Compute format

GET /api/event-criterias/:id?format=compute

Returns an event filter as a flat list of conditions, instead of the full tree. It is the other branch of the format query param of GET /event-criterias/:id.

Auth: Required — event-criteria:view permission

This representation is narrower than the definition: it carries row conditions over a single query block. A filter whose definition does not fit in it is rejected with 422 and a code that says exactly what does not fit — a degraded version of the filter is never returned.

Path Parameters

Query Parameters

Example

curl "https://api.reten.ai/api/event-criterias/9f3a1c22-4b7e-4c31-9a8d-2f5e6b7c8d90?format=compute" \
  -H "Authorization: Bearer <token>" \
  -H "x-tenant-id: <tenant-id>" \
  -H "x-project-id: <project-id>"

import axios from 'axios';

const id = '9f3a1c22-4b7e-4c31-9a8d-2f5e6b7c8d90';
const response = await axios.get(
  `https://api.reten.ai/api/event-criterias/${id}`,
  {
    headers: { Authorization: 'Bearer <token>', 'x-tenant-id': '<tenant-id>', 'x-project-id': '<project-id>' },
    params: { format: 'compute' },
  },
);
const computeCriteria = response.data;

Response shape 200 OK

IComputeEventCriteria:

Each element of conditions is one of two shapes, distinguishable by its operator:

Scalar condition

Record condition

The four cases

1 · Scalar condition

Events whose status is Facturada.

{
  "id": "9f3a1c22-4b7e-4c31-9a8d-2f5e6b7c8d90",
  "name": "Venta oficial — solo facturada",
  "conditions": [
    {
      "field": "status",
      "operator": "=",
      "value": "Facturada",
      "source": "event_transactions"
    }
  ]
}

2 · Record with two sub-filters

Events with an attribute whose key is documento_tipo and whose value is factura.

{
  "id": "1c7d2e83-5a90-4f62-8b31-6d4e5f70a1b2",
  "name": "Documentos tipo factura",
  "conditions": [
    {
      "field": "attributes",
      "operator": "CONTAINS_RECORD",
      "source": "event_transactions",
      "match_all": true,
      "sub_filters": [
        { "field": "key", "operator": "=", "value": "documento_tipo" },
        { "field": "value", "operator": "=", "value": "factura" }
      ]
    }
  ]
}

3 · Scalar + record

Invoiced sales with at least one line that is not a return.

{
  "id": "4b8e0f15-2c76-4a03-9d58-7e1f2a3b4c5d",
  "name": "Ventas con ítems no-devolución",
  "conditions": [
    {
      "field": "status",
      "operator": "=",
      "value": "Facturada",
      "source": "event_transactions"
    },
    {
      "field": "items",
      "operator": "CONTAINS_RECORD",
      "source": "event_transactions",
      "match_all": true,
      "sub_filters": [
        { "field": "category", "operator": "NOT IN", "value": ["devolucion"] }
      ]
    }
  ]
}

4 · Scalar + related table

Invoiced sales with at least one line whose product is in the Food Service or HORECA category. The condition over the related table arrives at the same level as the others, distinguished by its source: the path to product_properties does not travel.

{
  "id": "6d2a9c41-7e35-48b0-b1c9-0f8e7d6c5b4a",
  "name": "Proyecto Food Service",
  "conditions": [
    {
      "field": "status",
      "operator": "=",
      "value": "Facturada",
      "source": "event_transactions"
    },
    {
      "field": "categorization.category_name",
      "operator": "IN",
      "value": ["Food Service", "HORECA"],
      "source": "product_properties"
    }
  ]
}

What match_all means

It is where an integration most easily goes wrong, so it is spelled out:

1. match_all combines the sub_filters within the element, not over the elements. match_all: true means that there is one element of the array that satisfies all the sub_filters. An element that matches only on key or only on value does not satisfy it.
2. match_all: false is the same existential with OR inside: there is one element that satisfies at least one of the sub_filters. It never means "all the elements".
3. With a single sub_filter, match_all: true always travels. With fewer than two sub-filters the connective does not change the result, and true is the canonical shape.
4. CONTAINS_RECORD is existential with negative operators too. items with category NOT IN ["devolucion"] and match_all: true means "there is an item that is not a return", not "no item is a return". This representation cannot express universal quantification.
5. An empty array or NULL never satisfies a CONTAINS_RECORD, and a sub-field that is NULL does not match, not even with a negative operator.

What field means

The field is the full path relative to its scope, and the scope is never repeated:

Operator subset

The operators travel as SQL literals, not as the members of the definition's enum:

CONTAINS_RECORD does not translate any comparison operator: it is emitted by the record leaf.

JSON type of value

Which JSON type the value travels as is fixed by the column type in the catalog, not the shape in which the definition was saved:

Response 422 Unprocessable Entity

The filter is valid and compiles to SQL; what does not exist is its projection to this representation. The body carries the diagnosis:

{
  "code": "TIME_WINDOW_UNSUPPORTED",
  "path": "queryBlocks[0].timeWindow",
  "message": "No se puede proyectar queryBlocks[0].timeWindow (TIME_WINDOW_UNSUPPORTED): el wire no transporta la ventana IN_THE_LAST"
}

The twelve codes

Relationship with conditionCount

Error Responses