| PGObject(3pm) | User Contributed Perl Documentation | PGObject(3pm) |
PGObject - A toolkit integrating intelligent PostgreSQL dbs into Perl objects
Version 2.0.2
To use without caching:
use PGObject;
To use with caching:
use PGObject ':cache';
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'}],
);
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.
This function clears the info cache if this was loaded with caching enabled.
The cache is also automatically cleared when a function that was run could not be found (this could be caused by updating the db).
Arguments:
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:
Arguments:
These are aggregates, each one has appended 'OVER (ROWS UNBOUNDED PRECEDING)' to it.
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.
Creates a new registry if it does not exist. This is useful when segments of an application must override existing type mappings.
This is deprecated and throws a warning.
Use PGObject::Type::Registry->new_registry($registry_name) instead.
This no longer returns anything of significance.
DEPRECATED
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.
Use PGObject::Type::Registry->register_type() instead.
Deprecated.
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.
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() or a to_db() method, 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
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.
Note that 2.0 moves the registry to a service module which handles both registry and deserialization of database types. This is intended to be both cleaner and more flexible than the embedded system in 1.x.
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.
Backwards Incompatibilities from 1.x
Deserialization occurs in a context which specifies a registry. In 1.x there were no concerns about default mappings but now this triggers a warning. The most basic and frequently used portions of this have been kept but return values for registering types has changed. We no longer provide a return variable but throw an exception if the type cannot be safely registered.
This follows a philosophy of throwing exceptions when guarantees cannot be met.
We now throw warnings when the default registry is used.
Longer-run, deserializers should use the PGObject::Type::Registry interface directly.
PGObject is intended to be the database-facing side of a framework for objects. The intended structure is for three tiers of logic:
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.
It is important to remember, when writing PGObject top half frameworks that the catalog lookups may be memoized and may come back as a data structure. This means that changes to the structures returned from get_function_info() in this module and similar functions in other catalog-bound modules may not be safe to modify in arbitrary ways. Therefore we recommend that the return values from catalog-lookup functions are treated as immutable.
Normalizing output is safe provided there are no conflicts between naming conventions. This is usually true since different naming conventions would interfere withmapping. However, there could be cases where it is not true, for example, where two different mapping modules agree on a subset of normalization conventions but differ on some details. The two might safely handle the same conventions but normalize differently resulting in conflicts of both were used.
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:
Chris Travers, "<chris.travers at gmail.com>"
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.
You can find documentation for this module with the perldoc command.
perldoc PGObject
You can also look for information at:
<http://rt.cpan.org/NoAuth/Bugs.html?Dist=PGObject>
<http://annocpan.org/dist/PGObject>
<http://cpanratings.perl.org/d/PGObject>
<http://search.cpan.org/dist/PGObject/>
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 significant 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.
COPYRIGHT (C) 2013-2014 Chris Travers COPYRIGHT (C) 2014-2017 The LedgerSMB Core Team
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.
| 2017-10-01 | perl v5.26.0 |