Translate Haskell into English Manually, Part II

Okay, we're getting really close now--one more example. Let's transform makeCoolState.hs into something slightly more useful and “pretend” we're reading tokens. We'll naively treat each character as a Symbol token, and we'll simply read four tokens.

import Control.Monad.State

data TokenType = Identifier | Qualifier | Type | Symbol Char
  deriving (Show, Eq)

data Token = Token {
  tokenType :: TokenType,
  tokenValue :: String
} deriving Show

data ParseContext = ParseContext {
  input :: String,    -- The input that has not been parsed yet.
  output :: String,   -- The output generated so far.
  currTok :: Token,   -- The current token, if defined.
  stack :: [Token]    -- A stack of tokens we haven't dealt with yet.
} deriving Show

-- Read one token from input and put it on the stack.
getToken :: State ParseContext ()
getToken = do
  ctx@(ParseContext {input = c:input', stack = stack'}) <- get
  let tok = Token {tokenType = Symbol c, tokenValue = [c]} in
    put (ctx {input = input', currTok = tok, stack = tok:stack'})

-- This is a completely different way of writing getToken.
getTokenAnotherWay :: State ParseContext ()
getTokenAnotherWay = modify getTokenAnotherWay'
  where
    getTokenAnotherWay' ctx =
      ctx {input = input', currTok = tok, stack = tok:stack'}
      where
        c = head $ input $ ctx
        input' = tail $ input $ ctx
        tok = Token {tokenType = Symbol c, tokenValue = [c]}
        stack' = stack $ ctx

-- Call getToken a few times.
parse :: State ParseContext ()
parse = do
  getToken            -- I would not normally duplicate lines like this.
  getToken            -- I'm just trying to illustrate a point.
  getTokenAnotherWay
  getTokenAnotherWay

main = do
  s <- getContents    -- currTok is undefined below.  It's defined later.
  let ctx = ParseContext {input = s, output = "", stack = []} in
    print $ execState parse $ ctx

Running it:

$ echo "Haskell" | runhugs -98 monadPipes.hs
ParseContext {
  input = "ell\n",
  output = "",
  currTok = Token {tokenType = Symbol 'k', tokenValue = "k"},
  stack = [
    Token {tokenType = Symbol 'k', tokenValue = "k"},
    Token {tokenType = Symbol 's', tokenValue = "s"},
    Token {tokenType = Symbol 'a', tokenValue = "a"},
    Token {tokenType = Symbol 'H', tokenValue = "H"}]}

At the risk of cheating, I've manually added newlines and indentation to the output to make it more legible. Nonetheless, this output is really exciting! It's taken the string Haskell and treated the first four letters as Symbol tokens, placing each one on the stack.

This fascinating glimpse into the heart of the ParseContext was made possible simply by making sure to add deriving Show in the right places and calling print on the final ParseContext in the line:

print $ execState parse $ ctx

Notice a few other things. The do syntax is used to parse four tokens in a row:

parse = do
  getToken            -- I would not normally duplicate lines like this.
  getToken            -- I'm just trying to illustrate a point.
  getTokenAnotherWay
  getTokenAnotherWay

In getToken, I used get and put, but in getTokenAnotherWay, I used modify. In general, I've varied how the code was implemented in very minor ways to show different ways of doing things. After all, even C has three different constructs for a loop; you pick the most convenient tool for the job at hand.

As a last note, to experience how incredibly lazy Haskell is, instead of using echo to enter the value Haskell, try typing the input characters one at a time, pressing Return after each character. Haskell will start outputting the output before you are done typing the input. In fact, Haskell will produce as much output as possible for the input you have provided. In a sense, because the input is read lazily, the rest of the program can run “proactively”.

Now, take a deep breath, and prepare for the plunge. Here's what you have at your disposal:

At long last, here's the code:

{- Translate C type declarations into English.
  This exercise was taken from "Expert C Programming:  Deep C Secrets", p. 84.
   Example: echo -n "int *p;" | runhugs -98 cdecl.hs
-}

import Char
import Control.Monad.State

data TokenType = Identifier | Qualifier | Type | Symbol Char
  deriving (Show, Eq)

data Token = Token {
  tokenType :: TokenType,
  tokenValue :: String
} deriving Show

data ParseContext = ParseContext {
  input :: String,    -- The input that has not been parsed yet.
  output :: String,   -- The output generated so far.
  currTok :: Token,   -- The current token, if defined.
  stack :: [Token]    -- A stack of tokens we haven't dealt with yet.
} deriving Show

-- Convenience functions:
currTokType = tokenType . currTok
currTokValue = tokenValue . currTok
stackHead = head . stack

-- "Write" to a ParseContext's output.  Use ++ for simplicity.
writeOutput :: String -> State ParseContext ()
writeOutput s = modify (\ctx -> ctx {output = output ctx ++ s})

