# ------------------------------------------------------------------
# Petal - Perl Template Attribute Language
# ------------------------------------------------------------------
# Author: Jean-Michel Hiver
# Description: Front-end for all Petal templating functionality
# ------------------------------------------------------------------
package Petal;
use Petal::Hash;
use Petal::Cache::Disk;
use Petal::Cache::Memory;
use Petal::Parser;
use Petal::Canonicalizer::XML;
use Petal::Canonicalizer::XHTML;
use Petal::Functions;
use Petal::Entities;
use File::Spec;
use Carp;
use Data::Dumper;
use Scalar::Util;
use strict;
use warnings;
use MKDoc::XML::Decode;
use Petal::I18N;


BEGIN
{
    ($] > 5.007) and do {
        require Encode;
    };
    $@ and warn $@;
}



# these are used as local variables when the XML::Parser
# is crunching templates...
use vars qw /@tokens @nodeStack/;


# HTML errors?
our $HTML_ERRORS = undef;


# Encode / Decode info...
our $DECODE_CHARSET = 'utf8';
our $ENCODE_CHARSET = 'utf8'; # deprecated


# Prints as much info as possible when this is enabled.
our $DEBUG_DUMP = 1;


# Warn about uninitialised values in the template?
our $WARN_UNINIT = 0;


# What do we use to parse input?
our $INPUT  = 'XML';
our $INPUTS = {
    'XML'     => 'Petal::Parser',
    'HTML'    => 'Petal::Parser',
    'XHTML'   => 'Petal::Parser',
};


# What do we use to format output?
our $OUTPUT  = 'XML';
our $OUTPUTS = {
    'XML'   => 'Petal::Canonicalizer::XML',
    'HTML'  => 'Petal::Canonicalizer::XHTML',
    'XHTML' => 'Petal::Canonicalizer::XHTML',
};


# makes taint mode happy if set to 1
our $TAINT = undef;


# don't confess() errors if we access an undefined template variable
our $ERROR_ON_UNDEF_VAR = 1;


# where are our templates supposed to be?
our @BASE_DIR = ('.');
our $BASE_DIR = undef; # for backwards compatibility...


# vroom!
our $DISK_CACHE = 1;


# vroom vroom!
our $MEMORY_CACHE = 1;


# cache only mode
our $CACHE_ONLY = 0;


# prevents infinites includes...
our $MAX_INCLUDES = 30;
our $CURRENT_INCLUDES = 0;


# this is for CPAN
our $VERSION = '2.20';


# The CodeGenerator class backend to use.
# Change this only if you know what you're doing.
our $CodeGenerator = 'Petal::CodeGenerator';
our $CodeGeneratorLoaded = 0;

# Default language for multi-language mode.
# Change if you feel that English isn't a fair default.
our $LANGUAGE = 'en';


# this is for XML namespace support. Can't touch this :-)
our $NS = 'petal';
our $NS_URI = 'http://purl.org/petal/1.0/';

our $XI_NS = 'xi';
our $XI_NS_URI = 'http://www.w3.org/2001/XInclude';

our $MT_NS       = 'metal';
our $MT_NS_URI   = 'http://xml.zope.org/namespaces/metal';
our $MT_NAME_CUR = 'main';

# translation service, optional
our $TranslationService = undef;


# Displays the canonical template for template.xml.
# You can set $INPUT using by setting the PETAL_INPUT environment variable.
# You can set $OUTPUT using by setting the PETAL_OUTPUT environment variable.
sub main::canonical
{
    my $file = shift (@ARGV);
    local $Petal::DISK_CACHE = 0;
    local $Petal::MEMORY_CACHE = 0;
    local $Petal::CACHE_ONLY = 0;
    local $Petal::INPUT  = $ENV{PETAL_INPUT}  || 'XML';
    local $Petal::OUTPUT = $ENV{PETAL_OUTPUT} || 'XHTML';
    print ${Petal->new ($file)->_canonicalize()};
}


# Displays the perl code for template.xml.
# You can set $INPUT using by setting the PETAL_INPUT environment variable.
# You can set $OUTPUT using by setting the PETAL_OUTPUT environment variable.
sub main::code
{
    my $file = shift (@ARGV);
    local $Petal::DISK_CACHE = 0;
    local $Petal::MEMORY_CACHE = 0;
    local $Petal::CACHE_ONLY = 0;
    print Petal->new ($file)->_code_disk_cached;
}


# Displays the perl code for template.xml, with line numbers.
# You can set $INPUT using by setting the PETAL_INPUT environment variable.
# You can set $OUTPUT using by setting the PETAL_OUTPUT environment variable.
sub main::lcode
{
    my $file = shift (@ARGV);
    local $Petal::DISK_CACHE = 0;
    local $Petal::MEMORY_CACHE = 0;
    local $Petal::CACHE_ONLY = 0;
    print Petal->new ($file)->_code_with_line_numbers;
}


sub load_code_generator
{
    if (not $CodeGeneratorLoaded)
    {
	eval "require $CodeGenerator";
	confess "Failed to load $CodeGenerator, $@" if $@;
	$CodeGeneratorLoaded = 1;
    }
}


# Instanciates a new Petal object.
sub new
{
    my $class = shift;
    $class = ref $class || $class;
    unshift (@_, 'file') if (@_ == 1);
    my $self = bless { @_ }, $class;
    $self->_initialize();
    
    return $self;
}


# (multi language mode)
# if the language has been specified, let's try to
# find which template we can use.
sub _initialize
{
    my $self  = shift;
    my $file  = $self->{file};
    if ($file =~ /#/)
    {
	my ($file, $macro) = split /#/, $file, 2;
	$self->{file}  = $file;
	$self->_initialize_lang();
	$self->{file} .= "#$macro";
    }
    else
    {
	$self->_initialize_lang();
    }
}


sub _initialize_lang
{
    my $self = shift;
    my $lang = $self->language() || return;
    my @dirs = $self->base_dir();
    @dirs    = map { File::Spec->canonpath ("$_/$self->{file}") } @dirs;
    
    $self->{file}  =~ s/\/$//;
    my $filename   = Petal::Functions::find_filename ($lang, @dirs);
    $self->{file} .= "/$filename" if ($filename);
}


