mtp/vidchk: write ffprobe metadata to file

bin/mtag/vidchk.py

@@ -1,6 +1,7 @@
 #!/usr/bin/env python3
 
 import json
+import re
 import sys
 import subprocess as sp
 
@@ -14,6 +15,7 @@ except:
 
 _ = r"""
 inspects video files for errors and such
+plus stores a bunch of metadata to filename.ff.json
 
 usage:
   -mtp vidchk=t600,ay,p,bin/mtag/vidchk.py
@@ -33,6 +35,29 @@ FAST = True  # parse entire file at container level
 # FAST = False  # fully decode audio and video streams
 
 
+# warnings to ignore
+harmless = re.compile("^Unsupported codec with id ")
+
+
+def wfilter(lines):
+    return [x for x in lines if not harmless.search(x)]
+
+
+def errchk(so, se, rc):
+    if rc:
+        err = (so + se).decode("utf-8", "replace").split("\n", 1)
+        err = wfilter(err) or err
+        return f"ERROR {rc}: {err[0]}"
+
+    if se:
+        err = se.decode("utf-8", "replace").split("\n", 1)
+        err = wfilter(err)
+        if err:
+            return f"Warning: {err[0]}"
+
+    return None
+
+
 def main():
     fp = sys.argv[1]
     zb = sys.stdin.buffer.read()
@@ -46,6 +71,26 @@ def main():
     except:
         return "could not determine resolution"
 
+    # grab streams/format metadata + the final keyframe (and the next 2 seconds of frames after it)
+    zs = "ffprobe -hide_banner -v warning -of json -show_streams -show_format -show_packets -show_data_hash crc32 -read_intervals 999999%+2"
+    cmd = zs.encode("ascii").split(b" ") + [fsenc(fp)]
+    p = sp.Popen(cmd, stdout=sp.PIPE, stderr=sp.PIPE)
+    so, se = p.communicate()
+
+    # spaces to tabs, drops filesize from 69k to 48k
+    so = b"\n".join(
+        [
+            b"\t" * int((len(x) - len(x.lstrip())) / 4) + x.lstrip()
+            for x in (so or b"").split(b"\n")
+        ]
+    )
+    with open(fsenc(f"{fp}.ff.json"), "wb") as f:
+        f.write(so)
+
+    err = errchk(so, se, p.returncode)
+    if err:
+        return err
+
     if min(w, h) < 1080:
         return "resolution too small"
 
@@ -66,16 +111,7 @@ def main():
 
     p = sp.Popen(cmd, stdout=sp.PIPE, stderr=sp.PIPE)
     so, se = p.communicate()
-    rc = p.returncode
-    if rc:
-        err = (so + se).decode("utf-8", "replace").split("\n", 1)[0]
-        return f"ERROR {rc}: {err}"
-
-    if se:
-        err = se.decode("utf-8", "replace").split("\n", 1)[0]
-        return f"Warning: {err}"
-
-    return None
+    return errchk(so, se, p.returncode)
 
 
 if __name__ == "__main__":