HTML::Template - Perl module to use HTML-like templating language


First you make a template - this is just a normal HTML file with a few extra tags, the simplest being <TMPL_VAR>

For example, test.tmpl:

    <head><title>Test Template</title></head>
    My Home Directory is <TMPL_VAR NAME=HOME>
    My Path is set to <TMPL_VAR NAME=PATH>

Now you can use it in a small CGI program:

    #!/usr/bin/perl -w
    use HTML::Template;
    # open the html template
    my $template = HTML::Template->new(filename => 'test.tmpl');
    # fill in some parameters
    $template->param(HOME => $ENV{HOME});
    $template->param(PATH => $ENV{PATH});
    # send the obligatory Content-Type and print the template output
    print "Content-Type: text/html\n\n", $template->output;

If all is well in the universe this should show something like this in your browser when visiting the CGI:

    My Home Directory is /home/some/directory
    My Path is set to /bin;/usr/bin


This module attempts to make using HTML templates simple and natural. It extends standard HTML with a few new HTML-esque tags - <TMPL_VAR> <TMPL_LOOP>, <TMPL_INCLUDE>, <TMPL_IF>, <TMPL_ELSE> and <TMPL_UNLESS>. The file written with HTML and these new tags is called a template. It is usually saved separate from your script - possibly even created by someone else! Using this module you fill in the values for the variables, loops and branches declared in the template. This allows you to separate design - the HTML - from the data, which you generate in the Perl script.

This module is licensed under the same terms as Perl. See the LICENSE section below for more details.


If you're new to HTML::Template, I suggest you start with the introductory article available on Perl Monks:


Please see HTML::Template::FAQ


It is true that there are a number of packages out there to do HTML templates. On the one hand you have things like HTML::Embperl which allows you freely mix Perl with HTML. On the other hand lie home-grown variable substitution solutions. Hopefully the module can find a place between the two.

One advantage of this module over a full HTML::Embperl-esque solution is that it enforces an important divide - design and programming. By limiting the programmer to just using simple variables and loops in the HTML, the template remains accessible to designers and other non-perl people. The use of HTML-esque syntax goes further to make the format understandable to others. In the future this similarity could be used to extend existing HTML editors/analyzers to support HTML::Template.

An advantage of this module over home-grown tag-replacement schemes is the support for loops. In my work I am often called on to produce tables of data in html. Producing them using simplistic HTML templates results in programs containing lots of HTML since the HTML itself cannot represent loops. The introduction of loop statements in the HTML simplifies this situation considerably. The designer can layout a single row and the programmer can fill it in as many times as necessary - all they must agree on is the parameter names.

For all that, I think the best thing about this module is that it does just one thing and it does it quickly and carefully. It doesn't try to replace Perl and HTML, it just augments them to interact a little better. And it's pretty fast.




The <TMPL_VAR> tag is very simple. For each <TMPL_VAR> tag in the template you call:

    $template->param(PARAMETER_NAME => "VALUE") 

When the template is output the <TMPL_VAR> is replaced with the VALUE text you specified. If you don't set a parameter it just gets skipped in the output.

You can also specify the value of the parameter as a code reference in order to have "lazy" variables. These sub routines will only be referenced if the variables are used. See "LAZY VALUES" for more information.


The following "attributes" can also be specified in template var tags:



The <TMPL_LOOP> tag is a bit more complicated than <TMPL_VAR>. The <TMPL_LOOP> tag allows you to delimit a section of text and give it a name. Inside this named loop you place <TMPL_VAR>s. Now you pass to param() a list (an array ref) of parameter assignments (hash refs) for this loop. The loop iterates over the list and produces output from the text block for each pass. Unset parameters are skipped. Here's an example:

In the template:

      Name: <TMPL_VAR NAME=NAME> <br>
      Job:  <TMPL_VAR NAME=JOB>  <p>

In your Perl code:

        EMPLOYEE_INFO => [{name => 'Sam', job => 'programmer'}, {name => 'Steve', job => 'soda jerk'}]
    print $template->output();

The output is:

    Name: Sam
    Job: programmer

    Name: Steve
    Job: soda jerk

As you can see above the <TMPL_LOOP> takes a list of variable assignments and then iterates over the loop body producing output.

