Dave Rolsky

NAME

Alzabo - A data modelling tool and RDBMS-OO mapper

SYNOPSIS

  Cannot be summarized here.

DESCRIPTION

What is Alzabo?

Alzabo is a program and a module, with two core functions. Its first use is as a data modelling tool. Through either a schema creation interface or a perl program, you can create a set of schema, table, column, etc. objects to represent your data model. Alzabo is also capable of reverse engineering your data model from an existing system.

Its second function is as an RDBMS to object mapping system. Once you have created a schema, you can use the Alzabo::Runtime::Table and Alzabo::Runtime::Row classes to access its data. These classes offer a high level interface to common operations such as SQL SELECT, INSERT, DELETE, and UPDATE commands.

A higher level interface can be created through the use of the Alzabo::MethodMaker module. This module takes a schema object and auto-generates useful methods based on the tables, columns, and relationships it finds in the module. The code is generates can be integrated with your own code quite easily.

To take it a step further, you could then aggregate a set of rows from different tables into a larger container object which could understand the logical relationship between these tables.

This is not yet intended to be a total object replacement for SQL, as there is not yet complete suppot for more operations such as joins or arbitrary SQL functions such as AVG or MAX, though that is coming. However, it can still replace the drudgery of the most common operations. In addition, the Alzabo::Runtime::Row objects support the use of a caching system. Two caching modules, Alzabo::ObjectCache and Alzabo::ObjectCacheIPC, are included. However, you may substitute any caching system you like provided it has the appropriate method interface.

What to Read?

Alzabo has a lot of documentation. If you are primarily interested in using Alzabo as an RDBMS-OO wrapper, much of the documentation can be skipped. This assumes that you will create your schema via the schema creation interface or via reverse engineering.

Here is the suggested reading order:

Alzabo - Alzabo concepts

Alzabo - Rows and cursors

Alzabo - How to use Alzabo

Alzabo - Exceptions

Alzabo - Usage Examples

The section for your RDBMS:

Alzabo::Runtime::Schema - The most important parts here are those related to loading a schema and connecting to a database. Also be sure to read about the join method.

Alzabo::Runtime::Table - This contains most of the methods used to fetch rows from the database, as well as the insert method.

Alzabo::Runtime::Row - The row objects are how data is updated, deleted, and retrieved from the database. Its also important to read the section on its import, as this is used to determine how caching is done (for now).

Alzabo::Runtime::Cursor - The most important part of the documentation here is the HANDLING ERRORS section.

Alzabo::Runtime::RowCursor - A cursor object that returns only a single row.

Alzabo::Runtime::JoinCursor - A cursor object that returns multiple rows.

Alzabo::MethodMaker - One of the most useful parts of Alzabo. This module can be used to auto-generate methods based on the structure of your schema.

Alzabo::ObjectCache - This describes how to select the caching modules you want to use. It contains a number of scenarios and descibes how they are affected by caching. If you plan on using Alzabo in a multi-process environment (such as mod_perl) this is very important.

Alzabo::Exceptions - Describes the nature of all the exceptions used in Alzabo.

Alzabo::FAQ

Alzabo::QuickRef - A quick reference for the various methods of the Alzabo objects.

Alzabo concepts

Instantiation

Every schema keeps track of whether it has been instantiated or not. A schema that is instantiated is one that exists in an RDBMS backend. This can be done explicitly by calling the schema's create method. It is also implicitly set when a schema is created as the result of reverse engineering.

Instantiation has several effects. The most important part of this is to realize that once a schema is instantiated, the way it generates SQL for itself changes. Before it is instantiated, if you ask it to generate SQL via Alzabo::Create::Schema->make_sql, it will generate the set SQL statements that are needed to create the schema in the RDBMS.

After is instantiated, the schema will instead generate the SQL necessary to convert the version in the RDBMS backend to match the object's current state. This can be though of as a SQL 'diff'.

While this feature is quite useful, it can be confusing too. The most surprising aspect of this is that if you create a schema via reverse engineering and then call Alzabo::Create::Schema->make_sql, you will not get any SQL. This is because the schema knows that it is instantiated and it also knows that it is the same as the version in the RDBMS, so no SQL is necessary.

