System.Directory.Tree

filesystem, as well as useful functions for reading and writing file and directory structures in the IO monad.

Errors are caught in a special constructor in the DirTree type.

Defined instances of Functor, Traversable and Foldable allow for easily operating on a directory of files. For example, you could use Foldable.foldr to create a hash of the entire contents of a directory.

The functions readDirectoryWithL and buildL allow for doing directory-traversing IO lazily as required by the execution of pure code. This allows you to treat large directories the same way as you would a lazy infinite list.

The AnchoredDirTree type is a simple wrapper for DirTree to keep track of a base directory context for the DirTree.

Please send me any requests, bugs, or other feedback on this module!

Data types for representing directory trees

data DirTree a

the String in the name field is always a file name, never a full path. The free type variable is used in the File constructor and can hold Handles, Strings representing a file's contents or anything else you can think of. We catch any IO errors in the Failed constructor. an Exception can be converted to a String with show.

Constructors

Failed

Fields

name :: FileName

err :: IOException

Dir

Fields

name :: FileName

contents :: [DirTree a]

File

Fields

name :: FileName

file :: a

Instances

Functor DirTree

Foldable DirTree

Traversable DirTree

Eq a => Eq (DirTree a)

Two DirTrees are equal if they have the same constructor, the same name (and in the case of Dirs) their sorted contents are equal:

(Ord a, Eq a) => Ord (DirTree a)

First compare constructors: Failed < Dir < File... Then compare name... Then compare free variable parameter of File constructors

Show a => Show (DirTree a)

data AnchoredDirTree a

a simple wrapper to hold a base directory name, which can be either an absolute or relative path. This lets us give the DirTree a context, while still letting us store only directory and file names (not full paths) in the DirTree. (uses an infix constructor; don't be scared)

Constructors

:/

Fields

anchor :: FilePath

dirTree :: DirTree a

Instances

Functor AnchoredDirTree

Eq a => Eq (AnchoredDirTree a)

Ord a => Ord (AnchoredDirTree a)

Show a => Show (AnchoredDirTree a)

type FileName = String

an element in a FilePath:

High level IO functions

readDirectory :: FilePath -> IO (AnchoredDirTree String)

build an AnchoredDirTree, given the path to a directory, opening the files using readFile. Uses readDirectoryWith internally and has the effect of traversing the entire directory structure. See readDirectoryWithL for lazy production of a DirTree structure.

readDirectoryWith :: (FilePath -> IO a) -> FilePath -> IO (AnchoredDirTree a)

same as readDirectory but allows us to, for example, use ByteString.readFile to return a tree of ByteStrings.

readDirectoryWithL :: (FilePath -> IO a) -> FilePath -> IO (AnchoredDirTree a)

A lazy version of readDirectoryWith that does IO operations as needed i.e. as the tree is traversed in pure code. NOTE: This function uses unsafePerformIO under the hood. I believe our use here is safe, but this function is experimental in this release:

writeDirectory :: AnchoredDirTree String -> IO (AnchoredDirTree ())

write a DirTree of strings to disk. Clobbers files of the same name. Doesn't affect files in the directories (if any already exist) with different names. Returns a new AnchoredDirTree where failures were lifted into a Failed constructor:

writeDirectoryWith :: (FilePath -> a -> IO b) -> AnchoredDirTree a -> IO (AnchoredDirTree b)

writes the directory structure to disk and uses the provided function to write the contents of Files to disk. The return value of the function will become the new contents of the returned, where IO errors at each node are replaced with Failed constructors. The returned tree can be compared to the passed tree to see what operations, if any, failed:

Lower level functions

build :: FilePath -> IO (AnchoredDirTree FilePath)

builds a DirTree from the contents of the directory passed to it, saving the base directory in the Anchored* wrapper. Errors are caught in the tree in the Failed constructor. The file fields initially are populated with full paths to the files they are abstracting.

buildL :: FilePath -> IO (AnchoredDirTree FilePath)

identical to build but does directory reading IO lazily as needed:

openDirectory :: FilePath -> IOMode -> IO (AnchoredDirTree Handle)

a simple application of readDirectoryWith openFile:

writeJustDirs :: AnchoredDirTree a -> IO (AnchoredDirTree a)

writes the directory structure (not files) of a DirTree to the anchored directory. Returns a structure identical to the supplied tree with errors replaced by Failed constructors:

Manipulating FilePaths

zipPaths :: AnchoredDirTree a -> DirTree (FilePath, a)

tuple up the complete file path with the file contents, by building up the path, trie-style, from the root. The filepath will be relative to "anchored" directory.

This allows us to, for example, mapM_ uncurry writeFile over a DirTree of strings, although writeDirectory does a better job of this.

free :: AnchoredDirTree a -> DirTree a

Deprecated: Use record dirTree

DEPRECATED. Use record dirTree instead.

Utility functions

Shape comparison and equality

equalShape :: DirTree a -> DirTree b -> Bool

Tests equality of two trees, ignoring their free variable portion. Can be used to check if any files have been added or deleted, for instance.

comparingShape :: DirTree a -> DirTree b -> Ordering

a compare function that ignores the free file type variable:

Handling failure

successful :: DirTree a -> Bool

True if there are no Failed constructors in the tree

anyFailed :: DirTree a -> Bool

True if any Failed constructors in the tree

failed :: DirTree a -> Bool

returns true if argument is a Failed constructor:

failures :: DirTree a -> [DirTree a]

returns a list of Failed constructors only:

failedMap :: (FileName -> IOException -> DirTree a) -> DirTree a -> DirTree a

maps a function to convert Failed DirTrees to Files or Dirs

Tree Manipulations

flattenDir :: DirTree a -> [DirTree a]

Flattens a DirTree into a (never empty) list of tree constructors. Dir constructors will have [] as their contents:

sortDir :: Ord a => DirTree a -> DirTree a

Recursively sort a directory tree according to the Ord instance

sortDirShape :: DirTree a -> DirTree a

Recursively sort a tree as in sortDir but ignore the file contents of a File constructor

filterDir :: (DirTree a -> Bool) -> DirTree a -> DirTree a

applies the predicate to each constructor in the tree, removing it (and its children, of course) when the predicate returns False. The topmost constructor will always be preserved:

Low-level

transformDir :: (DirTree a -> DirTree a) -> DirTree a -> DirTree a

At Dir constructor, apply transformation function to all of directory's contents, then remove the Nothing's and recurse. This always preserves the topomst constructor.

Navigation

dropTo :: FileName -> AnchoredDirTree a -> Maybe (AnchoredDirTree a)

If the argument is a Dir containing a sub-DirTree matching FileName then return that subtree, appending the name of the old root Dir to the anchor of the AnchoredDirTree wrapper. Otherwise return Nothing.

Operators

(</$>) :: Functor f => (DirTree a -> DirTree b) -> f (AnchoredDirTree a) -> f (AnchoredDirTree b)

Allows for a function on a bare DirTree to be applied to an AnchoredDirTree within a Functor. Very similar to and useful in combination with <$>:

Lenses

These are compatible with the lens library

_contents :: Applicative f => ([DirTree a] -> f [DirTree a]) -> DirTree a -> f (DirTree a)

_err :: Applicative f => (IOException -> f IOException) -> DirTree a -> f (DirTree a)

_file :: Applicative f => (a -> f a) -> DirTree a -> f (DirTree a)

_name :: Functor f => (FileName -> f FileName) -> DirTree a -> f (DirTree a)

_anchor :: Functor f => (FilePath -> f FilePath) -> AnchoredDirTree a -> f (AnchoredDirTree a)

_dirTree :: Functor f => (DirTree t -> f (DirTree a)) -> AnchoredDirTree t -> f (AnchoredDirTree a)