NaturalNode / natural / 40116079416

87%

push

github

Pr 784 (#795)

* fix: do not match abbreviations inside other words in SentenceTokenizer

The abbreviation pattern had no word boundary, so with abbreviations such
as "St." and "Inc." configured, the end of "best.", "list." or "zinc." was
taken for an abbreviation and the sentence was not split. Only match an
abbreviation when it does not directly follow a letter or a digit.

---------

Authored-by: Osama Ansar <osama.ansar.pk@gmail.com>

2369 of 3073 branches covered (77.09%)

Branch coverage included in aggregate %.

1 of 1 new or added line in 1 file covered. (100.0%)

9023 of 9994 relevant lines covered (90.28%)

12843.2 hits per line