Hash Columns

Columns > Hash columns... adds one column holding a hash of the columns you pick, row by row. It is the usual way to build a hash key in a data warehouse, for example MD5(UPPER(TRIM(first_name)) || '|' || birth_date).

Any column type works. Each value is turned into text first, the way Octa shows it in the grid, then the values are joined and hashed.

Options

Option	What it does
Columns, in order	Pick columns with Add a column; move them with ^ / v, take one out with x. The order is part of the hash: a, b and b, a give different results.
Algorithm	MD5 (32 hex characters, the default), SHA-256 (64) or SHA-512 (128).
Delimiter	Put between the values, \| by default. Any text, also empty. Without one, ab + c and a + bc hash the same, which is why a delimiter is the default.
NULL as	The text that stands in for an empty cell. Empty by default; some teams use a marker such as <NULL> so that an empty cell and an empty string differ.
Trim whitespace	Strip spaces at the start and end of each value first.
Upper-case	Upper-case each value first, so abc and ABC hash the same.
New column	The name of the result, hash_<columns> by default. It must not exist yet.

The Preview shows the first rows' hashes. Hover one to see the exact text that was hashed. The hash is lowercase hex. Edit > Undo removes the column again.

On a live database

On a database tab that holds only part of its table, the database computes the hash, for every row:

• Add column reads the table again from the database with the hash filled in, so the selection, marks and Undo history start fresh. Every page you scroll to, and Load whole table, carries the hash too. Save or undo your changes in the tab first; until then Add column is greyed.
• The Preview comes from the database: a short pause after you change a setting, it shows the hashes of the first rows the database returns.
• Analyses that run on the database (Summary, Value frequency, the filters, sorting) see the hash column over every row, so finding duplicate keys covers the whole table.
• The hash cells cannot be typed into. Other changes to them, such as Find and replace, are never saved, and the next read from the database puts the database's hash back. Delete the column to remove it; Refresh reads the table fresh, without it.
• Each value is the database's own text of the cell. Dates and decimals can look different from the grid, so a hash can differ from the same row in a file. Trim whitespace strips spaces only on the database. SQL Server needs version 2019 or later (UTF-8 text).
• If the database refuses, the table shows why, with Try again and Use the loaded rows; with the loaded rows the hash column stays empty. So it does after switching Run analyses on the database off, which reads the plain table again.

Good to know

• A tab that holds only part of its source (a file cut at the row limit, or a database table with Run analyses on the database switched off) cannot be hashed yet: the result would cover the loaded rows only. Use Data > Load whole table... first.
• A database renders dates and decimals as text its own way, so the same values can hash differently in Octa and in a database query. When the two must match, cast to text yourself in SQL in the same format Octa shows.
• The same function is available from the command line as --hash-columns and to AI agents as the MCP tool hash_columns.