dbt Generic Tests¶

Truthound's main dbt authoring surface is a set of generic tests. They let you keep validation next to the model contract instead of moving the logic into separate SQL files or Python orchestration code.

Quick Model-Level Example¶

version: 2

models:
  - name: test_model_valid
    tests:
      - truthound.truthound_check:
          arguments:
            rules:
              - column: id
                check: not_null
              - column: id
                check: unique
              - column: email
                check: email_format
              - column: user_uuid
                check: uuid_format
              - column: url
                check: url_format

Column-Level Convenience Tests¶

Truthound also ships shorthand tests for common cases:

columns:
  - name: id
    tests:
      - truthound.truthound_not_null
      - truthound.truthound_unique

  - name: email
    tests:
      - truthound.truthound_email_format

  - name: age
    tests:
      - truthound.truthound_range:
          arguments:
            min: 0
            max: 150

Supported Rule Families¶

The tested first-party project exercises the following families:

completeness: not_null, not_empty
uniqueness: unique
numeric checks: range, positive, non_negative
format checks: email_format, uuid_format, url_format
set membership: in_set
relational checks: referential_integrity

Warning vs Error Behavior¶

Truthound uses normal dbt test config semantics.

Use severity: warn when:

you are validating intentionally bad fixtures
you want visibility before hard enforcement
you are migrating an existing model toward a stricter contract

tests:
  - truthound.truthound_check:
      arguments:
        rules:
          - column: status
            check: in_set
            values: ["active", "inactive", "pending"]
      config:
        severity: warn

Modern YAML Shape¶

dbt 1.10 warns when test arguments are supplied as top-level keys instead of under arguments. The package still compiles in that shape today, but new docs should use the modern form shown above.

Source Testing¶

The same package works on dbt sources:

version: 2

sources:
  - name: raw
    tables:
      - name: users
        tests:
          - truthound.truthound_check:
              arguments:
                rules:
                  - column: user_id
                    check: not_null
                  - column: email
                    check: email_format

Production Guidance¶

keep grouped model-level tests for business-critical datasets
use column-level convenience tests for obvious single-column guarantees
prefer explicit package qualification everywhere
use warnings only intentionally and document why they are warnings

Failure Modes¶

undefined generic tests usually mean the package was not installed or names were not resolved the way you expected
passing compilation with unexpected runtime failures often indicates an adapter edge case rather than a YAML problem