\t\t{audio}", NORMAL_STYLE)
+            if self.subtitles:
+                np(f"\t    Subtitles:", INFO_STYLE)
             for subtitle in self.subtitles:
                 np(f"\t\t{subtitle}", NORMAL_STYLE)
             print()