Chris Travers > PGObject-1.400.1 > PGObject

Download:
PGObject-1.400.1.tar.gz

Dependencies

Annotate this POD

View/Report Bugs
Module Version: 1.400.1   Source  

NAME ^

PGObject - A toolkit integrating intelligent PostgreSQL dbs into Perl objects

VERSION ^

Version 1.4.1

SYNPOSIS ^

To get basic info from a function

  my $f_info = PGObject->function_info(
      dbh        =>  $dbh,
      funcname   =>  $funcname,
      funcschema =>  'public',
  );

To get info about a function, filtered by first argument type

  my $f_info = PGObject->function_info(
      dbh        =>  $dbh,
      funcname   =>  $funcname,
      funcschema =>  'public',
      funcprefix =>  'test__',
      objtype    =>  'invoice',
      objschema  =>  'public',
  );

To call a function with enumerated arguments

  my @results = PGObject->call_procedure(
      dbh          =>  $dbh,
      funcname     => $funcname,
      funcprefix =>  'test__',
      funcschema   => $funcname,
      args         => [$arg1, $arg2, $arg3],
  );

To do the same with a running total

  my @results = PGObject->call_procedure(
      dbh           =>  $dbh,
      funcname      => $funcname,
      funcschema    => $funcname,
      args          => [$arg1, $arg2, $arg3],
      running_funcs => [{agg => 'sum(amount)', alias => 'running_total'}],
  );

DESCRIPTION ^

PGObject contains the base routines for object management using discoverable stored procedures in PostgreSQL databases. This module contains only common functionality and support structures, and low-level API's. Most developers will want to use more functional modules which add to these functions.

The overall approach here is to provide the basics for a toolkit that other modules can extend. This is thus intended to be a component for building integration between PostgreSQL user defined functions and Perl objects.

Because decisions such as state handling are largely outside of the scope of this module, this module itself does not do any significant state handling. Database handles (using DBD::Pg 2.0 or later) must be passed in on every call. This decision was made in order to allow for diversity in this area, with the idea that wrapper classes would be written to implement this.

FUNCTIONS ^

function_info(%args)

Arguments:

dbh (required)

Database handle

funcname (required)

function name

funcschema (optional, default 'public')

function schema

funcprefix (optiona, default '')

Prefix for the function. This can be useful for separating functions by class.

argtype1 (optional)

Name of first argument type. If not provided, does not filter on this criteria.

argschema (optional)

Name of first argument type's schema. If not provided defaults to 'public'

This function looks up basic mapping information for a function. If more than one function is found, an exception is raised. This function is primarily intended to be used by packages which extend this one, in order to accomplish stored procedure to object mapping.

Return data is a hashref containing the following elements:

args

This is an arrayref of hashrefs, each of which contains 'name' and 'type'

name

The name of the function

num_args

The number of arguments

call_procedure(%args)

Arguments:

funcname

The function name

funcschema

The schema in which the function resides

funcprefix (optiona, default '')

Prefix for the function. This can be useful for separating functions by class.

args

This is an arrayref. Each item is either a literal value, an arrayref, or a hashref of extended information. In the hashref case, the type key specifies the string to use to cast the type in, and value is the value.

orderby

The list (arrayref) of columns on output for ordering.

running_funcs

An arrayref of running windowed aggregates. Each contains two keys, namely 'agg' for the aggregate and 'alias' for the function name.

These are aggregates, each one has appended 'OVER (ROWS UNBOUNDED PRECEDING)' to it.

registry

This is the name of the registry used for type conversion. It can be omitted and defaults to 'default.' Note that use of a non-standard registry currently does *not* merge changes from the default registry, so you need to reregister types in non-default registries when you create them.

Please note, these aggregates are not intended to be user-supplied. Please only allow whitelisted values here or construct in a tested framework elsewhere. Because of the syntax here, there is no sql injection prevention possible at the framework level for this parameter.

process_type($val, $type, $registry)

If $type is registered, returns "$type"->from_db($val). Otherwise returns $val. If $val is an arrayref, loops through it for every item and strips trialing [] from $type.

This module should generally only be used by type handlers or by this module.

new_registry($registry_name)

Creates a new registry if it does not exist. This is useful when segments of an application must override existing type mappings.

Returns 1 on creation, 2 if already exists.

register_type(pgtype => $tname, registry => $regname, perl_class => $pm)

Registers a type as a class. This means that when an attribute of type $pg_type is returned, that PGObject will automatically return whatever $perl_class->from_db returns. This allows you to have a db-specific constructor for such types.

