Parse::Template - Processor for templates containing Perl expressions


  use Parse::Template;

  my %template = 
     'TOP' =>  q!Text before %%$self->eval('DATA')%% text after!,
     'DATA' => q!Insert data: ! .
               q!1. List: %%"@list$N"%%! .
               q!2. Hash: %%"$hash{'key'}$N"%%! .
               q!3. File content: %%<FH>%%! .
               q!4. Sub: %%&SUB()$N%%!
  my $tmplt = new Parse::Template (%template);
  open FH, "< foo";

  $tmplt->env('var' => '(value!)');
  $tmplt->env('list' => [1, 2, 10], 
              'N' => "\n",
              'FH' => \*FH,
              'SUB' => sub { "->content generated by a sub<-" },
              'hash' => { 'key' => q!It\'s an hash value! });
  print $tmplt->eval('TOP'), "\n";


The Parse::Template class evaluates Perl expressions placed within a text. This class can be used as a code generator, or a generator of documents in various document formats (HTML, XML, RTF, etc.).

The principle of template-based text generation is simple. A template consists of a text which includes expressions to be evaluated. Interpretation of these expressions generates text fragments which are substituted in place of the expressions. In the case of Parse::Template the expressions to be evaluated are Perl expressions placed within two %%.

Evaluation takes place within an environment in which, for example, you can place data structures which will serve to generate the parts to be completed.

           Text + Perl Expression 
                +-----> Evaluation ----> Text(document or program)
           Subs + Data structures

The Parse::Template class permits decomposing a template into parts. These parts are defined by a hash passed as an argument to the class constructor: Parse::Template->new('someKey', '... text with expressions to evaluate ...'). Within a part, a sub-part can be included by means of an expression of the form:


$self designates the instance of the Parse::Template class. In an expression you can also use the $part which contains the part of the template where the expression is found.

Within an expression it is possible to specify only the name of a part to be inserted. In this case a subroutine with the name of this part is generated dynamically. In the example given in the synopsis, the insertion of the TOP part can thus be rewritten as follows:

  'TOP' => q!Text before %%DATA()%% text after!

DATA() is placed within %% and is in effect treated as an expression to be evaluated.

The subroutines take arguments. In the following example, the argument is used to control the depth of recursive calls of a template:

  print Parse::Template->new(
    'TOP' => q!%%$_[0] < 10 ? '[' . TOP($_[0] + 1) . ']' : ''%%!
   )->eval('TOP', 0);

$_[0] initially contains 0. TOP is included as long as the argument is less than 10. For each inclusion, 1 is added to the argument.

The env() method permits constructing the environment required for evaluation of a template. Each entry to be defined within this environment must be specified using a key consisting of the name of the symbol to be created, associated with a reference whose type is that of the entry to be created within this environment (for example, a reference to an array to create an array). A scalar variable is defined by associating the name of the variable with its value. A scalar variable containing a reference is defined by writing 'var'=>\$variable, where $variable is a lexical variable that contains the reference.

Each instance of Parse::Template is defined within a specific class, a subclass of Parse::Template. The subclass contains the environment specific to the template and inherits methods from the Parse::Template class.

If a template is created from an existing template (i.e. calling new as a method of the existing template), it inherits all the parts defined by its ancestor.

In case of a syntax error in the evalutaion of an expression, Parse::Template tries to indicate the template part and the expression that is "incriminated". If the variable $Parse::Template::CONFESS contains the value TRUE, the stack of evaluations is printed.


new HASH

Constructor for the class. HASH is a hash which defines the template text.


  use Parse::Template;
  $t = new Parse::Template('key' => 'associated text');
env HASH

Permits defining the environment that is specific to a template.

env(SYMBOL) returns the reference associated with the symbol, or undef if the symbol is not defined. The reference that is returned is of the type indicated by the character (&, $, %, @, *) that prefixes the symbol.


  $tmplt->env('LIST' => [1, 2, 3])}   Defines a list

  @{$tmplt->env('*LIST')}             Returns the list

  @{$tmplt->env('@LIST')}             Ditto

Evaluates the template part designated by PART_NAME. Returns the string resulting from this evaluation.


Returns the designated part of the template.

ppregexp REGEXP

Preprocesses a regular expression so that it can be inserted into a template where the regular expression delimiter is either a "/" or a "!".


setPart() permits defining a new entry in the hash that defines the contents of the template.


The Parse::Template class can be used in all sorts of amusing ways. Here are a few illustrations.