The way to deal with this is to call the Alzabo::Create::Schema->set_instantiated method with a false value. Use this method with care.

Rows and cursors

In Alzabo, data is returned in the form of a row object. This object can be used to access the data for an individual row.

Unless you are retrieving a row via a unique identifier (usually its primary key), you will be given a cursor object. This is quite similar to how DBI uses statement handles and is done for similar reasons.

How to use Alzabo

The first thing you'll want to do is create a schema. The easiest way to do this is via the included web based schema creation interface, which requires the HTML::Mason package from CPAN (www.cpan.org) to run.

This interface can be installed during the normal installation process, and you will be prompted as to whether or not you want to use it.

The other way to create a schema is via a perl script. Here's the beginning of such a script:

  use Alzabo::Create::Schema;

  eval
  {
      my $s = Alzabo::Create::Schema->new( name => 'foo',
                                           rdbms => 'MySQL' );

      my $table = $s->make_table( name => 'some_table' );

      my $a_col = $table->make_column( name => 'a_column',
                                       type => 'int',
                                       nullable => 0,
                                       sequenced => 0,
                                       attributes => [ 'unsigned' ] );

      $table->add_primary_key($a_col);

      my $b_col = $table->make_column( name => 'b_column',
                                       type => 'varchar',
                                       length => 240,
                                       nullable => 0 );

      $table->make_index( columns => [ column => $b_col,
                                       prefix => 10 ] );

      ...

      $s->save_to_file;
  };

  if ($@) { handle exceptions }

Exceptions

Alzabo uses exceptions as its error reporting mechanism. This means that pretty much all calls to its methods should be wrapped in eval{}. This is less onerous than it sounds. In general, there's no reason not to wrap all of your calls in one eval, rather than each one in a seperate eval. Then at the end of the block simply check the value of $@. See the code of the included HTML::Mason based interface for examples.

Also see the Alzabo::Exceptions documentation, which lists all of the different exception used by Alzabo.

Usage Examples

Alzabo is a powerful tool but as with many powerful tools it can also be a bit overwhelming at first. The easiest way to understand some of its basic capabilities is through some examples.. Let's first assume that you've created the following schema:

  TABLE: Movie
  movie_id                 tinyint      -- primary key
  title                    varchar(200)
  release_year             year

  TABLE: Person
  person_id                tinyint      -- primary key
  name                     varchar(200)
  birthdate                date
  birthplace_location_id   tinyint      -- foreign key to location

  TABLE: Job
  job_id                   tinyint      -- primary key
  job                      varchar(200) -- something like 'actor' or 'director'

  TABLE: Credit
  movie_id                 tinyint      -- primary key part 1, foreign key to movie
  person_id                tinyint      -- primary key part 2, foreign key to person
  job_id                   tinyint      -- primary key part 3, foreign key to job

  TABLE: Location
  location_id              tinyint      -- primary key
  location                 varchar(200) -- 'New York City' or 'USA'
  parent_location_id       tinyint      -- foreign key to location

This is a vastly scaled down version of the 90+ table database that Alzabo was written to support.

Fetching data

First of all, let's do something simple. Let's assume I have a person_id value and I want to find all the movies that they were in and print the title, year of release, and the job they did in the movie. Here's what it looks like:

  my $schema = Alzabo::Runtime::Schema->load_from_file( name => 'movies' );

  my $person_t = $schema->table('Person');
  my $credit_t = $schema->table('Credit');
  my $movie_t  = $schema->table('Movie');
  my $job_t    = $schema->table('Job');

  # returns a row representing this person.
  my $person = $person_t->row_by_pk( pk => 42 );

  # all the rows in the credit table that have the person_id of 42.
  my $cursor = $person->rows_by_foreign_key( fk => $person_t->foreign_keys_by_table($credit_t) );

  print $person->select('name'), " was in the following films:\n\n";
  while (my $credit = $cursor->next_row)
  {
      # rows_by_foreign_key returns a RowCursor object.  We immediately
      # call its next_row method, knowing it will only have one row (if
      # it doesn't then our referential integrity is in trouble!)
      my $movie =
          $credit->rows_by_foreign_key( fk => $credit_t->foreign_keys_by_table($movie_t) )->next_row;

      my $job =
          $credit->rows_by_foreign_key( fk => $credit_t->foreign_keys_by_table($job_t) )->next_row;

      print $movie->select('title'), " released in ", $movie->select('release_year'), "\n";
      print '  ', $job->('job'), "\n";
  }

A more sophisticated version of this code would take into account that a person can do more than one job in the same movie.

The method names are admittedly verbose but the end result code is significantly simpler to read than the equivalent using raw SQL and DBI calls.

Let's redo the example to use Alzabo::MethodMaker;

  # I'm assuming that the pluralize_english subroutine pluralizes
  # things as one would expect.
  use Alzabo::MethodMaker( schema      => 'movies',
                           all         => 1,
                           name_maker  => \&method_namer );

  my $schema = Alzabo::Runtime::Schema->load_from_file( name => 'movies' );

  # instantiates a row representing this person.
  my $person = $schema->Person->row_by_pk( pk => 42 );

  # all the rows in the credit table that have the person_id of 42.
  my $cursor = $person->Credits;

  print $person->name, " was in the following films:\n\n";
  while (my $credit = $cursor->next_row)
  {
      my $movie = $credit->Movie;

      my $job = $credit->Job;

      print $movie->title, " released in ", $movie->release_year, "\n";
      print '  ', $job->job, "\n";
  }

Validating data

Let's assume that we've been passed a hash of values representing an update to the location table. Here's a way of making sure that that this update won't lead to a loop in terms of the parent/child relationships.

  sub update_location
  {
      my $self = shift; # this is the row object
      my %data = @_;
      if ( $data{parent_location_id} )
      {
          my $parent_location_id = $data{parent_location_id};
          my $location_t = $schema->table('Location');
          while ( my $location = eval { $location_t->row_by_pk( pk => $parent_location_id ) } )
          {
              die "Insert into location would create loop"
                  if $location->select('parent_location_id') == $data{location_id};

              $parent_location_id = $location->select('parent_location_id');
          }
      }
  }

Once again, let's rewrite the code to use Alzabo::MethodMaker:

  sub update_location
  {
      my $self = shift; # this is the row object
      my %data = @_;
      if ( $data{parent_location_id} )
      {
          my $location = $self;
          while ( my $location = eval { $location->parent } )
          {
              die "Insert into location would create loop"
                  if $location->parent_location_id == $data{location_id};
          }
      }
  }

Changing the schema

In MySQL, there are a number of various types of integers. The type TINYINT can hold values from -128 to 127. But what if have more than 127 movies? And if that's the case we might have more than 127 people too.

For safety's sake, it might be best to make all of the primary key integer columns INT columns instead. And while we're at it we want to make them UNSIGNED as well, as we don't need to insert negative numbers into these columns.

You could break out the RDBMS manual (because you probably forgot the exact ALTER TABLE syntax you'll need). Or you could use Alzabo. Note that this time it is an Alzabo::Create::Schema object, not Alzabo::Runtime::Schema.

  my $schema = Alzabo::Create::Schema->load_from_file( name => 'movies' );

  foreach my $t ( $schema->tables )
  {
      foreach my $c ( $t->columns )
      {
           if ( $c->is_primary_key and lc $c->type eq 'tinyint' )
           {
                $c->set_type('int');
                $c->add_attribute('unsigned');
           }
      }
  }
  $schema->create( user => 'user', password => 'password' );
  $schema->save_to_file;

Multiple RDBMS Support

Alzabo aims to be as cross-platform as possible. To that end, RDBMS specific operations are contained in several module hierarchies.

The first, the Alzabo::Driver::* hierarchy, is used to handle communication with the database. It uses DBI and the appropriate DBD::* module to handle communications. It provides a higher level of abstraction than DBI, requiring that the RDBMS specific modules implement methods to do such things as create databases or return the next value in a sequence.

The second, the Alzabo::RDBMSRules::* hierarchy, is used during schema creation in order to validate user input such as schema and table names. It also generates SQL to create the database or turn one schema into another (sort of a SQL diff). Finally, it also handles reverse engineering an existing database.

The this, the Alzabo::SQLMaker::* hierarchy, is used to generate SQL and handle bound parameters for select, insert, update, and delete operations.

