man IO::All () - IO::All of it to Graham and Damian!


IO::All - IO::All of it to Graham and Damian!


    use IO::All;                                # Let the madness begin...

    # Some of the many ways to read a whole file into a scalar
    io('file.txt') > $contents;                 # Overloaded "arrow"
    $contents < io 'file.txt';                  # Flipped but same operation
    $io = io 'file.txt';                        # Create a new IO::All object
    $contents = $$io;                           # Overloaded scalar dereference
    $contents = $io->all;                       # A method to read everything
    $contents = $io->slurp;                     # Another method for that
    $contents = join '', $io->getlines;         # Join the separate lines
    $contents = join '', map "$_\n", @$io;      # Same. Overloaded array deref
    $io->tie;                                   # Tie the object as a handle
    $contents = join '', <$io>;                 # And use it in builtins
    # and the list goes on ...

    # Other file operations:
    @lines = io('file.txt')->slurp;             # List context slurp
    $content > io('file.txt');                  # Print to a file
    io('file.txt')->print($content, $more);     # (ditto)
    $content >> io('file.txt');                 # Append to a file
    io('file.txt')->append($content);           # (ditto)
    $content << $io;                            # Append to a string
    io('copy.txt') < io('file.txt');            $ Copy a file
    io('file.txt') > io('copy.txt');            # Invokes File::Copy
    io('more.txt') >> io('all.txt');            # Add on to a file

    # Print the path name of a file:
    print $io->name;                            # The direct method
    print "$io";                                # Object stringifies to name
    print $io;                                  # Quotes not needed here
    print $io->filename;                        # The file portion only

    # Read all the files/directories in a directory:
    $io = io('my/directory/');                  # Create new directory object
    @contents = $io->all;                       # Get all contents of dir
    @contents = @$io;                           # Directory as an array
    @contents = values %$io;                    # Directory as a hash
    push @contents, $subdir                     # One at a time
      while $subdir = $io->next;

    # Print the name and file type for all the contents above:
    print "$_ is a " . $_->type . "\n"          # Each element of @contents
      for @contents;                            # is an IO::All object!!

    # Print first line of each file:
    print $_->getline                           # getline gets one line
      for io('dir')->all_files;                 # Files only

    # Print names of all files/dirs three directories deep:
    print "$_\n" for $io->all(3);               # Pass in the depth. Default=1

    # Print names of all files/dirs recursively:
    print "$_\n" for $io->all(0);               # Zero means all the way down
    print "$_\n" for $io->All;                  # Capitalized shortcut
    print "$_\n" for $io->deep->all;            # Another way

    # There are some special file names:
    print io('-');                              # Print STDIN to STDOUT
    io('-') > io('-');                          # Do it again
    io('-') < io('-');                          # Same. Context sensitive.
    "Bad puppy" > io('=');                      # Message to STDERR
    $string_file = io('$');                     # Create IO::String Object
    $temp_file = io('?');                       # Create a temporary file

    # Socket operations:
    $server = io('localhost:5555')->fork;       # Create a daemon socket
    $connection = $server->accept;              # Get a connection socket
    $input < $connection;                       # Get some data from it
    "Thank you!" > $connection;                 # Thank the caller
    $connection->close;                         # Hang up
    io(':6666')->accept->slurp > io->devnull;   # Take a complaint and file it

    # DBM database operations:
    $dbm = io 'my/database';                    # Create a database object
    print $dbm->{grocery_list};                 # Hash context makes it a DBM
    $dbm->{todo} = $new_list;                   # Write to database
    $dbm->dbm('GDBM_file');                     # Demand specific DBM
    io('mydb')->mldbm->{env} = \%ENV;           # MLDBM support

    # Tie::File support:
    $io = io 'file.txt';
    $io->[42] = 'Line Forty Three';             # Change a line
    print $io->[@$io / 2];                      # Print middle line
    @$io = reverse @$io;                        # Reverse lines in a file

    # Stat functions:
    printf "%s %s %s\n",                        # Print name, uid and size of 
      $_->name, $_->uid, $_->size               # contents of current directory
        for io('.')->all;
    print "$_\n" for sort                       # Use mtime method to sort all
      {$b->mtime <=> $a->mtime}                 # files under current directory
        io('.')->All_Files;                     # by recent modification time.