Often you'll want to generate a <TMPL_LOOP>'s contents programmatically. Here's an example of how this can be done (many other ways are possible!):

    # a couple of arrays of data to put in a loop:
    my @words     = qw(I Am Cool);
    my @numbers   = qw(1 2 3);
    my @loop_data = ();              # initialize an array to hold your loop
    while (@words and @numbers) {
        my %row_data;      # get a fresh hash for the row data
        # fill in this row
        $row_data{WORD}   = shift @words;
        $row_data{NUMBER} = shift @numbers;
        # the crucial step - push a reference to this row into the loop!
        push(@loop_data, \%row_data);
    # finally, assign the loop data to the loop param, again with a reference:
    $template->param(THIS_LOOP => \@loop_data);

The above example would work with a template like:

      Word: <TMPL_VAR NAME="WORD">     
      Number: <TMPL_VAR NAME="NUMBER">

It would produce output like:

    Word: I
    Number: 1

    Word: Am
    Number: 2

    Word: Cool
    Number: 3

<TMPL_LOOP>s within <TMPL_LOOP>s are fine and work as you would expect. If the syntax for the param() call has you stumped, here's an example of a param call with one nested loop:

        LOOP => [
                name      => 'Bobby',
                nicknames => [{name => 'the big bad wolf'}, {name => 'He-Man'}],

Basically, each <TMPL_LOOP> gets an array reference. Inside the array are any number of hash references. These hashes contain the name=>value pairs for a single pass over the loop template.

Inside a <TMPL_LOOP>, the only variables that are usable are the ones from the <TMPL_LOOP>. The variables in the outer blocks are not visible within a template loop. For the computer-science geeks among you, a <TMPL_LOOP> introduces a new scope much like a perl subroutine call. If you want your variables to be global you can use global_vars option to new() described below.


    <TMPL_INCLUDE NAME="filename.tmpl">

This tag includes a template directly into the current template at the point where the tag is found. The included template contents are used exactly as if its contents were physically included in the master template.

The file specified can be an absolute path (beginning with a '/' under Unix, for example). If it isn't absolute, the path to the enclosing file is tried first. After that the path in the environment variable HTML_TEMPLATE_ROOT is tried, if it exists. Next, the "path" option is consulted, first as-is and then with HTML_TEMPLATE_ROOT prepended if available. As a final attempt, the filename is passed to open() directly. See below for more information on HTML_TEMPLATE_ROOT and the path option to new().

As a protection against infinitely recursive includes, an arbitrary limit of 10 levels deep is imposed. You can alter this limit with the max_includes option. See the entry for the max_includes option below for more details.



The <TMPL_IF> tag allows you to include or not include a block of the template based on the value of a given parameter name. If the parameter is given a value that is true for Perl - like '1' - then the block is included in the output. If it is not defined, or given a false value - like '0' - then it is skipped. The parameters are specified the same way as with <TMPL_VAR>.

Example Template:

      Some text that only gets displayed if BOOL is true!

Now if you call $template->param(BOOL => 1) then the above block will be included by output.

<TMPL_IF> </TMPL_IF> blocks can include any valid HTML::Template construct - VARs and LOOPs and other IF/ELSE blocks. Note, however, that intersecting a <TMPL_IF> and a <TMPL_LOOP> is invalid.

    Not going to work:

If the name of a <TMPL_LOOP> is used in a <TMPL_IF>, the IF block will output if the loop has at least one row. Example:

      This will output if the loop is not empty.


