bruno-portfolio / agrobr

What is agrobr?

agrobr is a Python package that pulls Brazilian agricultural data from 13 public sources like CEPEA, CONAB, IBGE, USDA, and ABIove, delivering clean, standardized, validated pandas (or Polars) DataFrames in one async line of code. It handles prices, crop yields, exports, rural credit, and fertilizer deliveries—solving the mess of scraping inconsistent sites, parsing PDFs/Excel, and ensuring data quality. Install with pip install agrobr (add [all] for extras like pdfplumber or BigQuery fallback), then grab soja prices via datasets.preco_diario("soja").

Why is it gaining traction?

Unlike manual scripts or fragmented tools like agrobrain or agrobricks, agrobr offers a semantic layer with automatic fallbacks (CEPEA to Noticias Agricolas), deterministic mode for reproducible pipelines, and DuckDB caching that builds historical datasets over time. CLI commands like agrobr cepea indicador soja --ultimo make quick checks dead simple, while contracts guarantee schema stability and health checks monitor source uptime. Async-first design scales for Airflow/Prefect without boilerplate.

Who should use this?

Agri-tech data engineers building commodity dashboards, quant traders modeling soja/milho prices, or researchers analyzing CONAB safras and IBGE PAM data. Ideal for devs tired of desinstalar pacote python scraps and instalar pacote python for each source, especially those needing validated exports for reports or ML features.

Verdict

Strong prototype for Brazilian ag data (docs shine, tests pass daily), but 13 stars and 1.0% credibility score signal early maturity—fine for prototyping, monitor for production scale. pip install agrobr now if you're in commodities.

(198 words)