Util.lhs 27 KB
Newer Older
1
%
Simon Marlow's avatar
Simon Marlow committed
2
% (c) The University of Glasgow 2006
3
% (c) The University of Glasgow 1992-2002
4 5 6 7
%
\section[Util]{Highly random utility functions}

\begin{code}
8
{-# OPTIONS -w #-}
9 10 11
-- The above warning supression flag is a temporary kludge.
-- While working on this module you are encouraged to remove it and fix
-- any warnings in the module. See
Ian Lynagh's avatar
Ian Lynagh committed
12
--     http://hackage.haskell.org/trac/ghc/wiki/Commentary/CodingStyle#Warnings
13 14
-- for details

15
module Util (
sof's avatar
sof committed
16

17
	-- general list processing
18
	zipEqual, zipWithEqual, zipWith3Equal, zipWith4Equal,
19
        zipLazy, stretchZipWith,
20
	mapFst, mapSnd,
21
	mapAndUnzip, mapAndUnzip3,
22
	nOfThem, filterOut, partitionWith, splitEithers,
23
        foldl1',
24 25 26 27

	lengthExceeds, lengthIs, lengthAtLeast, 
	listLengthCmp, atLength, equalLength, compareLength,

28
	isSingleton, only, singleton,
29
	notNull, snocView,
sof's avatar
sof committed
30

31 32
	isIn, isn'tIn,

33 34 35
	-- for-loop
	nTimes,

36
	-- sorting
37
	sortLe, sortWith, on,
38 39 40 41 42

	-- transitive closures
	transitiveClosure,

	-- accumulating
43
	foldl2, count, all2,
sof's avatar
sof committed
44
	
45
	takeList, dropList, splitAtList, split,
46 47

	-- comparisons
48
	isEqual, eqListBy, 
49
	thenCmp, cmpList, maybePrefixMatch,
50
	removeSpaces,
51

52
	-- strictness
53
	seqList,
54

55
	-- pairs
56
	unzipWith,
57

58
	global, consIORef,
59 60 61

	-- module names
	looksLikeModuleName,
sof's avatar
sof committed
62
	
63 64 65 66
	toArgs,

	-- Floating point stuff
	readRational,
67 68 69 70

	-- IO-ish utilities
	createDirectoryHierarchy,
	doesDirNameExist,
71
	modificationTimeIfExists,
72 73 74 75 76

	later, handleDyn, handle,

	-- Filename utils
	Suffix,
77
	splitFilename, suffixOf, basenameOf, joinFileExt,
78
	splitFilenameDir, joinFileName,
79 80
	splitFilename3,
	splitLongestPrefix,
81 82 83
	replaceFilenameSuffix, directoryOf, filenameOf,
	replaceFilenameDirectory,
	escapeSpaces, isPathSeparator,
84
	parseSearchPath,
85
	normalisePath, platformPath, pgmPath,
86 87
    ) where

88 89
#include "HsVersions.h"

90
import FastTypes
91

92 93
#if defined(DEBUG) || __GLASGOW_HASKELL__ < 604
import Panic
94 95
#endif

Simon Marlow's avatar
Simon Marlow committed
96 97 98 99 100 101
import Control.Exception ( Exception(..), finally, catchDyn, throw )
import qualified Control.Exception as Exception
import Data.Dynamic	( Typeable )
import Data.IORef	( IORef, newIORef )
import System.IO.Unsafe	( unsafePerformIO )
import Data.IORef	( readIORef, writeIORef )
102
import Data.List        hiding (group)
103

104
import qualified Data.List as List ( elem )
105
#ifdef DEBUG
106
import qualified Data.List as List ( notElem )
107
#endif
sof's avatar
sof committed
108

Simon Marlow's avatar
Simon Marlow committed
109 110 111 112 113 114 115
import Control.Monad	( when )
import SYSTEM_IO_ERROR as IO ( catch, isDoesNotExistError )
import System.Directory	( doesDirectoryExist, createDirectory,
                          getModificationTime )
import Data.Char	( isUpper, isAlphaNum, isSpace, ord, isDigit )
import Data.Ratio	( (%) )
import System.Time	( ClockTime )
116

117
infixr 9 `thenCmp`
118 119
\end{code}

120 121 122 123 124 125 126 127 128 129 130 131 132 133
%************************************************************************
%*									*
\subsection{A for loop}
%*									*
%************************************************************************

\begin{code}
-- Compose a function with itself n times.  (nth rather than twice)
nTimes :: Int -> (a -> a) -> (a -> a)
nTimes 0 _ = id
nTimes 1 f = f
nTimes n f = f . nTimes (n-1) f
\end{code}