    # File::Spec support:
    $contents < io->catfile(qw(dir file.txt));  # Portable IO operation

    # Miscellaneous:
    @lines = io('file.txt')->chomp->slurp;      # Chomp as you slurp
    @chunks = 
      io('file.txt')->separator('xxx')->slurp;  # Use alternnate record sep
    $binary = io('file.bin')->binary->all;      # Read a binary file
    io('a-symlink')->readlink->slurp;           # Readlink returns an object
    print io('foo')->absolute->pathname;        # Print absolute path of foo

    # IO::All External Plugin Methods
    io("myfile") > io->("");     # Upload a file using ftp
    $html < io->http("");         # Grab a web page
    io('')->print($spam); # Email a "friend"

    # This is just the beginning, read on...


Graham Barr for doing it all. Damian Conway for doing it all different.

IO::All combines all of the best Perl IO modules into a single Spiffy object oriented interface to greatly simplify your everyday Perl IO idioms. It exports a single function called CWio, which returns a new IO::All object. And that object can do it all!

The IO::All object is a proxy for IO::File, IO::Dir, IO::Socket, IO::String, Tie::File, File::Spec, File::Path and File::ReadBackwards; as well as all the DBM and MLDBM modules. You can use most of the methods found in these classes and in IO::Handle (which they inherit from). IO::All adds dozens of other helpful idiomatic methods including file stat and manipulation functions.

IO::All is pluggable, and modules like IO::All::LWP and IO::All::Mailto add even more functionality. Optionally, every IO::All object can be tied to itself. This means that you can use most perl IO builtins on it: readline, <>, getc, print, printf, syswrite, sysread, close.

The distinguishing magic of IO::All is that it will automatically open (and close) files, directories, sockets and other IO things for you. You never need to specify the mode ('<', '>>', etc), since it is determined by the usage context. That means you can replace this:

    open STUFF, '<', './mystuff'
      or die "Can't open './mystuff' for input:\n$!";
    local $/;
    my $stuff = <STUFF>;
    close STUFF;

with this:

    my $stuff < io"./mystuff";

And that is a good thing!


Here is an alphabetical list of all the public methods that you can call on an IO::All object.

CWabs2rel, CWabsolute, CWaccept, CWAll, CWall, CWAll_Dirs, CWall_dirs, CWAll_Files, CWall_files, CWAll_Links, CWall_links, CWappend, CWappendf, CWappendln, CWassert, CWatime, CWautoclose, CWautoflush, CWbackwards, CWbcc, CWbinary, CWbinmode, CWblksize, CWblocks, CWblock_size, CWbuffer, CWcanonpath, CWcase_tolerant, CWcatdir, CWcatfile, CWcatpath, CWcc, CWchdir, CWchomp, CWclear, CWclose, CWconfess, CWcontent, CWctime, CWcurdir, CWdbm, CWdeep, CWdevice, CWdevice_id, CWdevnull, CWdir, CWdomain, CWempty, CWeof, CWerrors, CWfile, CWfilename, CWfileno, CWfilepath, CWfilter, CWfork, CWfrom, CWftp, CWget, CWgetc, CWgetline, CWgetlines, CWgid, CWhandle, CWhead, CWhttp, CWhttps, CWinode, CWio_handle, CWis_absolute, CWis_dir, CWis_dbm, CWis_executable, CWis_file, CWis_link, CWis_mldbm, CWis_open, CWis_pipe, CWis_readable, CWis_socket, CWis_stdio, CWis_string, CWis_temp, CWis_writable, CWjoin, CWlength, CWlink, CWlock, CWmailer, CWmailto, CWmkdir, CWmkpath, CWmldbm, CWmode, CWmodes, CWmtime, CWname, CWnew, CWnext, CWnlink, CWopen, CWpassword, CWpath, CWpathname, CWperms, CWpipe, CWport, CWprint, CWprintf, CWprintln, CWput, CWrdonly, CWrdwr, CWread, CWreaddir, CWreadlink, CWrecv, CWrel2abs, CWrelative, CWrename, CWrequest, CWresponse, CWrmdir, CWrmtree, CWrootdir, CWscalar, CWseek, CWsend, CWseparator, CWshutdown, CWsize, CWslurp, CWsocket, CWsort, CWsplitdir, CWsplitpath, CWstat, CWstdio, CWstderr, CWstdin, CWstdout, CWstring, CWstring_ref, CWsubject, CWsysread, CWsyswrite, CWtail, CWtell, CWtemp, CWtie, CWtmpdir, CWto, CWtouch, CWtruncate, CWtype, CWuser, CWuid, CWunlink, CWunlock, CWupdir, CWuri, CWutf8, CWutime and CWwrite.

Each method is documented further below.


