Stupidly-simple speed profiling tool for python
A stupidly-simple tool to see where your time is going in Python programs
Trying to figure out where the time's going in your python code? Tired of writing
elapsed = time.time() - start_time? You can find out with just a few lines of code after you
pip install timebudget
The simplest way
With just two lines of code (one is the import), you can see how long something takes...
from timebudget import timebudget with timebudget("Loading and processing the file"): raw = open(filename,'rt').readlines() lines = [line.rstrip() for line in raw]
Loading and processing the file took 1.453sec
Record times and print a report
To get a report on the total time from functions you care about, just annotate those functions:
from timebudget import timebudget timebudget.set_quiet() # don't show measurements as they happen timebudget.report_at_exit() # Generate report when the program exits @timebudget # Record how long this function takes def possibly_slow(): ... @timebudget # ... and this function too def should_be_fast(): ...
And now when you run your program, you'll see how much time was spent in each annotated function:
timebudget report... possibly_slow: 901.12ms for 3 calls should_be_fast: 61.35ms for 2 calls
Or instead of calling
report_at_exit() you can manually call
timebudget.report(reset=True) # print out the report now, and reset the statistics
If you don't set
reset=True then the statistics will accumulate into the next report.
You can also wrap specific blocks of code to be recorded in the report, and optionally override
set_quiet choice for any block:
with timebudget("load-file", quiet=False): text = open(filename,'rt').readlines()
Percent of time in a loop
If you are doing something repeatedly, and want to know the percent of time doing different things, time the loop itself, and pass the name to report:
@timebudget def outer_loop(): if sometimes(): possibly_slow() should_be_fast() should_be_fast() timebudget.report('outer_loop')
Then the report looks like:
timebudget report per outer_loop cycle... outer_loop: 100.0% 440.79ms/cyc @ 1.0 calls/cyc possibly_slow: 40.9% 180.31ms/cyc @ 0.6 calls/cyc should_be_fast: 13.7% 60.19ms/cyc @ 2.0 calls/cyc
Here, the times in milliseconds are the totals (averages per cycle), not the average time per call. So in the above example,
should_be_fast is taking about 30ms per call, but being called twice per loop. Similarly,
possibly_slow is still about 300ms each time it's called, but it's only getting called on 60% of the cycles on average, so on average it's using 41% of the time in
outer_loop or 180ms.
Needs Python 3.6 or higher. Because f-strings and type annotations are awesome, and it's 2019, and python 2.7 is on its deathbed.
This tool is inspired by TQDM, the awesome progress bar. TQDM is stupidly simple to add to your code, and just makes it better. I aspire to imitate that.
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size timebudget-0.7-py3-none-any.whl (9.2 kB)||File type Wheel||Python version py3||Upload date||Hashes View hashes|
|Filename, size timebudget-0.7.tar.gz (4.7 kB)||File type Source||Python version None||Upload date||Hashes View hashes|