The RDBMS to be used is specified when creating the schema. Currently, there is no easy way to convert a schema from one RDBMS to another, though this is a future goal.

MySQL

Alzabo does not provide support for all possible MySQL features. One notable feature that is supported in column prefixes in indexes, which are necessary to allow indexes on blob/text columns.

PostgreSQL

Postgres support in Alzabo is currently missing several features.

First, reverse engineering does not handle constraints (including foreign keys). This will change in the future.

Second, reverse engineering cannot determine from the existence of a sequence that a sequence is meant to be used for a particular column unless the sequence was created as a result of assigning the serial type to a column.

Third, there is no support for large objects. This was considered but given that 7.1 should support rows larger than 8K it was determined that supporting large objects was not worth the amount of effort reqiured.

Architecture

The general design of Alzabo is as follows.

There are objects representing the schema, which contains table objects. Table objects contain column, foreign key, and index objects. Column objects contain column definition objects. A single column definition may be shared by multiple columns, but has only one owner.

This is a diagram of these inheritance relationships:

  Alzabo::* (::Schema, ::Table, ::Column, ::ColumnDefinition, ::ForeignKey, ::Index)
                   /   \
                is parent to
                 /       \
 Alzabo::Create::*   Alzabo::Runtime::*

This a diagram of how objects contain other objects:

                      Schema
                     /      \
              contains       contains--Alzabo::Driver subclass object (1)
                  |                 \
               Table (0 or more)     Alzabo::RDBMSRules subclass object (1)
                /  \                  (* Alzabo::Create::Schema only)
               /    \
              contains--------------------
             /        \                   \
            /          \                   \
     ForeignKey      Column (0 or more)    Index (0 or more)
     (0 or more)       |
                    contains
                       |
                  ColumnDefinition (1)

Note that more than one column _may_ share a single definition object (this is explained in the Alzabo::Create::ColumnDefinition documentation). This is only relevant if you are writing a schema creation interface.

Other classes/objects used in Alzabo include:

  • Alzabo::Config

    This class is generated by Makefile.PL during installation and contains information such as what directory contains saved schemas and other configuration information.

  • Alzabo::ChangeTracker

    This object provides a method for an object to register a series to backout from multiple changes. This is done by providing the ChangeTracker object with a callback after a change is succesfully made to an object or objects. If a future change in a set of operations fail, the tracker can be told to back the changes out. This is used primarily in Alzabo::Create::Schema.

  • Alzabo::MethodMaker

    This module can auto-generate useful methods for you schema, table, and row objects based on the structure of your schema.

  • Alzabo::Exceptions

    This object creates the exception subclasses used by Alzabo.

  • Alzabo::ObjectCache and Alzabo::ObjectCacheIPC

    These are object caching modules. The latter uses IPC to make sure that cached objects expired in one process get expired in any other process using the same caching module. This can be quite useful when running under mod_perl, for example.

  • Alzabo::Runtime::Row

    This object represents a row from a table. These objects are created by Alzabo::Runtime::Table, Alzabo::Runtime::RowCursor, and Alzabo::Runtime::JoinCursor objects. It is the sole interface by which actual data is retrieved, updated, or deleted in a table.

  • Alzabo::Runtime::JoinCursor and Alzabo::Runtime::RowCursor

    This object is a cursor that returns row objects. Using a cursor saves a lot of memory for big selects.

  • Alzabo::Util

    Contains simple subroutines shared by various modules.

Why the subdivision between Alzabo::*, Alzabo::Create::*, and Alzabo::Runtime::*?

There are several reasons for doing this:

  • In some environments (mod_perl) we would like to optimize for memory. For an application that uses an existing schema, all we need is to be able read object information, rather than needing to change the schema's definition. This means there is no reason to have the overhead of compiling all the methods used when creating and modifying objects.

  • In other environments (for example, when running as a separately spawned CGI process) compile time is important.

  • For most people using Alzabo, they will use one of the existing schema creation interfaces and then write an application using that schema. At the simplest level, they would only need to learn how to instantiate Alzabo::Runtime::Row objects and how that class's methods work. For more sophisticated users, they can still avoid having to ever look at documentation on methods that alter the schema and its contained objects.

AUTHOR

Dave Rolsky, <autarch@urth.org>