Ffmpeg-python: Obtain stream information

[elaye]

Hello,
Is there a way to get the size (width and height) of an input video stream?
Thanks for the good work!

[Depau]

ffmpeg-python does not do anything with the input files you specify. In fact it doesn't even check if they exist or are valid (try ffmpeg.input("file_that_does_not_exist.doc")).

If you want to retrieve information about some video you can run ffprobe and parse its output, i.e.:

$ ffprobe -hide_banner some_file.mp4
Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'some_file.mp4':
  Metadata:
    major_brand     : isom
    minor_version   : 512
    compatible_brands: isomiso2avc1mp41
    encoder         : Lavf57.71.100
  Duration: 00:03:00.07, start: 0.000000, bitrate: 373 kb/s
    Stream #0:0(eng): Video: h264 (High) (avc1 / 0x31637661), yuv420p, 640x360 [SAR 1:1 DAR 16:9], 237 kb/s, 25 fps, 25 tbr, 12800 tbn, 50 tbc (default)
    Metadata:
      handler_name    : VideoHandler
    Stream #0:1(und): Audio: aac (LC) (mp4a / 0x6134706D), 44100 Hz, stereo, fltp, 129 kb/s (default)
    Metadata:
      handler_name    : SoundHandler

Then you can use re to parse the output, for example:

import re, subprocess

o = subprocess.check_output(["ffprobe", "-hide_banner", "some_file.mp4"])
match = re.search(r"Video:.+?(\d+x\d+) \[SAR.+", o)
v_size = match.group(1)

```python

print(v_size)
640x360

Please note that the regex may need to be adapted if the video has multiple streams: the current one will only match the resolution of the first video.

These are some regular expressions that I use in one of my projects to extract some information, which can then be accessed with `match.groupdict()` (they still match the first stream only):

```python
mediaprobe_re = re.compile(
    r"Duration:\s+(?P<dur>(?:(?:\d:?)+[.]?\d*)|N/A)(?:.+start:\s+(?P<start>\d+[.]\d+))?.+bitrate:\s+(?P<bitrate>(?:\d+\s*..[/]s)|N/A)")
streamprobe_re = re.compile(
    r"\s*Stream.+:\s+Video:.+\s+(?P<res>\d+x\d+)(?:.*,\s*(?P<fps>\d+[.]?\d*)\sfps)?(?:.+\(default\))?")
audioprobe_re = re.compile(
    r"\s*Stream.+:\s+Audio:.*")
fftime_re = re.compile(
    r"(?P<h>\d+):(?P<m>\d+):(?P<s>\d+)\.(?P<fract>\d+)")

Note that you must do the parsing on your own: ffmpeg-python does not support this.
However, this may be an idea for a new feature; @kkroening what do you think?

[elaye]

Ok, thanks a lot for the detailed answer. I'll look into that.

[kkroening]

It'd be nice to be able to call ffprobe from python. No specific plans to implement it, but wouldn't be opposed to it. I can see lots of cases where it'd be useful though for sure.

[kkroening]

Here's an example of calling ffprobe on the extra-examples2 branch: jffprobe.py.

It's a command-line tool that works like this:

python jffprobe.py *.mp3

It produces a JSON representation on stdout of the ffprobe data for all the filenames listed and any error/warning messages on stderr.

But you could probably take the run_ffprobe function and put it in another script. I might add this into the core of ffmpeg-python at some point, but again no specific plans as of this moment.

[kkroening]

Actually I decided to add an ffmpeg.probe function in #67.

Example to print duration of a media file:

import ffmpeg
print(ffmpeg.probe('in.mp4')['format']['duration']))

This will go out in the next release.

[elaye]

That's awesome! I didn't know ffprobe could output json, that's quite handy.

[BBI-YggyKing]

For completeness, here's how to use the ffmpeg.probe function to get width and height of a video:

probe = ffmpeg.probe(args.file)
video = next((stream for stream in probe['streams'] if stream['codec_type'] == 'video'), None)
width = int(video['width'])
height = int(video['height'])