134 135 136 137 138 139
%************************************************************************
%*									*
\subsection[Utils-lists]{General list processing}
%*									*
%************************************************************************

140 141 142 143 144 145
\begin{code}
filterOut :: (a->Bool) -> [a] -> [a]
-- Like filter, only reverses the sense of the test
filterOut p [] = []
filterOut p (x:xs) | p x       = filterOut p xs
		   | otherwise = x : filterOut p xs
146 147 148 149 150 151 152 153 154

partitionWith :: (a -> Either b c) -> [a] -> ([b], [c])
partitionWith f [] = ([],[])
partitionWith f (x:xs) = case f x of
			   Left  b -> (b:bs, cs)
			   Right c -> (bs, c:cs)
		       where
			 (bs,cs) = partitionWith f xs

155 156 157 158 159 160 161
splitEithers :: [Either a b] -> ([a], [b])
splitEithers [] = ([],[])
splitEithers (e : es) = case e of
			  Left x -> (x:xs, ys)
			  Right y -> (xs, y:ys)
		      where
			(xs,ys) = splitEithers es
162 163
\end{code}

164 165 166
A paranoid @zip@ (and some @zipWith@ friends) that checks the lists
are of equal length.  Alastair Reid thinks this should only happen if
DEBUGging on; hey, why not?
167 168

\begin{code}
169 170 171 172
zipEqual	:: String -> [a] -> [b] -> [(a,b)]
zipWithEqual	:: String -> (a->b->c) -> [a]->[b]->[c]
zipWith3Equal	:: String -> (a->b->c->d) -> [a]->[b]->[c]->[d]
zipWith4Equal	:: String -> (a->b->c->d->e) -> [a]->[b]->[c]->[d]->[e]
173 174

#ifndef DEBUG
175 176 177 178
zipEqual      _ = zip
zipWithEqual  _ = zipWith
zipWith3Equal _ = zipWith3
zipWith4Equal _ = zipWith4
179
#else
180 181
zipEqual msg []     []     = []
zipEqual msg (a:as) (b:bs) = (a,b) : zipEqual msg as bs
182
zipEqual msg as     bs     = panic ("zipEqual: unequal lists:"++msg)
183 184 185

zipWithEqual msg z (a:as) (b:bs)=  z a b : zipWithEqual msg z as bs
zipWithEqual msg _ [] []	=  []
186
zipWithEqual msg _ _ _		=  panic ("zipWithEqual: unequal lists:"++msg)
187 188 189 190

zipWith3Equal msg z (a:as) (b:bs) (c:cs)
				=  z a b c : zipWith3Equal msg z as bs cs
zipWith3Equal msg _ [] []  []	=  []
191
zipWith3Equal msg _ _  _   _	=  panic ("zipWith3Equal: unequal lists:"++msg)
192 193 194 195

zipWith4Equal msg z (a:as) (b:bs) (c:cs) (d:ds)
				=  z a b c d : zipWith4Equal msg z as bs cs ds
zipWith4Equal msg _ [] [] [] []	=  []
196
zipWith4Equal msg _ _  _  _  _	=  panic ("zipWith4Equal: unequal lists:"++msg)
197 198 199
#endif
\end{code}

200 201 202 203 204 205 206 207
\begin{code}
-- zipLazy is lazy in the second list (observe the ~)

zipLazy :: [a] -> [b] -> [(a,b)]
zipLazy [] ys = []
zipLazy (x:xs) ~(y:ys) = (x,y) : zipLazy xs ys
\end{code}

208 209

\begin{code}
210 211 212 213 214 215 216 217 218 219
stretchZipWith :: (a -> Bool) -> b -> (a->b->c) -> [a] -> [b] -> [c]
-- (stretchZipWith p z f xs ys) stretches ys by inserting z in 
-- the places where p returns *True*

stretchZipWith p z f [] ys = []
stretchZipWith p z f (x:xs) ys
  | p x       = f x z : stretchZipWith p z f xs ys
  | otherwise = case ys of
		  []     -> []
		  (y:ys) -> f x y : stretchZipWith p z f xs ys
220 221 222
\end{code}


223
\begin{code}
224 225 226 227 228 229
mapFst :: (a->c) -> [(a,b)] -> [(c,b)]
mapSnd :: (b->c) -> [(a,b)] -> [(a,c)]

mapFst f xys = [(f x, y) | (x,y) <- xys]
mapSnd f xys = [(x, f y) | (x,y) <- xys]

