File::Flat - Implements a flat filesystem

File::Flat implements a flat filesystem. A flat filesystem is a filesystem in which directories do not exist. It provides an abstraction over any normal filesystem which makes it appear as if directories do not exist. In effect, it will automatically create directories as needed. This is create for things like install scripts and such, as you never need to worry about the existance of directories, just write to a file, no matter where it is.

The implementation of File::Flat is extremely comprehensive in scope. It has methods for all stardard file interaction taks, the -X series of tests, and some other things, such as slurp.

All methods are statically called, for example, to write some stuff to a file.

  use File::Flat;
  File::Flat->write( 'filename', 'file contents' );

File::Flat tries to use more task orientated modules wherever possible. This includes the use of File::Copy, File::NCopy, File::Remove and others. These are mostly loaded on-demand. Pruning is a technique where empty directories are assumed to be useless, and thus empty removed whenever one is created. Thus, when some other task has the potential to leave an empty directory, it is checked and deleted if it is empty.

By default File::Flat does not prune, and pruning must be done explicitly, via either the prune in File::Flat method, or by setting the second argument to the remove in File::Flat method to be true.

However by setting the global CW$AUTO_PRUNE variable to true, File::Flat will automatically prune directories at all times. You should generally use this locally, such as in the following example.

  #!/usr/bin/perl

  use strict;
  use File::Flat;

  delete_files(@ARGV);
  exit();

  # Recursively delete and prune all files provided on the command line
  sub delete_files {
        local $File::Flat::AUTO_PRUNE = 1;
        foreach my $file ( @_ ) {
                File::Flat->remove( $file ) or die "Failed to delete $file";
        }
  }

File::Flat itself should be completely capable of handling any platform through its exclusive use of File::Spec.

However, some of the modules it relies upon, particularly File::Remove, and possible File::NCopy are not File::Spec happy yet. Results may wary on non Unix platforms. Users of non-Unix platforms are invited to patch File::Remove ( and possibly File::NCopy ) and File::Flat should work.

Tests for the existance of the file. This is an exact duplicate of the -e function. Tests whether CWfilename is a file. This is an exact duplicate of the -f function. Test whether CWfilename is a directory. This is an exact duplicate of the -d function. Does the file or directory exist, and can we read from it. Does the file or directory exist, and can we write to it OR can we create the file or directory. Does a file or directory exist, and can we both read and write it. Does a file or directory exist, and can we execute it. Is this something we can open a filehandle to. Returns true if filename exists, is a file, and we can read from it. Can we remove the file or directory. Does the file CWfilename exist, and is it a text file. Does the file CWfilename exist, and is it a binary file. If the file exists, returns its size in bytes. Returns undef if the file does not exist. Rough analogue of the open function, but creates directories on demand as needed. Supports most of the normal options to the normal open function.

In the single argument form, it takes modes in the form [mode]filename. For example, all the following are valid.

  File::Flat->open( 'filename' );
  File::Flat->open( '<filename' );
  File::Flat->open( '>filename' );
  File::Flat->open( '>>filename' );
  File::Flat->open( '+<filename' );

In the two argument form, it takes the following

  File::Flat->open( '<', 'filename' );
  File::Flat->open( '>', 'filename' );
  File::Flat->open( '>>', 'filename' );
  File::Flat->open( '+<', 'filename' );

It does not support the more esoteric forms of open, such us opening to a pipe or other such things.

On successfully opening the file, it returns it as an IO::File object. Returns undef on error. The same as File::Flat->open( '<', 'filename' ) The same as File::Flat->open( '>', 'filename' ) The same as File::Flat->open( '>>', 'filename' ) The same as File::Flat->open( '+<', 'filename' ) Opens and reads in an entire file, chomping as needed.

In array context, it returns an array containing each line of the file. In scalar context, it returns a reference to an array containing each line of the file. It returns undef on error. The CWslurp method 'slurps' a file in. That is it attempts to read the entire file into a variable in as quick and memory efficient method as possible.

On success, returns a reference to a scalar, containing the entire file. Returns undef on error. The CWwrite method is the main method for writing content to a file. It takes two arguments, the location to write to, and the content to write, in several forms.

If the file already exists, it will be clobered before writing starts. If the file doesn't exists, the file and any directories will be created as needed.

Content can be provided in three forms. The contents of a scalar argument will be written directly to the file. You can optionally pass a reference to the scalar. This is recommended when the file size is bigger than a few thousand characters, is it does not duplicate the file contents in memory. Alternatively, you can pass the content as a reference to an array containing the contents. To ensure uniformity, CWwrite will add a newline to each line, replacing any existing newline as needed.

Returns true on success, and undef on error. This method is the same as CWwrite, except that it appends to the end of an existing file ( or creates the file as needed ).

This is the method you should be using to write to log files, etc. Performs an atomic write over a file. It does this by writing to a temporary file, and moving the completed file over the top of the existing file ( or creating a new file as needed ). When writing to a file that is on the same partition as /tmp, this should always be atomic.

This method otherwise acts the same as CWwrite. The CWcopy method attempts to copy a file or directory from the source to the target. New directories to contain the target will be created as needed.

For example CW<File::Flat-( './this', './a/b/c/d/that' );>> will create the directory structure required as needed.

In the file copy case, if the target already exists, and is a writable file, we replace the existing file, retaining file mode and owners. If the target is a directory, we do NOT copy into that directory, unlike with the 'cp' unix command. And error is instead returned.

CWcopy will also do limited recursive copying or directories. If source is a directory, and target does not exists, a recursive copy of source will be made to target. If target already exists ( file or directory ), CWcopy will returns with an error. The CWmove method follows the conventions of the 'mv' command, with the exception that the directories containing target will of course be created on demand. The CWremove method will remove a file, or recursively remove a directory.

If a second (true) argument is provided, then once the file or directory has been deleted, the method will the automatically work its way upwards pruning (deleting) empty and thus assumably useless directories.

Returns true if the deletion (and pruning if requested) was a success, or CWundef otherwise. For a file that has already been delete, CWprune will work upwards, removing any empty directories it finds.

For anyone familiar with CVS, it is similar to the CWupdate -P flag.

Returns true, or CWundef on error. The CWtruncate method will truncate an existing file to partular size. A size of 0 ( zero ) is used if no size is provided. If the file does not exists, it will be created, and set to 0. Attempting to truncate a directory will fail.

Returns true on success, or undef on error. In the case where you do actually have to create a directory only, the CWmakeDirectory method can be used to create a directory or any depth.

An optional file mode ( default 0755 ) can be provided.

Returns true on success, returns undef on error.

Function interface to be written, like File::Spec::Functions, to provide importable functions.

There's something bigger here too, I'm not exactly sure what it is, but I think there might be the beginings of a unified filesystem interface here... FSI.pm

Bugs should be filed at via the CPAN bug tracker at:

<http://rt.cpan.org/NoAuth/ReportBug.html?Queue=File%3A%3AFlat>

For other issues or comments, contact the author

Adam Kennedy (Maintainer), <http://ali.as/>, cpan@ali.as

File::Spec

Copyright (c) 2002 - 2004 Adam Kennedy. All rights reserved. This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.

The full text of the license can be found in the LICENSE file included with this module.