SUNAT: aduanet

Cobertura publicada

Registros

9,040,254

Última actualización

21 set. 2026

Frecuencia

weekly

Frescura

CURRENT

Identificador

ruc

Búsqueda por nombre

No publicada

Conteo y frescura consultados desde api.latinfo.dev/sources durante el build del 2026-09-28.

Fuente primaria

• Institution: SUNAT / ADUANET.
• Data: Weekly bulk files for definitive customs regimes. This importer stores recent import (MA, MAM) and export (X) line items keyed by juridical RUC.
• Format: HTML index with ZIP links. Each ZIP contains one DBF file encoded as Latin-1 / dBase.

Cómo se procesa

• Downloads the official ADUANET index page.
• Extracts only bulk ZIP links whose file names start with MA, MAM, or X.
• Downloads ZIPs with curl -4; ADUANET is slower and less reliable through Node HTTP from the Mint runner.
• Extracts the DBF with system unzip.
• Parses DBF records directly from the fixed-width dBase header and record layout.
• Stores only juridical RUCs (20*). Natural-person RUCs and other document types are skipped before build.
• Emits a multi-record TSV keyed by RUC, sorted with LC_ALL=C sort, then builds the standard Latinfo binary index and content-addressed manifest.
• Builds a second analytical output from the exact same canonical TSV: strict typed Parquet, partitioned by regime and declaration year, with raw ZIP lineage and an atomic lake manifest.
• Does not bypass CAPTCHA and does not run millions of form queries. It consumes the official bulk files published for download.
• Script: src/imports/pe-sunat-aduanet.ts.

Campos publicados

Limitaciones conocidas

• The official page only exposes roughly the last two months. Historical coverage requires archiving weekly from now or obtaining older official files.
• MB files are not ingested in v1 because they do not carry RUC directly; they require a declaration-level join to MA/MAM.
• Smoke IDs can age out if the company stops appearing in the rolling two-month window. Replace fixture anchors with current records after a new import if needed.
• ADUANET downloads can be slow. The importer caches ZIPs in /tmp/pe-sunat-aduanet-zips and validates cached ZIPs with unzip -tq.
• The official DBFs contain byte-identical duplicate lines. The lake removes only exact canonical duplicates under a bounded, documented policy; the raw files remain immutable.