The registry argument is optional and defaults to 'default'

If the registry does not exist, an error is raised. if the pg_type is already registered to a different type, this returns 0. Returns 1 on success.

get_registered(registry => $registry, pg_type => $pg_type)

This is a public interface to the registry, which can be useful for composite types decoding themselves from tuple data, and so forth.

unregister_type(pgtype => $tname, registry => $regname)

Tries to unregister the type. If the type does not exist, returns 0, otherwise returns 1. This is mostly useful for when a specific type must make sure it has the slot. This is rarely desirable. It is usually better to use a subregistry instead.

$hashref = get_type_registry()

Returns the type registry. Mostly useful for debugging.

WRITING PGOBJECT-AWARE HELPER CLASSES ^

One of the powerful features of PGObject is the ability to declare methods in types which can be dynamically detected and used to serialize data for query purposes. Objects which contain a pgobject_to_db(), that method will be called and the return value used in place of the object. This can allow arbitrary types to serialize themselves in arbitrary ways.

For example a date object could be set up with such a method which would export a string in yyyy-mm-dd format. An object could look up its own definition and return something like :

   { cast => 'dbtypename', value => '("A","List","Of","Properties")'}

If a scalar is returned that is used as the serialized value. If a hashref is returned, it must follow the type format:

  type  => variable binding type,
  cast  => db cast type
  value => literal representation of type, as intelligible by DBD::Pg

REQUIRED INTERFACES

Registered types MUST implement a $class->from_db function accepts the string from the database as its only argument, and returns the object of the desired type.

Any type MAY present an $object->to_db() interface, requiring no arguments, and returning a valid value. These can be hashrefs as specified above, arrayrefs (converted to PostgreSQL arrays by DBD::Pg) or scalar text values.

UNDERSTANDING THE REGISTRY SYSTEM

The registry system allows Perl classes to "claim" PostgreSQL types within a certain domain. For example, if I want to ensure that all numeric types are turned into Math::BigFloat objects, I can build a wrapper class with appropriate interfaces, but PGObject won't know to convert numeric types to this new class, so this is what registration is for.

By default, these mappings are fully global. Once a class claims a type, unless another type goes through the trouble of unregisterign the first type and making sure it gets the authoritative spot, all items of that type get turned into the appropriate Perl object types. While this is sufficient for the vast number of applications, however, there may be cases where names conflict across schemas or the like. To address this application components may create their own registries. Each registry is fully global, but application components can specify non-standard registries when calling procedures, and PGObject will use only those components registered on the non-standard registry when checking rows before output.

WRITING TOP-HALF OBJECT FRAMEWORKS FOR PGOBJECT ^

PGObject is intended to be the database-facing side of a framework for objects. The intended structure is for three tiers of logic:

Database facing, low-level API's
Object management modules
Application handlers with things like database connection management.

By top half, we are referring to the second tier. The third tier exists in the client application.

The PGObject module provides only low-level API's in that first tier. The job of this module is to provide database function information to the upper level modules.

We do not supply type information, If your top-level module needs this, please check out https://code.google.com/p/typeutils/ which could then be used via our function mapping APIs here.

A BRIEF GUIDE TO THE NAMESPACE LAYOUT ^

Most names underneath PGObject can be assumed to be top-half modules and modules under those can be generally assumed to be variants on those. There are, however, a few reserved names:

::Debug is reserved for debugging information. For example, functions which retrieve sources of functions, or grab diagnostics, or the like would go here.
::Test is reserved for test framework extensions applible only here
::Type is reserved for PG aware type classes.

For example, one might have PGObject::Type::BigFloat for a Math::Bigfloat wrapper, or PGObject::Type::DateTime for a DateTime wrapper.

::Util is reserved for utility functions and classes.

AUTHOR ^

Chris Travers, <chris.travers at gmail.com>

BUGS ^

Please report any bugs or feature requests to bug-pgobject at rt.cpan.org, or through the web interface at http://rt.cpan.org/NoAuth/ReportBug.html?Queue=PGObject. I will be notified, and then you'll automatically be notified of progress on your bug as I make changes.

SUPPORT ^

You can find documentation for this module with the perldoc command.

    perldoc PGObject

You can also look for information at:

ACKNOWLEDGEMENTS ^

This code has been loosely based on code written for the LedgerSMB open source accounting and ERP project. While that software uses the GNU GPL v2 or later, this is my own reimplementation, based on my original contributions to that project alone, and it differs in signficant ways. This being said, without LedgerSMB, this module wouldn't exist, and without the lessons learned there, and the great people who have helped make this possible, this framework would not be half of what it is today.

SEE ALSO ^

PGObject::Simple - Simple mapping of object properties to stored proc args
PGObject::Simple::Role - Moose-enabled wrapper for PGObject::Simple

COPYRIGHT ^

COPYRIGHT (C) 2013 Chris Travers

Redistribution and use in source and compiled forms with or without modification, are permitted provided that the following conditions are met:

THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

syntax highlighting: