NAME

SQL::Translator::Parser - base object for parsers

DESCRIPTION

Parser modules that get invoked by SQL::Translator need to implement a single function: parse. This function will be called by the SQL::Translator instance as $class::parse($tr, $data_as_string), where $tr is a SQL::Translator instance. Other than that, the classes are free to define any helper functions, or use any design pattern internally that make the most sense.

FORMAT OF THE DATA STRUCTURE

The data structure returned from the parse function has a very particular format.

o

The data structure should be a reference to a hash, the keys of which are table names.

o

The values associated with each table should also be a reference to a hash. This hash should have several keys, enumerated below.

type

This is the type of the table, if applicable, as a string, or undef if not (for example, if the database does not have multiple options). For MySQL, this value might include MyISAM, HEAP, or similar.

indices

The indices keys is a reference to an array of hashrefs. Each hashref defines one index, and has the keys 'name' (if defined, it will be a string), 'type' (a string), and 'fields' (a reference to another array). For example, a table in a MySQL database with two indexes, created as:

PRIMARY KEY (id),
KEY foo_idx (foo),
KEY foo_bar_idx (foo, bar),

would be described in the indices element as:

[
  {
    'type' => 'primary_key',
    'fields' => [
                  'id'
                ],
    'name' => undef,
  },
  {
    'type' => 'normal',
    'fields' => [
                  'foo'
                ],
    'name' => 'foo_idx',
  },
  {
    'type' => 'normal',
    'fields' => [
                  'foo',
                  'bar',
                ],
    'name' => 'foo_bar_idx',
  },
]

fields

The fields element is a refernce to a hash; the keys of this hash are the row names from the table, and each value fills in this template:

{ 
  type           => 'field',
  order          => 1,      # the order in the original table
  name           => '',     # same as the key
  data_type      => '',     # in the db's jargon,
                            # i.e., MySQL => int, Oracale => INTEGER
  size           => '',     # int
  null           => 1 | 0,  # boolean
  default        => '',
  is_auto_inc    => 1 1 0,  # boolean
  is_primary_key => 1 | 0,  # boolean
} 

So a row defined as:

username CHAR(8) NOT NULL DEFAULT 'nobody',
KEY username_idx (username)

would be represented as:

'fields => {
  'username' => { 
    type           => 'field',
    order          => 1,
    name           => 'username',
    data_type      => 'char',
    size           => '8',
    null           => undef,
    default        => 'nobody',
    is_auto_inc    => undef,
    is_primary_key => undef,
  },
},
'indices' => [
  {
    'name' => 'username_idx',
    'fields' => [
                  'username'
                ],
    'type' => 'normal',
  },
],

AUTHORS

Ken Y. Clark, <kclark@cpan.org<gt>, darren chamberlain <darren@cpan.org>.

SEE ALSO

perl(1).