Personal tools

Haskell Quiz/Bytecode Compiler/Solution Michael Sloan

From HaskellWiki

< Haskell Quiz | Bytecode Compiler(Difference between revisions)
Jump to: navigation, search
Current revision (10:43, 13 January 2007) (edit) (undo)
(sharpen cat)
 
(8 intermediate revisions not shown.)
Line 1: Line 1:
-
<haskell>import Data.Map (Map)
+
[[Category:Haskell Quiz solutions|Bytecode Compiler]]
-
import qualified Data.Map as Map
+
-
import Data.List (intersect)
+
-
import Maybe
+
-
import Char(digitToInt)
+
-
data Op = Add | Sub | Mul | Div | Pow | Mod | Psh Int | LPar | RPar
+
This implementation's only deficiancy is no handling of the negation operator. Proper handling isn't implemented in the ruby quiz tests/solns, however at least basic negative numbers should be included.
-
deriving (Show, Eq)
+
-
instance Ord Op where
+
<haskell>import Char(digitToInt)
-
x <= y = precedence x <= precedence y
+
import Debug.Trace(trace)
-
precedence (Psh _) = 5
+
data Tok = C Char
-
precedence Add = 1
+
| N Int
-
precedence Sub = 1
+
| Op Char
-
precedence Mul = 2
+
| App Tok Tok Tok
-
precedence Mod = 2
+
| G {members :: [Tok]}
-
precedence Div = 3
+
-
precedence Pow = 3
+
--Debugging porpoises
-
precedence LPar = 4
+
instance Show Tok where
-
precedence RPar = 4
+
show (C ch) = ch : ""
-
opFromChar '+' = Add
+
show (N i) = show i
-
opFromChar '-' = Sub
+
show (Op ch) = ch : ""
-
opFromChar '*' = Mul
+
show (App a o b) = '(' : (show a) ++ (show o) ++ (show b) ++ ")"
-
opFromChar '/' = Div
+
show (G xs) = '[' : (concatMap (\x -> show x) xs) ++ "]"
-
opFromChar '^' = Pow
+
-
opFromChar '%' = Mod
+
compile = bytes . parse
-
opFromChar '(' = LPar
+
-
opFromChar ')' = RPar
+
parse = stripG . G . foldr ((.) . prec) id "+-*/%^" . pToks . parseNOp . map C
-
 
+
-
leftAssoc Pow = False
+
parseNOp [] = []
-
leftAssoc RPar = False
+
parseNOp ((N num):(C ch):inp)
-
leftAssoc _ = True
+
| ch `elem` "0123456789" = parseNOp (N (num * 10 + digitToInt ch) : inp)
-
 
+
parseNOp ((C ch):inp) | ch `elem` "0123456789" = parseNOp (N (digitToInt ch) : inp)
-
parse = infixParse [] []
+
parseNOp ((C ch):inp) | ch `elem` "+-*/^%" = parseNOp (Op ch : inp)
-
infixParse o s [] = if (length $ intersect [LPar, RPar] o) > 0 then error "Mismatched parenthesis" else reverse s ++ o
+
parseNOp ((C ch):inp) | ch `elem` " \t\r\n" = parseNOp inp
-
infixParse (RPar:LPar:o) s xs = infixParse o s xs
+
parseNOp ((C ch):inp) | ch `elem` "()" = (C ch) : (parseNOp inp)
-
infixParse o s (x:xs)
+
parseNOp ((C ch):inp) = trace ("Skipped char '"++[ch]) (parseNOp inp)
-
| x `elem` "+-*/^%)" = if not (null o) && (if leftAssoc op then (<=) else (<)) op (o !! 0) then infixParse (tail o) ((o !! 0):s) (x:xs) else infixParse (op:o) s xs
+
parseNOp (x:inp) = x : (parseNOp inp)
-
| x `elem` "0123456789" = case (if null s then Add else s !! 0) of
+
-
(Psh cur) -> infixParse ((Psh $ cur * 10 + digitToInt x) : (tail o)) s xs
+
--Groups an array of tokens if it needs to be
-
_ -> infixParse ((Psh $ digitToInt x):o) s xs
+
g (x:xs) | null xs = x
-
| x == '(' = infixParse (LPar:o) s xs
+
g xs = G xs
-
| x `elem` " \t\r\n" = infixParse o s xs
+
-
where op = opFromChar x
+
--Returns parsed tokens/stream remainder
-
 
+
pToks = fst . parseG
-
getByte v i = (v `div` 2^(i*8)) `mod` 256
+
pRem = snd . parseG
-
toByteArray x n = map (\i -> getByte x i) [(n-1),(n-2)..0]
+
-
 
