Source code for ubelt.util_time

# -*- coding: utf-8 -*-
from __future__ import absolute_import, division, print_function, unicode_literals
import time
import math
import sys
import gc
import itertools as it

__all__ = ['Timer', 'Timerit', 'timestamp']

if sys.version_info.major == 2:
    default_timer = time.clock if sys.platform.startswith('win32') else time.time
else:
    default_timer = time.perf_counter


[docs]class Timer(object): """ Measures time elapsed between a start and end point. Can be used as a with-statement context manager, or using the tic/toc api. Args: label (str): identifier for printing defaults to '' verbose (int): verbosity flag, defaults to True if label is given newline (bool): if False and verbose, print tic and toc on the same line Attributes: elapsed (float): number of seconds measured by the context manager tstart (float): time of last `tic` reported by `default_timer()` Example: >>> # Create and start the timer using the the context manager >>> timer = Timer('Timer test!', verbose=1) >>> with timer: >>> math.factorial(10000) >>> assert timer.elapsed > 0 Example: >>> # Create and start the timer using the tic/toc interface >>> timer = Timer().tic() >>> elapsed1 = timer.toc() >>> elapsed2 = timer.toc() >>> elapsed3 = timer.toc() >>> assert elapsed1 <= elapsed2 >>> assert elapsed2 <= elapsed3 """ def __init__(self, label='', verbose=None, newline=True): if verbose is None: verbose = bool(label) self.label = label self.verbose = verbose self.newline = newline self.tstart = -1 self.elapsed = -1 self.write = sys.stdout.write self.flush = sys.stdout.flush
[docs] def tic(self): """ starts the timer """ if self.verbose: self.flush() self.write('\ntic(%r)' % self.label) if self.newline: self.write('\n') self.flush() self.tstart = default_timer() return self
[docs] def toc(self): """ stops the timer """ elapsed = default_timer() - self.tstart if self.verbose: self.write('...toc(%r)=%.4fs\n' % (self.label, elapsed)) self.flush() return elapsed
def __enter__(self): self.tic() return self def __exit__(self, ex_type, ex_value, trace): self.elapsed = self.toc() if trace is not None: return False
[docs]class Timerit(object): """ Reports the average time to run a block of code. Unlike `timeit`, `Timerit` can handle multiline blocks of code Args: num (int): number of times to run the loop label (str): identifier for printing bestof (int): takes the max over this number of trials verbose (int): verbosity flag, defaults to True if label is given CommandLine: python -m utool.util_time Timerit python -m utool.util_time Timerit:0 python -m utool.util_time Timerit:1 Example: >>> num = 15 >>> t1 = Timerit(num, verbose=2) >>> for timer in t1: >>> # <write untimed setup code here> this example has no setup >>> with timer: >>> # <write code to time here> for example... >>> math.factorial(10000) >>> # <you can now access Timerit attributes> >>> print('t1.total_time = %r' % (t1.total_time,)) >>> assert t1.total_time > 0 >>> assert t1.n_loops == t1.num >>> assert t1.n_loops == num Example: >>> num = 10 >>> # If the timer object is unused, time will still be recorded, >>> # but with less precision. >>> for _ in Timerit(num, 'imprecise'): >>> math.factorial(10000) >>> # Using the timer object results in the most precise timings >>> for timer in Timerit(num, 'precise'): >>> with timer: math.factorial(10000) """ def __init__(self, num, label=None, bestof=3, verbose=None): if verbose is None: verbose = bool(label) self.num = num self.label = label self.times = [] self.verbose = verbose self.total_time = None self.n_loops = None self.bestof = bestof
[docs] def call(self, func, *args, **kwargs): """ Alternative way to time a simple function call using condensed syntax. Returns: self (Timerit): Use `ave_secs`, `min`, or `mean` to get a scalar. Example: >>> ave_sec = Timerit(num=10).call(math.factorial, 50).ave_secs >>> assert ave_sec > 0 """ for timer in self: with timer: func(*args, **kwargs) return self
def __iter__(self): if self.verbose >= 2: if self.label is None: print('Timing for %d loops' % self.num) else: print('Timing %s for %d loops.' % (self.label, self.num,)) self.n_loops = 0 self.total_time = 0 # Create a foreground and background timer bg_timer = Timer(verbose=0) # (ideally this is unused) fg_timer = Timer(verbose=0) # (used directly by user) # disable the garbage collector while timing with ToggleGC(False): # Core timing loop for i in it.repeat(None, self.num): # Start background timer (in case the user doesn't use fg_timer) # Yield foreground timer to let the user run a block of code # When we return from yield the user code will have just finished # Then record background time + loop overhead bg_timer.tic() yield fg_timer bg_time = bg_timer.toc() # Check if the fg_timer object was used, but fallback on bg_timer if fg_timer.elapsed >= 0: block_time = fg_timer.elapsed # higher precision else: block_time = bg_time # low precision # record timings self.times.append(block_time) self.total_time += block_time self.n_loops += 1 # Timing complete, print results assert len(self.times) == self.num, 'incorrectly recorded times' if self.verbose > 0: self._print_report(self.verbose) @property def ave_secs(self): """ The expected execution time of the timed code snippet in seconds. This is the minimum value recorded over all runs. SeeAlso: self.min self.mean self.std """ return self.min()