-- Pop the stack.
pop :: State ParseContext ()
pop = modify (\ctx -> ctx {stack = tail $ stack $ ctx})

-- Write the value of the top of the stack and then pop it.
popAndWrite :: State ParseContext ()
popAndWrite = do
  top <- gets stackHead
  writeOutput (tokenValue top)
  pop

-- Classify a string into a Token.
classifyString :: String -> Token
classifyString "const"  = Token Qualifier "read-only"
classifyString "*"      = Token (Symbol '*') "pointer to"
classifyString [c]
  | not (isAlphaNum c)  = Token (Symbol c) [c]
classifyString s        = Token tokType s
  where
    tokType = case s of
      "volatile" -> Qualifier
      x | x `elem` ["void", "char", "signed", "unsigned", "short",
                    "int", "long", "float", "double", "struct",
                    "union", "enum"] -> Type
      x -> Identifier

-- Read the next token into currTok.
getToken :: State ParseContext ()
getToken = modify getToken'
  where
    getToken' ctx@(ParseContext {input = s}) =
      ctx {currTok = token, input = theRest}
      where
        (token, theRest) = lexString $ lstrip $ s
        lstrip s = dropWhile isSpace s

-- Read a token.  Return it and the left-over portion of the string.
lexString :: String -> (Token, String)
lexString s@(c:cs) | isAlphaNum c = (token, theRest)
  where
    (tokString, theRest) = span isAlphaNum s
    token = classifyString tokString
lexString (c:cs) = (classifyString [c], cs)

-- Put tokens on the stack until we reach the first identifier.
readToFirstIdentifier :: State ParseContext ()
readToFirstIdentifier = do
  getToken
  pushUntilIdentifier
  afterIdentifier <- get
  let s = identifier ++ " is "
      identifier = currTokValue afterIdentifier in
    put (afterIdentifier {output = s})
  getToken

-- Keep pushing tokens until we hit an identifier.
pushUntilIdentifier :: State ParseContext ()
pushUntilIdentifier = do
  ctx <- get
  if currTokType ctx == Identifier
    then return ()                      -- Leave things as they are.
    else do
      put (ctx {stack = (currTok ctx) : (stack ctx)})
      getToken
      pushUntilIdentifier
      return ()

-- Deal with arrays.
dealWithArrays :: State ParseContext ()
dealWithArrays = do
  ctx <- get
  case currTokType ctx of
    Symbol '[' -> do
      writeOutput "array "
      getToken
      writeIfNumber
      getToken
      writeOutput "of "
      dealWithArrays
    _ -> return ()                      -- Recurse until we get past the ['s.
  where
    writeIfNumber = do                  -- Call writeSize if a number.
      tokValue <- gets currTokValue
      if isDigit $ head $ tokValue
        then do
          writeSize
          getToken
        else return ()
    writeSize = do                      -- Output the array size.
      tokValue <- gets currTokValue
      let num = show $ (+ -1) $ read $ tokValue
          s = "0.." ++ num ++ " " in    -- Can't use where instead of let here.
        writeOutput s

-- Deal with function arguments.
dealWithFunctionArgs :: State ParseContext ()
dealWithFunctionArgs = do
  getUntilParen
  getToken
  writeOutput "function returning "
  where
    getUntilParen = do                  -- Read tokens until we hit ).
      ctx <- get
      case currTokType ctx of
        Symbol ')' -> return ()
        _ -> do
          getToken
          getUntilParen

-- Deal with pointers.
dealWithPointers :: State ParseContext ()
dealWithPointers = do
  top <- gets stackHead
  case tokenType top of
    Symbol '*' -> do
      popAndWrite
      writeOutput " "
      dealWithPointers
    _ -> return ()                      -- Recurse until we get past the *'s.

-- Process tokens that we stacked while reading to identifier.
dealWithStack :: State ParseContext ()
dealWithStack = do
  stack' <- gets stack
  case stack' of
    [] -> return ()
    (x:xs) ->
      case tokenType x of
        Symbol '(' -> do
          pop
          getToken
          dealWithDeclarator
        _ -> popAndWrite

-- Do all parsing after first identifier.
dealWithDeclarator :: State ParseContext ()
dealWithDeclarator = do
  tokType <- gets currTokType
  case tokType of
    Symbol '[' -> dealWithArrays
    Symbol '(' -> dealWithFunctionArgs
    _ -> return ()                      -- "Exit" the case, not the function.
  dealWithPointers
  dealWithStack

-- Do all parsing.
parse :: State ParseContext ()
parse = do
  readToFirstIdentifier
  dealWithDeclarator

-- Translate a C type declaration into English.
translate :: String -> String
translate s =
  output $ execState parse $ ctx        -- Change "output" to "show" to debug.
  where ctx = ParseContext {input = s, output = "", stack = []}

main :: IO ()
main = do
  input <- getContents
  putStrLn $ translate $ input