CsvFlex

Loading DuckDB…

Tables

none

Open or drop CSV, TSV, JSON, Parquet or Excel files. Each one becomes a table you can query. Nothing is uploaded.

Import options

Delimiter Header row Encoding Excel sheet

Used for the next files you open, and by Re-import on a table.

Table

Columns

Group by

Work out

Columns from

Rows

Values

Filter

Sort by Limit

SQL

Choose a table.

Table Clean-up

Columns

Written as

Split on into columns

Separator New column

Find

Replace with

Regular expression (\1 in the replacement for a group) Match case

SQL

Choose a table.

The query failed

Drop a file to start

CSV, TSV, JSON, Parquet or Excel - millions of rows are fine. It is read by a database running in this tab and is never uploaded.

Column	Type	Empty	Distinct	Min	Max	Average

Run a query or pick a table to profile its columns.

Using CsvFlex

Opening files

Open or drop CSV, TSV, JSON, Parquet or Excel files. Each becomes a table named after the file (Orders 2026.csv → orders_2026). The delimiter, header, encoding and types are detected; change them under Import options and use Re-import.

Querying

SQL is DuckDB's: joins, window functions, PIVOT, SUMMARIZE. Query a table by name, or a file by its name in quotes: FROM 'orders.csv'.
Builder writes the SQL for you: pick columns, filter, group, pivot.
Clean up changes a table: trim, dedupe, fix dates, split, merge, find & replace. It shows the SQL and how many rows it will touch first.
Ctrl+Enter runs the SQL, or only the selected part of it.

The grid

Click a heading to sort; Shift+click to sort by more than one column. Drag a heading's edge to resize it.
Filter boxes take plain text (contains), or =x, !=x, >10, <=5, ^starts, $ends, empty, !empty.
Export downloads exactly what the grid shows, filtered and sorted, in any size.

After a reload

Your SQL and settings come back, but the files do not - they are never stored. Open them again.

Drop a file to start

Screen too small