IO::All objects overload a small set of Perl operators to great effect. The overloads are limited to <, <<, >, >>, dereferencing operations, and stringification.

Even though relatively few operations are overloaded, there is actually a huge matrix of possibilities for magic. That's because the overloading is sensitive to the types, position and context of the arguments, and an IO::All object can be one of many types.

The most important overload to grok is stringification. IO::All objects stringify to their file or directory name. Here we print the contents of the current directory:

    perl -MIO::All -le 'print for io(".")->all'

is the same as:

    perl -MIO::All -le 'print $_->name for io(".")->all'

Stringification is important because it allows IO::All operations to return objects when they might otherwise return file names. Then the recipient can use the result either as an object or a string.

'>' and '<' move data between objects in the direction pointed to by the operator.

    $content1 < io('file1');
    $content1 > io('file2');
    io('file2') > $content3;
    io('file3') < $content3;
    io('file3') > io('file4');
    io('file5') < io('file4');

'>>' and '<<' do the same thing except the recipent string or file is appended to.

An IO::All file used as an array reference becomes tied using Tie::File:

    $file = io"file";
    # Print last line of file
    print $file->[-1];
    # Insert new line in middle of file
    $file->[$#$file / 2] = 'New line';

An IO::All file used as a hash reference becomes tied to a DBM class:

    io('mydbm')->{ingy} = 'YAML';

An IO::All directory used as an array reference, will expose each file or subdirectory as an element of the array.

    print "$_\n" for @{io 'dir'};

IO::All directories used as hash references have file names as keys, and IO::All objects as values:

    print io('dir')->{'foo.txt'}->slurp;

Files used as scalar references get slurped:

    print ${io('dir')->{'foo.txt'}};

Not all combinations of operations and object types are supported. Some just haven't been added yet, and some just don't make sense. If you use an invalid combination, an error will be thrown.


This section describes some various things that you can easily cook up with IO::All.

File Locking

IO::All makes it very easy to lock files. Just use the CWlock method. Here's a standalone program that demonstrates locking for both write and read:

    use IO::All;
    my $io1 = io('myfile')->lock;
    $io1->println('line 1');

    fork or do {
        my $io2 = io('myfile')->lock;
        print $io2->slurp;

    sleep 1;
    $io1->println('line 2');
    $io1->println('line 3');

There are a lot of subtle things going on here. An exclusive lock is issued for CW$io1 on the first CWprintln. That's because the file isn't actually opened until the first IO operation.

When the child process tries to read the file using CW$io2, there is a shared lock put on it. Since CW$io1 has the exclusive lock, the slurp blocks.

The parent process sleeps just to make sure the child process gets a chance. The parent needs to call CWunlock or CWclose to release the lock. If all goes well the child will print 3 lines.

Round Robin

This simple example will read lines from a file forever. When the last line is read, it will reopen the file and read the first one again.

    my $io = io'file1.txt';
    while (my $line = $io->getline || $io->getline) {
        print $line;

Reading Backwards

If you call the CWbackwards method on an IO::All object, the CWgetline and CWgetlines will work in reverse. They will read the lines in the file from the end to the beginning.

    my @reversed;
    my $io = io('file1.txt');
    while (my $line = $io->getline) {
        push @reversed, $line;

or more simply:

    my @reversed = io('file1.txt')->backwards->getlines;

The CWbackwards method returns the IO::All object so that you can chain the calls.

NOTE: This operation requires that you have the File::ReadBackwards module installed.

Client/Server Sockets

IO::All makes it really easy to write a forking socket server and a client to talk to it.

In this example, a server will return 3 lines of text, to every client that calls it. Here is the server code:

    use IO::All;

    my $socket = io(':12345')->fork->accept;
    $socket->print($_) while <DATA>;

    On your mark,
    Get set,

Here is the client code:

    use IO::All;

    my $io = io('localhost:12345');
    print while $_ = $io->getline;

You can run the server once, and then run the client repeatedly (in another terminal window). It should print the 3 data lines each time.

Note that it is important to close the socket if the server is forking, or else the socket won't go out of scope and close.

A Tiny Web Server

Here is how you could write a simplistic web server that works with static and dynamic pages:

    perl -MIO::All -e 'io(":8080")->fork->accept->(sub { $_[0] < io(-x $1 ? "./$1 |" : $1) if /^GET \/(.*) / })'

There is are a lot of subtle things going on here. First we accept a socket and fork the server. Then we overload the new socket as a code ref. This code ref takes one argument, another code ref, which is used as a callback.

