ora2pg

Langue: en

Autres versions - même langue

Version: 2008-10-27 (ubuntu - 07/07/09)

Section: 1 (Commandes utilisateur)

NAME

Ora2Pg - Oracle to PostgreSQL database schema converter

SYNOPSIS

         Ora2pg as a companion script called ora2pg.pl when use in
         conjonction with a custom version of ora2pg.conf do what
         I'm trying to explain bellow :-) See content of the ora2pg.conf
         file for more explanation on configuration directives.
 
         BEGIN {
                 $ENV{ORACLE_HOME} = '/usr/local/oracle/oracle816';
         }
 
         use strict;
 
         use Ora2Pg;
 
         # Init the database connection
         my $dbsrc = 'dbi:Oracle:host=testdb.mydom.fr;sid=TEST;port=1521';
         my $dbuser = 'system';
         my $dbpwd = 'manager';
 
         # Create an instance of the Ora2Pg perl module
         my $schema = new Ora2Pg (
                 datasource => $dbsrc,   # Database DBD datasource
                 user => $dbuser,        # Database user
                 password => $dbpwd,     # Database password
         );
 
         # Create the PostgreSQL representation of all tables
         $schema->export_schema("output.sql");
 
         exit(0);
 
 

or if you only want to extract some tables:

         # Create an instance of the Ora2Pg perl module
         my @tables = ('t1', 't2', 't3');
         my $schema = new Ora2Pg (
                 datasource => $dbsrc,   # Database DBD datasource
                 user => $dbuser,        # Database user
                 password => $dbpwd,     # Database password
                 tables => \@tables,
         or                              
                 tables => [('tab1','tab2')],  # Tables to extract
                 debug => 1                    # Verbose running.
         );
 
 

or if you only want to extract the 10 first tables:

         # Create an instance of the Ora2Pg perl module
         my $schema = new Ora2Pg (
                 datasource => $dbsrc,   # Database DBD datasource
                 user => $dbuser,        # Database user
                 password => $dbpwd,     # Database password
                 max => 10               # 10 first tables to extract
         );
 
 

or if you only want to extract tables 10 to 20:

         # Create an instance of the Ora2Pg perl module
         my $schema = new Ora2Pg (
                 datasource => $dbsrc,   # Database DBD datasource
                 user => $dbuser,        # Database user
                 password => $dbpwd,     # Database password
                 min => 10,              # Begin extraction at indice 10
                 max => 20               # End extraction at indice 20
         );
 
 

Setting showtableid to 1 will display a table and their indices without any extraction. This will help you to set values of min/max options.

To choose a particular Oracle schema to export just set the following option to your schema name:

         schema => 'APPS'
 
 

This schema definition can also be needed when you want to export data. If export failed and complain that the table doesn't exists use this to prefix the table name by the schema name.

If you want to use PostgreSQL 7.4 schema support activate the init option 'export_schema' set to 1. Default is no schema export.

To know at which indices tables can be found during extraction use the option:

         showtableid => 1
 
 

You can now process multiple types of extraction at the samse time by setting the value to a space separated liste of the following keywords.

To extract all views set the type option as follow:

         type => 'VIEW'
 
 

To extract all grants set the type option as follow:

         type => 'GRANT'
 
 

To extract all sequences set the type option as follow:

         type => 'SEQUENCE'
 
 

To extract all triggers set the type option as follow:

         type => 'TRIGGER'
 
 

To extract all functions set the type option as follow:

         type => 'FUNCTION'
 
 

To extract all procedures set the type option as follow:

         type => 'PROCEDURE'
 
 

To extract all packages and body set the type option as follow:

         type => 'PACKAGE'
 
 

Default is table extraction

         type => 'TABLE'
 
 

To extract tablespace (PostgreSQL >= v8):

         type => 'TABLESPACE'
 
 

To extract all data from table extraction as INSERT statement use:

         type => 'DATA'
 
 

To extract all data from table extraction as COPY statement use:

         type => 'COPY'
 
 

and data_limit => n to specify the max tuples to return. If you set this options to 0 or nothing, no limitation are used. Additional option 'table', 'min' and 'max' can also be used. This is usefull only when data are send to Pg backend directly, not when when dumping to file.

When use of COPY or DATA you MUST export data by calling method:

$schema->export_data(``output.sql'');

Data are dumped to the given filename or to stdout with no argument. You can also send these datas directly to a PostgreSQL backend using the following method:

$schema->send_to_pgdb($destdatasrc,$destuser,$destpasswd);

In this case you must call export_data() without argument after the call to method send_to_pgdb().

If you set type to COPY and you want to dump data directly to a PG database, you must call method send_to_pgdb but data will not be sent via DBD::Pg but they will be load to the database using the psql command. Calling this method is still required to be able to extract database name, hostname and port information. Edit the $PSQL variable to match the path of your psql command (nothing to edit if psql is in your path).

