"TypeError: byte indices must be integers or slices, not str" Converting bytes to ints

Question

I am using a different program (ffmpeg) to grab the length of a youtube video that was downloaded in order to randomize a specific point in the video. However I am getting this error when I am trying to execute this code:

def grabTimeOfDownloadedYoutubeVideo(youtubeVideo):
    process = subprocess.Popen(['/usr/local/bin/ffmpeg', '-i', youtubeVideo], stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
    stdout, stderr = process.communicate()
    matches = str(re.search(b"Duration:\s{1}(?P<hours>\d+?):(?P<minutes>\d+?):(?P<seconds>\d+\.\d+?),", stdout, re.DOTALL).groupdict()).encode()
    print(matches)
    hours = int(matches['hours'])
    minutes = int(matches['minutes'])
    seconds = int(matches['seconds'])
    total = 0
    total += 60 * 60 * hours
    total += 60 * minutes
    total += seconds
    print(total)

The matches variable prints out to this:

b"{'minutes': b'04', 'hours': b'00', 'seconds': b'24.94'}"

So all of the output comes out with a 'b' at the beginning of it. How do I remove the 'b' and just get the number?

Full error message here:

Traceback (most recent call last):
  File "bot.py", line 87, in <module>
    grabTimeOfDownloadedYoutubeVideo("videos/1.mp4")
  File "bot.py", line 77, in grabTimeOfDownloadedYoutubeVideo
    hours = int(matches['hours'])
TypeError: byte indices must be integers or slices, not str

Solution 1

matches = str(re.search(b"Duration:\s{1}(?P<hours>\d+?):(?P<minutes>\d+?):(?P<seconds>\d+\.\d+?),", stdout, re.DOTALL).groupdict()).encode()

is weird. By converting the result of the regex match to a string, you're causing the error (because now matches['hours'] will fail).

By encoding that string to a bytes object (why?), you're complicating things even further.

matches = re.search(r"Duration:\s(?P<hours>\d+?):(?P<minutes>\d+?):(?P<seconds>\d+\.\d+?),", stdout).groupdict()

should do (although I'm uncertain about using stdout as input...)

Solution 2

It seems you have a byte object there. In order to use it you can do the following**:

Decode it:

matches = matches.decode("utf-8")

Then, by using ast.literal_eval, translate the str to what it truly is, a dict:

matches = ast.literal_eval(matches)

Then you can access the contents of matches as you normally would:

int(matches['hours']) # returns 0

**Of course this simply fixes an error that really shouldn't be here in the first place as @Tim points out.

Comments

[Tim Pietzcker]

Isn't that completely backwards? He created that bytes object himself for no obvious reason.

[Dimitris Fasarakis Hilliard]

Sure, I'm just filling this in because the title indicates an error that will always be able to get fixed by following the steps I suggested.

[Lewis Menelaws]

Thanks for the reply! I actually converted it to a bytes object reading this answer stackoverflow.com/questions/5184483/python-typeerror-on-rege‌​x It was the answer that somehow got me to get past that error and thought it may be an easier solution to what im asking

[Lewis Menelaws]

Whenever I use what you used I get another TypeError TypeError: cannot use a string pattern on a bytes-like object

[Tim Pietzcker]

That's probably because you're accessing stdout. You need to .decode() that using the correct encoding (which depends on your platform). If you're lucky, using str(stdout) instead of stdout will already work.

[Tim Pietzcker]

Even better, call Popen using the universal_newlines=True option. That way, you get strings instead of bytes objects, already converted using the system encoding.

[Lewis Menelaws]

Thank you for the help Tim.