The callback is called once for every line read on the socket. The line is put into CW$_ and the socket itself is passed in to the callback.

Our callback is scanning the line in CW$_ for an HTTP GET request. If one is found it parses the file name into CW$1. Then we use CW$1 to create an new IO::All file object... with a twist. If the file is executable (CW-x), then we create a piped command as our IO::All object. This somewhat approximates CGI support.

Whatever the resulting object is, we direct the contents back at our socket which is in CW$_[0]. Pretty simple, eh?

DBM Files

IO::All file objects used as a hash reference, treat the file as a DBM tied to a hash. Here I write my DB record to STDERR:

    io("names.db")->{ingy} > io'=';

Since their are several DBM formats available in Perl, IO::All picks the first one of these that is installed on your system:

    DB_File GDBM_File NDBM_File ODBM_File SDBM_File

You can override which DBM you want for each IO::All object:

    my @keys = keys %{io('mydbm')->dbm('SDBM_File')};

File Subclassing

Subclassing is easy with IO::All. Just create a new module and use IO::All as the base class. Since IO::All is a Spiffy module, you do it like this:

    package NewModule;
    use IO::All -base;

You need to do it this way so that IO::All will export the CWio function. Here is a simple recipe for subclassing:

IO::Dumper inherits everything from IO::All and adds an extra method called CWdump, which will dump a data structure to the file we specify in the CWio function. Since it needs Data::Dumper to do the dumping, we override the CWopen method to CWrequire Data::Dumper and then pass control to the real CWopen.

First the code using the module:

    use IO::Dumper;


And next the IO::Dumper module itself:

    package IO::Dumper;
    use IO::All -base;
    use Data::Dumper;

    sub dump {
        my $self = shift;
        Dumper(@_) > $self;


Inline Subclassing

This recipe does the same thing as the previous one, but without needing to write a separate module. The only real difference is the first line. Since you don't use IO::Dumper, you need to still call its CWimport method manually.


    package IO::Dumper;
    use IO::All -base;
    use Data::Dumper;

    sub dump {
        my $self = shift;
        Dumper(@_) > $self;


This section gives a full description of all of the methods that you can call on IO::All objects. The methods have been grouped into subsections based on object construction, option settings, configuration, action methods and support for specific modules.

Object Construction and Initialization Methods

* new
There are three ways to create a new IO::All object. The first is with the special function CWio which really just calls CWIO::All->new. The second is by calling CWnew as a class method. The third is calling CWnew as an object instance method. In this final case, the new objects attributes are copied from the instance object.
All three forms take a single argument, a file descriptor. A file descriptor can be any of the following:
    - A file name
    - A file handle
    - A directory name
    - A directory handle
    - A typeglob reference
    - A piped shell command. eq '| ls -al'
    - A socket domain/port.  eg ''
    - '-' means STDIN or STDOUT (depending on usage)
    - '=' means STDERR
    - '$' means an IO::String object
    - '?' means a temporary file
    - A URI including: http, https, ftp and mailto
    - An IO::All object
If you provide an IO::All object, you will simply get that same object returned from the constructor. If no file descriptor is provided, an object will still be created, but it must be defined by one of the following methods before it can be used for I/O:
* file
Using the CWfile method sets the type of the object to file and sets the pathname of the file if provided. It might be important to use this method if you had a file whose name was CW'-', or if the name might otherwise be confused with a directory or a socket. In this case, either of these statements would work the same:
    my $file = io('-')->file;
    my $file = io->file('-');
* dir
Make the object be of type directory.
* socket
Make the object be of type socket.
* link
Make the object be of type link.
* pipe
Make the object be of type pipe. The following two statements are equivalent:
    my $io = io('ls -l |');
    my $io = io('ls -l')->pipe;
    my $io = io->pipe('ls -l');
* dbm
This method takes the names of zero or more DBM modules. The first one that is available is used to process the dbm file.
    io('mydbm')->dbm('NDBM_File', 'SDBM_File')->{author} = 'ingy';
If no module names are provided, the first available of the following is used:
    DB_File GDBM_File NDBM_File ODBM_File SDBM_File
* mldbm
Similar to the CWdbm method, except create a Multi Level DBM object using the MLDBM module. This method takes the names of zero or more DBM modules and an optional serialization module. The first DBM module that is available is used to process the MLDBM file. The serialization module can be Data::Dumper, Storable or FreezeThaw.
    io('mymldbm')->mldbm('GDBM_File', 'Storable')->{author} = 
      {nickname => 'ingy'};
* string
Make the object be a IO::String object. These are equivalent:
    my $io = io('$');
    my $io = io->string;
* temp
Make the object represent a temporary file. It will automatically be open for both read and write.
* stdio
Make the object represent either STDIN or STDOUT depending on how it is used subsequently. These are equivalent:
    my $io = io('-');
    my $io = io->stdin;
* stdin
Make the object represent STDIN.
* stdout
Make the object represent STDOUT.
* stderr
Make the object represent STDERR.
* handle
Forces the object to be created from an pre-existing IO handle. You can chain calls together to indicate the type of handle:
    my $file_object = io->file->handle($file_handle);
    my $dir_object = io->dir->handle($dir_handle);
* http
Make the object represent an http uri. Requires IO-All-LWP.
* https
Make the object represent an https uri. Requires IO-All-LWP.
* ftp
Make the object represent a ftp uri. Requires IO-All-LWP.
* mailto
Make the object represent a mailto uri. Requires IO-All-Mailto.

