process-video: fps video filter for invalid fps metadata videos

xklb/createdb/av.py

@@ -220,25 +220,6 @@ def munge_av_tags(args, media) -> dict:
 
     streams = probe.streams
 
-    def parse_framerate(string) -> float | None:
-        top, bot = string.split("/")
-        bot = float(bot)
-        if bot == 0:
-            return None
-        return float(top) / bot
-
-    fps = iterables.safe_unpack(
-        [
-            parse_framerate(s.get("avg_frame_rate"))
-            for s in streams
-            if s.get("avg_frame_rate") is not None and "/0" not in s.get("avg_frame_rate")
-        ]
-        + [
-            parse_framerate(s.get("r_frame_rate"))
-            for s in streams
-            if s.get("r_frame_rate") is not None and "/0" not in s.get("r_frame_rate")
-        ],
-    )
     width = iterables.safe_unpack([s.get("width") for s in streams])
     height = iterables.safe_unpack([s.get("height") for s in streams])
 
@@ -279,7 +260,7 @@ def parse_framerate(string) -> float | None:
         "other_count": other_count,
         "width": width,
         "height": height,
-        "fps": fps,
+        "fps": probe.fps,
         "duration": nums.safe_int(duration),
         "language": language,
         "corruption": None if corruption is None else int(corruption * 100),

xklb/mediafiles/process_ffmpeg.py

@@ -164,15 +164,39 @@ def process_path(args, path, **kwargs):
                 ]
             )
 
+            video_filters = []
+            if probe.fps and probe.fps > 240:
+                log.info("fps>240 is not supported by AV1 and it seems suspect. Ignoring metadata %s", path)
+                frames = nums.safe_int(
+                    processes.cmd(
+                        "ffprobe",
+                        "-v",
+                        "fatal",
+                        "-select_streams",
+                        "v:0",
+                        "-count_frames",
+                        "-show_entries",
+                        "stream=nb_read_frames",
+                        "-of",
+                        "default=noprint_wrappers=1:nokey=1",
+                        path,
+                    ).stdout
+                )
+                if frames and probe.duration:
+                    actual_fps = frames / probe.duration
+                    video_filters.append(f"fps={actual_fps}")
+
             width = int(video_stream.get("width"))
             height = int(video_stream.get("height"))
 
             if width > (args.max_width * (1 + args.max_width_buffer)):
-                ff_opts.extend(["-vf", f"scale={args.max_width}:-2"])
+                video_filters.append(f"scale={args.max_width}:-2")
             elif height > (args.max_height * (1 + args.max_height_buffer)):
-                ff_opts.extend(["-vf", f"scale=-2:{args.max_height}"])
+                video_filters.append(f"scale=-2:{args.max_height}")
             else:  # make sure input raster is even for YUV_420 colorspace
-                ff_opts.extend(["-vf", "pad='if(mod(iw,2),iw+1,iw)':'if(mod(ih,2),ih+1,ih)'"])
+                video_filters.append("pad='if(mod(iw,2),iw+1,iw)':'if(mod(ih,2),ih+1,ih)'")
+
+            ff_opts.extend(["-vf", ",".join(video_filters)])
 
     elif album_art_stream:
         ff_opts.extend(["-map", "0:v", "-c:v", "copy"])
@@ -214,7 +238,7 @@ def process_path(args, path, **kwargs):
                 opus_rate = 24000
             else:
                 opus_rate = 16000
-            ff_opts.extend(["-c:a", "libopus", "-ar", str(opus_rate), "-filter:a", "loudnorm=i=-18:tp=-3:lra=17"])
+            ff_opts.extend(["-c:a", "libopus", "-ar", str(opus_rate), "-af", "loudnorm=i=-18:tp=-3:lra=17"])
 
         if is_split:
             try:

xklb/utils/processes.py

@@ -361,6 +361,27 @@ def __init__(self, path, *args):
                 else:
                     self.duration -= start
 
+        self.fps = iterables.safe_unpack(
+            [
+                self.parse_framerate(s.get("avg_frame_rate"))
+                for s in self.streams
+                if s.get("avg_frame_rate") is not None and "/0" not in s.get("avg_frame_rate")
+            ]
+            + [
+                self.parse_framerate(s.get("r_frame_rate"))
+                for s in self.streams
+                if s.get("r_frame_rate") is not None and "/0" not in s.get("r_frame_rate")
+            ],
+        )
+
+    @staticmethod
+    def parse_framerate(string) -> float | None:
+        top, bot = string.split("/")
+        bot = float(bot)
+        if bot == 0:
+            return None
+        return float(top) / bot
+
 
 def unar_out_path(archive_path):
     output_path = str(Path(archive_path).with_suffix(""))