The London Perl and Raku Workshop takes place on 26th Oct 2024. If your company depends on Perl, please consider sponsoring and/or attending.


Text::Hogan::Compiler - parse templates and output Perl code


version 2.03


    use Text::Hogan::Compiler;

    my $compiler = Text::Hogan::Compiler->new;

    my $text = "Hello, {{name}}!";

    my $tokens   = $compiler->scan($text);
    my $tree     = $compiler->parse($tokens, $text);
    my $template = $compiler->generate($tree, $text);

    say $template->render({ name => "Alex" });



Takes nothing, returns a Compiler object.

    my $compiler = Text::Hogan::Compiler->new;


Takes template text and returns an arrayref which is a list of tokens.

    my $tokens = $compiler->scan("Hello, {{name}}!");

Optionally takes a hashref with options. 'delimiters' is a string which represents different delimiters, split by white-space. You should never need to pass this directly, it it used to implement the in-template delimiter-switching functionality.

    # equivalent to the above call with mustaches
    my $tokens = Text::Hogan::Compiler->new->scan("Hello, <% name %>!", { delimiters => "<% %>" });

'allow_whitespace_before_hashmark' is a boolean. If true,tags are allowed to have space(s) between the delimiters and the opening sigil ('#', '/', '^', '<', etc.).

    my $tokens = Text::Hogan::Compiler->new->scan("Hello{{ # foo }}, again{{ / foo }}.", { allow_whitespace_before_hashmark => 1 });


Takes the tokens returned by scan, along with the original text, and returns a tree structure ready to be turned into Perl code.

    my $tree = $compiler->parse($tokens, $text);

Optionally takes a hashref that can have a key called "section_tags" which should be an arrayref. I don't know what it does. Probably something internal related to recursive calls that you don't need to worry about.

Note that a lot of error checking on your input gets done in this method, and it is pretty much the only place exceptions might be thrown. Exceptions which may be thrown include: "Closing tag without opener", "Missing closing tag", "Nesting error" and "Illegal content in < super tag".


Takes the parsed tree and the original text and returns a Text::Hogan::Template object that you can call render on.

    my $template = $compiler->generate($tree, $text);

Optionally takes a hashref that can have

- a key "as_string". If that is passed then instead of getting a template object back you get some stringified Perl code that you can cache somewhere on disk as part of your build process.

    my $perl_code_as_string = $compiler->generate($tree, $text, { 'as_string' => 1 });

- a key "numeric_string_as_string". If that is passed output that looks like a number is NOT converted into a number (ie "01234" is NOT converted to "1234")

    my $perl_code_as_string = $compiler->generate($tree, $text, { 'numeric_string_as_string' => 1 });

The options hashref can have other keys which will be passed to Text::Hogan::Template::new among other places.


Takes a template string and calls scan, parse and generate on it and returns you the Text::Hogan::Template object.

    my $template = $compiler->compile("Hello, {{name}}!");

Also caches templates by a sensible cache key, which can be useful if you're not stringifying and storing on disk or in memory anyway.

Optionally takes a hashref that will be passed on to scan, parse, and generate.

    my $perl_code_as_string = $compiler->compile(
            delimiters => "<% %>",
            as_string => 1,
            allow_whitespace_before_hashmark => 1,


As long as you are consistent with your use of encoding in your template variables and your context variables, everything should just work. You can use byte strings or character strings and you'll get what you expect.

The only danger would be if you use byte strings of a multi-byte encoding and you happen to get a clash with your delimiters, eg. if your 4 byte kanji character happens to contain the ASCII '<' and '%' characters next to each other. I have no idea what the likelihood of that is, but hopefully if you're working with non-ASCII character sets you're also using Perl's unicode character strings features.

Compiling long character string inputs with Text::Hogan used to be extremely slow but an optimisation added in version 2.00 has made the overhead much more manageable.


Started out statement-for-statement copied from hogan.js by Twitter!

Initial translation by Alex Balhatchet (

Further improvements from:

Ed Freyfogle Mohammad S Anwar Ricky Morse Jerrad Pierce Tom Hukins Tony Finch Yanick Champoux