If you need to use the same options to create a lot of objects, and don't want to duplicate the code, just create a dummy object with the options you want, and use that object to spawn other objects.

    my $lt = io->lock->tie;
    my $io1 = $lt->new('file1');
    my $io2 = $lt->new('file2');

Since the new method copies attributes from the calling object, both CW$io1 and CW$io2 will be locked and tied.

Option Setting Methods

The following methods don't do any actual I/O, but they specify options about how the I/O should be done.

Each option can take a single argument of 0 or 1. If no argument is given, the value 1 is assumed. Passing 0 turns the option off.

All of these options return the object reference that was used to invoke them. This is so that the option methods can be chained together. For example:

    my $io = io('path/file')->tie->assert->chomp->lock;
* absolute
Indicates that the CWpathname for the object should be made absolute.
* assert
This method ensures that the path for a file or directory actually exists before the file is open. If the path does not exist, it is created.
* autoclose
By default, IO::All will close an object opened for input when EOF is reached. By closing the handle early, one can immediately do other operations on the object without first having to close it. This option is on by default, so if you don't want this behaviour, say so like this:
The object will then be closed when CW$io goes out of scope, or you manually call CW$io->close.
* autoflush
Proxy for IO::Handle::autoflush
* backwards
Sets the object to 'backwards' mode. All subsequent CWgetline operations will read backwards from the end of the file. Requires the File::ReadBackwards CPAN module.
* binary
Indicates the file has binary content and should be opened with CWbinmode.
* chdir
chdir() to the pathname of a directory object. When object goes out of scope, chdir back to starting directory.
* chomp
Indicates that all operations that read lines should chomp the lines. If the CWseparator method has been called, chomp will remove that value from the end of each record.
* confess
Errors should be reported with the very detailed Carp::confess function.
* deep
Indicates that calls to the CWall family of methods should search directories as deep as possible.
* fork
Indicates that the process should automatically be forked inside the CWaccept socket method.
* lock
Indicate that operations on an object should be locked using flock.
* rdonly
This option indicates that certain operations like DBM and Tie::File access should be done in read-only mode.
* rdwr
This option indicates that DBM and MLDBM files should be opened in read- write mode.
* relative
Indicates that the CWpathname for the object should be made relative.
* sort
Indicates whether objects returned from one of the CWall methods will be in sorted order by name. True by default.
* tie
Indicate that the object should be tied to itself, thus allowing it to be used as a filehandle in any of Perl's builtin IO operations.
    my $io = io('foo')->tie;
    @lines = <$io>;
* utf8
Indicates that IO should be done using utf8 encoding. Calls binmode with CW:utf8 layer.

Configuration Methods

The following methods don't do any actual I/O, but they set specific values to configure the IO::All object.

If these methods are passed no argument, they will return their current value. If arguments are passed they will be used to set the current value, and the object reference will be returned for potential method chaining.

