Personal tools

Cookbook

From HaskellWiki

(Difference between revisions)
Jump to: navigation, search
m (the ++ operator works for any list type)
m (Added some more subheadings for Strings.)
Line 96: Line 96:
 
"foobar"
 
"foobar"
 
</haskell>
 
</haskell>
  +
  +
=== Regular expressions ===
  +
=== Interpolation ===
  +
=== Performance ===
  +
=== Unicode (?) ===
  +
   
 
== Numbers ==
 
== Numbers ==

Revision as of 15:10, 26 February 2007

We need to start a GOOD (aka, not a PLEAC clone) Haskell cookbook.

This page is based on the Scheme Cookbook at http://schemecookbook.org/Cookbook/WebHome

Contents

1 Prelude

A lot of functions are defined in the "Prelude". Also, if you ever want to search for a function, based on the name, type or module, take a look at the excellent Hoogle. This is for a lot of people a must-have while debugging and writing Haskell programs.

2 GHCi/Hugs

2.1 GHCi interaction

To start GHCi from a command prompt, simply type `ghci'

   $ ghci
      ___         ___ _
     / _ \ /\  /\/ __(_)
    / /_\// /_/ / /  | |      GHC Interactive, version 6.6, for Haskell 98.
   / /_\\/ __  / /___| |      http://www.haskell.org/ghc/
   \____/\/ /_/\____/|_|      Type :? for help.
   
   Loading package base ... linking ... done.
   Prelude>

Prelude is the "base" library of Haskell.

To create variables at the GHCi prompt, use `let'

Prelude> let x = 5
Prelude> x
5
Prelude> let y = 3
Prelude> y
3
Prelude> x + y
8

3 Types

To check the type of an expression or function, use the command `:t'

Prelude> :t x
x :: Integer
Prelude> :t y
y :: Integer

Haskell has the following types defined in the Standard Prelude.

    Int         -- bounded, word-sized integers
    Integer     -- unbounded integers
    Double      -- floating point values
    Char        -- characters
    String      -- strings are lists of characters
    ()          -- the unit type
    Bool        -- booleans
    [a]         -- lists
    (a,b)       -- tuples / product types
    Either a b  -- sum types
    Maybe a     -- optional values

4 Strings

4.1 Input

Strings can be read as input using getLine.

Prelude> getLine
Foo bar baz
"Foo bar baz"

4.2 Output

Strings can be output in a number of different ways.

Prelude> putStr "Foo"
FooPrelude>

As you can see, putStr does not include the newline character `\n'. We can either use putStr like this:

Prelude> putStr "Foo\n"
Foo

Or use putStrLn, which is already in the Standard Prelude

Prelude> putStrLn "Foo"
Foo

We can also use print to print a string, including the quotation marks.

Prelude> print "Foo"
"Foo"

4.3 Concatenation

Concatenation of strings (or any other list) is done with the `++' operator.

Prelude> "foo" ++ "bar"
"foobar"

4.4 Regular expressions

4.5 Interpolation

4.6 Performance

4.7 Unicode (?)

5 Numbers

Numbers in Haskell can be of the type
Int, Integer, Float, Double, or Rational
.

5.1 Random numbers

6 Dates and time

Use System.Time.getClockTime to get a properly formatted date stamp.

Prelude> System.Time.getClockTime
Wed Feb 21 20:05:35 CST 2007

7 Lists

In Haskell, lists are what Arrays are in most other languages. Haskell has all of the general list manipulation functions, see also
Data.List
.
Prelude> head [1,2,3]
1
 
Prelude> tail [1,2,3]
[2,3]
 
Prelude> length [1,2,3]
3

Furthermore, Haskell supports some neat concepts.

7.1 Infinite lists

Prelude> [1..]

The list of all squares:

square x = x*x
squares = map square [1..]
But in the end, you probably don't want to use infinite lists, but make them finite. You can do this with
take
:
Prelude> take 10 squares
[1,4,9,16,25,36,49,64,81,100]

7.2 List Comprehensions

The list of all squares can also be written in a more comprehensive way, using list comprehensions:

squares = [x*x | x <- [1..]]

8 Pattern matching

Haskell does implicit pattern matching.

A good example of pattern matching is done in the fact function for finding a factorial.

fact :: Integer -> Integer
fact 0 = 1
fact n = n * fact (n - 1)
In this function,
fact :: Integer -> Integer
is the functions type definition. The next line,
fact 0 = 1
is a pattern match, so when the argument to the function fact is 0, the return value is 1.

The 3rd and final line of this function is another pattern match, which says that, whatever number was entered as the argument, is multiplied by the factorial of that number, minus 1. Notice this function is recursive.