When copying tables, Ora2Pg normally exports constraints as they are; if they are non-deferrable they are exported as non-deferrable. However, non-deferrable constraints will probably cause problems when attempting to import data to PostgreSQL. The option

        fkey_deferrable => 1
 
 

will cause all foreign key constraints to be exported as deferrable, even if they are non-deferrable. In addition,

        defer_fkey => 1
 
 

when exporting data will add a command to actually defer all constraints before importing data.

To non perl gurus, you can use the configuration file and run ora2pg.pl. You will find all information into the ora2pg.conf to be able to set it correctly.

DESCRIPTION

Ora2Pg is a perl OO module used to export an Oracle database schema to a PostgreSQL compatible schema.

It simply connect to your Oracle database, extract its structure and generate a SQL script that you can load into your PostgreSQL database.

I'm not a Oracle DBA so I don't really know something about its internal structure so you may find some incorrect things. Please tell me what is wrong and what can be better.

It currently dump the database schema (tables, views, sequences, indexes, grants), with primary, unique and foreign keys into PostgreSQL syntax without editing the SQL code generated.

It now can dump Oracle data into PostgreSQL DB as online process. You can choose what columns can be exported for each table.

Functions, procedures and triggers PL/SQL code generated must be reviewed to match the PostgreSQL syntax. Some usefull recommandation on porting Oracle to PostgreSQL can be found at ``http://techdocs.postgresql.org/'' under the ``Converting from other Databases to PostgreSQL'' Oracle part. I just notice one thing more is that the trunc() function in Oracle is the same for number or date so be carefull when porting to PostgreSQL to use trunc() for number and date_trunc() for date.

ABSTRACT

The goal of the Ora2Pg perl module is to cover all part needed to export an Oracle database to a PostgreSQL database without other thing that provide the connection parameters to the Oracle database.

Features include:

     - Database schema export (tables, views, sequences, indexes),
       with unique, primary and foreign key, check constraint.
     - Grants/privileges export by user and group.
     - Table selection (by name and max table) export.
     - Export Oracle schema to PostgreSQL 7.3+ schema.
     - Predefined functions/triggers/procedures/packages export.
     - Data export.
     - Sql query converter (todo)
 
 

See ora2pg.conf for more information on use.

My knowledge regarding database is really poor especially for Oracle so contribution is welcome.

REQUIREMENT

You just need the DBI, DBD::Pg and DBD::Oracle perl module to be installed

PUBLIC METHODS

new HASH_OPTIONS

Creates a new Ora2Pg object.

Supported options are (See ora2pg.conf for more details):

     - datasource : DBD datasource (required)
     - user : DBD user (optional with public access)
     - password : DBD password (optional with public access)
     - schema : Oracle internal schema to extract
     - type : Type of data to extract, can be TABLE,VIEW,GRANT,SEQUENCE,
       TRIGGER,FUNCTION,PROCEDURE,DATA,COPY,PACKAGE,TABLESPACE
       or a combinaison of these keywords.
     - debug : Print the current state of the parsing
     - export_schema     : Export Oracle schema to PostgreSQL 7.3 schema
     - tables : Extract only the given tables (arrayref) and set the
       extracting order
     - exclude : Exclude the given tables from extract (arrayref)
     - showtableid : Display only the table indice during extraction
     - min : Indice to begin extraction. Default to 0
     - max : Indice to end extraction. Default to 0 mean no limits
     - data_limit : Number max of tuples to return during data extraction
       (default 0 no limit)
     - case_sensitive: Allow to preserve Oracle object name as they are
       written. Default is not.
     - skip_fkeys : Skip foreign key constraints extraction. Default to 0
       (extraction)
     - skip_pkeys : Skip primary keys extraction. Default to 0 (extraction)
     - skip_ukeys : Skip unique column constraints extraction. Default to 0
       (extraction)
     - skip_indices : Skip all other index types extraction. Default to 0
       (extraction)
     - skip_checks : Skip checks constraints extraction. Default to 0
       (extraction)
     - keep_pkey_names : By default, primary key names in the source database
       are ignored, and default key names are created in the target database.
       If this is set to true, primary key names are kept.
     - bzip2     : Path to the Bzip2 program to compress data export. Default
       /usr/bin/bzip2
     - gen_user_pwd : When set to 1 this will replace default password
       'change_my_secret' with a random string. Need String::Random
     - fkey_deferrable: Force foreign key constraints to be exported as
       deferrable. Default 0: asis.
     - defer_fkey : Force all foreign key constraints to be deferred during
       data import. Default 0: asis.
     - pg_numeric_type: Convert Oracle NUMBER data type to internal PG data
       type instead of use of the slow numeric(p,s) data type.
     - default_numeric: NUMBER(x) without precision are by default converted
       to float. You can overwrite this value by any PG type.
     - keep_pkey_names: preserve oracle primary keys name. Default ignored.
     - pg_supports_inout: Allow PG support of in/ou/inout function parameter
       Must be used with PostgreSQL > 8.1. Default no support (backward
       compatibility).
     - pg_supports_role: Allow PG support of roles instead of user/group.
       Default no support (backward compatibility).
     - disable_triggers: Disables triggers on all tables in COPY or
       DATA mode.
     - disable_sequence: Disables alter sequence on all tables in COPY or
       DATA mode.
     - noescape: disable character escaping during data export.
     - datatype: allow to redefine datatype Oracle to PostgreSQl conversion.
     - binmode: force perl to use a given binary mode for output. Default is
       ':raw';
     - sysusers: allow to add other system users to the default exclusion
       list (SYS,SYSTEM,DBSNMP,OUTLN,PERFSTAT).
 
