View on
MetaCPAN is shutting down
For details read Perl NOC. After June 25th this page will redirect to
Ryo Anazawa > CGI-Header-0.40 > CGI::Header



Annotate this POD

View/Report Bugs
Module Version: 0.40   Source   Latest Release: CGI-Header-0.63


CGI::Header - Adapter for CGI::header() function


  use CGI;
  use CGI::Header;

  my $query = CGI->new;

  # HTTP header properties
  my $header = {
      -attachment => 'foo.gif',
      -charset    => 'utf-7',
      -cookie     => [ $cookie1, $cookie2 ], # CGI::Cookie objects
      -expires    => '+3d',
      -nph        => 1,
      -p3p        => [qw/CAO DSP LAW CURa/],
      -target     => 'ResultsWindow',
      -type       => 'image/gif'

  # create a CGI::Header object
  my $h = CGI::Header->new( $header, $query );

  # update $header
  $h->set( 'Content-Length' => 3002 );
  $h->delete( 'Content-Disposition' );

  $h->header; # same reference as $header


This document refers to CGI::Header version 0.40.


This module is compatible with 3.51 or higher.


This module is a utility class to manipulate a hash reference received by the header() function of This class is, so to speak, a subclass of Hash, while Perl5 doesn't provide a built-in class called Hash.

This module isn't the replacement of the CGI::header() function. If you're allowed to replace the function with other modules like HTTP::Headers, you should do so.

This module can be used in the following situation:

1. $header is a hash reference which represents CGI response headers

For example, CGI::Application implements header_add() method which can be used to add HTTP header properties. Instances of CGI applications often hold those properties.

  my $header = { -type => 'text/plain' };
2. Manipulates $header using CGI::Header

Since property names are case-insensitive, application developers have to normalize them manually when they specify header properties. CGI::Header normalizes them automatically.

  use CGI::Header;

  my $h = CGI::Header->new( $header );
  $h->set( 'Content-Length' => 3002 ); # add Content-Length header

  # => {
  #     -type => 'text/plain',
  #     -content_length => '3002',
  # }
3. Passes $header to CGI::header() to stringify the variable
  use CGI;

  print CGI::header( $header );
  # Content-length: 3002
  # Content-Type: text/plain; charset=ISO-8859-1

header() function just stringifies given header properties. This module can be used to generate PSGI-compatible header array references. See CGI::Header::PSGI.


$header = CGI::Header->new( { -type => 'text/plain', ... }[, $query] )

Given a header hash reference, returns a CGI::Header object which holds a reference to the original given argument:

  my $header = { -type => 'text/plain' };
  my $h = CGI::Header->new( $header );
  $h->header; # same reference as $header

The object updates the reference when called write methods like set(), delete() or clear():

  # updates $header
  $h->set( 'Content-Length' => 3002 );
  $h->delete( 'Content-Disposition' );

You can also pass your query object, preceded by the header hash ref.:

  my $query = CGI->new;
  my $h = CGI::Header->new( $header, $query );
  $h->query; # => $query

NOTE: In this case, new() doesn't check whether property names of $header are normalized or not at all, and so you have to rehash() the header hash reference explicitly when you aren't sure that they are normalized.

$header = CGI::Header->new( -type => 'text/plain', ... )

It's roughly equivalent to:

  my $h = CGI::Header->new({ -type => 'text/plain', ... })->rehash;

Unlike rehash(), if a property name is duplicated, that property will be overwritten silently:

  my $h = CGI::Header->new(
      -Type        => 'text/plain',
      Content_Type => 'text/html'

  $h->header->{-type}; # => "text/html"

In addition to HTTP header properties, you can specify '-query' property which represents your query object:

  my $query = CGI->new;

  my $h = CGI::Header->new(
      -type  => 'text/plain',
      -query => $query,

  $h->header; # => { -type => 'text/plain' }
  $h->query;  # => $query
$header = CGI::Header->new( $media_type )

A shortcut for:

  my $header = CGI::Header->new({ -type => $media_type });
CGI::Header->lc( $str )

This method is obsolete and will be removed in 0.36.

Returns the lowercased version of $str. Unlike CORE::lc, this method gets rid of an initial dash, and also transliterates dashes into underscores in $str.

  my $str = CGI::Header->lc( "Foo-Bar" ); # => "foo_bar"


$query = $header->query

Returns your current query object. query() defaults to the Singleton instance of ($CGI::Q).

