llvm-project/llvm/tools/opt-viewer/optpmap.py

57 lines
1.7 KiB
Python

import sys
import multiprocessing
_current = None
_total = None
def _init(current, total):
global _current
global _total
_current = current
_total = total
def _wrapped_func(func_and_args):
func, argument, should_print_progress = func_and_args
if should_print_progress:
with _current.get_lock():
_current.value += 1
sys.stdout.write('\r\t{} of {}'.format(_current.value, _total.value))
sys.stdout.flush()
return func(argument)
def pmap(func, iterable, processes, should_print_progress, *args, **kwargs):
"""
A parallel map function that reports on its progress.
Applies `func` to every item of `iterable` and return a list of the
results. If `processes` is greater than one, a process pool is used to run
the functions in parallel. `should_print_progress` is a boolean value that
indicates whether a string 'N of M' should be printed to indicate how many
of the functions have finished being run.
"""
global _current
global _total
_current = multiprocessing.Value('i', 0)
_total = multiprocessing.Value('i', len(iterable))
func_and_args = [(func, arg, should_print_progress,) for arg in iterable]
if processes == 1:
result = map(_wrapped_func, func_and_args, *args, **kwargs)
else:
pool = multiprocessing.Pool(initializer=_init,
initargs=(_current, _total,),
processes=processes)
result = pool.map(_wrapped_func, func_and_args, *args, **kwargs)
pool.close()
pool.join()
if should_print_progress:
sys.stdout.write('\r')
return result