 

Attempt that this list should grow a little more because all initialization is done by this way.

Specials configuration options to handle character encoding: ------------------------------------------------------------

NLS_LANG

If you experience any issue where mutibyte characters are being substituted with replacement characters during export try to set the NLS_LANG configutation directive to the Oracle encoding. This may help a lot especially with UTF8 encoding.

BINMODE

If you experience the perl message: ``Wide character in print'', this warning happens when you output a Unicode string to a non-unicode filehandle. You can force perl to use binary mode for output by setting the BINMODE configuration option to a given encoding. If you set it to 'utf8', it will force printing like this: binmode OUTFH, ``:utf8''; By default Ora2Pg open the output file in 'raw' binary mode.

export_data FILENAME

Print SQL data output to a filename or to STDOUT if no file is given.

Must be used only if type option is set to DATA or COPY

export_sql FILENAME

Print SQL conversion output to a filename or simply return these data if no file is given.

send_to_pgdb DEST_DATASRC DEST_USER DEST_PASSWD

Open a DB handle to a PostgreSQL database

modify_struct TABLE_NAME ARRAYOF_FIELDNAME

Modify a table structure during export. Only given fieldname will be exported.

replace_tables HASH

Modify tables table names during export.

replace_cols HASH

Modify column names during export.

set_where_clause HASH

Add WHERE clause during data export on specifics table and/or all table

PRIVATE METHODS

_init HASH_OPTIONS

Initialize a Ora2Pg object instance with a connexion to the Oracle database.

_grants

This function is used to retrieve all privilege information.

It extract all Oracle's ROLES to convert them as Postgres groups and search all users associated to these roles.

Set the main hash $self->{groups}. Set the main hash $self->{grants}.

_sequences

This function is used to retrieve all sequences information.

Set the main hash $self->{sequences}.

_triggers

This function is used to retrieve all triggers information.

Set the main hash $self->{triggers}.

_functions

This function is used to retrieve all functions information.

Set the main hash $self->{functions}.

_packages

This function is used to retrieve all packages information.

Set the main hash $self->{packages}.

_tables

This function is used to retrieve all table information.

Set the main hash of the database structure $self->{tables}. Keys are the names of all tables retrieved from the current database. Each table information compose an array associated to the table_info key as array reference. In other way:

     $self->{tables}{$class_name}{table_info} = [(OWNER,TYPE)];
 
 

DBI TYPE can be TABLE, VIEW, SYSTEM TABLE, GLOBAL TEMPORARY, LOCAL TEMPORARY, ALIAS, SYNONYM or a data source specific type identifier. This only extract TABLE type.

It also get the following informations in the DBI object to affect the main hash of the database structure :

     $self->{tables}{$class_name}{field_name} = $sth->{NAME};
     $self->{tables}{$class_name}{field_type} = $sth->{TYPE};
 
 

It also call these other private subroutine to affect the main hash of the database structure :

     @{$self->{tables}{$class_name}{column_info}} = $self->_column_info($class_name, $owner);
     %{$self->{tables}{$class_name}{unique_key}}  = $self->_unique_key($class_name, $owner);
     @{$self->{tables}{$class_name}{foreign_key}} = $self->_foreign_key($class_name, $owner);
     %{$self->{tables}{$class_name}{check_constraint}}  = $self->_check_constraint($class_name, $owner);
 
 

_views

This function is used to retrieve all views information.

Set the main hash of the views definition $self->{views}. Keys are the names of all views retrieved from the current database values are the text definition of the views.

It then set the main hash as follow:

     # Definition of the view
     $self->{views}{$table}{text} = $view_infos{$table};
 
 

_tablespaces

This function is used to retrieve all Oracle Tablespaces information.

Set the main hash $self->{tablespaces}.

_get_sql_data

Returns a string containing the entire SQL Schema definition compatible with PostgreSQL

_extract_sequence_info

This function adjusts the last used number returned from the sequence used by the source database. Result is a SQL script assigning the new start value to the sequences found.

