Audit Tables (aud_*)

The aud_* tables are the configuration backbone of the platform. They live in BigQuery and define every aspect of pipeline behavior at runtime.

Table Hierarchy

Table Reference

aud_modelo

Top-level grouping. Represents a business domain or data model.

Column	Type	Description
`modelo_id`	STRING	Unique identifier, used as DAG lookup key
`nombre`	STRING	Human-readable name
`descripcion`	STRING	Business context
`chain_dag`	STRING	Which chain DAG owns this model
`is_active`	BOOL	Whether this model is processed
`owner_team`	STRING	Responsible team
`owner_email`	STRING	Contact for alerts

aud_entidad

One row per table/entity extracted from a source.

Column	Type	Description
`entidad_id`	STRING	Unique identifier
`modelo_id`	STRING	FK to aud_modelo
`nombre`	STRING	Entity name (maps to table/file name)
`source_type`	STRING	`oracle` / `api_rest` / `csv` / `pdf`
`extraction_mode`	STRING	`full` / `incremental`
`executor_type`	STRING	`airflow` / `dataflow`
`dataflow_template`	STRING	Template name if executor = dataflow
`gcs_destination`	STRING	GCS bucket/path for raw Avro
`bq_dataset_history`	STRING	BQ dataset for history table
`bq_dataset_active`	STRING	BQ dataset for active table
`is_active`	BOOL	Whether this entity is processed

Valid values:

yaml

source_type: [oracle, api_rest, csv, pdf]
extraction_mode: [full, incremental]
executor_type: [airflow, dataflow]

aud_columna

Schema definition per entity. Drives MERGE keys and validations.

Column	Type	Description
`columna_id`	STRING	Unique identifier
`entidad_id`	STRING	FK to aud_entidad
`nombre_columna`	STRING	Column name in source and BQ
`tipo_dato`	STRING	BQ data type
`es_pk`	BOOL	Used as MERGE key in active layer
`es_partition`	BOOL	Partition column (fecha_lote)
`es_requerida`	BOOL	NOT NULL validation

aud_dependencia

Defines execution dependencies between entities.

Column	Type	Description
`dependencia_id`	STRING	Unique identifier
`entidad_id`	STRING	Entity that depends on another
`entidad_dependiente_id`	STRING	Entity that must complete first
`tipo_dependencia`	STRING	`data` / `execution`
`descripcion`	STRING	Reason for dependency

aud_calendario

Controls scheduling and chain placement.

Column	Type	Description
`calendario_id`	STRING	Unique identifier
`entidad_id`	STRING	FK to aud_entidad
`schedule_cron`	STRING	Cron expression
`grupo_dag`	STRING	Which Group DAG this entity belongs to
`chain_dag`	STRING	Which Chain DAG owns this group
`orden_en_cadena`	INT	Sequential order within the chain
`excluir_feriados`	BOOL	Skip execution on holidays
`timezone`	STRING	e.g. `America/Argentina/Buenos_Aires`

aud_validacion

Validation rules evaluated pre and post extraction.

Column	Type	Description
`validacion_id`	STRING	Unique identifier
`entidad_id`	STRING	FK to aud_entidad
`tipo_validacion`	STRING	`row_count` / `schema` / `not_null` / `business_rule`
`momento`	STRING	`pre_extraction` / `post_extraction` / `post_load`
`expresion`	STRING	SQL expression or threshold
`severidad`	STRING	`warning` / `error` (error halts pipeline)
`mensaje_error`	STRING	Human-readable error description
`is_active`	BOOL	Whether this rule is evaluated

How the Framework Reads These Tables

At DAG startup, the framework executes a single parameterized query:

sql

SELECT
    e.*,
    ARRAY_AGG(c) AS columnas,
    ARRAY_AGG(v) AS validaciones,
    cal.grupo_dag,
    cal.orden_en_cadena
FROM aud_entidad e
JOIN aud_modelo m ON e.modelo_id = m.modelo_id
JOIN aud_calendario cal ON e.entidad_id = cal.entidad_id
LEFT JOIN aud_columna c ON e.entidad_id = c.entidad_id
LEFT JOIN aud_validacion v ON e.entidad_id = v.entidad_id AND v.is_active = TRUE
WHERE cal.grupo_dag = @dag_name
  AND e.is_active = TRUE
  AND m.is_active = TRUE
GROUP BY e.entidad_id, ..., cal.grupo_dag, cal.orden_en_cadena

The result is a fully resolved config object per entity, ready for task generation.

Audit Tables (aud_*) ​

Table Hierarchy ​

Table Reference ​

aud_modelo ​

aud_entidad ​

aud_columna ​

aud_dependencia ​

aud_calendario ​

aud_validacion ​

How the Framework Reads These Tables ​