# (multi language mode)
# returns the current preferred language.
sub language
{
    my $self = shift;
    return $self->{language} || $self->{lang};
}


sub default_language   { exists $_[0]->{default_language}   ? $_[0]->{default_language}   : $LANGUAGE           }
sub input              { exists $_[0]->{input}              ? $_[0]->{input}              : $INPUT              }
sub output             { exists $_[0]->{output}             ? $_[0]->{output}             : $OUTPUT             }
sub taint              { exists $_[0]->{taint}              ? $_[0]->{taint}              : $TAINT              }
sub error_on_undef_var { exists $_[0]->{error_on_undef_var} ? $_[0]->{error_on_undef_var} : $ERROR_ON_UNDEF_VAR }
sub disk_cache         { exists $_[0]->{disk_cache}         ? $_[0]->{disk_cache}         : $DISK_CACHE         }
sub memory_cache       { exists $_[0]->{memory_cache}       ? $_[0]->{memory_cache}       : $MEMORY_CACHE       }
sub cache_only         { exists $_[0]->{cache_only}         ? $_[0]->{cache_only}         : $CACHE_ONLY         }
sub max_includes       { exists $_[0]->{max_includes}       ? $_[0]->{max_includes}       : $MAX_INCLUDES       }


sub base_dir
{
    my $self = shift;
    return map { defined $_ ? $_ : () } $self->_base_dir();
}


sub _base_dir
{
    my $self = shift;
    if (exists $self->{base_dir})
    {
	my $base_dir = $self->{base_dir};
	if (ref $base_dir) { return @{$base_dir} }
	else
	{
	    die '\$self->{base_dir} is not defined' unless (defined $base_dir);
	    return $base_dir;
	}
    }
    else
    {
	if (defined $BASE_DIR) { return ( $BASE_DIR, @BASE_DIR ) }
	else                   { return @BASE_DIR                }
    }
}


# _include_compute_path ($path);
# ------------------------------
# Computes the new absolute path from the current
# path and $path
sub _include_compute_path
{
    my $self  = shift;
    my $file  = shift;
    
    # this is for metal self-includes
    if ($file =~ /^#/)
    {
	$file = $self->{file} . $file;
    }
    
    return $file unless ($file =~ /^\./);
    
    my $path = $self->{file};
    ($path)  = $path =~ /(.*)\/.*/;
    $path  ||= '.';
    $path .= '/';
    $path .= $file;
    
    my @path = split /\//, $path;
    my @new_path = ();
    while (scalar @path)
    {
	my $next = shift (@path);
	next if $next eq '.';
	
	if ($next eq '..')
	{
	    die "Cannot go above base directory: $file" if (scalar @new_path == 0);
	    pop (@new_path);
	    next;
	}
	
	push @new_path, $next;
    }
    
    return join '/', @new_path;
}


# Processes the current template object with the information contained in
# %hash. This information can be scalars, hash references, array
# references or objects.
#
# Example:
#
#   my $data_out = $template->process (
#     user   => $user,
#     page   => $page,
#     basket => $shopping_basket,    
#   );
#
# print "Content-Type: text/html\n\n";
# print $data_out;
sub process
{
    my $self = shift;
    $self->_process_absolutize_pathes();
    
    # ok, from there on we need to override any global variable with stuff
    # that might have been specified when constructing the object
    local $TAINT              = defined $self->{taint}               ? $self->{taint}               : $TAINT;
    local $ERROR_ON_UNDEF_VAR = defined $self->{error_on_undef_var}  ? $self->{error_on_undef_var}  : $ERROR_ON_UNDEF_VAR;
    local $DISK_CACHE         = defined $self->{disk_cache}          ? $self->{disk_cache}          : $DISK_CACHE;
    local $MEMORY_CACHE       = defined $self->{memory_cache}        ? $self->{memory_cache}        : $MEMORY_CACHE;
    local $CACHE_ONLY         = defined $self->{cache_only}          ? $self->{cache_only}          : $CACHE_ONLY;
    local $MAX_INCLUDES       = defined $self->{max_includes}        ? $self->{max_includes}        : $MAX_INCLUDES;
    local $INPUT              = defined $self->{input}               ? $self->{input}               : $INPUT;
    local $OUTPUT             = defined $self->{output}              ? $self->{output}              : $OUTPUT;
    local $BASE_DIR           = defined $self->{base_dir} ? do { ref $self->{base_dir} ? undef : $self->{base_dir} } : $BASE_DIR;
    local @BASE_DIR           = defined $self->{base_dir} ? do { ref $self->{base_dir} ? @{$self->{base_dir}} : () } : @BASE_DIR;
    local $LANGUAGE           = defined $self->{default_language}    ? $self->{default_language}    : $LANGUAGE;
    local $DEBUG_DUMP         = defined $self->{debug_dump}          ? $self->{debug_dump}          : $DEBUG_DUMP;
    local $DECODE_CHARSET     = defined $self->{decode_charset}      ? $self->{decode_charset}      : $DECODE_CHARSET;
    local $TranslationService = defined $self->{translation_service} ? $self->{translation_service} : $TranslationService;
    # local $ENCODE_CHARSET     = defined $self->{encode_charset}     ? $self->{encode_charset}     : $ENCODE_CHARSET;

    # prevent infinite includes from happening...
    my $current_includes = $CURRENT_INCLUDES;
    return "ERROR: MAX_INCLUDES : $CURRENT_INCLUDES" if ($CURRENT_INCLUDES > $MAX_INCLUDES);
    local $CURRENT_INCLUDES = $current_includes + 1;
    
    my $res = undef;
    eval {
	my $hash = undef;
	if (ref $_[0] eq 'Petal::Hash') { $hash = shift }
	elsif (ref $_[0] eq 'HASH')     { $hash = new Petal::Hash (%{shift()}) }
	else                            { $hash = new Petal::Hash (@_)         }
	
	my $coderef = $self->_code_memory_cached;
	die "\$coderef is undefined\n\n" unless $coderef;
	die "\$hash is undefined\n\n" unless $hash;
	$res = $coderef->($hash);
    };
    
    if ( $CACHE_ONLY == 1 ){ return 1; }
   
    if (defined $@ and $@) { $res = $self->_handle_error ($@) }
    elsif (defined $TranslationService && $CURRENT_INCLUDES == 1) { $res = Petal::I18N->process ($res) } 

    return $res;
}


# File::Spec->rel2abs() is pretty slow since it uses Cwd which does a
# super-ugly backtick. Hence this method absolutizes base directories
# only once. It is necessary to work with absolute base directories to
# avoid cache conflicts.
sub _process_absolutize_pathes
{
    my $self = shift;
    
    if (defined $BASE_DIR)
    {
	$BASE_DIR = File::Spec->rel2abs ($BASE_DIR) unless (
	    File::Spec->file_name_is_absolute ($BASE_DIR)
	     );
    }
    
