Loading Data

Load CSV, Parquet, JSON files from local files or remote URLs.

Supported Formats

  • CSV - Comma-separated values
  • Parquet - Columnar storage format
  • JSON - JSON and newline-delimited JSON (NDJSON)
  • Excel - .xlsx files (via extension)

Drag and Drop

The easiest way to load a file is to drag and drop it directly onto the application window. The file will be loaded into memory and appear in the schema explorer.

Loading from URLs

You can query remote files directly using DuckDB functions:

-- CSV from URL
SELECT * FROM read_csv('https://example.com/data.csv');

-- Parquet from URL
SELECT * FROM read_parquet('https://example.com/data.parquet');

-- JSON from URL
SELECT * FROM read_json('https://example.com/data.json');

Creating Tables

To persist data for multiple queries, create a table:

CREATE TABLE users AS
SELECT * FROM read_csv('users.csv');

-- Now query the table
SELECT * FROM users WHERE age > 30;

Auto Schema Detection

DuckDB automatically detects column names and data types from your files. For CSVs, it samples the first rows to infer types.

File Persistence

Files uploaded via the file picker button can be persisted across sessions using the File System Access API. Drag-and-drop files are always session-only and will need to be re-uploaded after a page refresh.