LGTM! Yes, I concur there are a lot of inefficient loops and calls happening in abbreviation_replacer.py. Since I was porting it from ruby, I thought of keeping the same loops and calls in python to get tests to pass. In a few cases, I tried to be idiomatic wherever possible, but still, there are a lot of places where we can surely improve pysbd.

It would be nice if you could list those speed up calls in the comment here so we can tackle them one by one.

I was focusing on the parts that were the majority of the time spent. Figured no need to optimize parts that weren't taking up much of the time. Here is some of the output from cProfile:

   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
        1    0.000    0.000    1.484    1.484 <string>:1(<module>)
      366    0.003    0.000    0.258    0.001 abbreviation_replacer.py:10(replace_pre_number_abbr)
     1492    0.004    0.000    1.145    0.001 abbreviation_replacer.py:103(scan_for_replacements)
      222    0.002    0.000    0.157    0.001 abbreviation_replacer.py:19(replace_prepositive_abbr)
      904    0.007    0.000    0.726    0.001 abbreviation_replacer.py:28(replace_period_of_abbr)
        1    0.000    0.000    0.013    0.013 abbreviation_replacer.py:43(replace_abbreviation_as_sentence_boundary)
        1    0.000    0.000    0.000    0.000 abbreviation_replacer.py:54(__init__)
        1    0.000    0.000    1.400    1.400 abbreviation_replacer.py:58(replace)
        1    0.000    0.000    0.001    0.001 abbreviation_replacer.py:70(replace_multi_period_abbreviations)
        4    0.000    0.000    0.000    0.000 abbreviation_replacer.py:71(mpa_replace)
        1    0.005    0.005    1.380    1.380 abbreviation_replacer.py:83(search_for_abbreviations_in_string)

Didn't want to cause code churn/potentially introduce bugs by over optimizing other pieces of code. I also don't actually have clear ideas at the moment for how to make those slow pieces even faster.

Any hesitation on releasing this as is?

Although, I wanted to make more optimizations in abbreviation_replacer.py but didn't find much time to do it thoroughly so releasing it for now and we can do further optimizations in the future.

Just to segregate scripts as per purpose, moving your test_timing_script.py to examples for further reference.