230 231 232 233 234 235 236 237 238
mapAndUnzip :: (a -> (b, c)) -> [a] -> ([b], [c])

mapAndUnzip f [] = ([],[])
mapAndUnzip f (x:xs)
  = let
	(r1,  r2)  = f x
	(rs1, rs2) = mapAndUnzip f xs
    in
    (r1:rs1, r2:rs2)
239 240 241 242 243 244 245 246 247 248

mapAndUnzip3 :: (a -> (b, c, d)) -> [a] -> ([b], [c], [d])

mapAndUnzip3 f [] = ([],[],[])
mapAndUnzip3 f (x:xs)
  = let
	(r1,  r2,  r3)  = f x
	(rs1, rs2, rs3) = mapAndUnzip3 f xs
    in
    (r1:rs1, r2:rs2, r3:rs3)
249 250
\end{code}

251 252
\begin{code}
nOfThem :: Int -> a -> [a]
sof's avatar
sof committed
253
nOfThem n thing = replicate n thing
254

sof's avatar
sof committed
255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276
-- 'atLength atLen atEnd ls n' unravels list 'ls' to position 'n';
-- specification:
--
--  atLength atLenPred atEndPred ls n
--   | n < 0         = atLenPred n
--   | length ls < n = atEndPred (n - length ls)
--   | otherwise     = atLenPred (drop n ls)
--
atLength :: ([a] -> b)
         -> (Int -> b)
         -> [a]
         -> Int
         -> b
atLength atLenPred atEndPred ls n 
  | n < 0     = atEndPred n 
  | otherwise = go n ls
  where
    go n [] = atEndPred n
    go 0 ls = atLenPred ls
    go n (_:xs) = go (n-1) xs

-- special cases.
277
lengthExceeds :: [a] -> Int -> Bool
278
-- (lengthExceeds xs n) = (length xs > n)
sof's avatar
sof committed
279
lengthExceeds = atLength notNull (const False)
sof's avatar
sof committed
280 281

lengthAtLeast :: [a] -> Int -> Bool
sof's avatar
sof committed
282
lengthAtLeast = atLength notNull (== 0)
sof's avatar
sof committed
283 284 285 286 287 288 289 290 291 292 293 294 295 296

lengthIs :: [a] -> Int -> Bool
lengthIs = atLength null (==0)

listLengthCmp :: [a] -> Int -> Ordering 
listLengthCmp = atLength atLen atEnd 
 where
  atEnd 0      = EQ
  atEnd x
   | x > 0     = LT -- not yet seen 'n' elts, so list length is < n.
   | otherwise = GT

  atLen []     = EQ
  atLen _      = GT
297

298 299 300 301 302 303 304 305 306 307 308 309
equalLength :: [a] -> [b] -> Bool
equalLength [] []         = True
equalLength (_:xs) (_:ys) = equalLength xs ys
equalLength xs    ys      = False

compareLength :: [a] -> [b] -> Ordering
compareLength [] []         = EQ
compareLength (_:xs) (_:ys) = compareLength xs ys
compareLength [] _ys        = LT
compareLength _xs []        = GT

----------------------------
310 311 312
singleton :: a -> [a]
singleton x = [x]

313 314 315
isSingleton :: [a] -> Bool
isSingleton [x] = True
isSingleton  _  = False
316

sof's avatar
sof committed
317 318 319 320
notNull :: [a] -> Bool
notNull [] = False
notNull _  = True

321 322 323 324 325 326
only :: [a] -> a
#ifdef DEBUG
only [a] = a
#else
only (a:_) = a
#endif
327 328 329
\end{code}

Debugging/specialising versions of \tr{elem} and \tr{notElem}
330

331 332 333 334 335 336 337 338 339 340 341 342 343 344
\begin{code}
isIn, isn'tIn :: (Eq a) => String -> a -> [a] -> Bool

# ifndef DEBUG
isIn    msg x ys = elem__    x ys
isn'tIn msg x ys = notElem__ x ys

--these are here to be SPECIALIZEd (automagically)
elem__ _ []	= False
elem__ x (y:ys)	= x==y || elem__ x ys

notElem__ x []	   =  True
notElem__ x (y:ys) =  x /= y && notElem__ x ys

ross's avatar
ross committed
345
# else /* DEBUG */
346
isIn msg x ys
347
  = elem (_ILIT 0) x ys
348 349 350
  where
    elem i _ []	    = False
    elem i x (y:ys)
351 352 353
      | i ># _ILIT 100 = trace ("Over-long elem in " ++ msg) $
			 x `List.elem` (y:ys)
      | otherwise      = x == y || elem (i +# _ILIT(1)) x ys
354 355

isn'tIn msg x ys
356
  = notElem (_ILIT 0) x ys
357 358 359
  where
    notElem i x [] =  True
    notElem i x (y:ys)
360 361 362
      | i ># _ILIT 100 = trace ("Over-long notElem in " ++ msg) $
		         x `List.notElem` (y:ys)
      | otherwise      =  x /= y && notElem (i +# _ILIT(1)) x ys
ross's avatar
ross committed
363
# endif /* DEBUG */
364 365
\end{code}