+
parseG [] = ([], [])
-
bytecode [] = []
+
parseG [(G ts)] = (ts, [])
-
bytecode (op:xs) = (case op of Add -> [10]; Mul -> [11]; Pow -> [12]; Div -> [14]; Mod -> [15]; (Psh x) -> if (abs x < 2^15) then (1 : toByteArray x 2) else (2 : toByteArray x 4)) ++ bytecode xs</haskell>
+
parseG ((C '(') : inp) = (g (pToks inp) : (pToks $ pRem inp), [])
 +
parseG ((C ')') : inp) = ([], inp)
 +
parseG (i:inp) = (i : (pToks inp), pRem inp)
 +
 +
aPrec mo a (Op o) b = App (g $ prec mo [a]) (Op o) (g $ prec mo [b])
 +
 +
--Traverses groups, applies operators to their immediate arguments
 +
prec mo [] = []
 +
prec mo (a:(Op o):b:inp) | o == mo = prec mo ((aPrec mo a (Op o) b) : inp)
 +
prec mo ((G xs):inp) = (g $ prec mo xs) : (prec mo inp)
 +
prec mo ((App a o b):inp) = (aPrec mo a o b) : (prec mo inp)
 +
prec mo (x:inp) = x : (prec mo inp)
 +
 +
--Removes any vestigial grouping
 +
stripG (App a o b) = App (stripG a) o (stripG b)
 +
stripG (G (x:xs)) | null xs = stripG x
 +
stripG (G (x:xs)) = error $ "error: " ++ show (x:xs) ++ " <- improper infix nesting here"
 +
stripG x = x
 +
 +
--Converts to a byte format
 +
bytes (C x) = error ("Invalid character: " ++ [x])
 +
bytes (N n) = if abs n < 2^15 then 1 : toBytes 2 n else 2 : toBytes 4 n
 +
bytes (Op ch) = case ch of '+' -> [10]; '-' -> [11]; '*' -> [12]; '^' -> [13]; '/' -> [14]; '%' -> [15]
 +
bytes (App a o b) = (bytes a) ++ (bytes b) ++ (bytes o)
 +
bytes x = error ("Error, invalid: " ++ show x)
 +
 +
toBytes n x = map (\i -> (x `div` 2^(i*8)) `mod` 256 ) [(n-1),(n-2)..0]</haskell>

Current revision


This implementation's only deficiancy is no handling of the negation operator. Proper handling isn't implemented in the ruby quiz tests/solns, however at least basic negative numbers should be included. 

import Char(digitToInt)
import Debug.Trace(trace)
 
data Tok = C   Char
         | N   Int
         | Op  Char
         | App Tok Tok Tok
         | G {members :: [Tok]}
 
--Debugging porpoises
instance Show Tok where
    show (C ch) = ch : ""
    show (N i) = show i
    show (Op ch) = ch : ""
    show (App a o b) = '(' : (show a) ++ (show o) ++ (show b) ++ ")"
    show (G xs) = '[' : (concatMap (\x -> show x) xs) ++ "]"
 
compile = bytes . parse
 
parse = stripG . G . foldr ((.) . prec) id "+-*/%^" . pToks . parseNOp . map C
 
parseNOp [] = []
parseNOp ((N num):(C ch):inp)
                      | ch `elem` "0123456789" = parseNOp (N (num * 10 + digitToInt ch) : inp)
parseNOp ((C ch):inp) | ch `elem` "0123456789" = parseNOp (N (digitToInt ch) : inp)
parseNOp ((C ch):inp) | ch `elem` "+-*/^%"     = parseNOp (Op ch : inp)
parseNOp ((C ch):inp) | ch `elem` " \t\r\n"    = parseNOp inp
parseNOp ((C ch):inp) | ch `elem` "()"         = (C ch) : (parseNOp inp)
parseNOp ((C ch):inp)                          = trace ("Skipped char '"++[ch]) (parseNOp inp)
parseNOp (x:inp)                               = x : (parseNOp inp)
 
--Groups an array of tokens if it needs to be
g (x:xs) | null xs = x
g xs = G xs
 
--Returns parsed tokens/stream remainder
pToks = fst . parseG
pRem  = snd . parseG
 
parseG []              = ([], [])
parseG [(G ts)]        = (ts, [])
parseG ((C '(') : inp) = (g (pToks inp) : (pToks $ pRem inp), []) 
parseG ((C ')') : inp) = ([], inp)
parseG (i:inp)         = (i : (pToks inp), pRem inp)
 
aPrec mo a (Op o) b = App (g $ prec mo [a]) (Op o) (g $ prec mo [b])
 
--Traverses groups, applies operators to their immediate arguments
prec mo [] = []
prec mo (a:(Op o):b:inp) | o == mo = prec mo ((aPrec mo a (Op o) b) : inp)
prec mo ((G xs):inp)               = (g $ prec mo xs) : (prec mo inp)
prec mo ((App a o b):inp)          = (aPrec mo a o b) : (prec mo inp)
prec mo (x:inp) = x : (prec mo inp)
 
--Removes any vestigial grouping
stripG (App a o b) = App (stripG a) o (stripG b)
stripG (G (x:xs)) | null xs = stripG x
stripG (G (x:xs)) = error $ "error: " ++ show (x:xs) ++ " <- improper infix nesting here"
stripG x = x
 
--Converts to a byte format
bytes (C x) = error ("Invalid character: " ++ [x])
bytes (N n) = if abs n < 2^15 then 1 : toBytes 2 n else 2 : toBytes 4 n
bytes (Op ch) = case ch of '+' -> [10]; '-' -> [11]; '*' -> [12]; '^' -> [13]; '/' -> [14]; '%' -> [15]
bytes (App a o b) = (bytes a) ++ (bytes b) ++ (bytes o)
bytes x = error ("Error, invalid: " ++ show x)
 
toBytes n x = map (\i -> (x `div` 2^(i*8)) `mod` 256 ) [(n-1),(n-2)..0]
Retrieved from "http://www.haskell.org/haskellwiki/Haskell_Quiz/Bytecode_Compiler/Solution_Michael_Sloan"

Category: Haskell Quiz solutions