I am using ANSI color codes in the progress bar description string to emphasize success, failure etc.

Unfortunately, the bar length calculation doesn't work as expected with these color codes.

In the example below, the bar stops short of the terminal width by 13 characters, which is the combined length of the ANSI control sequences.

from tqdm import tqdm

ansi = {'BOLD': '\033[1m',
        'RED': '\033[91m',
        'END': '\033[0m'}

text = '{BOLD}{RED}Failed{END}: ...'.format(**ansi)

pbar = tqdm(total=100, desc=text)
pbar.close()

Output (example):

Failed: ...:   0%|                                  | 0/100 [00:00<?, ?it/s]

Expected:

Failed: ...:   0%|                                               | 0/100 [00:00<?, ?it/s]

This should be as easy to fix as stripping the ANSI codes away as in the code below when doing the length calculations here: https://github.com/tqdm/tqdm/blob/master/tqdm/_tqdm.py#L365

import re
re_ansi = re.compile(r"\x1b\[[;\d]*[A-Za-z]")
stripped = re_ansi.sub('', text)
print(len(text))  # 24
print(len(stripped))  # 11

I'd be happy to submit a PR, but I don't know if this is a desired feature. Performance should not be an issue, as the stripping executes in < 1 µs.

 >>> print(tqdm.__version__)
4.24.0
>>> print(sys.version)
3.7.0 (default, Jun 28 2018, 07:39:16) 
[Clang 4.0.1 (tags/RELEASE_401/final)]
>>> print(sys.platform)
darwin