366 367 368 369 370 371 372 373 374 375
foldl1' was added in GHC 6.4

\begin{code}
#if __GLASGOW_HASKELL__ < 604
foldl1'                  :: (a -> a -> a) -> [a] -> a
foldl1' f (x:xs)         =  foldl' f x xs
foldl1' _ []             =  panic "foldl1'"
#endif
\end{code}

376 377 378 379 380 381 382 383 384 385 386 387
%************************************************************************
%*									*
\subsubsection[Utils-Carsten-mergesort]{A mergesort from Carsten}
%*									*
%************************************************************************

\begin{display}
Date: Mon, 3 May 93 20:45:23 +0200
From: Carsten Kehler Holst <kehler@cs.chalmers.se>
To: partain@dcs.gla.ac.uk
Subject: natural merge sort beats quick sort [ and it is prettier ]

388
Here is a piece of Haskell code that I'm rather fond of. See it as an
389 390
attempt to get rid of the ridiculous quick-sort routine. group is
quite useful by itself I think it was John's idea originally though I
391
believe the lazy version is due to me [surprisingly complicated].
392 393
gamma [used to be called] is called gamma because I got inspired by
the Gamma calculus. It is not very close to the calculus but does
394 395 396
behave less sequentially than both foldr and foldl. One could imagine
a version of gamma that took a unit element as well thereby avoiding
the problem with empty lists.
397 398 399 400 401 402

I've tried this code against

   1) insertion sort - as provided by haskell
   2) the normal implementation of quick sort
   3) a deforested version of quick sort due to Jan Sparud
403
   4) a super-optimized-quick-sort of Lennart's
404 405 406 407

If the list is partially sorted both merge sort and in particular
natural merge sort wins. If the list is random [ average length of
rising subsequences = approx 2 ] mergesort still wins and natural
408
merge sort is marginally beaten by Lennart's soqs. The space
409
consumption of merge sort is a bit worse than Lennart's quick sort
410 411 412
approx a factor of 2. And a lot worse if Sparud's bug-fix [see his
fpca article ] isn't used because of group.

413
have fun
414 415 416 417 418
Carsten
\end{display}

\begin{code}
group :: (a -> a -> Bool) -> [a] -> [[a]]
419 420 421 422 423 424 425
-- Given a <= function, group finds maximal contiguous up-runs 
-- or down-runs in the input list.
-- It's stable, in the sense that it never re-orders equal elements
--
-- Date: Mon, 12 Feb 1996 15:09:41 +0000
-- From: Andy Gill <andy@dcs.gla.ac.uk>
-- Here is a `better' definition of group.
426

427 428 429 430 431
group p []     = []
group p (x:xs) = group' xs x x (x :)
  where
    group' []     _     _     s  = [s []]
    group' (x:xs) x_min x_max s 
432 433
	|      x_max `p` x  = group' xs x_min x (s . (x :)) 
	| not (x_min `p` x) = group' xs x x_max ((x :) . s) 
434
	| otherwise         = s [] : group' xs x x (x :) 
435 436
	-- NB: the 'not' is essential for stablity
	--	x `p` x_min would reverse equal elements
437 438 439 440

generalMerge :: (a -> a -> Bool) -> [a] -> [a] -> [a]
generalMerge p xs [] = xs
generalMerge p [] ys = ys
441
generalMerge p (x:xs) (y:ys) | x `p` y   = x : generalMerge p xs (y:ys)
442
			     | otherwise = y : generalMerge p (x:xs) ys
443 444 445 446 447 448 449 450 451 452 453 454

-- gamma is now called balancedFold

balancedFold :: (a -> a -> a) -> [a] -> a
balancedFold f [] = error "can't reduce an empty list using balancedFold"
balancedFold f [x] = x
balancedFold f l  = balancedFold f (balancedFold' f l)

balancedFold' :: (a -> a -> a) -> [a] -> [a]
balancedFold' f (x:y:xs) = f x y : balancedFold' f xs
balancedFold' f xs = xs

455 456
generalNaturalMergeSort p [] = []
generalNaturalMergeSort p xs = (balancedFold (generalMerge p) . group p) xs
457