    @BASE_DIR = ( map { File::Spec->file_name_is_absolute ($_) ? $_ : File::Spec->rel2abs ($_) }
		  map { defined $_ ? $_ : () } @BASE_DIR );
    
    if (defined $self->{base_dir})
    {
	if (ref $self->{base_dir})
	{
	    $self->{base_dir} = [
		map { File::Spec->file_name_is_absolute ($_) ? $_ : File::Spec->rel2abs ($_) }
		map { defined $_ ? $_ : () } @{$self->{base_dir}}
	       ] if (defined $self->{base_dir});
	}
	else
	{
	    $self->{base_dir} = File::Spec->rel2abs ($self->{base_dir}) unless (
		File::Spec->file_name_is_absolute ($self->{base_dir})
		 );
	}
    }
}


sub _handle_error
{
    my $self = shift;
    my $error = shift;

    $Petal::HTML_ERRORS and do {
        my $res  = '<pre>';
        $res    .= "Error: $error\n";
        $res    .= "=============\n";
        $res    .= "\n\n";
        $res    .= "Petal object dump:\n";
        $res    .= "==================\n";
	$res    .= Dumper ($self);
        $res    .= "\n\n";
	$res    .= "Stack trace:\n";
	$res    .= "============\n";
	$res    .= Carp::longmess();
	$res    .= "\n\n";
        $res    .= "Template perl code dump:\n";
        $res    .= "========================\n";

        my $dump = eval { $self->_code_with_line_numbers() };
        $res    .= ($dump) ? $dump : "(no dump available)";
        
        $res .= '</pre>';
        return $res;
    };
    
    $Petal::DEBUG_DUMP and do {
	my $tmpdir  = File::Spec->tmpdir();
	my $tmpfile = $$ . '.' . time() . '.' . ( join '', map { chr (ord ('a') + int (rand (26))) } 1..10 );
	my $debug   = "$tmpdir/petal_debug.$tmpfile";
	
	open ERROR, ">$debug" || die "Cannot write-open \">$debug\"";
	
	print ERROR "Error: $error\n";
	ref $error and do {
	    print ERROR "=============\n";
	};
	print ERROR "\n";
	
	print ERROR "Petal object dump:\n";
	print ERROR "==================\n";
	print ERROR Dumper ($self);
	print ERROR "\n\n";
	
	print ERROR "Stack trace:\n";
	print ERROR "============\n";
	print ERROR Carp::longmess();
	print ERROR "\n\n";
	
	print ERROR "Template perl code dump:\n";
	print ERROR "========================\n";
	my $dump = eval { $self->_code_with_line_numbers() };
	($dump) ? print ERROR $dump : print ERROR "(no dump available)";
	
	die "[PETAL ERROR] $error. Debug info written in $debug";
    };
    
    ! $Petal::DEBUG_DUMP and do {
	die "[PETAL ERROR] $error. No debug info written.";
    };
}


# $self->code_with_line_numbers;
# ------------------------------
#   utility method to return the Perl code, each line being prefixed with
#   its number... handy for debugging templates. The nifty line number padding
#   patch was provided by Lucas Saud <lucas.marinho@uol.com.br>.
sub _code_with_line_numbers
{
    my $self = shift;
    my $code = $self->_code_disk_cached;

    # get lines of code
    my @lines = split(/\n/, $code);

    # add line numbers
    my $count = 0;
    @lines = map {
	my $cur_line = $_;
	$count++;
	
	# space padding so the line numbers nicely line up with each other
	my $line_num = sprintf ("%" . length(scalar(@lines)) . "d", $count);
	
	# put line number and line back together
	"${line_num}. ${cur_line}";
    } @lines;
    
    return join("\n", @lines);
}


# $self->_file;
# -------------
#   setter / getter for the 'file' attribute
sub _file
{
    my $self = shift;
    $self->{file} = shift if (@_);
    $self->{file} =~ s/^\///;
    return $self->{file};
}


sub _macro
{
    my $self = shift;
    my $file = $self->_file;
    $file =~ s/^.*#// || return;
    return $file;
}


sub _file_path_with_macro
{
    my $self  = shift;
    my $file  = $self->_file_path;
    my $macro = $self->_macro;
    my $res   = $macro ? "$file#$macro" : $file;
    return $res;
}


# $self->_file_path;
# ------------------
#   computes the file of the absolute path where the template
#   file should be fetched
sub _file_path
{
    my $self = shift;
    my $file = $self->_file;
    $file =~ s/#.*$//;
    my @dirs = $self->base_dir;
    
    foreach my $dir (@dirs)
    {
	# my $base_dir  = File::Spec->canonpath ($dir);
	# $base_dir     = File::Spec->rel2abs ($base_dir) unless ($base_dir =~ /^\//);
	my $base_dir  = $dir;
	$base_dir     =~ s/\/$//;
	my $file_path = File::Spec->canonpath ($base_dir . '/' . $file);
	return $file_path if (-e $file_path and -r $file_path);
    }
    
    Carp::confess ("Cannot find $file in @dirs. (typo? permission problem?)");
}


# $self->_file_data_ref;
# ----------------------
# slurps the template data into a variable and returns a
# reference to that variable
sub _file_data_ref
{
    my $self      = shift;
    my $file_path = $self->_file_path;
    $file_path =~ s/#.*$//;
    
    if ($] > 5.007)
    {
	my $encoding = Encode::resolve_alias ($DECODE_CHARSET) || 'utf8';
	open FP, "<:encoding($encoding)", "$file_path" or die "Cannot read-open $file_path";
    }
    else
    {
	open FP, "<$file_path" || die 'Cannot read-open $file_path';
    }
    
    my $res = join '', <FP>;
    close FP;
    
