Updating Schema in SQLite3

By R. S. Doiel, 2020-04-16

SQLite3 is a handy little database as single file tool. You can interact with the file through largely standard SQL commands and embed it easily into applications via the C libraries that the project supports. It is also available from various popular scripting languages like Python, PHP, and Lua. One of the things I occasionally need to do and always seems to forget it how to is modify a table schema where I need to remove a column¹. So here are some of the basics do I can quickly find them later and avoid reading various articles tutorials because the search engines doesn’t return the page in the SQLite documentation.

In the next sections I’ll be modeling a simple person object with a id, uname, display_name, role and updated fields.

Creating a person table

CREATE TABLE IF NOT EXISTS "person" 
        ("id" INTEGER NOT NULL PRIMARY KEY, 
        "uname" VARCHAR(255) NOT NULL, 
        "role" VARCHAR(255) NOT NULL, 
        "display_name" VARCHAR(255) NOT NULL, 
        "updated" INTEGER NOT NULL);

Adding a column

We will create a junk column which we will remove later.

.schema person
ALTER TABLE person ADD COLUMN junk VARCHAR(255) NOT NULL;
.schema person

Dropping a column

To drop a column in SQLite you need to actually create a new table, migrate the data into it then drop the old table and finally rename it. It is best to wrap this in a transaction.

BEGIN TRANSACTION;
    CREATE TABLE IF NOT EXISTS "person_new" 
           ("id" INTEGER NOT NULL PRIMARY KEY, 
           "uname" VARCHAR(255) NOT NULL, 
           "role" VARCHAR(255) NOT NULL, 
           "display_name" VARCHAR(255) NOT NULL, 
           "updated" INTEGER NOT NULL);
    INSERT INTO person_new
           SELECT id, uname, role, display_name, updated
           FROM person;
    DROP TABLE person;
    ALTER TABLE person_new RENAME TO person;
COMMIT;