458
#if NOT_USED
459 460 461
generalMergeSort p [] = []
generalMergeSort p xs = (balancedFold (generalMerge p) . map (: [])) xs

462 463 464 465 466 467
mergeSort, naturalMergeSort :: Ord a => [a] -> [a]

mergeSort = generalMergeSort (<=)
naturalMergeSort = generalNaturalMergeSort (<=)

mergeSortLe le = generalMergeSort le
468 469
#endif

470 471
sortLe :: (a->a->Bool) -> [a] -> [a]
sortLe le = generalNaturalMergeSort le
472 473 474 475 476

sortWith :: Ord b => (a->b) -> [a] -> [a]
sortWith get_key xs = sortLe le xs
  where
    x `le` y = get_key x < get_key y	
477 478 479 480

on :: (a -> a -> Ordering) -> (b -> a) -> b -> b -> Ordering
on cmp sel = \x y -> sel x `cmp` sel y

481 482 483 484 485 486 487 488 489 490 491 492 493
\end{code}

%************************************************************************
%*									*
\subsection[Utils-transitive-closure]{Transitive closure}
%*									*
%************************************************************************

This algorithm for transitive closure is straightforward, albeit quadratic.

\begin{code}
transitiveClosure :: (a -> [a])		-- Successor function
		  -> (a -> a -> Bool)	-- Equality predicate
494
		  -> [a]
495 496 497
		  -> [a]		-- The transitive closure

transitiveClosure succ eq xs
498
 = go [] xs
499
 where
500 501 502
   go done [] 			   = done
   go done (x:xs) | x `is_in` done = go done xs
   		  | otherwise      = go (x:done) (succ x ++ xs)
503 504 505 506 507 508 509 510 511 512 513 514

   x `is_in` []                 = False
   x `is_in` (y:ys) | eq x y    = True
  		    | otherwise = x `is_in` ys
\end{code}

%************************************************************************
%*									*
\subsection[Utils-accum]{Accumulating}
%*									*
%************************************************************************

515 516 517 518 519 520
A combination of foldl with zip.  It works with equal length lists.

\begin{code}
foldl2 :: (acc -> a -> b -> acc) -> acc -> [a] -> [b] -> acc
foldl2 k z [] [] = z
foldl2 k z (a:as) (b:bs) = foldl2 k (k z a b) as bs
521 522 523 524 525 526 527

all2 :: (a -> b -> Bool) -> [a] -> [b] -> Bool
-- True if the lists are the same length, and 
-- all corresponding elements satisfy the predicate
all2 p []     []     = True
all2 p (x:xs) (y:ys) = p x y && all2 p xs ys
all2 p xs     ys     = False
528 529 530 531 532 533 534 535 536 537 538
\end{code}

Count the number of times a predicate is true

\begin{code}
count :: (a -> Bool) -> [a] -> Int
count p [] = 0
count p (x:xs) | p x       = 1 + count p xs
	       | otherwise = count p xs
\end{code}

sof's avatar
sof committed
539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562
@splitAt@, @take@, and @drop@ but with length of another
list giving the break-off point:

\begin{code}
takeList :: [b] -> [a] -> [a]
takeList [] _ = []
takeList (_:xs) ls = 
   case ls of
     [] -> []
     (y:ys) -> y : takeList xs ys

dropList :: [b] -> [a] -> [a]
dropList [] xs    = xs
dropList _  xs@[] = xs
dropList (_:xs) (_:ys) = dropList xs ys


splitAtList :: [b] -> [a] -> ([a], [a])
splitAtList [] xs     = ([], xs)
splitAtList _ xs@[]   = (xs, xs)
splitAtList (_:xs) (y:ys) = (y:ys', ys'')
    where
      (ys', ys'') = splitAtList xs ys

563 564 565 566 567 568 569 570 571
snocView :: [a] -> Maybe ([a],a)
	-- Split off the last element
snocView [] = Nothing
snocView xs = go [] xs
	    where
		-- Invariant: second arg is non-empty
	      go acc [x]    = Just (reverse acc, x)
	      go acc (x:xs) = go (x:acc) xs

572 573 574 575 576
split :: Char -> String -> [String]
split c s = case rest of
		[]     -> [chunk] 
		_:rest -> chunk : split c rest
  where (chunk, rest) = break (==c) s
sof's avatar
sof committed
577 578
\end{code}

579

580 581 582 583 584 585
%************************************************************************
%*									*
\subsection[Utils-comparison]{Comparisons}
%*									*
%************************************************************************

