Stage Brazil CVM Filings Metadata with dbt

Want dbt CVM filings models? Land JSON from apicvm into raw tables, then build stg_cvm_documents and marts for coverage. dbt does not download PDFs for you — your extractor/loader does.

The problem

Analytics engineers need reproducible tables: which documents exist for which ticker/year. Copy-pasting from the CVM portal breaks lineage. API UUIDs become primary keys.

ELT sketch

loader (Prefect/Airflow) → raw.cvm_documents_json → dbt stg → marts.filing_coverage

-- models/staging/stg_cvm_documents.sql
select
  document_id,
  ticker,
  document_type,
  filing_year,
  document_name,
  loaded_at
from {{ source('apicvm', 'cvm_documents') }}

Loader side (Python):

import os, requests

BASE = os.environ["APICVM_URL"]
H = {"Authorization": f"Bearer {os.environ['APICVM_KEY']}"}
r = requests.get(
    f"{BASE}/v1/documents",
    headers=H,
    params={"ticker": "CPFE3", "type": "DFP", "year": 2024, "perPage": 50},
)
r.raise_for_status()
# upsert r.json()["data"] into raw.cvm_documents

Mart ideas

filing_coverage: ticker × type × year with boolean present
fre_sections: filtered where name ILIKE '%risco%' (landed from name query param results)
Never hard-code counts in YAML seeds — refresh from API

Current limitations

dbt does not call apicvm; you need a loader
Text extraction stays outside dbt (async callbacks)
Empty API pages are valid corpus gaps

Next steps

Ready to integrate?

Get an API key and start querying Brazilian CVM filings programmatically.

Read the API docs Get API key