![Datacoves](images/datacoves-light.png)
The Datacoves platform helps enterprises overcome their data delivery challenges quickly using dbt and Airflow, implementing best practices from the start without the need for multiple vendors or costly consultants.
Hosted VS Code, dbt-core, SqlFluff, and Airflow, find out more at Datacoves.com.
dbt-checkpoint provides pre-commit hooks to ensure the quality of your dbt projects.
dbt
is awesome, but when the number of models, sources, and macros in a project grows, it becomes challenging to maintain the same level of quality across developers. Users forget to update columns in property (yml) files or add table and column descriptions. Without automation the reviewer workload increases and unintentional errors may be missed. dbt-checkpoint allows organizations to add automated validations improving your code review and release process.
dbt-checkpoint has telemetry built into some of its hooks to help the maintainers from Datacoves understand which hooks are being used and which are not to prioritize future development of dbt-checkpoint. We do not track credentials nor details of your dbt execution such as model names. We also do not track any of the dbt hooks, such as for generating documentation. The one detail we do use related to dbt is the anonymous user_id generated by dbt to help us identify distinct projects.
By default this is turned on – you can opt out of event tracking at any time by adding the following to your .dbt-checkpoint.yaml
file:
version: 1
disable-tracking: true
You can specify a dbt project root directory for all hooks. This is particularly useful when your dbt project is not located at the root of your repository but in a sub-directory of it.
In that situation, you previously had to specify a --manifest
flag in each hook.
Now, you can avoid repeating yourself by adding the dbt-project-dir
key to your .dbt-checkpoint.yaml
config file:
version: 1
dbt-project-dir: my_dbt_project
This way, we will automatically look for the required manifest/catalog inside your my_dbt_project
project folder.
Since dbt-checkpoint 1.1.0
, certain hooks implement an implicit logic that "discover" their sql/yml equivalent for checking.
For a complete background please refer to #118.
Since the root-level exclude
statement is handled by pre-commit, when those hooks discover their related sql/yml files, this root exclusion is ommitted (dbt-checkpoint re-includes files that may have been excluded). To exclude files from being discovered by this logic, the exclude
path/regex must be provided in each hook (#119)
💡 Click on hook name to view the details.
Model checks:
check-column-desc-are-same
: Check column descriptions are the same.check-column-name-contract
: Check column name abides to contract.check-model-columns-have-desc
: Check the model columns have description.check-model-has-all-columns
: Check the model has all columns in the properties file.check-model-has-contract
: Check the model has contract enforced.check-model-has-constraints
: Check the model has specific constraints defined.check-model-has-description
: Check the model has description.check-model-has-meta-keys
: Check the model has keys in the meta part.check-model-has-labels-keys
: Check the model has keys in the labels part.check-model-has-properties-file
: Check the model has properties file.check-model-has-tests-by-name
: Check the model has a number of tests by test name.check-model-has-tests-by-type
: Check the model has a number of tests by test type.check-model-has-tests-by-group
: Check the model has a number of tests from a group of tests.check-model-has-tests
: Check the model has a number of tests.check-model-name-contract
: Check model name abides to contract.check-model-parents-and-childs
: Check the model has a specific number (max/min) of parents or/and childs.check-model-parents-database
: Check the parent model has a specific database.check-model-parents-name-prefix
: Check the parent model names have a specific prefix.check-model-parents-schema
: Check the parent model has a specific schema.check-model-tags
: Check the model has valid tags.check-model-materialization-by-childs
: Check the materialization of models given a threshold of child models.
Script checks:
check-script-semicolon
: Check the script does not contain a semicolon.check-script-has-no-table-name
: Check the script has not table name (is not usingsource()
orref()
macro for all tables).check-script-ref-and-source
: Check the script has only existing refs and sources.
Source checks:
check-source-columns-have-desc
: Check for source column descriptions.check-source-has-all-columns
: Check the source has all columns in the properties file.check-source-table-has-description
: Check the source table has description.check-source-has-freshness
: Check the source has the freshness.check-source-has-loader
: Check the source has loader option.check-source-has-meta-keys
: Check the source has keys in the meta part.check-source-has-labels-keys
: Check the source has keys in the labels part.check-source-has-tests-by-group
: Check the source has a number of tests from a group of tests.check-source-has-tests-by-name
: Check the source has a number of tests by test name.check-source-has-tests-by-type
: Check the source has a number of tests by test type.check-source-has-tests
: Check the source has a number of tests.check-source-tags
: Check the source has valid tags.check-source-childs
: Check the source has a specific number (max/min) of childs.
Macro checks:
check-macro-has-description
: Check the macro has description.check-macro-arguments-have-desc
: Check the macro arguments have description.
