Cabal/src/Distribution/Lex.hs

   1 -----------------------------------------------------------------------------
   2
   3 -- |
   4 -- Module      :  Distribution.Lex
   5 -- Copyright   :  Ben Gamari 2015-2019
   6 --
   7 -- Maintainer  :  cabal-devel@haskell.org
   8 -- Portability :  portable
   9 --
  10 -- This module contains a simple lexer supporting quoted strings
  11 module Distribution.Lex
  12   ( tokenizeQuotedWords
  13   ) where
  14
  15 import Distribution.Compat.DList
  16 import Distribution.Compat.Prelude
  17 import Prelude ()
  18
  19 -- | A simple parser supporting quoted strings.
  20 --
  21 -- Please be aware that this will only split strings when seeing whitespace
  22 -- outside of quotation marks, i.e, @"foo\"bar baz\"qux quux"@ will be
  23 -- converted to @["foobar bazqux", "quux"]@.
  24 --
  25 -- This behavior can be useful when parsing text like
  26 -- @"ghc-options: -Wl,\"some option with spaces\""@, for instance.
  27 tokenizeQuotedWords :: String -> [String]
  28 tokenizeQuotedWords = filter (not . null) . go False mempty
  29   where
  30     go
  31       :: Bool
  32       -- \^ in quoted region
  33       -> DList Char
  34       -- \^ accumulator
  35       -> String
  36       -- \^ string to be parsed
  37       -> [String]
  38     -- \^ parse result
  39     go _ accum []
  40       | [] <- accum' = []
  41       | otherwise = [accum']
  42       where
  43         accum' = runDList accum
  44     go False accum (c : cs)
  45       | isSpace c = runDList accum : go False mempty cs
  46       | c == '"' = go True accum cs
  47     go True accum (c : cs)
  48       | c == '"' = go False accum cs
  49     go quoted accum (c : cs) =
  50       go quoted (accum `mappend` singleton c) cs