slug: embedded-files title: “Embedded: query local and remote files” description: “Run SQE’s engine in-process with sqe-cli –embedded and query CSV, JSON, and Parquet files directly with read_csv / read_json / read_parquet. No server, no catalog, local files or HTTPS URLs.”

Embedded: query local and remote files

SQE’s engine runs in-process. sqe-cli --embedded --memory starts the query engine and the file-reader table-valued functions in a single binary — no coordinator, no workers, no network listeners, no catalog. The read_csv, read_json, and read_parquet TVFs read files directly, whether they live on local disk or behind an HTTPS URL.

This is the fastest way to explore data with SQL. There is no Docker stack to bring up; this quickstart just runs the CLI.

How it works

• --embedded runs the engine in-process. --memory disables the persistent catalog — the session is ephemeral and nothing is written to disk.
• Sample data files (CSV, JSON, Parquet) in the data/ directory are mounted into the container at runtime.
• read_csv, read_json, and read_parquet accept a local path, an https:// URL, or an s3:// URI — the same object-store layer backs all three.
• run.sh exercises local CSV aggregation, a cross-format JOIN (CSV + Parquet), and a remote HTTPS Parquet read, and captures the output.

What it demonstrates

• Querying CSV, JSON, and Parquet files with SQL in a single binary, no server.
• Cross-format joins: read_csv(...) and read_parquet(...) in one query.
• Remote file reads over HTTPS: read_parquet('https://...') streams the file via range requests.
• The --memory flag: no state survives the process (for persistent local tables, see the embedded-sqlite-catalog quickstart).

Status: validated (2026-06-06).

Run it

Full sample data, queries, and captured output are in the repo:

→ quickstart/embedded-files/

cd quickstart/embedded-files
cp .env.example .env
./run.sh