Use re instead of pyparsing for "cleaning color"

[zrquan]

Description

The overhead of pyparsing is quite high; we can use re to implement the clean_color function.

I ran a simple benchmark using hyperfine.

Benchmark 1: python use_re.py
  Time (mean ± σ):      48.0 ms ±   3.2 ms    [User: 39.0 ms, System: 8.8 ms]
  Range (min … max):    40.7 ms …  56.8 ms    59 runs

Benchmark 2: python use_pyparsing.py
  Time (mean ± σ):      3.542 s ±  0.088 s    [User: 3.521 s, System: 0.014 s]
  Range (min … max):    3.471 s …  3.770 s    10 runs

Summary
  python use_re.py ran
   73.75 ± 5.27 times faster than python use_pyparsing.py

use_re.py

import re

_ansi_escape = re.compile(r"\x1B(?:[@-Z\\-_]|\[[0-?]*[ -/]*[@-~])")

def clean_color(msg):
    return _ansi_escape.sub("", msg)

for _ in range(10000):
    msg = "\x1b[1;31m\x1b[40mthis is a message for testing"
    clean_color(msg)

use_pyparsing.py

import string
from pyparsing import Combine, Literal, Optional, Suppress, Word, delimitedList, oneOf

_escape_seq = Combine(
    Literal("\x1b")
    + "["
    + Optional(delimitedList(Word(string.digits), ";"))
    + oneOf(list(string.ascii_letters))
)

def clean_color(msg):
    return Suppress(_escape_seq).transformString(msg)

for _ in range(10000):
    msg = "\x1b[1;31m\x1b[40mthis is a message for testing"
    clean_color(msg)

Requirements

• Add your name to CONTRIBUTORS.md
• If this is a new feature, then please add some additional information about it to CHANGELOG.md