    # kill template comments
    $res =~ s/\<!--\?.*?\-->//gsm;
    
    my $decode = ($OUTPUT =~ /HTML$/i or $INPUT =~ /HTML$/i) ?
        new MKDoc::XML::Decode ('numeric', 'xhtml') :
	new MKDoc::XML::Decode ('numeric');
    
    $res = $decode->process ($res);    
    return \$res;
}


# $self->_code_disk_cached;
# -------------------------
# Returns the Perl code data, using the disk cache if possible
sub _code_disk_cached
{
    my $self = shift;
    my $code = (defined $DISK_CACHE and $DISK_CACHE) ? Petal::Cache::Disk->get ($self->_file_path_with_macro, $self->language) : undef;
    unless (defined $code)
    {
	my $macro = $self->_macro() || $MT_NAME_CUR;
	
	local ($MT_NAME_CUR);
	$MT_NAME_CUR = $macro;
	
	my $data_ref = $self->_canonicalize;
	load_code_generator();
	$code = $CodeGenerator->process ($data_ref, $self);
	Petal::Cache::Disk->set ($self->_file_path_with_macro, $code, $self->language) if (defined $DISK_CACHE and $DISK_CACHE);
    }
    
    return $code;
}


# $self->_code_memory_cached;
# ---------------------------
# Returns the Perl code data, using the disk cache if possible
sub _code_memory_cached
{
    my $self = shift;
    my $code = (defined $MEMORY_CACHE and $MEMORY_CACHE) ? Petal::Cache::Memory->get ($self->_file_path_with_macro, $self->language) : undef;
    unless (defined $code)
    {
	my $code_perl = $self->_code_disk_cached;
	my $VAR1 = undef;
	
	eval "$code_perl";
	confess ($@ . "\n" . $self->_code_with_line_numbers) if $@;
	$code = $VAR1;
	
	Petal::Cache::Memory->set ($self->_file_path_with_macro, $code, $self->language) if (defined $MEMORY_CACHE and $MEMORY_CACHE);	
    }
    
    return $code;
}


# $self->_code_cache;
# -------------------
#   Returns TRUE if this object uses the code cache, FALSE otherwise
sub _memory_cache
{
    my $self = shift;
    return $self->{memory_cache} if (defined $self->{memory_cache});
    return $MEMORY_CACHE;
}


# $self->_canonicalize;
# ---------------------
#   Returns the canonical data which will be sent to the
#   Petal::CodeGenerator module
sub _canonicalize
{
    my $self = shift;
    my $parser_type        = $INPUTS->{$INPUT}   || confess "unknown \$Petal::INPUT = $INPUT";
    my $canonicalizer_type = $OUTPUTS->{$OUTPUT} || confess "unknown \$Petal::OUTPUT = $OUTPUT";
    
    my $data_ref = $self->_file_data_ref;
    my $parser = $parser_type->new;
    return $canonicalizer_type->process ($parser, $data_ref);
}


1;


=head1 NAME

Petal - Perl Template Attribute Language - TAL for Perl!


=head1 SYNOPSIS

in your Perl code:

  use Petal;
  my $template = new Petal ('foo.xhtml');
  print $template->process (bar => 'BAZ');


in foo.xhtml

  <html xmlns:tal="http://purl.org/petal/1.0/">
    <body tal:content="bar">Dummy Content</body>
  </html>


and you get something like:

  <html>
    <body>BAZ</body>
  </html>


=head1 SUMMARY

Petal is a XML based templating engine that is able to process any
kind of XML, XHTML and HTML.

Petal borrows a lot of good ideas from the Zope Page Templates TAL
specification, it is very well suited for the creation of WYSIWYG XHTML
editable templates.

The idea is to further enforce the separation of logic from presentation. With
Petal, graphic designers can use their favorite WYSIWYG editor to easily edit
templates without having to worry about the loops and ifs which happen behind
the scene.


=head1 NAMESPACE

Although this is not mandatory, Petal templates should include use the namespace
L<http://purl.org/petal/1.0/>. Example:

    <html xml:lang="en"
          lang="en"
          xmlns="http://www.w3.org/1999/xhtml"
          xmlns:tal="http://purl.org/petal/1.0/">

      Blah blah blah...
      Content of the file
      More blah blah...
    </html>

If you do not specify the namespace, Petal will by default try to use the
C<petal:> prefix. However, in all the examples of this POD we'll use the
C<tal:> prefix to avoid too much typing.


=head1 KICKSTART

Let's say you have the following Perl code:

    use Petal;
    local $Petal::OUTPUT = 'XHTML';

    my $template = new Petal ('foo.xhtml');
    $template->process ( my_var => some_object() );

some_object() is a subroutine that returns some kind of object, may it be a scalar,
object, array referebce or hash reference. Let's see what we can do...


=head2 Version 1: WYSIWYG friendly prototype.

Using TAL you can do:

    This is the variable 'my_var' :
    <span tal:replace="my_var/hello_world">Hola, Mundo!</span>

Now you can open your template in any WYSIWYG tool (mozilla composer,
frontpage, dreamweaver, adobe golive...) and work with less risk of damaging
your petal commands.


=head2 Version 2: Object-oriented version

Let's now say that C<my_var> is actually an object with a method hello_world()
that returns I<Hello World>. To output the same result, your line, which was:

    <span tal:replace="my_var/hello_world">Hola, Mundo!</span>

Would need to be... EXACTLY the same. Petal lets you access hashes and objects
in an entirely transparent way and tries to automagically do The Right Thing
for you.

This high level of polymorphism means that in most cases you can maintain your
code, swap hashes for objects, and not change a single line of your template
code.


=head2 Version 3: Personalizable

Now let's say that your method hello_world() can take an optional
argument so that C<$some_object-E<gt>hello_world ('Jack')> returns I<Hello Jack>.