* bcc
Set the Bcc field for a mailto object.
* binmode
Proxy for binmode. Requires a layer to be passed. Use CWbinary for plain binary mode.
* block_size
The default length to be used for CWread and CWsysread calls. Defaults to 1024.
* buffer
Returns a reference to the internal buffer, which is a scalar. You can use this method to set the buffer to a scalar of your choice. (You can just pass in the scalar, rather than a reference to it.) This is the buffer that CWread and CWwrite will use by default. You can easily have IO::All objects use the same buffer:
    my $input = io('abc');
    my $output = io('xyz');
    my $buffer;
    $output->write while $input->read;
* cc
Set the Cc field for a mailto object.
* content
Get or set the content for an LWP operation manually.
* domain
Set the domain name or ip address that a socket should use.
* errors
Use this to set a subroutine reference that gets called when an internal error is thrown.
* filter
Use this to set a subroutine reference that will be used to grep which objects get returned on a call to one of the CWall methods. For example:
    my @odd = io->curdir->filter(sub {$_->size % 2})->All_Files;
CW@odd will contain all the files under the current directory whose size is an odd number of bytes.
* from
Indicate the sender for a mailto object.
* mailer
Set the mailer program for a mailto transaction. Defaults to 'sendmail'.
* mode
Set the mode for which the file should be opened. Examples:
* name
Set or get the name of the file or directory represented by the IO::All object.
* password
Set the password for an LWP transaction.
* perms
Sets the permissions to be used if the file/directory needs to be created.
* port
Set the port number that a socket should use.
* request
Manually specify the request object for an LWP transaction.
* response
Returns the resulting reponse object from an LWP transaction.
* separator
Sets the record (line) separator to whatever value you pass it. Default is \n. Affects the chomp setting too.
* string_ref
Proxy for IO::String::string_ref Returns a reference to the internal string that is acting like a file.
* subject
Set the subject for a mailto transaction.
* to
Set the recipient address for a mailto request.
* uri
Direct access to the URI used in LWP transactions.
* user
Set the user name for an LWP transaction.

IO Action Methods

These are the methods that actually perform I/O operations on an IO::All object. The stat methods and the File::Spec methods are documented in separate sections below.

