Numbered Auth Notes, Finish/Test Connection wording, optional file Auth. Wizard downloads DuckDB JDBC 1.4.5.0 (LTS) from Maven and the Titanic Parquet sample from GitHub. In-memory URL plus SELECT from the sample file. Initial version. Query Parquet with DuckDB JDBC from Power BI, SQL Server Linked Server via ZappySys Data Gateway, Crystal Reports, SSRS, Azure Data Factory, and other ODBC tools.

  1. The wizard downloads DuckDB JDBC 1.4.5.0 (LTS) and the Titanic sample Parquet into {DownloadFolder}.
  2. This Auth is in-memory (jdbc:duckdb:memory:). After Finish, query the sample with SELECT * FROM "{DownloadFolder}\titanic.parquet".
  3. Use the File Auth if you want a persistent .duckdb database instead of memory.
  4. Finish the wizard. Then click Test Connection on the DSN Main UI (the wizard does not run Test Connection).
  5. Done. You can query from C#, Python, PowerShell, and ODBC apps.
]]> Persistent DuckDB file from Power BI, SQL Server Linked Server via ZappySys Data Gateway, Crystal Reports, SSRS, Azure Data Factory, and other ODBC tools.

  1. The wizard still downloads the JDBC JAR (and Titanic sample) into {DownloadFolder}.
  2. After Finish, point the JDBC URL at your .duckdb file if you do not want mydb.duckdb.
  3. You can still query Parquet with SELECT * FROM '{DownloadFolder}\titanic.parquet'.
  4. Finish the wizard. Then click Test Connection on the DSN Main UI (the wizard does not run Test Connection).
  5. Done. You can query from C#, Python, PowerShell, and ODBC apps.
]]> Reads the Titanic sample Parquet the wizard downloads next to the JDBC JAR ({DownloadFolder}\titanic.parquet). Connection is in-memory; the path is the data source.

]]> Reads a remote Parquet file directly over HTTPS. Use read_parquet() with a URL; DuckDB fetches and queries the file without a local copy. Returns all rows and columns from the file.

]]> Reads a local Parquet file by specifying its path in the FROM clause. DuckDB automatically uses the optimized Parquet engine. Replace the path with your file location.

]]> Reads one or more Parquet files using read_parquet() with a path and optional arguments such as filename=TRUE (add a source filename column) and binary_as_string=TRUE. Wildcards in the path match multiple files.

]]> Shows the schema of a Parquet file: column names and data types. Use DESCRIBE with a SELECT from the file path. Useful before writing queries or mapping columns.

]]> Reads multiple Parquet files with the same schema by passing a list of paths to read_parquet(). Returns a single result set. Use when you have a fixed set of files rather than a wildcard pattern.

]]> Reads all Parquet files matching a wildcard pattern (e.g. *.parquet) by putting the pattern in the FROM clause. DuckDB treats them as one logical table. Simpler than read_parquet() when you only need a path pattern.

]]> Reads multiple Parquet files via a wildcard and adds a filename column indicating the source file for each row. In DuckDB 1.3.0 and later this is the default when using read_parquet() with multiple files.

]]> Reads all matching Parquet files from several folders by passing a list of wildcard paths to read_parquet(). Results are merged into one dataset. Use when data is split across directories.

]]> Returns detailed Parquet metadata: schema, row groups, and column statistics. Use parquet_metadata() with a file path or URL. Helpful for tuning or understanding file layout.

]]> Returns low-level Parquet file metadata: row group details, encodings, and column-level statistics. Use parquet_file_metadata() when you need implementation-level details for debugging or optimization.

]]> Reads user-defined key-value metadata from the Parquet file footer. Use parquet_kv_metadata() with the file path when the file stores custom metadata (e.g. from a writer that writes key-value pairs).

]]> Returns the Parquet file schema: column names, types, and nested structure. Use parquet_schema() when you need the schema definition without reading data. Complements DESCRIBE SELECT for programmatic use.

]]> Writes the result of a query or table to a Parquet file. Use COPY with a SELECT (or table name) and a path; specify FORMAT parquet. The target path must be writable.

]]> Writes data to a Parquet file with options such as STRING_DICTIONARY_PAGE_SIZE_LIMIT to control dictionary page size (default 1 MB). Use when you need to tune file size or compatibility.

]]> Writes a Parquet file using Zstandard (zstd) compression. Set COMPRESSION zstd and optionally COMPRESSION_LEVEL (1 is fastest). Use when you want smaller files and can trade write speed for size.

]]> Reads data from a CSV file with read_csv() and writes it to a Parquet file using COPY ... TO ... (FORMAT parquet). Use for one-off conversion or ETL; DuckDB handles type inference from CSV.

]]>