adding-a-db-checklist.md

How to Add a Database to db.py

Before you start

There are basically 3 types of relational databases connections:

• Postgres and Postgres variants (Redshift, Greenplum, Vertica, etc.)
• MySQL and MySQL variants (MariaDB, MemSQL, etc.)
• ODBC (MS SQL, Oracle, etc.)

There are versions of each of these already in db.py so be sure to check if the database you're adding is compatible with one of these. It might save you a lot of work.

Databases currently supported

• Postgres
• SQLite
• MySQL
• MS SQL Server
• Redshift

Databases Requested

• Oracle
• Teradata

Checklist

For this example we'll be adding a database called foosql.

Queries

You'll need to port the required queries into whatever flavor of SQL your database uses. If you're even somewhat familiar with SQL, this shouldn't be too difficult since you can go off of one of other databases (they're almost all the same). Here's a checklist of what you'll need to do:

• column
  • head: select first N values of column
  • unique: select distinct values of column
  • all: select all values of column
  • sample: randomly select N values of column
• table
  • head: select first N rows of table
  • select: select specified columns from table
  • unique: select distinct columns specified from table
  • all: select all rows from table
  • sample: randomly select N rows of table
• system
  • schema_no_system
  • schema_with_system
  • schema_specified
  • foreign_keys_for_table
  • foreign_keys_for_column
  • ref_keys_for_table

Create a file called foosql.py and put it in the queries directory. Once you've written the neccessary queries, add an import statement to the db.py file. It will look something like this:

from .queries import foosql as foosql_templates

You'll also need to add your templates to the queries_templates:

queries_templates = {
  "mysql": mysql_templates,
  "postgres": postgres_templates,
  "redshift": postgres_templates,
  "sqlite": sqlite_templates,
  "mssql": mssql_templates,
  "foosql": foosql_templates
}

Connecting

Client Library db.py relies on the PEP 249 spec for connecting and querying. If the database you're using requires a new library for connecting, be sure it is PEP 249 compliant.

If you do need to use a new library, you can add it at the top of db.py. Just make sure to put a try/catch around it.

Alterations to DB You'll need to update the DB class to handle default connection parameters for your database. For example, if foosql defaulted to running on port 6789, I would need to update the if port is None... statement by adding a condition for foosql:

if port is None:
    if dbtype=="postgres":
        port = 5432
    ...
    elif dbtype=="foosql":
        port = 6789

There will also be some docstrings to update as well. Please add examples accordingly.

Lastly you'll need to actually make the connection. This is where you'll actually invoke the client library. For example, adding in foosql would look like this:

if self.dbtype=="postgres" or self.dbtype=="redshift":
    if not HAS_PG:
        raise Exception("Couldn't find psycopg2 library. Please ensure it is installed")
    self.con = pg.connect(user=self.username, password=self.password,
                          host=self.hostname, port=self.port, dbname=self.dbname)
    self.cur = self.con.cursor()
elif self.dbtype=="foosql":
    if not HAS_FOOSQL:
        raise Exception("Couldn't find foosql library. Please ensure it is installed")
    self.con = pg.connect(user=self.username, password=self.password,
                          host=self.hostname, port=self.port, dbname=self.dbname)
    self.cur = self.con.cursor()

Wrap Up

That's pretty much it. It's actually pretty easy to add databases. If you're having troubles, visit the issues page on github.