586
\begin{code}
587 588 589 590 591 592 593 594 595 596 597
isEqual :: Ordering -> Bool
-- Often used in (isEqual (a `compare` b))
isEqual GT = False
isEqual EQ = True
isEqual LT = False

thenCmp :: Ordering -> Ordering -> Ordering
{-# INLINE thenCmp #-}
thenCmp EQ   any = any
thenCmp other any = other

598 599 600 601 602
eqListBy :: (a->a->Bool) -> [a] -> [a] -> Bool
eqListBy eq []     []     = True
eqListBy eq (x:xs) (y:ys) = eq x y && eqListBy eq xs ys
eqListBy eq xs     ys     = False

603
cmpList :: (a -> a -> Ordering) -> [a] -> [a] -> Ordering
604 605
    -- `cmpList' uses a user-specified comparer

606 607 608
cmpList cmp []     [] = EQ
cmpList cmp []     _  = LT
cmpList cmp _      [] = GT
609
cmpList cmp (a:as) (b:bs)
610
  = case cmp a b of { EQ -> cmpList cmp as bs; xxx -> xxx }
611 612
\end{code}

613
\begin{code}
614 615
-- This (with a more general type) is Data.List.stripPrefix from GHC 6.8.
-- This definition can be removed once we require at least 6.8 to build.
616 617 618 619 620 621 622
maybePrefixMatch :: String -> String -> Maybe String
maybePrefixMatch []    rest = Just rest
maybePrefixMatch (_:_) []   = Nothing
maybePrefixMatch (p:pat) (r:rest)
  | p == r    = maybePrefixMatch pat rest
  | otherwise = Nothing

623 624
removeSpaces :: String -> String
removeSpaces = reverse . dropWhile isSpace . reverse . dropWhile isSpace
625 626 627 628 629 630 631 632 633 634 635 636 637
\end{code}

%************************************************************************
%*									*
\subsection[Utils-pairs]{Pairs}
%*									*
%************************************************************************

\begin{code}
unzipWith :: (a -> b -> c) -> [(a, b)] -> [c]
unzipWith f pairs = map ( \ (a, b) -> f a b ) pairs
\end{code}

638 639 640 641
\begin{code}
seqList :: [a] -> b -> b
seqList [] b = b
seqList (x:xs) b = x `seq` seqList xs b
642
\end{code}
643 644 645 646 647 648 649

Global variables:

\begin{code}
global :: a -> IORef a
global a = unsafePerformIO (newIORef a)
\end{code}
650

651 652 653 654 655 656 657
\begin{code}
consIORef :: IORef [a] -> a -> IO ()
consIORef var x = do
  xs <- readIORef var
  writeIORef var (x:xs)
\end{code}

658 659 660
Module names:

\begin{code}
Ian Lynagh's avatar
Ian Lynagh committed
661
looksLikeModuleName :: String -> Bool
662
looksLikeModuleName [] = False
663 664 665 666
looksLikeModuleName (c:cs) = isUpper c && go cs
  where go [] = True
	go ('.':cs) = looksLikeModuleName cs
	go (c:cs)   = (isAlphaNum c || c == '_') && go cs
667
\end{code}
sof's avatar
sof committed
668

669 670 671 672
Akin to @Prelude.words@, but acts like the Bourne shell, treating
quoted strings and escaped characters within the input as solid blocks
of characters.  Doesn't raise any exceptions on malformed escapes or
quoting.
sof's avatar
sof committed
673 674 675 676 677

\begin{code}
toArgs :: String -> [String]
toArgs "" = []
toArgs s  =
678 679 680
  case dropWhile isSpace s of  -- drop initial spacing
    [] -> []  -- empty, so no more tokens
    rem -> let (tok,aft) = token rem [] in tok : toArgs aft
sof's avatar
sof committed
681
 where
682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707
   -- Grab a token off the string, given that the first character exists and
   -- isn't whitespace.  The second argument is an accumulator which has to be
   -- reversed at the end.
  token [] acc = (reverse acc,[])            -- out of characters
  token ('\\':c:aft) acc                     -- escapes
               = token aft ((escape c) : acc)
  token (q:aft) acc | q == '"' || q == '\''  -- open quotes
               = let (aft',acc') = quote q aft acc in token aft' acc'
  token (c:aft) acc | isSpace c              -- unescaped, unquoted spacing
               = (reverse acc,aft)
  token (c:aft) acc                          -- anything else goes in the token
               = token aft (c:acc)

   -- Get the appropriate character for a single-character escape.
  escape 'n' = '\n'
  escape 't' = '\t'
  escape 'r' = '\r'
  escape c   = c