HTML Generator

The first example shows how to generate an HTML document by using a data structure placed within the evaluation environment. The template consists of two parts, DOC and SECTION. The SECTION part is called within the DOC part to generate as many sections as there are elements in the array section_content.

  my %template = ('DOC' => <<'END_OF_DOC;', 'SECTION' => <<'END_OF_SECTION;');
  my $content;
  for (my $i = 0; $i <= $#section_content; $i++) {
    $content .= SECTION($i);
  $section_content[$_[0]]->{Content} =~ s/^/<p>/mg;
  join '', '<H1>', $section_content[$_[0]]->{Title}, '</H1>',

  my $tmplt = new Parse::Template (%template);

  $tmplt->env('section_content' => [
        Title => 'First Section',
        Content => 'Nothing to write'
        Title => 'Second section',
        Content => 'Nothing else to write'

  print $tmplt->eval('DOC'), "\n";

HTML generation using functional notation

The second example shows how to generate an HTML document using a functional notation, in other words, obtaining the text:

  <P><B>text in bold</B><I>text in italic</I></P>


  P(B("text in bold"), I("text in italic"))

The functions P(), B() and I() are defined as parts of a template. The Perl expression that permits producing the content of an element is very simple, and reduces to:

  join '', @_

The content to be evaluated is the same regardless of the tag and can therefore be placed within a variable. We therefore obtain the following template:

  my $ELT_CONTENT = q!%%join '', @_%%!;
  my $HTML_T1 = new Parse::Template(
       'DOC' => '%%P(B("text in bold"), I("text in italic"))%%',
       'P'   => qq!<P>$ELT_CONTENT</P>!,
       'B'   => qq!<B>$ELT_CONTENT</B>!,
       'I'   => qq!<I>$ELT_CONTENT</I>!,
  print $HTML_T1->eval('DOC'), "\n";

We can go further by making use of the $part variable, which is defined by default in the environment of evaluation of the template:

  my $ELT_CONTENT = q!%%"<$part>" . join('', @_) . "</$part>"%%!;
  my $HTML_T2 = new Parse::Template(
       'DOC' => '%%P(B("text in bold"), I("text in italic"))%%',
       'P'   => qq!$ELT_CONTENT!,
       'B'   => qq!$ELT_CONTENT!,
       'I'   => qq!$ELT_CONTENT!,
  print $HTML_T2->eval('DOC'), "\n";

Let's look at another step which automates the production of expressions from the list of HTML tags which are of interest to us:

  my $DOC = q!P(B("text in bold"), I("text in italic"))!;
  my $ELT_CONTENT = q!%%"<$part>" . join('', @_) . "</$part>"%%!;
  my $HTML_T3 = new Parse::Template(
       'DOC' => qq!%%$DOC%%!,
       map { $_ => $ELT_CONTENT } qw(P B I)
  print $HTML_T3->eval('DOC'), "\n";

To benefit from the possibility of using the template parts as procedures, we can inherit from the generated template class:

  use Parse::Template;
  my $ELT_CONTENT = q!%%"<$part>" . join('', @_) . "</$part>"%%!;
  my $G = new Parse::Template(
       map { $_ => $ELT_CONTENT } qw(H1 B I)
  @main::ISA = ref($G);
  *AUTOLOAD = \&Parse::Template::AUTOLOAD;
  print H1(B("text in bold"), I("text in italic"));

The reference to Parse::Template::AUTOLOAD avoids the warning message:

  Use of inherited AUTOLOAD for non-method %s() is deprecated

Not very elegant.

HTML generation by method call

With a slight transformation it is possible to use a method-invocation notation:

  my $ELT_CONTENT = q!%%shift(@_); "<$part>" . join('', @_) . "</$part>"%%!;
  my $HTML_T4 = new Parse::Template(
       map { $_ => $ELT_CONTENT } qw(P B I)
  print $HTML_T4->P(
                    $HTML_T4->B("text in bold"),
                    $HTML_T4->I("text in italic")
                   ), "\n";

The shift(@_) permits getting rid of the template object, which we don't need within the expression.

Inheritance of parts

In the following example the child template $C inherits the parts defined in its parent template $A:

  my %ancestor = 
     'TOP' => q!%%"Use the $part model and -> " . CHILD()%%!,
     'ANCESTOR' => q!ANCESTOR %%"'$part' part\n"%%!,

  my %child = 
     'CHILD' => q!CHILD %%"'$part' part"%% -> %%ANCESTOR() . "\n"%%!,
  my $A = new Parse::Template (%ancestor);
  my $C = $A->new(%child);
  print $C->TOP();

The part <TOP> defined in $A can be called directly from $C, that derives from $A.

Other examples

Parse::Template was initially created to serve as a code generator for the Parse::Lex class. You will find other examples of its use in the classes Parse::Lex, Parse::CLex and Parse::Token.


I would be very interested to receive your comments and suggestions.


Instances are not destroyed. Therefore, do not use this class to create a large number of instances.


Philippe Verdret (with translation of documentation into English by Ocrat)


Copyright (c) 1995-2001 Philippe Verdret. All rights reserved. This module is free software; you can redistribute it and/or modify it under the same terms as Perl itself.