Update timing parsing regex to account for scientific notation.

[chinandrew]

Submission Checklist

• Run unit tests
• Declare copyright holder and open-source license: see below

Summary

The regex in parse_timing_lines() has a bug where if scientific notation is present, only the exponent is captured:

In [1]: from cmdstanpy.utils.stancsv import parse_timing_lines

In [2]: lines = [
   ...:     b'# \n',
   ...:     b'#  Elapsed Time: 630099 seconds (Warm-up)\n',
   ...:     b'#                435856 seconds (Sampling)\n',
   ...:     b'#                1.06595e+06 seconds (Total)\n',
   ...:     b'# \n'
   ...: ]

In [3]: parse_timing_lines(lines)
Out[3]: {'Warm-up': 630099.0, 'Sampling': 435856.0, 'Total': 6.0}

This change updates the regex for the first capture group to include e, +, and -, so the output becomes

In [3]: parse_timing_lines(lines)
Out[3]: {'Warm-up': 630099.0, 'Sampling': 435856.0, 'Total': 1065950.0}

Not sure if support for - is really needed in practice, but I added it for completeness. In keeping with the existing escaped . in the regex, I also escape + and -, though I believe no escapes are actually needed in the character class, i.e. ([\d.e+-]+) is valid.

Note that this can lead to the case where the Warm-Up + Sampling != Total due to truncation from the scientific notation. In this example, 1.06595e+06 gets parsed to 1065950.0 but Warm-Up + Sampling is 630099.0 + 435856.0 = 1065955.0. Arguably this is not a parsing issue since it's correctly reading the comment, but perhaps some additional logic to set Total = Warm-Up + Sampling might be useful.

Copyright and Licensing

Please list the copyright holder for the work you are submitting (this will be you or your assignee, such as a university or company):

Myself

By submitting this pull request, the copyright holder is agreeing to license the submitted work under the following licenses:

• Code: BSD 3-clause (https://opensource.org/licenses/BSD-3-Clause)

[WardBrian]

Thanks for the contribution @chinandrew. Is this a hypothetical issue or did you actually get the new test output from a run of cmdstan?

[chinandrew]

The example above is from an actual cmdstanpy run:

b'#  Elapsed Time: 630099 seconds (Warm-up)\n',
b'#                435856 seconds (Sampling)\n',
b'#                1.06595e+06 seconds (Total)\n',

For the unit test, I modified 435856 to 4358560.0e-1 to capture the - case, though I just ran an extremely short run and small values appear to get rounded to 0:

#  Elapsed Time: 0.015 seconds (Warm-up)
#                0 seconds (Sampling)
#                0.015 seconds (Total)

so perhaps the - case isn't even needed.

[WardBrian]

Thanks!