You would write:

    <span tal:replace="my_var/hello_world 'Jack'">Hola, Mundo!</span>


Optionally, you can get rid of the quotes by using two dashes, a la GNU
command-line option:

    <span tal:replace="my_var/hello_world --Jack">Hola, Mundo!</span>


So you can pass parameters to methods using double dashes or quotes.
Now let us say that your C<my_var> object also has a method current_user()
that returns the current user real name. You can do:

    <span tal:replace="my_var/hello_world my_var/current_user">Hola, Mundo!</span>


TRAP:

You cannot write nested expressions such as:

    ${my_var/hello_world ${my_var/current_user}}

This will NOT work. At least, not yet.


=head2 Version 4: Internationalized

Let's say that you have a directory called C<hello_world> with the following
files:

    hello_world/en.xhtml
    hello_world/fr.xhtml
    hello_world/es.xhtml

You can use Petal as follows in your Perl code:

    use Petal;
    local $Petal::OUTPUT = 'XHTML';

    my $template = new Petal ( file => 'hello_world', lang => 'fr-CA' );
    print $template->process ( my_var => some_object() );

What will happen is that the C<$template> object will look in the
C<hello_world> directory and try to find a file named C<fr-CA.xhtml>, then
C<fr.xhtml>, then will default to C<en.xhtml>. It works fine for includes, too!

These internationalized templates can have whatever file-extension you like,
Petal searches on the first part of the filename.  So you can call them
C<fr.html>, C<fr.xml>, C<fr.xhtml> or use whatever convention suits you.

NOTE: There is now support for ZPT-like i18n attributes, which should
provide a much nicer framework. See L<Petal::I18N> for details.


TIP:

If you feel that 'en' should not be the default language, you can specify a
different default:

    my $template = new Petal (
        file             => 'hello_world',
        language         => 'zh',
        default_language => 'fr' # vive la France!
    );


TRAP:

If you do specify the C<lang> option, you MUST use a path to a template
directory, not a file directory.

Conversely, if you do not specify a C<lang> option, you MUST use a path to a
template file, not a directory.


=head1 OPTIONS

When you create a Petal template object you can specify various options using
name => value pairs as arguments to the constructor.  For example:

  my $template = Petal->new(
    file     => 'gerbils.html',
    base_dir => '/var/www/petshop',
    input    => 'HTML',
    output   => 'HTML',
  );

The recognized options are:


=head2 file => I<filename>

The template filename.  This option is mandatory and has no default.

Note: If you also use 'language' this option should point to a directory.


=head2 base_dir => I<pathname> | [ I<pathname list> ] (default: '.')

The directories listed in this option will be searched in turn to locate the
template file.  A single directory can be specified as a scalar.  For a
directory list use an arrayref.


=head2 input => 'HTML' | 'XHTML' | 'XML' (default: 'XML')

Defines the format of the template files.  Recognised values are:

  'HTML'  - Alias for 'XHTML'
  'XHTML' - Petal will use Petal::Parser to parse the template
  'XML'   - Petal will use Petal::Parser to parse the template


=head2 output => 'HTML' | 'XHTML' | 'XML' (default: 'XML')

Defines the format of the data generated as a result of processing the template
files.  Recognised values are:

  'HTML'  - Petal will output XHTML, self-closing certain tags
  'XHTML' - Alias for 'HTML'
  'XML'   - Petal will output generic XML 


=head2 language => I<language code>

For internationalized applications, you can use the 'file' option to point to a
I<directory> and select a language-specific template within that directory
using the 'language' option.  Languages are selected using a two letter code
(eg: 'fr') optionally followed by a hyphen and a two letter country code (eg:
'fr-CA').


=head2 default_language => I<language code> (default: 'en')

This language code will be used if no template matches the selected
language-country or language.


=head2 taint => I<true> | I<false> (default: I<false>)

If set to C<true>, makes perl taint mode happy.


=head2 error_on_undef_var => I<true> | I<false> (default: I<true>)

If set to C<true>, Petal will confess() errors when trying to access undefined
template variables, otherwise an empty string will be returned.


=head2 disk_cache => I<true> | I<false> (default: I<true>)

If set to C<false>, Petal will not use the C<Petal::Cache::Disk> module.


=head2 memory_cache => I<true> | I<false> (default: I<true>)

If set to C<false>, Petal will not use the C<Petal::Cache::Memory> module.


=head2 cache_only => I<true> | I<false> (default: I<false>)

If set to C<true>, Petal will return true after having compiled a template into
perl code and a subroutine , and optionally using disk_cache or memory_cache if
either is set.


=head2 max_includes => I<number> (default: 30)

The maximum number of recursive includes before Petal stops processing.  This
is to guard against accidental infinite recursions.


=head2 debug_dump => I<true> | I<false> (default: I<true>)

If this option is true, when Petal cannot process a template it will
output lots of debugging information in a temporary file which you can
inspect.  The location for this file is wherever File::Spec->tmpdir()
specifies as a temp directory (usually /tmp on a unix system).


=head2 encode_charset => I<charset> (default: undef)

This option is _DEPRECATED_ as of Petal 2.01.
Petal will now always return results in Perl's internal form.

It doesn't guarantee that the result will be in UTF-8 or in your
local encoding, but at least the UTF-8 flag should be set properly.

If you want to encode the results for a specific charset, you
should look at the module L<Encode>.


=head2 decode_charset => I<charset> (default: undef)

This option will work only if you use Perl 5.8 or greater.

If specified, Petal will assume that the template to be processed (and its
sub-templates) are in the character set I<charset>.

I<charset> can be any character set that can be used with the module L<Encode>. 


=head1 TAL SYNTAX

This functionality is directly and shamelessly stolen from the excellent TAL
specification: L<http://www.zope.org/Wikis/DevSite/Projects/ZPT/TAL>.


=head2 define

Abstract

  <tag tal:define="variable_name EXPRESSION">

Evaluates C<EXPRESSION> and assigns the returned value to C<variable_name>.

Example

  <!--? sets document/title to 'title' -->
  <span tal:define="title document/title">

Why?

This can be useful if you have a C<very/very/long/expression>.  You can set it
to let's say C<vvle> and then use C<vvle> instead of using
C<very/very/long/expression>.


=head2 condition (ifs)

Abstract

  <tag tal:condition="true:EXPRESSION">
     blah blah blah
  </tag>

Example

  <span tal:condition="true:user/is_authenticated">
    Yo, authenticated!
  </span>

Why?