[docs] def min(self): """ The best time overall. This is typically the best metric to consider when evaluating the execution time of a function. To understand why consider this quote from the docs of the original timeit module: ''' In a typical case, the lowest value gives a lower bound for how fast your machine can run the given code snippet; higher values in the result vector are typically not caused by variability in Python's speed, but by other processes interfering with your timing accuracy. So the min() of the result is probably the only number you should be interested in. ''' Example: >>> self = Timerit(num=10, verbose=0) >>> self.call(math.factorial, 50) >>> assert self.min() > 0 """ return min(self.times)
[docs] def mean(self): """ The mean of the best results of each trial. Note: This is typically less informative than simply looking at the min Example: >>> self = Timerit(num=10, verbose=0) >>> self.call(math.factorial, 50) >>> assert self.mean() > 0 """ from ubelt.util_list import chunks chunks = chunks(self.times, self.bestof) times = list(map(min, chunks)) mean = sum(times) / len(times) return mean
[docs] def std(self): """ The standard deviation of the best results of each trial. Note: As mentioned in the timeit source code, the standard deviation is not often useful. Typically the minimum value is most informative. Example: >>> self = Timerit(num=10, verbose=1) >>> self.call(math.factorial, 50) >>> assert self.std() >= 0 """ from ubelt.util_list import chunks chunks = chunks(self.times, self.bestof) times = list(map(min, chunks)) mean = sum(times) / len(times) std = math.sqrt(sum((t - mean) ** 2 for t in times) / len(times)) return std
def _seconds_str(self): """ CommandLine: python -m ubelt.util_time Timerit._seconds_str Example: >>> self = Timerit(num=100, bestof=10, verbose=0) >>> self.call(lambda : sum(range(100))) >>> print(self._seconds_str()) ... 'best=3.423 µs, ave=3.451 ± 0.027 µs' """ units = [ ('s', 1e0), ('ms', 1e-3), (_trychar('µs', 'us'), 1e-6), ('ns', 1e-9), ] mean = self.mean() for unit, mag in units: # pragma: nobranch if mean > mag: break unit_min = self.min() / mag unit_mean = mean / mag precision = 4 # Is showing the std useful? It probably doesn't hurt. std = self.std() unit_std = std / mag pm = _trychar('±', '+-') fmtstr = ('best={min:.{pr1}} {unit}, ' 'mean={mean:.{pr1}} {pm} {std:.{pr2}} {unit}') pr1 = precision pr2 = max(precision - 2, 1) unit_str = fmtstr.format(min=unit_min, unit=unit, mean=unit_mean, pm=pm, std=unit_std, pr1=pr1, pr2=pr2) return unit_str def _report(self, verbose=1): """ Creates a human readable report """ report_lines = [] pline = report_lines.append if self.label is None: pline('Timed for: %d loops, best of %d' % ( self.n_loops, min(self.n_loops, self.bestof))) else: pline('Timed %s for: %d loops, best of %d' % ( self.label, self.n_loops, min(self.n_loops, self.bestof))) if verbose > 2: pline(' body took: %s seconds' % self.total_time) pline(' time per loop: %s' % (self._seconds_str(),)) return '\n'.join(report_lines) def _print_report(self, verbose=1): """ Prints human readable report using the print function """ print(self._report(verbose=verbose))
class ToggleGC(object): """ Context manager to disable garbage collection. Example: >>> import gc >>> prev = gc.isenabled() >>> with ToggleGC(False): >>> assert not gc.isenabled() >>> with ToggleGC(True): >>> assert gc.isenabled() >>> assert not gc.isenabled() >>> assert gc.isenabled() == prev """ def __init__(self, flag): self.flag = flag self.prev = None def __enter__(self): self.prev = gc.isenabled() if self.flag: gc.enable() else: gc.disable() def __exit__(self, ex_type, ex_value, trace): if self.prev: gc.enable() else: gc.disable()
[docs]def timestamp(method='iso8601'): """ make an iso8601 timestamp CommandLine: python -m ubelt.util_time timestamp Example: >>> stamp = timestamp() >>> print('stamp = {!r}'.format(stamp)) ...-...-...T... """ if method == 'iso8601': # ISO 8601 # datetime.datetime.utcnow().isoformat() # datetime.datetime.now().isoformat() # utcnow tz_hour = time.timezone // 3600 utc_offset = '-' + str(tz_hour) if tz_hour < 0 else '+' + str(tz_hour) stamp = time.strftime('%Y-%m-%dT%H%M%S') + utc_offset return stamp else: raise ValueError('only iso8601 is accepted for now')
def _trychar(char, fallback): # nocover """ CommandLine: python -m ubelt.util_time _trychar pytest ubelt/util_time.py::_trychar:0 -s Example: >>> char = _trychar('µs', 'us') >>> print('char = {}'.format(char)) """ # Logic from ipython timeit to handle terminals that cant show mu if hasattr(sys.stdout, 'encoding') and sys.stdout.encoding: # pragma: nobranch try: char.encode(sys.stdout.encoding) except Exception: # nocover pass else: return char return fallback # nocover if __name__ == '__main__': """ CommandLine: python -m ubelt.util_time python -m ubelt.util_time all """ import xdoctest as xdoc xdoc.doctest_module(__file__)