   -- Read into accumulator until a quote character is found.
  quote qc =
    let quote' [] acc                  = ([],acc)
        quote' ('\\':c:aft) acc        = quote' aft ((escape c) : acc)
        quote' (c:aft) acc | c == qc   = (aft,acc)
        quote' (c:aft) acc             = quote' aft (c:acc)
    in quote'
sof's avatar
sof committed
708
\end{code}
709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757

-- -----------------------------------------------------------------------------
-- Floats

\begin{code}
readRational__ :: ReadS Rational -- NB: doesn't handle leading "-"
readRational__ r = do 
     (n,d,s) <- readFix r
     (k,t)   <- readExp s
     return ((n%1)*10^^(k-d), t)
 where
     readFix r = do
	(ds,s)  <- lexDecDigits r
	(ds',t) <- lexDotDigits s
	return (read (ds++ds'), length ds', t)

     readExp (e:s) | e `elem` "eE" = readExp' s
     readExp s			   = return (0,s)

     readExp' ('+':s) = readDec s
     readExp' ('-':s) = do
			(k,t) <- readDec s
			return (-k,t)
     readExp' s	      = readDec s

     readDec s = do
        (ds,r) <- nonnull isDigit s
        return (foldl1 (\n d -> n * 10 + d) [ ord d - ord '0' | d <- ds ],
                r)

     lexDecDigits = nonnull isDigit

     lexDotDigits ('.':s) = return (span isDigit s)
     lexDotDigits s       = return ("",s)

     nonnull p s = do (cs@(_:_),t) <- return (span p s)
                      return (cs,t)

readRational :: String -> Rational -- NB: *does* handle a leading "-"
readRational top_s
  = case top_s of
      '-' : xs -> - (read_me xs)
      xs       -> read_me xs
  where
    read_me s
      = case (do { (x,"") <- readRational__ s ; return x }) of
	  [x] -> x
	  []  -> error ("readRational: no parse:"        ++ top_s)
	  _   -> error ("readRational: ambiguous parse:" ++ top_s)
758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788


-----------------------------------------------------------------------------
-- Create a hierarchy of directories

createDirectoryHierarchy :: FilePath -> IO ()
createDirectoryHierarchy dir = do
  b <- doesDirectoryExist dir
  when (not b) $ do
	createDirectoryHierarchy (directoryOf dir)
	createDirectory dir

-----------------------------------------------------------------------------
-- Verify that the 'dirname' portion of a FilePath exists.
-- 
doesDirNameExist :: FilePath -> IO Bool
doesDirNameExist fpath = doesDirectoryExist (directoryOf fpath)

-- -----------------------------------------------------------------------------
-- Exception utils

later = flip finally

handleDyn :: Typeable ex => (ex -> IO a) -> IO a -> IO a
handleDyn = flip catchDyn

handle :: (Exception -> IO a) -> IO a -> IO a
handle h f = f `Exception.catch` \e -> case e of
    ExitException _ -> throw e
    _               -> h e

789 790 791 792 793 794 795 796 797 798
-- --------------------------------------------------------------
-- check existence & modification time at the same time

modificationTimeIfExists :: FilePath -> IO (Maybe ClockTime)
modificationTimeIfExists f = do
  (do t <- getModificationTime f; return (Just t))
	`IO.catch` \e -> if isDoesNotExistError e 
			then return Nothing 
			else ioError e

799 800 801
-- --------------------------------------------------------------
-- Filename manipulation
		
802 803 804 805
-- Filenames are kept "normalised" inside GHC, using '/' as the path
-- separator.  On Windows these functions will also recognise '\\' as
-- the path separator, but will generally construct paths using '/'.

806 807 808 809 810
type Suffix = String

splitFilename :: String -> (String,Suffix)
splitFilename f = splitLongestPrefix f (=='.')

811 812 813 814 815
basenameOf :: FilePath -> String
basenameOf = fst . splitFilename

suffixOf :: FilePath -> Suffix
suffixOf = snd . splitFilename
816

817 818 819 820
joinFileExt :: String -> String -> FilePath
joinFileExt path ""  = path
joinFileExt path ext = path ++ '.':ext

821 822 823
-- "foo/bar/xyzzy.ext" -> ("foo/bar", "xyzzy.ext")
splitFilenameDir :: String -> (String,String)
splitFilenameDir str
824 825 826 827
   = let (dir, rest) = splitLongestPrefix str isPathSeparator
	 (dir', rest') | null rest = (".", dir)
		       | otherwise = (dir, rest)
     in  (dir', rest')
828 829 830 831