WARNING: Much of the benefit of HTML::Template is in decoupling your Perl and HTML. If you introduce numerous cases where you have TMPL_IFs and matching Perl ifs, you will create a maintenance problem in keeping the two synchronized. I suggest you adopt the practice of only using TMPL_IF if you can do so without requiring a matching if in your Perl code.



You can include an alternate block in your <TMPL_IF> block by using <TMPL_ELSE>. NOTE: You still end the block with </TMPL_IF>, not </TMPL_ELSE>!

      Some text that is included only if BOOL is true
      Some text that is included only if BOOL is false



This tag is the opposite of <TMPL_IF>. The block is output if the PARAMETER_NAME is set false or not defined. You can use <TMPL_ELSE> with <TMPL_UNLESS> just as you can with <TMPL_IF>.

      Some text that is output only if BOOL is FALSE.
      Some text that is output only if BOOL is TRUE.

If the name of a <TMPL_LOOP> is used in a <TMPL_UNLESS>, the <UNLESS> block output if the loop has zero rows.

      This will output if the loop is empty.



HTML::Template's tags are meant to mimic normal HTML tags. However, they are allowed to "break the rules". Something like:

    <img src="<TMPL_VAR IMAGE_SRC>">

is not really valid HTML, but it is a perfectly valid use and will work as planned.

The NAME= in the tag is optional, although for extensibility's sake I recommend using it. Example - <TMPL_LOOP LOOP_NAME> is acceptable.

If you're a fanatic about valid HTML and would like your templates to conform to valid HTML syntax, you may optionally type template tags in the form of HTML comments. This may be of use to HTML authors who would like to validate their templates' HTML syntax prior to HTML::Template processing, or who use DTD-savvy editing tools.


In order to realize a dramatic savings in bandwidth, the standard (non-comment) tags will be used throughout this documentation.



Call new() to create a new Template object:

    my $template = HTML::Template->new(
        filename => 'file.tmpl',
        option   => 'value',

You must call new() with at least one name = value> pair specifying how to access the template text. You can use filename => 'file.tmpl' to specify a filename to be opened as the template. Alternately you can use:

    my $t = HTML::Template->new(
        scalarref => $ref_to_template_text,
        option    => 'value',


    my $t = HTML::Template->new(
        arrayref => $ref_to_array_of_lines,
        option   => 'value',

These initialize the template from in-memory resources. In almost every case you'll want to use the filename parameter. If you're worried about all the disk access from reading a template file just use mod_perl and the cache option detailed below.

You can also read the template from an already opened filehandle, either traditionally as a glob or as a FileHandle:

    my $t = HTML::Template->new(filehandle => *FH, option => 'value');

The four new() calling methods can also be accessed as below, if you prefer.

    my $t = HTML::Template->new_file('file.tmpl', option => 'value');
    my $t = HTML::Template->new_scalar_ref($ref_to_template_text, option => 'value');
    my $t = HTML::Template->new_array_ref($ref_to_array_of_lines, option => 'value');
    my $t = HTML::Template->new_filehandle($fh, option => 'value');

And as a final option, for those that might prefer it, you can call new as:

    my $t = HTML::Template->new(
        type   => 'filename',
        source => 'file.tmpl',

Which works for all three of the source types.

If the environment variable HTML_TEMPLATE_ROOT is set and your filename doesn't begin with "/", then the path will be relative to the value of c<HTML_TEMPLATE_ROOT>.

Example - if the environment variable HTML_TEMPLATE_ROOT is set to /home/sam and I call HTML::Template->new() with filename set to "sam.tmpl", HTML::Template will try to open /home/sam/sam.tmpl to access the template file. You can also affect the search path for files with the path option to new() - see below for more information.