$hashref = $header->header

Returns the header hash reference associated with this CGI::Header object. You can always pass the header hash to CGI::header() function to generate CGI response headers:

  print CGI::header( $header->header );
$self = $header->rehash

Rebuilds the header hash to normalize parameter names without changing the reference. Returns this object itself. If parameter names aren't normalized, the methods listed below won't work as you expect.

  my $h1 = $header->header;
  # => {
  #     '-content_type'   => 'text/plain',
  #     'Set-Cookie'      => 'ID=123456; path=/',
  #     'expires'         => '+3d',
  #     '-target'         => 'ResultsWindow',
  #     '-content-length' => '3002'
  # }


  my $h2 = $header->header; # same reference as $h1
  # => {
  #     '-type'           => 'text/plain',
  #     '-cookie'         => 'ID=123456; path=/',
  #     '-expires'        => '+3d',
  #     '-target'         => 'ResultsWindow',
  #     '-content_length' => '3002'
  # }

Normalized parameter names are:

1. lowercased
  'Content-Length' -> 'content-length'
2. start with a dash
  'content-length' -> '-content-length'
3. use underscores instead of dashes except for the first character
  '-content-length' -> '-content_length'

CGI::header() also accepts aliases of parameter names. This module converts them as follows:

 '-content_type'  -> '-type'
 '-set_cookie'    -> '-cookie'
 '-cookies'       -> '-cookie'
 '-window_target' -> '-target'

If a property name is duplicated, throws an exception:

  # => {
  #     -Type        => 'text/plain',
  #     Content_Type => 'text/html',
  # }

  $header->rehash; # die "Property '-type' already exists"
$value = $header->get( $field )
$value = $header->set( $field => $value )

Get or set the value of the header field. The header field name ($field) is not case sensitive. You can use underscores as a replacement for dashes in header names.

  # field names are case-insensitive
  $header->get( 'Content-Length' );
  $header->get( 'content-length' );

The $value argument may be a plain string or a reference to an array of CGI::Cookie objects for the Set-Cookie header.

  $header->set( 'Content-Length' => 3002 );
  my $length = $header->get( 'Content-Length' ); # => 3002

  # $cookie1 and $cookie2 are CGI::Cookie objects
  $header->set( 'Set-Cookie' => [$cookie1, $cookie2] );
  my $cookies = $header->get( 'Set-Cookie' ); # => [ $cookie1, $cookie2 ]
$bool = $header->exists( $field )

Returns a Boolean value telling whether the specified field exists.

  if ( $header->exists('ETag') ) {
$value = $header->delete( $field )

Deletes the specified field form CGI response headers. Returns the value of the deleted field.

  my $value = $header->delete( 'Content-Disposition' ); # => 'inline'
$self = $header->clear

This will remove all header fields.

$bool = $header->is_empty

This method is obsolete and will be removed in 0.41.

Returns true if the header contains no key-value pairs.


  if ( $header->is_empty ) { # true
$clone = $header->clone

Returns a copy of this CGI::Header object. It's identical to:

  my %copy = %{ $header->header }; # shallow copy
  my $clone = CGI::Header->new( \%copy, $header->query );
$filename = $header->attachment
$header->attachment( $filename )

Can be used to turn the page into an attachment. Represents suggested name for the saved file.

  $header->attachment( 'genome.jpg' );
  my $filename = $header->attachment; # => "genome.jpg"

In this case, the outgoing header will be formatted as:

  Content-Disposition: attachment; filename="genome.jpg"
@tags = $header->p3p_tags
$header->p3p_tags( @tags )

This method will be renamed to p3p in 0.41.

Represents P3P tags. The parameter can be an array or a space-delimited string. Returns a list of P3P tags. (In scalar context, returns the number of P3P tags.)

  $header->p3p_tags(qw/CAO DSP LAW CURa/);
  # or
  $header->p3p_tags( 'CAO DSP LAW CURa' );

  my @tags = $header->p3p_tags; # => ("CAO", "DSP", "LAW", "CURa")
  my $size = $header->p3p_tags; # => 4

In this case, the outgoing header will be formatted as:

  P3P: policyref="/w3c/p3p.xml", CP="CAO DSP LAW CURa"
$format = $header->expires
$header->expires( $format )

The Expires header gives the date and time after which the entity should be considered stale. You can specify an absolute or relative expiration interval. The following forms are all valid for this field:

  $header->expires( '+30s' ); # 30 seconds from now
  $header->expires( '+10m' ); # ten minutes from now
  $header->expires( '+1h'  ); # one hour from now
  $header->expires( 'now'  ); # immediately
  $header->expires( '+3M'  ); # in three months
  $header->expires( '+10y' ); # in ten years time

  # at the indicated time & date
  $header->expires( 'Thu, 25 Apr 1999 00:40:33 GMT' );

If set to a true value, will issue the correct headers to work with a NPH (no-parse-header) script.

