Postgresql DDL Extension for StarUML 2

This extension for StarUML support to generate DDL (Data Definition Language) from PostgreSQL. Install this extension from Extension Manager of StarUML.

Usage

1. Click the menu (Tools > Postgresql Model Generation > Generate DDL...)
2. Select a directory where the generated code will be deployed
3. Save the ERD model if changes were perfomed during code generation.

If you want to change the generation options please follow the steps below.

1. Click the menu (Tools > Postgresql Model Generation > Configure...)
2. From Preferences/Postgresql DDL tab set your preferences and close
3. Follow the spteps above to re-generate with the new preferences.

Generation rules

Belows are the rules to convert from ERD elements to DDL.

All entities and columns are converted to create table statements as follow:

• Project: provides the database name, either derived from its name or through the database tag. If the tag is not present one will be created with the actual name.
• Data Model: part of a Project, it gives the schema where all member tables will be created. The schema name is provided by the schema tag, default value being public if none is defined.
• ER Diagram: groups all the Entities. Each entity will be transformed into a table and at this level the user can provide a table name prefix for them using the tag prefix.
• Entity: provides table description. The table name is composed of the ER Diagram prefix (optional) and the entity name. The table tag can be used to override the Entity name.
• Column: generates a table column. The column name is derived from the name but the column tag can be used to override it. The default value can be stored in the default tag.

Index and constraints creation

• primary key: each column marked as primary key will be part of the primary key constraint.
• foreign key: each column marked as foreign key will have its own index
• unique constraint: the columns marked as unique will be part of the table's unique contraint. With this version only one unique contraint is generated per table.
• reference constraints: for a foreign key column with the reference field populated, a reference contraint will be created.

Type mapping

The integer columns with length -1 will be mapped to the equivalent serial type. For instance BIGINT with length -1 will be mapped to BIGSERIAL.

StarUML Type	PostgreSQL type	Note
VARCHAR	varchar	length can be used for maximum size
BOOLEAN	boolean	length is not used
INTEGER	integer	serial if length is set to -1
CHAR	char	length can be used for size
BINARY	bytea	length is ignored
VARBINARY	bytea	length is ignored
BLOB	bytea	length is ignored
TEXT	text	length is not used
SMALLINT	smallint	smallserial with length -1
BIGINT	bigint	bigserial if length is -1
DECIMAL	numeric	length is used
NUMERIC	numeric	length is used
FLOAT	real	length is ignored
DOUBLE	double precision	length is ignored
BIT	bit	length is used
DATE	date	length is not used
TIME	time without time zone	length is not used
DATETIME	timestamp with time zone	length is not used
TIMESTAMPTZ	timestamp with time zone	length is not used
TIMESTAMP	timestamp without time zone	length is not used
POINT	point
POLYGON	polygon
enum	[table]_[column] type	enumeration type named with the provided pattern

Generated files name pattern

For ease of use the files are generated following specific patterns. In general the files are in pairs, one ending in _create for the object creation and one ending in _drop that will clear all objects from the database.

Type	Creation filename pattern	Removal filename pattern	Deployment notes
Database	db_create.sql	db_drop.sql	Unless the database name is changed these files needs to be run once
Schema	schema_create.sql	schema_drop.sql	If schema name changes these files need to be run followed by the table ones
Diagram Tables	<model name>_<diagram name>_create.sql	<model name>_<diagram name>_drop.sql	Code for all the children diagram entities
Schema tables	<model name>_table_create.sql	<model name>_table_drop.sql	Code for all the children data model entities

• Note: the difference between diagram and schema tables is given by the model owner: the diagram entities that are moved under a diagram are diagram tables, the others (default) are schema's.

File deployment

Once files are generated or changed, you have to deploy or re-deploy them into your database server. Connect to the server using psql or your favorite client and deploy the files based on the scenarios below.

• Note: if you choose to have reference constraints please deploy the file accordingly.

Initial deployment.

It is assumed that the DDL is generated for the first time. The files should be deployed in the following oder:

• deploy the database creation file, for instance from the psql prompt: \i db_create.sql
• deploy the schema creation files, \i schema_create.sql
• for each data model deploy the table file or files.

Remove the tables.

Tables should be removed if model changes are performed to them. Identify the specific data model or diagram where the table are and run the specific script ending with _drop.sql.

Clean up the database.

Run db_drop.sql.

User tags

It is possible to assist with the code generation through custom tags with specific names. The code generation will normally use the model names to generate the SQL ones but that can be overriden.

Tag	Level	Description
database	Project	Overrides the database name given by the normalized project name
prefix	ERDDiagram	Gives the prefix of all the table names of the diagram
schema	ERDDataModel	Indicates the schema where the tables will be created
table	Entity	Override the table name otherwise given by the normalized Entity name
column	Column	Override the column name, otherwise given by the Column name
default	Column	Gives the default column value. The string value is taken as is so user needs to be aware of specific database syntax
enum	Column	Gives the column comma separated enumeration elements

• Note: name normalization consists in replacing the space characters with underscores and converting to lower case.

The override tags are used to separate the high level model from the physical model. The model table name may be user/business friendly but the generated one may follow specific constraints that can be hard to follow at the high level. For example, the User table can be problematic in a real schema as User is a keyword but it makes perfect sense while building the model; the database name for this table may end up being cex_people_usr.

Enumeration types

Postgresql supports enumeration for column value and the plugin allows the user to define them using the enum custom tag. The user can declare the column type as enum and then add the custom enum tag on the column. The value text field should have the list of possible values. The plugin will subsequently create a dedicated enumration type with the name pattern table_column.

In order to facilitate the usage of enumeration values, an implicit character varying conversion cast is created; as such the user can use the enumeration values as strings without an explicit casting.

In the included example, the Employee status field is a (permanent, temporary) enumeration.

Options

The following preferences are currently available:

Preference	Description
Owner	Postgresql object owner, only one for entire model
Storage tablespace	Database storage tablespace name
Storage character set	Database storage character set
Database collation	Collation list
Generate Foreign Key Constraints	Generate foreign key constraints when column reference field is populated
Use tab	use TAB for indentation
Indent spaces	Number of spaces to be used for indentation if tab use is off

Contributions

Any contributions are welcome. If you find a bug or have a suggestion, please post as an issue.

Notes.

By default the entities when dropped into the diagram view are added in the model as diagram siblings. They can be manually moved into the diagram manually to achieve a similar structure with the sample included. However the generator accommodates for both cases.