* accept
For sockets. Opens a server socket (LISTEN => 1, REUSE => 1). Returns an IO::All socket object that you are listening on. If the CWfork method was called on the object, the process will automatically be forked for every connection.
* all
Read all contents into a single string.
    compare(io('file1')->all, io('file2')->all);
* all (For directories)
Returns a list of IO::All objects for all files and subdirectories in a directory. '.' and '..' are excluded. Takes an optional argument telling how many directories deep to search. The default is 1. Zero (0) means search as deep as possible. The filter method can be used to limit the results. The items returned are sorted by name unless CW->sort(0) is used.
* All
Same as CWall(0).
* all_dirs
Same as CWall, but only return directories.
* All_Dirs
Same as CWall_dirs(0).
* all_files
Same as CWall, but only return files.
* All_Files
Same as CWall_files(0).
* all_links
Same as CWall, but only return links.
* All_Links
Same as CWall_links(0).
* append
Same as print, but sets the file mode to '>>'.
* appendf
Same as printf, but sets the file mode to '>>'.
* appendln
Same as println, but sets the file mode to '>>'.
* clear
Clear the internal buffer. This method is called by CWwrite after it writes the buffer. Returns the object reference for chaining.
* close
Close will basically unopen the object, which has different meanings for different objects. For files and directories it will close and release the handle. For sockets it calls shutdown. For tied things it unties them, and it unlocks locked things.
* empty
Returns true if a file exists but has no size, or if a directory exists but has no contents.
* eof
Proxy for IO::Handle::eof
* exists
Returns whether or not the file or directory exists.
* filename
Return the name portion of the file path in the object. For example:
would return CWfile.txt.
* fileno
Proxy for IO::Handle::fileno
* filepath
Return the path portion of the file path in the object. For example:
would return CWmy/path.
* get
Perform an LWP GET request manually.
* getc
Proxy for IO::Handle::getc
* getline
Calls IO::File::getline. You can pass in an optional record separator.
* getlines
Calls IO::File::getlines. You can pass in an optional record separator.
* head
Return the first 10 lines of a file. Takes an optional argument which is the number of lines to return. Works as expected in list and scalar context. Is subject to the current line separator.
* io_handle
Direct access to the actual IO::Handle object being used on an opened IO::All object.
* is_dir
Returns boolean telling whether or not the IO::All object represents a directory.
* is_executable
Returns true if file or directory is executable.
* is_dbm
Returns boolean telling whether or not the IO::All object represents a dbm file.
* is_file
Returns boolean telling whether or not the IO::All object represents a file.
* is_link
Returns boolean telling whether or not the IO::All object represents a symlink.
* is_mldbm
Returns boolean telling whether or not the IO::All object represents a mldbm file.
* is_open
Indicates whether the IO::All is currently open for input/output.
* is_pipe
Returns boolean telling whether or not the IO::All object represents a pipe operation.
* is_readable
Returns true if file or directory is readable.
* is_socket
Returns boolean telling whether or not the IO::All object represents a socket.
* is_stdio
Returns boolean telling whether or not the IO::All object represents a STDIO file handle.
* is_string
Returns boolean telling whether or not the IO::All object represents an IO::String object.
* is_temp
Returns boolean telling whether or not the IO::All object represents a temporary file.
* is_writable
Returns true if file or directory is writable.
* length
Return the length of the internal buffer.
* mkdir
Create the directory represented by the object.
* mkpath
Create the directory represented by the object, when the path contains more than one directory that doesn't exist. Proxy for File::Path::mkpath.
* next
For a directory, this will return a new IO::All object for each file or subdirectory in the directory. Return undef on EOD.
* open
Open the IO::All object. Takes two optional arguments CWmode and CWperms, which can also be set ahead of time using the CWmode and CWperms methods. NOTE: Normally you won't need to call open (or mode/perms), since this happens automatically for most operations.
* pathname
Return the absolute or relative pathname for a file or directory, depending on whether object is in CWabsolute or CWrelative mode.
* print
Proxy for IO::Handle::print
* printf
Proxy for IO::Handle::printf
* println
Same as print, but adds newline to each argument unless it already ends with one.
* put
Perform an LWP PUT request manually.
* read
This method varies depending on its context. Read carefully (no pun intended). For a file, this will proxy IO::File::read. This means you must pass it a buffer, a length to read, and optionally a buffer offset for where to put the data that is read. The function returns the length actually read (which is zero at EOF). If you don't pass any arguments for a file, IO::All will use its own internal buffer, a default length, and the offset will always point at the end of the buffer. The buffer can be accessed with the CWbuffer method. The length can be set with the CWblock_size method. The default length is 1024 bytes. The CWclear method can be called to clear the buffer. For a directory, this will proxy IO::Dir::read.
* readdir
Similar to the Perl CWreaddir builtin. In scalar context, return the next directory entry (ie file or directory name), or undef on end of directory. In list context, return all directory entries. Note that CWreaddir does not return the special CW. and CW.. entries.
* readline
Same as CWgetline.
* readlink
Calls Perl's readlink function on the link represented by the object. Instead of returning the file path, it returns a new IO::All object using the file path.
* recv
Proxy for IO::Socket::recv
* rename
    my $new = $io->rename('new-name');
Calls Perl's rename function and returns an IO::All object for the renamed file. Returns false if the rename failed.
* rewind
Proxy for IO::Dir::rewind
* rmdir
Delete the directory represented by the IO::All object.
* rmtree
Delete the directory represented by the IO::All object and all the files and directories beneath it. Proxy for File::Path::rmtree.
* scalar
Deprecated. Same as CWall().
* seek
Proxy for IO::Handle::seek. If you use seek on an unopened file, it will be opened for both read and write.
* send
Proxy for IO::Socket::send
* shutdown
Proxy for IO::Socket::shutdown
* slurp
Read all file content in one operation. Returns the file content as a string. In list context returns every line in the file.
* stat
Proxy for IO::Handle::stat
* sysread
Proxy for IO::Handle::sysread
* syswrite
Proxy for IO::Handle::syswrite
* tail
Return the last 10 lines of a file. Takes an optional argument which is the number of lines to return. Works as expected in list and scalar context. Is subject to the current line separator.
* tell
Proxy for IO::Handle::tell
* throw
This is an internal method that gets called whenever there is an error. It could be useful to override it in a subclass, to provide more control in error handling.
* touch
Update the atime and mtime values for a file or directory. Creates an empty file if the file does not exist.
* truncate
Proxy for IO::Handle::truncate
* type
Returns a string indicated the type of io object. Possible values are:
Returns undef if type is not determinable.
* unlink
Unlink (delete) the file represented by the IO::All object. NOTE: You can unlink a file after it is open, and continue using it until it is closed.
* unlock
Release a lock from an object that used the CWlock method.
* utime
Proxy for the utime Perl function.
* write
Opposite of CWread for file operations only. NOTE: When used with the automatic internal buffer, CWwrite will clear the buffer after writing it.

Stat Methods

This methods get individual values from a stat call on the file, directory or handle represented by th IO::All object.