Pattern matching in Haskell evaluates the patterns in the order they are written, so
fact 0 = 1
is evaluated before
fact n = n * fact (n - 1)
.

9 Files

9.1 Simple IO

Using
interact :: (String -> String) -> IO ()
, you can easily do things with stdin and stdout.

A program to sum up numbers:

main = interact $ show . sum . map read . lines

A program that adds line numbers to each line:

main = interact numberLines
numberLines = unlines . zipWith combine [1..] . lines
 where combine lineNumber text = concat [show lineNumber, " ", text]

9.2 Reading from files

The System.IO library contains the functions needed for file IO. The program below displays the contents of the file c:\test.txt.

import System.IO
 
main = do
  h <- openFile "c:\\test.txt" ReadMode
  contents <- hGetContents h
  putStrLn contents
  hClose h

The same program, with some higher-lever functions:

main = do
  contents <- readFile "c:\\test.txt"
  putStrLn contents

9.3 Writing to files

The following program writes the first 100 squares to a file:

-- generate a list of squares with length 'num' in string-format.
numbers num = unlines $ take num $ map (show . \x -> x*x) [1..]
 
main = do
  writeFile "test.txt" (numbers 100)
  putStrLn "successfully written"
This will override the old contents of the file, or create a new file if the file doesn't exist yet. If you want to append to a file, you can use
appendFile
.

10 Data Structures

GHC comes with some handy data-structures by default. If you want to use a Map, use Data.Map. For sets, you can use Data.Set. A good way to find efficient data-structures is to take a look at the hierarchical libraries, see Haskell Hierarchical Libraries and scroll down to 'Data'.

10.1 Map

10.2 Set

10.3 Tree

10.4 ByteString

10.5 Arrays

Arrays are generally eschewed in Haskell. However, they are useful if you desperately need constant lookup or update or if you have huge amounts of raw data.

Immutable arrays like
Data.Array.IArray.Array i e
offer lookup in constant time but they get copied when you update an element. Use them if they can be filled in one go. The following example groups a list of numbers according to their residual after division by
n
in one go.
bucketByResidual :: Int -> [Int] -> Array Int [Int]
bucketByResidual n xs = accumArray (\xs x -> x:xs) [] (0,n-1) [(x `mod` n, x) | x <- xs]
 
Data.Arra.IArray> bucketByResidual 4 [x*x | x <- [1..10]]
array (0,3) [(0,[100,64,36,16,4]),(1,[81,49,25,9,1]),(2,[]),(3,[])]
 
Data.Arra.IArray> amap reverse it
array (0,3) [(0,[4,16,36,64,100]),(1,[1,9,25,49,81]),(2,[]),(3,[])]

Note that the array can fill itself up in a circular fashion. Useful for dynamic programming. Here is the edit distance between two strings without array updates.

editDistance :: Eq a => [a] -> [a] -> Int
editDistance xs ys = table ! (m,n)
    where
    (m,n) = (length xs, length ys)
    x     = array (1,m) (zip [1..] xs)
    y     = array (1,n) (zip [1..] ys)
 
    table :: Array (Int,Int) Int
    table = array bnds [(ij, dist ij) | ij <- range bnds]
    bnds  = ((0,0),(m,n))
 
    dist (0,j) = j
    dist (i,0) = i
    dist (i,j) = minimum [table ! (i-1,j) + 1, table ! (i,j-1) + 1,
        if x ! i == y ! j then table ! (i-1,j-1) else table ! (i-1,j-1)]


Mutable arrays like
Data.Array.IO.IOArray i e
are updated in place, but they have to live in the IO-monad or the ST-monad in order to not destroy referential transparency. There are also diff arrays like
Data.Array.Diff.DiffArray i e
that look like immutable arrays but do updates in place if used in a single threaded way. Here is depth first search with diff arrays that checks whether a directed graph contains a cycle. Note: this example really belongs to Map or Set.
import Control.Monad.State
type Node  = Int
data Color = White | Grey | Black 
 
hasCycle :: Array Node [Node] -> Bool
hasCycle graph = runState (mapDfs $ indices g) initSeen
    where
    initSeen :: DiffArray Node Color
    initSeen  = listArray (bounds graph) (repeat White)
    mapDfs    = fmap or . mapM dfs
    dfs node  = get >>= \seen -> case (seen ! node) of
        Black -> return False
        Grey  -> return True  -- we found a cycle
        White -> do
            modify $  \seen -> seen // [(node,Grey )]
            found  <- mapDfs (graph ! node)
            modify $  \seen -> seen // [(node,Black)]
            return found

11 Network Programming

12 XML

12.1 Parsing XML

13 Databases

13.1 MySQL

13.2 PostgreSQL

13.3 SQLite

14 FFI

14.1 How to interface with C