Conditions can be used to display something if an expression
is true. They can also be used to check that a list exists
before attempting to loop through it.


=head2 repeat (loops)

Abstract

  <tag tal:repeat="element_name EXPRESSION">
     blah blah blah
  </tag>

Why?

Repeat statements are used to loop through a list of values,
typically to display the resulting records of a database query.

Example:

  <li tal:repeat="user system/user_list">$user/real_name</li>

A select list with one item selected:

  <select>
    <option
      selected="selected"
      tal:attributes="value self/selected_lang/value"
      tal:content="self/selected_lang/label"
    >English</option>
    <option
      value="i-klingon"
      tal:repeat="lang self/unselected_langs;"
      tal:attributes="value lang/value"
      tal:content="lang/label"
    >Klingon</option>
  </select>

A table with rows of alternating colours set via CSS:

  <table>
    <div
      tal:omit-tag=""
      tal:repeat="audience self/audiences"
    >
      <tr 
        class="odd"
        tal:condition="repeat/odd"
      >
        <td>
          This a odd row, it comes before the even row.
        </td>
      </tr>
      <tr 
        class="even"
        tal:condition="repeat/even"
      >
        <td> 
          This a even row.
        </td>
      </tr>
    </div>
  </table>

I<repeat> is a local temporary object that only exists within a
petal:repeat loop.  It has a bunch of methods useful for selecting
different positions in the loop:

=head3 repeat/index

I<index> returns the numeric position of this item within the loop, starts with
one not zero.

=head3 repeat/number

I<number> is an alias for I<index>.

=head3 repeat/even

I<even> is true if the position is even (0, 2, 4 ...)

=head3 repeat/odd

I<odd> is true is the position is odd (1, 3, 5 ...)

=head3 repeat/start

I<start> is true if this is the first item.

=head3 repeat/end

I<end> is true if this is the last item.

=head3 repeat/inner

I<inner> is true if this is not the I<start> or I<end>.

=head2 attributes

Abstract

  <tag tal:attributes="attr1 EXPRESSION_1; attr2 EXPRESSION_2"; ...">
     blah blah blah
  </tag>

Example

  <a href="http://www.gianthard.com"
     lang="en-gb"
     tal:attributes="href document/href_relative; lang document/lang">

Why?

Attributes statements can be used to template a tag's attributes.


=head2 content

Abstract

  <tag tal:content="EXPRESSION">Dummy Data To Replace With EXPRESSION</tag>

By default, the characters greater than, lesser than, double quote and
ampersand are encoded to the entities I<&lt;>, I<&gt;>, I<&quot;> and I<&amp;>
respectively.  If you don't want them to (because the result of your expression
is already encoded) you have to use the C<structure> keyword.

Example

  <span tal:content="title">Dummy Title</span>

  <span tal:content="structure some/variable">
     blah blah blah
  </span>

Why?

It lets you replace the contents of a tag with whatever value the evaluation of
EXPRESSION returned. This is handy because you can fill your templates with
dummy content which will make them usable in a WYSIWYG tool.


=head2 replace

Abstract

  <tag tal:replace="EXPRESSION">
    This time the entire tag is replaced
    rather than just the content!
  </tag>

Example

  <span tal:replace="title">Dummy Title</span>

Why?

Similar reasons to C<content>. Note however that C<tal:content> and
C<tal:replace> are *NOT* aliases. The former will replace the contents of the
tag, while the latter will replace the whole tag.

Indeed you cannot use C<tal:content> and C<tal:replace> in the same tag.


=head2 omit-tag

Abstract

  <tag tal:omit-tag="EXPRESSION">Some contents</tag>

Example

  <b tal:omit-tag="not:bold">I may not be bold.</b>

If C<not:bold> is evaluated as I<TRUE>, then the <b> tag will be omited.
If C<not:bold> is evaluated as I<FALSE>, then the <b> tag will stay in place.

Why?

omit-tag statements can be used to leave the contents of a tag in place while
omitting the surrounding start and end tags if the expression which is
evaluated is TRUE.

TIP:

If you want to ALWAYS remove a tag, you can use C<omit-tag="string:1">


=head2 on-error

Warning: this is currently only partially implemented.  C<on-error> may be used
in Petal templates, but the expression isn't evaluated - Petal simply prints
the expression as a string.

Abstract

  <tag on-error="EXPRESSION">...</tag>

Example

  <p on-error="string:Cannot access object/method!!">
    $object/method
  </p>

Why?

When Petal encounters an error, it usually dies with some obscure error
message. The C<on-error> statement lets you trap the error and replace it
with a proper error message.


=head2 using multiple statements

You can do things like:

  <p tal:define="children document/children"
     tal:condition="children"
     tal:repeat="child children"
     tal:attributes="lang child/lang; xml:lang child/lang"
     tal:content="child/data"
     tal:on-error="string:Ouch!">Some Dummy Content</p>

Given the fact that XML attributes are not ordered, withing the same tag
statements will be executed in the following order:

    define
    condition
    repeat
        attributes
        content
    OR
        replace
    OR
        omit-tag
        content


TRAP:

Don't forget that the default prefix is C<petal:> NOT C<tal:>, until
you set the petal namespace in your HTML or XML document as follows:

    <html xmlns:tal="http://purl.org/petal/1.0/">


=head1 METAL MACROS

Petal supports an implementation of the METAL specification, which is a very
WYSIWYG compatible way of doing template includes.


=head2 define-macro

In order to define a macro inside a file (i.e. a fragment to be included), you
use the metal:define-macro directive. For example:

  File foo.xml
  ============

  <html xmlns:metal="http://xml.zope.org/namespaces/metal">
    <body>
      <p metal:define-macro="footer">
        (c) Me (r)(tm) (pouet pouet)
      </p>
    </body>
  </html>


=head2 use-macro

In order to use a previously defined macro, you use the metal:use-macro directive.
For example:

  File bar.xml
  ============

  <html xmlns:metal="http://xml.zope.org/namespaces/metal">
    <body>
      ... plenty of content ...