  $header->nph( 1 );
@fields = $header->field_names

This method is obsolete and will be removed in 0.41.

Returns the list of distinct field names present in the header in a random order. The field names have case as returned by CGI::header().

  my @fields = $header->field_names;
  # => ( 'Set-Cookie', 'Content-length', 'Content-Type' )
$self = $header->each( \&callback )

This method is obsolete and will be removed in 0.41.

Apply a subroutine to each header field in turn. The callback routine is called with two parameters; the name of the field and a value. If the Set-Cookie header is multi-valued, then the routine is called once for each value. Any return values of the callback routine are ignored.

  my @lines;
  $header->each(sub {
      my ( $field, $value ) = @_;
      push @lines, "$field: $value";

  print join @lines, "\n";
  # Content-length: 3002
  # Content-Type: text/plain
@headers = $header->flatten

Returns pairs of fields and values.

  # $cookie1 and $cookie2 are CGI::Cookie objects
  my $header = CGI::Header->new( -cookie => [$cookie1, $cookie2] );

  # => (
  #     "Set-Cookie" => "$cookie1",
  #     "Set-Cookie" => "$cookie2",
  #     ...
  # )

A shortcut for:

  $header->query->header( $header->header );


  use CGI::Header;

  my $header = { -type => 'text/plain' };
  tie my %header => 'CGI::Header' => $header;

  # update $header
  $header{'Content-Length'} = 3002;
  delete $header{'Content-Disposition'};
  %header = ();

  tied( %header )->header; # same reference as $header

Above methods are aliased as follows:

  TIEHASH -> new
  FETCH   -> get
  STORE   -> set
  DELETE  -> delete
  CLEAR   -> clear
  EXISTS  -> exists
  SCALAR  -> !is_empty

You can also iterate through the tied hash:

  my @fields = keys %header;
  my @values = values %header;
  my ( $field, $value ) = each %header;

See also perltie.



The following plugin just adds the Content-Length header to CGI response headers sent by blosxom.cgi:

  package content_length;
  use CGI::Header;

  sub start {

  sub last {
      my $h = CGI::Header->new( $blosxom::header );
      $h->set( 'Content-Length' => length $blosxom::output );


Since Blosxom depends on the procedural interface of, you don't have to pass $query to new() in this case.


  use CGI::Header;
  use HTTP::Headers;

  my @header_props = ( -type => 'text/plain', ... );
  my $h = HTTP::Headers->new( CGI::Header->new(@header_props)->flatten );
  $h->header( 'Content-Type' ); # => "text/plain"


Since the following strings conflict with property names, you can't use them as field names ($field):


You can set the Content-Type header to neither undef nor an empty:

  # wrong
  $header->set( 'Content-Type' => undef );
  $header->set( 'Content-Type' => q{} );

Use delete() instead:


If one of the following conditions is met, the Date header will be set automatically, and also the header field will become read-only:

  if ( $header->nph or $header->get('Set-Cookie') or $header->expires ) {
      my $date = $header->get('Date'); # => HTTP-Date (current time)
      $header->set( 'Date' => 'Thu, 25 Apr 1999 00:40:33 GMT' ); # wrong
      $header->delete('Date'); # wrong

You can't assign to the Expires header directly because the following behavior will surprise us:

  # wrong
  $header->set( 'Expires' => '+3d' );

  my $value = $header->get('Expires');
  # => "Thu, 25 Apr 1999 00:40:33 GMT" (not "+3d")

Use expires() instead:


You can't assign to the P3P header directly:

  # wrong
  $header->set( 'P3P' => '/path/to/p3p.xml' );

CGI::header() restricts where the policy-reference file is located, and so you can't modify the location (/w3c/p3p.xml). You're allowed to set P3P tags using p3p().


If the following condition is met, the Pragma header will be set automatically, and also the header field will become read-only:

  if ( $header->query->cache ) {
      my $pragma = $header->get('Pragma'); # => 'no-cache'
      $header->set( 'Pragma' => 'no-cache' ); # wrong
      $header->delete('Pragma'); # wrong

If the following condition is met, the Server header will be set automatically, and also the header field will become read-only:

  if ( $header->nph ) {
      my $server = $header->get('Server');
      # => $header->query->server_software

      $header->set( 'Server' => 'Apache/1.3.27 (Unix)' ); # wrong
      $header->delete( 'Server' ); # wrong


CGI, Plack::Util::headers(), HTTP::Headers


There are no known bugs in this module. Please report problems to ANAZAWA ( Patches are welcome.


Ryo Anazawa (


This module is free software; you can redistibute it and/or modify it under the same terms as Perl itself. See perlartistic.

syntax highlighting: