Documentation

Split a path in components. Repeated "/" characters don't lead to empty components. "." path components are removed. If the path is absolute, the first component will start with "/". ".." components are left intact. They can't be simply removed, because the preceding component might be a symlink. In this case, realpath is probably what you need.

The case that the path is empty, is probably an error. However, it is treated like ".", yielding an empty path components list.

Examples:

slice_path "/"        = ["/"]
slice_path "/foo/bar" = ["/foo","bar"]
slice_path "..//./"   = [".."]
slice_path "."        = []

See unslice_path, realpath, realpath_s.

Form a path from path components. This isn't the inverse of slice_path, since unslice_path . slice_path normalises the path.

See slice_path.

Normalise a path. This is done by reducing repeated / characters to one, and removing . path components. .. path components are left intact, because of possible symlinks.

normalise_path = unslice_path . slice_path

Split a file name in components. This are the base file name and the suffixes, which are separated by dots. If the name starts with a dot, it is regarded as part of the base name. The result is a list of file name components. The filename may be a path. In this case, everything up to the last path component will be returned as part of the base file name. The path gets normalised thereby.

No empty suffixes are returned. If the file name contains several consecutive dots, they are regared as part of the preceding file name component.

Concateneting the name components and adding dots, reproduces the original name, with a normalised path: concat . intersperse "." . slice_filename == normalise.

Note that the last path component might be "..". Then it is not possible to deduce the refered directory's name from the path. An IO action for getting the real path is then necessary.

Examples:

slice_filename "a.b//./.foo.tar.gz" == ["a.b/.foo","tar","gz"]
slice_filename ".x..y."             == [".x.", "y."]

See unslice_filename, slice_filename'.

This is a variant of slice_filename. It is like slice_filename, except for being more efficient, and the filename must not contain any preceding path, since this case isn't considered.

See slice_filename, unslice_filename.

Form file name from file name components, interspersing dots. This is the inverse of slice_filename, except for normalisation of any path.

unslice_filename = concat . intersperse "."

See slice_filename.

Split a path in directory and file name. Only in the case that the supplied path is empty, both parts are empty strings. Otherwise, "." is filled in for the corresponding part, if necessary. Unless the path is empty, concatenating the returned path and file name components with a slash in between, makes a valid path to the file.

split_path splits off the last path component. This isn't the same as the text after the last /.

Note that the last path component might be "..". Then it is not possible to deduce the refered directory's name from the path. Then an IO action for getting the real path is necessary.

Examples:

split_path "/a/b/c"      == ("/a/b", "c")
split_path "foo"         == (".", "foo")
split_path "foo/bar"     == ("foo", "bar")
split_path "foo/.."      == ("foo", "..")
split_path "."           == (".", ".")
split_path ""            == ("", "")
split_path "/foo"        == ("/", "foo")
split_path "foo/"        == (".", "foo")
split_path "foo/."       == (".", "foo")
split_path "foo///./bar" == ("foo", "bar")

See slice_path.

Get the directory part of a path.

dir_part = fst . split_path

See split_path.

Get the last path component of a path.

filename_part = snd . split_path

Examples:

filename_part "foo/bar" == "bar"
filename_part "."       == "."

See split_path.

Inverse of split_path, except for normalisation.

This concatenates two paths, and takes care of "." and empty paths. When the two components are the result of split_path, then unsplit_path creates a normalised path. It is best documented by its definition:

unsplit_path (".", "") = "."
unsplit_path ("", ".") = "."
unsplit_path (".", q)  = q
unsplit_path ("", q)   = q
unsplit_path (p, "")   = p
unsplit_path (p, ".")  = p
unsplit_path (p, q)    = p ++ "/" ++ q

Examples:

unsplit_path ("", "")     == ""
unsplit_path (".", "")    == "."
unsplit_path (".", ".")   == "."
unsplit_path ("foo", ".") == "foo"

See split_path.

Split a file name in prefix and suffix. If there isn't any suffix in the file name, then return an empty suffix. A dot at the beginning or at the end is not regarded as introducing a suffix.

The last path component is what is being split. This isn't the same as splitting the string at the last dot. For instance, if the file name doesn't contain any dot, dots in previous path component's aren't mistaken as introducing suffixes.

The path part is returned in normalised form. This means, "." components are removed, and multiple "/"s are reduced to one.

Note that there isn't any plausibility check performed on the suffix. If the file name doesn't have a suffix, but happens to contain a dot, then this dot is mistaken as introducing a suffix.

Examples:

split_filename "path/to/foo.bar"                             = ("path/to/foo","bar")
split_filename "path/to/foo"                                 = ("path/to/foo","")
split_filename "/path.to/foo"                                = ("/path.to/foo","")
split_filename "a///./x"                                     = ("a/x","")
split_filename "dir.suffix/./"                               = ("dir","suffix")
split_filename "Photographie, Das 20. Jahrhundert (300 dpi)" = ("Photographie, Das 20", " Jahrhundert (300 dpi)")

See slice_path, 'split_filename''

Variant of split_filename. This is a more efficient version of split_filename, for the case that you know the string is is a pure file name without any slashes.

See split_filename.

Inverse of split_filename. Concatenate prefix and suffix, adding a dot in between, iff the suffix is not empty. The path part of the prefix is normalised.

See split_filename.

Split a path in directory, base file name and suffix.

Form path from directory, base file name and suffix parts.

Test a path for a specific suffix and split it off.

If the path ends with the suffix, then the result is Just prefix, where prefix is the normalised path without the suffix. Otherwise it's Nothing.

Make a path absolute, using the current working directory.

This makes a relative path absolute with respect to the current working directory. An absolute path is returned unmodified.

The current working directory is determined with getCurrentDirectory which means that symbolic links in it are expanded and the path is normalised. This is different from pwd.

Make a path absolute.

This makes a relative path absolute with respect to a specified directory. An absolute path is returned unmodified.

Make a path absolute.

This makes a relative path absolute with respect to a specified directory. An absolute path is returned unmodified.

The order of the arguments can be confusing. You should rather use absolute_path_by. absolute_path' is included for backwards compatibility.

Guess the ".."-component free form of a path, specified as a list of path components, by syntactically removing them, along with the preceding path components. This will produce erroneous results when the path contains symlinks. If the path contains leading ".." components, or more ".." components than preceeding normal components, then the ".." components can't be normalised away. In this case, the result is Nothing.

Guess the ".."-component free, normalised form of a path. The transformation is purely syntactic. ".." path components will be removed, along with their preceding path components. This will produce erroneous results when the path contains symlinks. If the path contains leading ".." components, or more ".." components than preceeding normal components, then the ".." components can't be normalised away. In this case, the result is Nothing.

guess_dotdot = fmap unslice_path . guess_dotdot_comps . slice_path