_get_data TABLE

This function implements a Oracle-native data extraction.

Return a list of array reference containing the data

_sql_type INTERNAL_TYPE LENGTH PRECISION SCALE

This function return the PostgreSQL datatype corresponding to the Oracle internal type.

_column_info TABLE OWNER

This function implements a Oracle-native column information.

Return a list of array reference containing the following informations for each column the given a table

[(
  column name,
  column type,
  column length,
  nullable column,
  default value )]

_unique_key TABLE OWNER

This function implements a Oracle-native unique (including primary) key column information.

Returns a hash of hashes in the following form:
    ( constraintname => (type => 'PRIMARY',
                         columns => ('a', 'b', 'c')),
      constraintname => (type => 'UNIQUE',
                         columns => ('b', 'c', 'd')),
      etc.
    )

_check_constraint TABLE OWNER

This function implements a Oracle-native check constraint information.

Return a hash of list of all column name defined as check constraint for the given table and constraint name.

_foreign_key TABLE OWNER

This function implements a Oracle-native foreign key reference information.

Return a list of hash of hash of array reference. Ouuf! Nothing very difficult. The first hash is composed of all foreign key name. The second hash just have two key known as 'local' and remote' corresponding to the local table where the foreign key is defined and the remote table where the key refer.

The foreign key name is composed as follow:

     'local_table_name->remote_table_name'
 
 

Foreign key data consist in two array representing at the same indice the local field and the remote field where the first one refer to the second. Just like this:

     @{$link{$fkey_name}{local}} = @local_columns;
     @{$link{$fkey_name}{remote}} = @remote_columns;
 
 

_get_users

This function implements a Oracle-native users information.

Return a hash of all users as an array.

_get_roles

This function implements a Oracle-native roles information.

Return a hash of all groups (roles) as an array of associated users.

_get_all_roles

This function retrieve all Oracle roles information.

Return a hash of all roles as an array of associated users.

_get_all_grants

This function implements a Oracle-native user privilege information.

Return a hash of all tables grants as an array of associated users.

_get_indexes TABLE OWNER

This function implements a Oracle-native indexes information.

Return hash of array containing all unique index and a hash of array of all indexes name which are not primary keys for the given table.

_get_sequences

This function implements a Oracle-native sequences information.

Return a hash of array of sequence name with MIN_VALUE, MAX_VALUE, INCREMENT and LAST_NUMBER for the given table.

_get_views

This function implements a Oracle-native views information.

Return a hash of view name with the SQL query it is based on.

_alias_info

This function implements a Oracle-native column information.

Return a list of array reference containing the following informations for each alias of the given view

[(
  column name,
  column id )]

_get_triggers

This function implements a Oracle-native triggers information.

Return an array of refarray of all triggers informations

_get_functions

This function implements a Oracle-native functions information.

Return a hash of all function name with their PLSQL code

_get_packages

This function implements a Oracle-native packages information.

Return a hash of all function name with their PLSQL code

_table_info

This function retrieve all Oracle-native tables information.

Return a handle to a DB query statement

_get_tablespaces

This function implements a Oracle-native tablespaces information.

Return a hash of array of tablespace name with system filepath for the given table.

_get_schema

This function return 1 if the requested schema is found in the database else return 0

AUTHOR

Gilles Darold <gilles@darold.net> Copyright (c) 2001-2008 Gilles Darold - All rights reserved.

This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.

BUGS

This perl module is in the same state as my knowledge regarding database, it can move and not be compatible with older version so I will do my best to give you official support for Ora2Pg. Your volontee to help construct it and your contribution are welcome.

SEE ALSO

DBI, DBD::Oracle, DBD::Pg

ACKNOWLEDGEMENTS

The following person help a lot with reports and help:
         Jason Servetar
         Jean-Francois Ripouteau
         Octavi Fors
         Adriano Bonat
         Thomas Reiss
         Bozkurt Erkut from SONY
 
 

The following person help a lot with code review and bug fix:

         Stephane Schildknecht
         Jean-Paul Argudo
         Jan Kester
         Paolo Mattioli
         Mike Wilhelm-hiltz
         Jefferson Medeiros
         Ian Boston
         Thomas Wegner
         Andreas Haumer
         Marco Lombardo
         Adam Sah and Zedo Inc
         Antonios Christofide and National Technical University of Athens
         Josian Larcheveque
         Stephane Silly
         David Cotter - Alatto Technologies Ltd
         Wojciech Szenajch
         Richard Chen
         Sergio Freire
         Matt Miller
         Rene Bentzen
         Schnabl Andrea
         Ugo Brunel - Bull
         Bernd Helmle - credativ GmbH
         Peter Eisentraut
 
 

My best regards to all of them.