Optimize ``textwrap.indent`` a bit more.

[picnixz]

The optimization introduced in #107374 can be improved as follows:

Current code

if predicate is None:
 predicate = lambda s: not s.isspace()

 prefixed_lines = []
 for line in text.splitlines(True):
 if predicate(line):
 prefixed_lines.append(prefix)
 prefixed_lines.append(line)

Improved code

prefixed_lines = []

if predicate is None:
 for line in text.splitlines(True):
 if not line.isspace():
 prefixed_lines.append(prefix)
 prefixed_lines.append(line)
else:
 for line in text.splitlines(True):
 if predicate(line):
 prefixed_lines.append(prefix)
 prefixed_lines.append(line)

Benchmarks

import timeit
import textwrap

with open("Objects/unicodeobject.c") as f:
 text = f.read()

print(f"indent {len(text.splitlines())} lines.")

it = timeit.Timer(lambda: textwrap.indent(text, ' ' * 4))
result = it.repeat(number=1000)
result.sort()
print(f"{result[0]:.4f}msec")

• Current: 7.9305msec
• After: 6.7143msec

However the results are very different between two runs (and my laptop may be dying) so I'd like some confirmation from others.

Linked PRs

• gh-107424: avoid using lambda functions in textwrap.indent() #107426

[hugovk]

cc @methane