You can modify the Template object's behavior with new(). The options are available:

Error Detection Options

Caching Options

Filesystem Options

Debugging Options

Miscellaneous Options


A package method that is used to set/get the global default configuration options. For instance, if you want to set the utf8 flag to always be on for every template loaded by this process you would do:

    HTML::Template->config(utf8 => 1);

Or if you wanted to check if the utf8 flag was on or not, you could do:

    my %config = HTML::Template->config;
    if( $config{utf8} ) {

Any configuration options that are valid for new() are acceptable to be passed to this method.


param() can be called in a number of ways

1 - To return a list of parameters in the template :
    my @parameter_names = $self->param();
2 - To return the value set to a param :
    my $value = $self->param('PARAM');
3 - To set the value of a parameter :
    # For simple TMPL_VARs:
    $self->param(PARAM => 'value');
    # with a subroutine reference that gets called to get the value
    # of the scalar.  The sub will receive the template object as a
    # parameter.
    $self->param(PARAM => sub { return 'value' });
    # And TMPL_LOOPs:
4 - To set the value of a number of parameters :
    # For simple TMPL_VARs:
        PARAM  => 'value',
        PARAM2 => 'value'
    # And with some TMPL_LOOPs:
        PARAM              => 'value',
        PARAM2             => 'value',
5 - To set the value of a a number of parameters using a hash-ref :
            PARAM              => 'value',
            PARAM2             => 'value',

An error occurs if you try to set a value that is tainted if the force_untaint option is set.


Sets all the parameters to undef. Useful internally, if nowhere else!


output() returns the final result of the template. In most situations you'll want to print this, like:

    print $template->output();

When output is called each occurrence of <TMPL_VAR NAME=name> is replaced with the value assigned to "name" via param(). If a named parameter is unset it is simply replaced with ''. <TMPL_LOOP>s are evaluated once per parameter set, accumulating output on each pass.

Calling output() is guaranteed not to change the state of the HTML::Template object, in case you were wondering. This property is mostly important for the internal implementation of loops.

You may optionally supply a filehandle to print to automatically as the template is generated. This may improve performance and lower memory consumption. Example:

    $template->output(print_to => *STDOUT);

The return value is undefined when using the print_to option.


This method allow you to get information about the template structure. It can be called in a number of ways. The simplest usage of query is simply to check whether a parameter name exists in the template, using the name option:

    if ($template->query(name => 'foo')) {
        # do something if a variable of any type named FOO is in the template

This same usage returns the type of the parameter. The type is the same as the tag minus the leading 'TMPL_'. So, for example, a TMPL_VAR parameter returns 'VAR' from query().

    if ($template->query(name => 'foo') eq 'VAR') {
        # do something if FOO exists and is a TMPL_VAR

Note that the variables associated with TMPL_IFs and TMPL_UNLESSs will be identified as 'VAR' unless they are also used in a TMPL_LOOP, in which case they will return 'LOOP'.

query() also allows you to get a list of parameters inside a loop (and inside loops inside loops). Example loop:


And some query calls:

    # returns 'LOOP'
    $type = $template->query(name => 'EXAMPLE_LOOP');
    # returns ('bop', 'bee', 'example_inner_loop')
    @param_names = $template->query(loop => 'EXAMPLE_LOOP');
    # both return 'VAR'
    $type = $template->query(name => ['EXAMPLE_LOOP', 'BEE']);
    $type = $template->query(name => ['EXAMPLE_LOOP', 'BOP']);
    # and this one returns 'LOOP'
    $type = $template->query(name => ['EXAMPLE_LOOP', 'EXAMPLE_INNER_LOOP']);
    # and finally, this returns ('inner_bee', 'inner_bop')
    @inner_param_names = $template->query(loop => ['EXAMPLE_LOOP', 'EXAMPLE_INNER_LOOP']);
    # for non existent parameter names you get undef this returns undef.
    $type = $template->query(name => 'DWEAZLE_ZAPPA');
    # calling loop on a non-loop parameter name will cause an error. This dies:
    $type = $template->query(loop => 'DWEAZLE_ZAPPA');

As you can see above the loop option returns a list of parameter names and both name and loop take array refs in order to refer to parameters inside loops. It is an error to use loop with a parameter that is not a loop.