Modifiers:
generate-missing-sources
: If any source is missing this hook tries to create it.generate-model-properties-file
: Generate model properties file.unify-column-description
: Unify column descriptions across all models.replace-script-table-names
: Replace table names withsource()
orref()
macros in the script.remove-script-semicolon
: Remove the semicolon at the end of the script.
dbt commands:
dbt-clean
: Rundbt clean
command.dbt-compile
: Rundbt compile
command.dbt-deps
: Rundbt deps
command.dbt-docs-generate
: Rundbt docs generate
command.dbt-parse
: Rundbt parse
command.dbt-run
: Rundbt run
command.dbt-test
: Rundbt test
command.
dbt checks:
check-database-casing-consistency
: compare Manifest and Catalog to ensure DB and Schemas have the same casing.
❗If you have a suggestion for a new hook or you find a bug, let us know❗
For detailed installation and usage, instructions see pre-commit.com site.
pip install pre-commit
- Create a file named
.pre-commit-config.yaml
in your project root folder. - Add list of hooks you want to run befor every commit. E.g.:
repos:
- repo: https://github.com/dbt-checkpoint/dbt-checkpoint
rev: v1.2.1
hooks:
- id: dbt-parse
- id: dbt-docs-generate
args: ["--cmd-flags", "++no-compile"]
- id: check-script-semicolon
- id: check-script-has-no-table-name
- id: check-model-has-all-columns
name: Check columns - core
files: ^models/core
- id: check-model-has-all-columns
name: Check columns - mart
files: ^models/mart
- id: check-model-columns-have-desc
files: ^models/mart
- Optionally, run
pre-commit install
to set up the git hook scripts. With this,pre-commit
will run automatically ongit commit
! You can also manually runpre-commit run
after youstage
all files you want to run. Orpre-commit run --all-files
to run the hooks against all of the files (not onlystaged
).
Unfortunately, you cannot natively use dbt-checkpoint
if you are using dbt Cloud. But you can run checks after you push changes into Github.
dbt-checkpoint
for the most of the hooks needs manifest.json
(see requirements section in hook documentation), that is in the target
folder. Since this target folder is usually in .gitignore
, you need to generate it. For that you need to run the dbt-parse
command.
To be able to parse dbt, you also need profiles.yml file with your credentials. To provide passwords and secrets use Github Secrets (see example).
Say you want to check that a model contains at least two tests, you would use this configuration:
repos:
- repo: https://github.com/dbt-checkpoint/dbt-checkpoint
rev: v1.2.1
hooks:
- id: check-model-has-tests
args: ["--test-cnt", "2", "--"]
To be able to run this in Github CI you need to modified it to:
repos:
- repo: https://github.com/dbt-checkpoint/dbt-checkpoint
rev: v1.2.1
hooks:
- id: dbt-parse
- id: check-model-has-tests
args: ["--test-cnt", "2", "--"]
First step is to create profiles.yml. E.g.
# example profiles.yml file
jaffle_shop:
target: dev
outputs:
dev:
type: snowflake
threads: 8
client_session_keep_alive: true
account: "{{ env_var('ACCOUNT') }}"
database: "{{ env_var('DATABASE') }}"
schema: "{{ env_var('SCHEMA') }}"
user: "{{ env_var('USER') }}"
password: "{{ env_var('PASSWORD') }}"
role: "{{ env_var('ROLE') }}"
warehouse: "{{ env_var('WAREHOUSE') }}"
and store this file in project root ./profiles.yml
.
- inside your Github repository create folder
.github/workflows
(unless it already exists). - create new file e.g.
pr.yml
- specify your workflow e.g.:
name: dbt-checkpoint
on:
push:
pull_request:
branches:
- main
jobs:
dbt-checkpoint:
runs-on: ubuntu-latest
env:
ACCOUNT: ${{ vars.ACCOUNT }}
DATABASE: ${{ vars.DATABASE }}
SCHEMA: ${{ vars.SCHEMA }}
USER: ${{ vars.USER }}
PASSWORD: ${{ secrets.PASSWORD }}
ROLE: ${{ vars.ROLE }}
WAREHOUSE: ${{ vars.WAREHOUSE }}
steps:
- name: Checkout code
uses: actions/checkout@v2
- name: Setup Python
uses: actions/setup-python@v2
- id: Get file changes
uses: trilom/[email protected]
with:
output: " "
- name: Run dbt checkpoint
uses: dbt-checkpoint/[email protected]
with:
extra_args: --files ${{ steps.get_file_changes.outputs.files}}
dbt_version: 1.6.3
dbt_adapter: dbt-snowflake
Thank you to Radek Tomšej for initial development and maintenance of this great package, and for sharing your work with the community!