-- "foo/bar/xyzzy.ext" -> ("foo/bar", "xyzzy", ".ext")
splitFilename3 :: String -> (String,String,Suffix)
splitFilename3 str
832 833 834 835 836 837 838 839 840
   = let (dir, rest) = splitFilenameDir str
	 (name, ext) = splitFilename rest
     in  (dir, name, ext)

joinFileName :: String -> String -> FilePath
joinFileName ""  fname = fname
joinFileName "." fname = fname
joinFileName dir ""    = dir
joinFileName dir fname = dir ++ '/':fname
841 842 843 844 845 846 847 848

-- split a string at the last character where 'pred' is True,
-- returning a pair of strings. The first component holds the string
-- up (but not including) the last character for which 'pred' returned
-- True, the second whatever comes after (but also not including the
-- last character).
--
-- If 'pred' returns False for all characters in the string, the original
849
-- string is returned in the first component (and the second one is just
850 851
-- empty).
splitLongestPrefix :: String -> (Char -> Bool) -> (String,String)
852 853 854 855 856 857
splitLongestPrefix str pred
  | null r_pre = (str,           [])
  | otherwise  = (reverse (tail r_pre), reverse r_suf)
	-- 'tail' drops the char satisfying 'pred'
  where 
    (r_suf, r_pre) = break pred (reverse str)
858 859

replaceFilenameSuffix :: FilePath -> Suffix -> FilePath
860
replaceFilenameSuffix file suf = basenameOf file `joinFileExt` suf
861 862 863 864 865 866 867 868 869 870 871 872

-- directoryOf strips the filename off the input string, returning
-- the directory.
directoryOf :: FilePath -> String
directoryOf = fst . splitFilenameDir

-- filenameOf strips the directory off the input string, returning
-- the filename.
filenameOf :: FilePath -> String
filenameOf = snd . splitFilenameDir

replaceFilenameDirectory :: FilePath -> String -> FilePath
873
replaceFilenameDirectory path dir = dir `joinFileName` filenameOf path
874 875 876 877 878 879 880 881 882 883 884

escapeSpaces :: String -> String
escapeSpaces = foldr (\c s -> if isSpace c then '\\':c:s else c:s) ""

isPathSeparator :: Char -> Bool
isPathSeparator ch =
#ifdef mingw32_TARGET_OS
  ch == '/' || ch == '\\'
#else
  ch == '/'
#endif
885

886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919
--------------------------------------------------------------
-- * Search path
--------------------------------------------------------------

-- | The function splits the given string to substrings
-- using the 'searchPathSeparator'.
parseSearchPath :: String -> [FilePath]
parseSearchPath path = split path
  where
    split :: String -> [String]
    split s =
      case rest' of
        []     -> [chunk] 
        _:rest -> chunk : split rest
      where
        chunk = 
          case chunk' of
#ifdef mingw32_HOST_OS
            ('\"':xs@(_:_)) | last xs == '\"' -> init xs
#endif
            _                                 -> chunk'

        (chunk', rest') = break (==searchPathSeparator) s

-- | A platform-specific character used to separate search path strings in 
-- environment variables. The separator is a colon (\":\") on Unix and Macintosh, 
-- and a semicolon (\";\") on the Windows operating system.
searchPathSeparator :: Char
#if mingw32_HOST_OS || mingw32_TARGET_OS
searchPathSeparator = ';'
#else
searchPathSeparator = ':'
#endif

920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953
-----------------------------------------------------------------------------
-- Convert filepath into platform / MSDOS form.

-- We maintain path names in Unix form ('/'-separated) right until 
-- the last moment.  On Windows we dos-ify them just before passing them
-- to the Windows command.
-- 
-- The alternative, of using '/' consistently on Unix and '\' on Windows,
-- proved quite awkward.  There were a lot more calls to platformPath,
-- and even on Windows we might invoke a unix-like utility (eg 'sh'), which
-- interpreted a command line 'foo\baz' as 'foobaz'.

normalisePath :: String -> String
-- Just changes '\' to '/'

pgmPath :: String		-- Directory string in Unix format
	-> String		-- Program name with no directory separators
				--	(e.g. copy /y)
	-> String		-- Program invocation string in native format

#if defined(mingw32_HOST_OS)
--------------------- Windows version ------------------
normalisePath xs = subst '\\' '/' xs
pgmPath dir pgm  = platformPath dir ++ '\\' : pgm
platformPath p   = subst '/' '\\' p

subst a b ls = map (\ x -> if x == a then b else x) ls
#else
--------------------- Non-Windows version --------------
normalisePath xs   = xs
pgmPath dir pgm    = dir ++ '/' : pgm
platformPath stuff = stuff
--------------------------------------------------------
#endif
954
\end{code}