* atime
Last access time in seconds since the epoch
* blksize
Preferred block size for file system I/O
* blocks
Actual number of blocks allocated
* ctime
Inode change time in seconds since the epoch
* device
Device number of filesystem
* device_id
Device identifier for special files only
* gid
Numeric group id of file's owner
* inode
Inode number
* modes
File mode - type and permissions
* mtime
Last modify time in seconds since the epoch
* nlink
Number of hard links to the file
* size
Total size of file in bytes
* uid
Numeric user id of file's owner

File::Spec Methods

These methods are all adaptations from File::Spec. Each method actually does call the matching File::Spec method, but the arguments and return values differ slightly. Instead of being file and directory names, they are IO::All objects. Since IO::All objects stringify to their names, you can generally use the methods just like File::Spec.

* abs2rel
Returns the relative path for the absolute path in the IO::All object. Can take an optional argument indicating the base path.
* canonpath
Returns the canonical path for the IO::All object.
* case_tolerant
Returns 0 or 1 indicating whether the file system is case tolerant. Since an active IO::All object is not needed for this function, you can code it like:
or more simply:
* catdir
Concatenate the directory components together, and return a new IO::All object representing the resulting directory.
* catfile
Concatenate the directory and file components together, and return a new IO::All object representing the resulting file.
    my $contents = io->catfile(qw(dir subdir file))->slurp;
This is a very portable way to read CWdir/subdir/file.
* catpath
Concatenate the volume, directory and file components together, and return a new IO::All object representing the resulting file.
* curdir
Returns an IO::All object representing the current directory.
* devnull
Returns an IO::All object representing the /dev/null file.
* is_absolute
Returns 0 or 1 indicating whether the CWname field of the IO::All object is an absolute path.
* join
Same as CWcatfile.
* path
Returns a list of IO::All directory objects for each directory in your path.
* rel2abs
Returns the absolute path for the relative path in the IO::All object. Can take an optional argument indicating the base path.
* rootdir
Returns an IO::All object representing the root directory on your file system.
* splitdir
Returns a list of the directory components of a path in an IO::All object.
* splitpath
Returns a volume directory and file component of a path in an IO::All object.
* tmpdir
Returns an IO::All object representing a temporary directory on your file system.
* updir
Returns an IO::All object representing the current parent directory.


Each IO::All object gets reblessed into an IO::All::* object as soon as IO::All can determine what type of object it should be. Sometimes it gets reblessed more than once:
    my $io = io('mydbm.db');
    $io->{foo} = 'bar';
In the first statement, CW$io has a reference value of 'IO::All::File', if CWmydbm.db exists. In the second statement, the object is reblessed into class 'IO::All::DBM'.
An IO::All object will automatically be opened as soon as there is enough contextual information to know what type of object it is, and what mode it should be opened for. This is usually when the first read or write operation is invoked but might be sooner.
The mode for an object to be opened with is determined heuristically unless specified explicitly.
For input, IO::All objects will automatically be closed after EOF (or EOD). For output, the object closes when it goes out of scope. To keep input objects from closing at EOF, do this:
You can always call CWopen and CWclose explicitly, if you need that level of control. To test if an object is currently open, use the CWis_open method.
Overloaded operations return the target object, if one exists. This would set CW$xxx to the IO::All object:
    my $xxx = $contents > io('file.txt');
While this would set CW$xxx to the content string:
    my $xxx = $contents < io('file.txt');


The goal of the IO::All project is to continually refine the module to be as simple and consistent to use as possible. Therefore, in the early stages of the project, I will not hesitate to break backwards compatibility with other versions of IO::All if I can find an easier and clearer way to do a particular thing.

IO is tricky stuff. There is definitely more work to be done. On the other hand, this module relies heavily on very stable existing IO modules; so it may work fairly well.

I am sure you will find many unexpected features. Please send all problems, ideas and suggestions to

Known Bugs and Deficiencies

Not all possible combinations of objects and methods have been tested. There are many many combinations. All of the examples have been tested. If you find a bug with a particular combination of calls, let me know.

If you call a method that does not make sense for a particular object, the result probably won't make sense. Little attempt is made to check for improper usage.


IO::Handle, IO::File, IO::Dir, IO::Socket, IO::String, File::Spec, File::Path, File::ReadBackwards, Tie::File

Also check out the Spiffy module if you are interested in extending this module.


A lot of people have sent in suggestions, that have become a part of IO::All. Thank you.

Special thanks to Ian Langworth for continued testing and patching.

Thank you Simon Cozens for tipping me off to the overloading possibilities.

Finally, thanks to Autrijus Tang, for always having one more good idea.

(It seems IO::All of it to a lot of people!)


Brian Ingerson <>


Copyright (c) 2004. Brian Ingerson. All rights reserved.

This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.

See <>