Note that all the names are returned in lowercase and the types are uppercase.

Just like param(), query() with no arguments returns all the parameter names in the template at the top level.


As mentioned above, both TMPL_VAR and TMPL_LOOP values can be code references. These code references are only executed if the variable or loop is used in the template. This is extremely useful if you want to make a variable available to template designers but it can be expensive to calculate, so you only want to do so if you have to.

Maybe an example will help to illustrate. Let's say you have a template like this:

    <tmpl_if we_care>
      <tmpl_if life_universe_and_everything>

If life_universe_and_everything is expensive to calculate we can wrap it's calculation in a code reference and HTML::Template will only execute that code if we_care is also true.

    $tmpl->param(life_universe_and_everything => sub { calculate_42() });

Your code reference will be given a single argument, the HTML::Template object in use. In the above example, if we wanted calculate_42() to have this object we'd do something like this:

    $tmpl->param(life_universe_and_everything => sub { calculate_42(shift) });

This same approach can be used for TMPL_LOOPs too:

    <tmpl_if we_care>
      <tmpl_loop needles_in_haystack>
        Found <tmpl_var __counter>!

And in your Perl code:

    $tmpl->param(needles_in_haystack => sub { find_needles() });

The only difference in the TMPL_LOOP case is that the subroutine needs to return a reference to an ARRAY, not just a scalar value.

Multiple Calls

It's important to recognize that while this feature is designed to save processing time when things aren't needed, if you're not careful it can actually increase the number of times you perform your calculation. HTML::Template calls your code reference each time it seems your loop in the template, this includes the times that you might use the loop in a conditional (TMPL_IF or TMPL_UNLESS). For instance:

    <tmpl_if we care>
      <tmpl_if needles_in_haystack>
          <tmpl_loop needles_in_haystack>
            Found <tmpl_var __counter>!
        No needles found!

This will actually call find_needles() twice which will be even worse than you had before. One way to work around this is to cache the return value yourself:

    my $needles;
    $tmpl->param(needles_in_haystack => sub { defined $needles ? $needles : $needles = find_needles() });


I am aware of no bugs - if you find one, join the mailing list and tell us about it. You can join the HTML::Template mailing-list by visiting:

Of course, you can still email me directly ( with bugs, but I reserve the right to forward bug reports to the mailing list.

When submitting bug reports, be sure to include full details, including the VERSION of the module, a test script and a test template demonstrating the problem!

If you're feeling really adventurous, HTML::Template has a publically available Git repository. See below for more information in the PUBLIC GIT REPOSITORY section.


This module was the brain child of my boss, Jesse Erlbaum ( at Vanguard Media ( . The most original idea in this module - the <TMPL_LOOP> - was entirely his.

Fixes, Bug Reports, Optimizations and Ideas have been generously provided by:



You can find information about HTML::Template and other related modules at:


HTML::Template now has a publicly accessible Git repository provided by GitHub ( You can access it by going to Give it a try!


Sam Tregar,


Michael Peters,


  HTML::Template : A module for using HTML Templates with Perl
  Copyright (C) 2000-2011 Sam Tregar (
  This module is free software; you can redistribute it and/or modify it
  under the same terms as Perl itself, which means using either:
  a) the GNU General Public License as published by the Free Software
  Foundation; either version 1, or (at your option) any later version,
  b) the "Artistic License" which comes with this module.
  This program is distributed in the hope that it will be useful,
  but WITHOUT ANY WARRANTY; without even the implied warranty of
  the GNU General Public License or the Artistic License for more details.
  You should have received a copy of the Artistic License with this
  module.  If not, I'll be glad to provide one.
  You should have received a copy of the GNU General Public License
  along with this program. If not, write to the Free Software
  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307