Hacker News Reader

Comment by gnuvince | Hacker News Reader

Your code fails for French text.

As did Knuth's; see below the discussion about that.

If you wanted to change it, to split "words" by blanks and punctuation characters, you would have

  tr -s [:blank:][:punct:] "\n"

and then that piece would fit into the pipeline, the rest unscathed. This design is far easier to reason about than a dozen pages of Pascal.

Those globs should be protected from the shell otherwise ./bc is going to alter tr's arguments.

How many modern programs comply with Unicode Text Segmentation http://unicode.org/reports/tr29/ for finding word boundaries or Unicode Collation Algorithm http://unicode.org/reports/tr10/ for sorting the words.