Pandoc Sentence Splitter (for diff output)

PDsentences.hs

#!/usr/bin/env runhaskell
{-
Splits sentences in Pandoc conversions. Useful for diff output.

Put PDsentences.hs in your path and specify it as a new filter for
pandoc. This allows cleaner diff views with most markdown files being primarily
written text.

.gitconfig:
[diff "markdown"]
      textconv = pandoc -t markdown_github --no-wrap --filter=PDsentences.hs
.gitattributes:
*.md diff=markdown
-}

import Text.Pandoc.JSON
import Text.Regex.Posix
import Text.Regex

endOfSentencePat = ".*[\\.?!;][)\"\'>»’”]?$"
ignoredPat = "^(e\\.g|i\\.e|mr|mrs|ms|dr|jr|sr|vs)\\.$"

makeBreak :: [Char] -> Bool
makeBreak s = s =~ endOfSentencePat && not commonWord
  where commonWord =
          match (mkRegexWithOpts ignoredPat True False) s

consumeSpace :: [Inline] -> [Inline]
consumeSpace (Space:is) = consumeSpace is
consumeSpace is = breakSentence is

breakSentence :: [Inline] -> [Inline]
breakSentence ((Str s):is) =
  (Str s):needsLineBreak
  where needsLineBreak = 
          if makeBreak s then    
            LineBreak:consumeSpace is
          else
            breakSentence is

breakSentence (i:is) = i:breakSentence is
breakSentence [] = []

splitSentences (Plain t) = Plain (breakSentence t)
splitSentences (Para t) = Para (breakSentence t)
splitSentences x = x

main :: IO ()
main = toJSONFilter splitSentences