NAME

SQL::Translator::Producer::TT::Base - TT (Template Toolkit) based Producer base class.

SYNOPSIS

 # Create a producer using a template in the __DATA__ section.
 package SQL::Translator::Producer::Foo;

 use base qw/SQL::Translator::Producer::TT::Base/;

 # Convert produce call into a method call on our new class
 sub produce { return __PACKAGE__->new( translator => shift )->run; };

 # Configure the Template object.
 sub tt_config { ( INTERPOLATE => 1 ); }

 # Extra vars to add to the template
 sub tt_vars { ( foo => "bar" ); }

 # Put template in DATA section (or use file with ttfile producer arg)
 __DATA__
 Schema

 Database: [% schema.database %]
 Foo: $foo
 ...

DESCRIPTION

A base class producer designed to be sub-classed to create new TT based producers cheaply - by simply giving the template to use and sprinkling in some extra template variables and config.

You can find an introduction to this module in SQL::Translator::Manual.

The 1st thing the module does is convert the produce sub routine call we get from SQL::Translator into a method call on an object, which we can then sub-class. This is done with the following code which needs to appear in all sub classes.

 # Convert produce call into an object method call
 sub produce { return __PACKAGE__->new( translator => shift )->run; };

See "PRODUCER OBJECT" below for details.

The upshot of this is we can make new template producers by sub classing this base class, adding the above snippet and a template. The module also provides a number of hooks into the templating process, see "SUB CLASS HOOKS" for details.

See the "SYNOPSIS" above for an example of creating a simple producer using a single template stored in the producers DATA section.

SUB CLASS HOOKS

Sub-classes can override these methods to control the templating by giving the template source, adding variables and giving config to the Tempate object.

tt_config

 sub tt_config { ( INTERPOLATE => 1 ); }

Return hash of Template config to add to that given to the Template "new" method.

tt_schema

 sub tt_schema { "foo.tt"; }
 sub tt_schema { local $/ = undef; \<DATA>; }

The template to use, return a file name or a scalar ref of TT source, or an IO::Handle. See Template for details, as the return from this is passed on to it's "produce" method.

The default implementation uses the producer arg "ttfile" as a filename to read the template from. If the arg isn't there it will look for a "__DATA__" section in the class, reading it as template source if found. Returns undef if both these fail, causing the produce call to fail with a 'no template!' error.

tt_vars

 sub tt_vars { ( foo => "bar" ); }

Return hash of template vars to use in the template. Nothing added here by default, but see "tt_default_vars" for the variables you get for free.

tt_default_vars

Return a hash-ref of the default vars given to the template. You wouldn't normally over-ride this, just inherit the default implementation, to get the "translator" & "schema" variables, then over-ride "tt_vars" to add your own.

The current default variables are:

schema: The schema to template.
translator: The SQL::Translator object.

pre_process_schema

WARNING: This method is Experimental so may change!

Called with the SQL::Translator::Schema object and should return one (it doesn't have to be the same one) that will become the "schema" variable used in the template.

Gets called from tt_default_vars.

PRODUCER OBJECT

The rest of the methods in the class set up a sub-classable producer object. You normally just inherit them.

new

 my $tt_producer = TT::Base->new( translator => $translator );

Construct a new TT Producer object. Takes a single, named arg of the SQL::Translator object running the translation. Dies if this is not given.

translator

Return the SQL::Translator object.

schema

Return the SQL::Translator::Schema we are translating. This is equivalent to "$tt_producer->translator->schema".

run

Called to actually produce the output, calling the sub class hooks. Returns the produced text.

args

Util wrapper method around "TT::Base->translator->producer_args" for (mostly) readonly access to the producer args. How it works depends on the number of arguments you give it and the context.

 No args - Return hashref (the actual hash in Translator) or hash of args.
 1 arg   - Return value of the arg with the passed name.
 2+ args - List of names. In list context returns values of the given arg
           names, returns as a hashref in scalar context. Any names given
           that don't exist in the args are returned as undef.

This is still a bit messy but is a handy way to access the producer args when you use your own to drive the templating.

TODO

- Add support for a sqlf template repository, set as an INCLUDE_PATH, so that sub-classes can easily include file based templates using relative paths.

- Pass in template vars from the producer args and command line.

- Merge in TT::Table.

- Hooks to pre-process the schema and post-process the output.

AUTHOR

Mark Addison <grommit@users.sourceforge.net>.