Convert Table Creation from MSSQL to PostgreSQL

Question

CREATE TABLE [dbo].[Addresses](
    [Id] [int] IDENTITY(1,1) NOT NULL,
    [UserId] [int] NULL,
    [City] [nvarchar](50) NULL,
    [State] [nvarchar](50) NULL,
    [CountryCode] [nvarchar](50) NULL,
    [PostalCode] [nvarchar](50) NULL,
    [Street1] [nvarchar](50) NULL,
    [Street2] [nvarchar](50) NULL,
PRIMARY KEY CLUSTERED 
(
    [Id] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]
) ON [PRIMARY]

What would the equivalent PostgreSQL table creation script look like, based off of the previous MSSQL create script? Looking for an approach/resource that can guide me on converting MSSQL schema to PostgreSQL ...

Answer 1

Phil's answer is correct. As you also asked for " for an approach ", here is a very short description to convert Microsoft T-SQL:

• always end your statements with a ; (in general: replace GO with ; ) - and no, the ; does not go in front of the with keyword for a common table expression
• remove the non-standard square bracket "quoting" [..] (I highly recommend to never use quoted identifiers). Details about the identifier syntax are in the manual
• identity columns map to serial or bigserial in Postgres. That implies the data type integer or bigint so you only need the keyword serial . For details see the manual
• there is no nvarchar in Postgres. All character columns use the same character set.

• other data types to convert (incomplete!):

   SQL Server Postgres --------------------------- image bytea binary bytea varbinary(max) bytea varchar(max) text nvarchar(max) text text text ntext text bit boolean datetime timestamp smalldatetime timestamp timestamp no equivalent uniqueidentifier uuid xml xml 

• there are not clustered indexes so remove the clustered attribute

• the whole with part defines physical properties of the table that are not relevant in Postgres
• the dbo schema "maps" to the public schema in Postgres. It is somewhat uncommon prefix tables with public though unless you changed your schema search path (there is also no need to use the schema prefix for your own functions as it is required in SQL Server)

Some things to watch out for:

• string comparison is case sensitive in Postgres. where name = 'Arthur' will return something different than where name = 'arthur' . Search this site, there are plenty of questions regarding this.
• the "poor man's regex" syntax of SQL Server's LIKE (eg like '[0-9]' ) does not work in Postgres. You need to use the regex operator.
• date handling is a bit different
• Postgres is much more picky about correct data types and value literals. where varchar_column = 2 fails in Postgres even if there are only numbers in that column

• In Postgres you can insert multiple null values into a unique index on a nullable column. The following fails in SQL Server but works in Postgres

   create table foo (col1 integer, col2 integer); create unique index idx_foo on foo (col1, col2); insert into foo (col1, col2) values (1, null); insert into foo (col1, col2) values (1, null); 

---

This list is by no means complete! There is a lot more to look out when converting from one DBMS to another.

Answer 2

Here you go:

CREATE TABLE Addresses(
    Id SERIAL NOT NULL PRIMARY KEY,
    UserId int NULL,
    City varchar(50) NULL,
    State varchar(50) NULL,
    CountryCode varchar(50) NULL,
    PostalCode varchar(50) NULL,
    Street1 varchar(50) NULL,
    Street2 varchar(50) NULL
);

SQL Fiddle: http://sqlfiddle.com/#!15/b76b6