dbt-oracle 0.4.3

An Oracle DBT Adapater

Oracle DBT

Tested with dbt==0.19.1

for dbt 0.18.0 and below, use dbt-oracle version 0.3.2

Installation

You need the oracle database client installed in your system in order for this to work, here (https://cx-oracle.readthedocs.io/en/latest/user_guide/installation.html) you can find the cx_oracle python driver installation instructions.

This (https://gist.github.com/tcnksm/7316877) gist is also a useful resource for installing the client in Ubuntu. It’s an old link for Ubuntu 12 but it still works at least for ubuntu 18.

Installing:

pip install dbt-oracle==0.4.3

Configure your profile

dbt-oracle supports three authentication methods. In any case, one must specifiy user, pass, dbname, and schema

• host: Setting host (and possibly port and service, if that differs from dbname)
• connection string: Setting connection_string
• TNS: Setting dbname only

Host example

dbt_oracle_test:
   target: dev
   outputs:
      dev:
         type: oracle
         host: localhost
         user: system
         pass: oracle
         port: 1522
         dbname: xe
         schema: system
         threads: 4

If you need to connect via a service name that doesn’t match the database name, then you may optionally specify service with the above, e.g.:

service: xe_ha.domain.tld

TNS example

Configuring your ORACLE_HOME environment variable so dbt-oracle can find the tnsnames.ora file. Let’s assume your tnsnames.ora file is placed at

/home/user/oracle/network/admin/tnsnames.ora

you could set

ORACLE_HOME=/home/user/oracle/

and your profile:

dbt_oracle_test:
   target: dev
   outputs:
      dev:
         type: oracle
         user: system
         pass: oracle
         dbname: xe
         schema: system
         threads: 4

Connection string example

dbt_oracle_test:
   target: dev
   outputs:
      dev:
         type: oracle
         user: system
         pass: oracle
         dbname: xe
         schema: system
         threads: 4
         connection_string: "(DESCRIPTION=(ADDRESS=(PROTOCOL=TCP)(HOST=localhost)(PORT=1521))(CONNECT_DATA=(SERVICE_NAME=xe_ha.domain.tld)))"

Supported Features

Materilizations

• table: OK
• view: OK
• incremental: OK
• ephemeral: not OK

Seeds

OK

Hooks

OK

Custom schemas

Not tested

Sources

Not tested

Testing & documentation

• Schema tests OK
• Relationship tests Not OK
• Data tests Not OK in general – These work as long as you do not use CTEs in your data tests. One strategy to manage this is to persist a useful CTE as an analysis.
• SQL Tests OK
• Docs generate Not OK

Snapshots

OK

Testing

There is a dummy dbt project called dbt_test_project for testing some things that the official dbt integration tests do not cover. For both dbt_test_project and dbt oficial adpter tests we are using a database user ‘dbt_test’ with password ‘dbt_test’ You have to either create this user os change the credentias at tests/oracle.dbtspec and dbt_test_project/profiles.yml

For running it first start an oracle database instance:

docker run \
    --name dbt-oracle-db \
    -d \
    -p 1522:1522 \
    epiclabs/docker-oracle-xe-11g

Install the project locally

python setup.py install

then run dbt seed and run (theres is a profile file compatible with oracle 11g docker defaults at the test dir)

cd dbt_test_project
dbt seed --profiles-dir ./
dbt run --profiles-dir ./
dbt test --profiles-dir ./

you can also run

make test

for running both dbt adapter tests and the dbt_test_project included in this repo

The following dbt adapter tests are passing:

tests/oracle.dbtspec::test_dbt_empty
tests/oracle.dbtspec::test_dbt_base
tests/oracle.dbtspec::test_dbt_ephemeral
tests/oracle.dbtspec::test_dbt_incremental
tests/oracle.dbtspec::test_dbt_snapshot_strategy_timestamp
tests/oracle.dbtspec::test_dbt_snapshot_strategy_check_cols
tests/oracle.dbtspec::test_dbt_schema_test

Known failing tests:

FAILED tests/oracle.dbtspec::test_dbt_data_test
FAILED tests/oracle.dbtspec::test_dbt_ephemeral_data_tests

The dbt adapter tests for data tests fails due to how the test_dbt_data_test sequence is implemented. You can use data tests with this adapter, there is an example at the test project in this repo.

Final Notes

This is a new project and any contribuitions are welcome.

History

0.1.0 (2020-06-02)

• First release on PyPI.

0.2.0 (2020-06-02)

• Added full macro implementation. Thanks Fabrice Etanchaud