kcud

Kcud is a simplified library that extracts string columns from tables in a database stored as a DuckDB database file. (UNDER CONSTRUCTION FOR MORE FEATURES!)

Dependencies

• DuckDB's C++ API (install it based on your OS)
• DuckDB's CLI (refer to this page)
  • If not working, then build from DuckDB's source
    • cd into duckdb.
    • Run make.
    • A build folder is created.
    • To run DuckDB as an in-memory database, run build/release/duckdb and you are done. To run DuckDB with a native database file (i.e., .db file), run build/release/duckdb <filename>.db. Refer to the How to Generate .db Files section for details related to generating database files for testing.

How to Generate .db Files

• Run DuckDB as an in-memory database.
• Ensure that the TPC-H extension is loaded in DuckDB. If not, run INSTALL tpch; and then LOAD tpch;.
• Run CALL dbgen(sf=<sf>);, where <sf> is the scale factor the TPC-H workload is generated.
• Run COPY (SELECT l_comment FROM lineitem) TO 'comment_sf<sf>.csv' (header, delimiter ','); to dump the lineitem.l_comment column to a CSV file.
• Run .exit to quit DuckDB and then re-run DuckDB, specifying the database file (no need to create one in advance, as DuckDB will do it if the file is not found).
• Run PRAGMA force_compression='uncompressed'; to disable compression on the string columns. This is for the purpose of benchmarking.
• Load the rows in the dumped CSV file with CREATE TABLE comment AS SELECT * FROM 'comment_sf<sf>.csv';. Now you have a table named comment that contains all strings from lineitem.l_comment column.