      <p metal:use-macro="foo.xml#footer">
        Page Footer.
      </p>
    </body>
  </html>


=head2 define-slot

In any given macro you can define slots, which are bits of macros that can be
overridden by something else using the fill-macro directive. To re-use the
example above, imagine that we want to be able to optionally override the
(pouet pouet) bit with something else:


  File foo.xml
  ============

  <html xmlns:metal="http://xml.zope.org/namespaces/metal">
    <body>
      <p metal:define-macro="footer">
        (c) Me (r)(tm) <span metal:define-slot="pouet">(pouet pouet)</span>
      </p>
    </body>
  </html>


=head2 fill-slot

Your including file can override any slot using the fill-slot instruction, i.e.

  File bar.xml
  ============

  <html xmlns:metal="http://xml.zope.org/namespaces/metal">
    <body>
      ... plenty of content ...

      <p metal:use-macro="foo.xml#footer">
        Page Footer. <span metal:fill-slot="pouet" petal:omit-tag="">(bar baz)</span>
      </p>
    </body>
  </html>

This would result in the macro 'foo.xml#footer' to produce:

  <html>
    <body>
      <p>
        (c) Me (r)(tm) (bar baz)
      </p>
    </body>
  </html>


=head2 self includes

In Zope, METAL macros are expanded first, and then the TAL instructions are processed.
However with Petal, METAL macros are expanded at run-time just like regular includes,
which allows for recursive macros.

This example templates a sitemap, which on a hierarchically organized site would
be recursive by nature:

  <html xmlns:metal="http://xml.zope.org/namespaces/metal"
        xmlns:petal="http://purl.org/petal/1.0/">
    <body>
      <p>Sitemap:</p>

      <li metal:define-macro="recurse">
        <a href="#"
           petal:attributes="href child/Full_Path" 
           petal:content="child/Title"
        >Child Document Title</a>
        <ul 
          petal:define="children child/Children"
          petal:condition="children"
          petal:repeat="child children"
        >
          <li metal:use-macro="#recurse">Dummy Child 1</li>
          <li petal:replace="nothing">Dummy Child 2</li>
          <li petal:replace="nothing">Dummy Child 3</li>
        </ul>
      </li>
    </body>
  </html>


=head1 EXPRESSIONS AND MODIFIERS

Petal has the ability to bind template variables to the following Perl
datatypes: scalars, lists, hash, arrays and objects. The article describes
the syntax which is used to access these from Petal templates.

In the following examples, we'll assume that the template is used as follows:

  my $hashref = some_complex_data_structure();
  my $template = new Petal ('foo.xml');
  print $template->process ( $hashref );

Then we will show how the Petal Expression Syntax maps to the Perl way of
accessing these values.  


=head2 accessing scalar values

Perl expression

  $hashref->{'some_value'};

Petal expression

  some_value

Example

  <!--? Replaces Hello, World with the contents of
        $hashref->{'some_value'}
  -->
  <span tal:replace="some_value">Hello, World</span>


=head2 accessing hashes & arrays

Perl expression

  $hashref->{'some_hash'}->{'a_key'};

Petal expression

  some_hash/a_key

Example

  <!--? Replaces Hello, World with the contents
        of $hashref->{'some_hash'}->{'a_key'}
  -->
  <span tal:replace="some_hash/a_key">Hello, World</span>


Perl expression

  $hashref->{'some_array'}->[12]

Petal expression

  some_array/12

Example

  <!--? Replaces Hello, World with the contents
       of $hashref->{'some_array'}->[12]
  -->
  <span tal:replace="some_array/12">Hello, World</span>

Note: You're more likely to want to loop through arrays:

  <!--? Loops trough the array and displays each values -->
  <ul tal:condition="some_array">
    <li tal:repeat="value some_array"
        tal:content="value">Hello, World</li>
  </ul>


=head2 accessing object methods

Perl expressions

  1. $hashref->{'some_object'}->some_method();
  2. $hashref->{'some_object'}->some_method ('foo', 'bar');
  3. $hashref->{'some_object'}->some_method ($hashref->{'some_variable'})  

Petal expressions

  1. some_object/some_method
  2a. some_object/some_method 'foo' 'bar'
  2b. some_object/some_method "foo" "bar"
  2c. some_object/some_method --foo --bar
  3. some_object/some_method some_variable

Note that the syntax as described in 2c works only if you use strings
which do not contain spaces.

Example

  <p>
    <span tal:replace="value1">2</span> times
    <span tal:replace="value2">2</span> equals
    <span tal:replace="math_object/multiply value1 value2">4</span>
  </p>
    

=head2 composing

Petal lets you traverse any data structure, i.e.

Perl expression

  $hashref->{'some_object'}
          ->some_method()
          ->{'key2'}
          ->some_other_method ( 'foo', $hash->{bar} );

Petal expression

  some_object/some_method/key2/some_other_method --foo bar


=head2 true:EXPRESSION

  If EXPRESSION returns an array reference
    If this array reference has at least one element
      Returns TRUE
    Else
      Returns FALSE

  Else
    If EXPRESSION returns a TRUE value (according to Perl 'trueness')
      Returns TRUE
    Else
      Returns FALSE

the C<true:> modifiers should always be used when doing Petal conditions.


=head2 false:EXPRESSION

I'm pretty sure you can work this one out by yourself :-)


=head2 set:variable_name EXPRESSION

Sets the value returned by the evaluation of EXPRESSION in
C<$hash-E<gt>{variable_name}>. For instance:

Perl expression:

  $hash->{variable_name} = $hash->{object}->method();

Petal expression:

  set:variable_name object/method


=head2 string:STRING_EXPRESSION

The C<string:> modifier lets you interpolate petal expressions within a string
and returns the value.

  string:Welcome $user/real_name, it is $date!

Alternatively, you could write:

  string:Welcome ${user/real_name}, it is ${date}!
  
The advantage of using curly brackets is that it lets you interpolate
expressions which invoke methods with parameters, i.e.

  string:The current CGI 'action' param is: ${cgi/param --action}


=head1 ADVANCED PETAL


=head2 writing your own modifiers

Petal lets you write your own modifiers, either using coderefs
or modules.


=head3 Coderefs

Let's say that you want to write an uppercase: modifier, which
would uppercase the result of an expression evaluation, as in:

  uppercase:string:Hello, World

Would return

  HELLO, WORLD

Here is what you can do:

  # don't forget the trailing colon in C<uppercase:> !!
  $Petal::Hash::MODIFIERS->{'uppercase:'} = sub {
      my $hash = shift;
      my $args = shift;

      my $result = $hash->fetch ($args);
      return uc ($result);
  };


=head3 Modules.

You might want to use a module rather than a coderef. Here is the example above
reimplemented as a module:

    package Petal::Hash::UpperCase;
    use strict;
    use warnings;
  
    sub process {
      my $class = shift;
      my $hash  = shift;
      my $args  = shift;

      my $result = $hash->fetch ($args);
      return uc ($result);
    }

    1;

As long as your module is in the namespace Petal::Hash::<YourModifierName>,
Petal will automatically pick it up and assign it to its lowercased
name, i.e. in our example C<uppercase:>.

If your modifier is OUTSIDE Petal::Hash::<YourModifierName>, you need to
make Petal aware of its existence as follows:

  use MyPetalModifier::UpperCase;
  $Petal::Hash::MODIFIERS->{'uppercase:'} = 'MyPetalModifier::UpperCase';


=head1 Expression keywords


=head3 XML encoding / structure keyword

By default Petal will encode C<&>, C<<>, C<>> and C<"> to C<&amp;>, C<&lt;>,
C<&gt> and C<&quot;> respectively. However sometimes you might want to display
an expression which is already encoded, in which case you can use the
C<structure> keyword.

  structure my/encoded/variable

Note that this is a language I<keyword>, not a modifier. It does not use a
trailing colon.


=head3 Petal::Hash caching and fresh keyword 

Petal caches the expressions which it resolves, i.e. if you write the
expression:

  string:$foo/bar, ${baz/buz/blah}

Petal::Hash will compute it once, and then for subsequent accesses to that
expression always return the same value. This is almost never a problem, even
for loops because a new Petal::Hash object is used for each iteration in order
to support proper scoping.

However, in some rare cases you might not want to have that behavior, in which
case you need to prefix your expression with the C<fresh> keyword, i.e. 

  fresh string:$foo/bar, ${baz/buz/blah}

You can use C<fresh> with C<structure> if you need to:

  fresh structure string:$foo/bar, ${baz/buz/blah}

However the reverse does not work:

  <!--? VERY BAD, WON'T WORK !!! -->
  structure fresh string:$foo/bar, ${baz/buz/blah}


=head2 TOY FUNCTIONS (For debugging or if you're curious)


=head3 perl -MPetal -e canonical template.xml

Displays the canonical template for template.xml.
You can set C<$Petal::INPUT> using by setting the PETAL_INPUT environment variable.
You can set C<$Petal::OUTPUT> using by setting the PETAL_OUTPUT environment variable.


=head3 perl -MPetal -e code template.xml

Displays the perl code for template.xml.
You can set C<$Petal::INPUT> using by setting the PETAL_INPUT environment variable.
You can set C<$Petal::OUTPUT> using by setting the PETAL_OUTPUT environment variable.


=head3 perl -MPetal -e lcode template.xml

Displays the perl code for template.xml, with line numbers.
You can set C<$Petal::INPUT> using by setting the PETAL_INPUT environment variable.
You can set C<$Petal::OUTPUT> using by setting the PETAL_OUTPUT environment variable.


=head2 What does Petal do internally?

The cycle of a Petal template is the following:

    1. Read the source XML template
    2. $INPUT (XML or HTML) throws XML events from the source file
    3. $OUTPUT (XML or HTML) uses these XML events to canonicalize the template
    4. Petal::CodeGenerator turns the canonical template into Perl code
    5. Petal::Cache::Disk caches the Perl code on disk
    6. Petal turns the perl code into a subroutine
    7. Petal::Cache::Memory caches the subroutine in memory
    8. Petal executes the subroutine
    9. (optional) Petal internationalizes the resulting output.

If you are under a persistent environment a la mod_perl, subsequent calls to
the same template will be reduced to step 8 until the source template changes.

Otherwise, subsequent calls will resume at step 6, until the source template
changes.

If you are using the mod_perl prefork MPM, you can precompile Petal templates
into Apache's shared memory at startup by using the cache_only option.  This
will allow you to run through steps 1-7 without passing any data to Petal.


=head1 DECRYPTING WARNINGS AND ERRORS


=head2 "Cannot import module $module. Reason: $@" (nonfatal)

Petal was not able to import one of the modules. This error warning will be
issued when Petal is unable to load a plugin because it has been badly install
or is just broken.


=head2 "Petal modifier encode: is deprecated" (nonfatal)

You don't need to use encode:EXPRESSION to XML-encode expression anymore,
Petal does it for you. encode: has been turned into a no-op.


=head2 Cannot find value for ... (FATAL)

You tried to invoke an/expression/like/this/one but Petal could not resolve
it. This could be because an/expression/like evaluated to undef and hence the
remaining this/one could not be resolved.

Usually Petal gives you a line number and a dump of your template as Perl
code. You can look at the perl code to try to determine the faulty bit in
your template.


=head2 not well-formed (invalid token) at ... (FATAL)

Petal was trying to parse a file that is not well-formed XML or that has strange
entities in it. Try to run xmllint on your file to see if it's well formed or
try to use the $Petal::INPUT = 'XHTML' option.


=head2 other errors

Either I've forgot to document it, or it's a bug. Send an email to the Petal
mailing list.


=head1 EXPORTS

None.


=head1 AUTHOR

Copyright 2003 - MKDoc Ltd.

Authors: Jean-Michel Hiver, 
         Fergal Daly <fergal@esatclear.ie>,
	 and others.

This module free software and is distributed under the same license as Perl
itself. Use it at your own risk.

Thanks to everybody on the list who contributed to Petal in the form of
patches, bug reports and suggestions. See README for a list of contributors.


=head1 SEE ALSO

Join the Petal mailing list:

  http://lists.webarch.co.uk/mailman/listinfo/petal

Mailing list archives:

  http://lists.webarch.co.uk/pipermail/petal


Have a peek at the TAL / TALES / METAL specs:

  http://www.zope.org/Wikis/DevSite/Projects/ZPT/TAL
  http://www.zope.org/Wikis/DevSite/Projects/ZPT/TALES
  http://www.zope.org/Wikis/DevSite/Projects/ZPT/METAL