Input · Files

CSV Reader

Reads a CSV or other delimited text file, such as a TSV. Column types are detected for you, and you can override them or read every column as text.

Also known as: Import, read_csv, TSV

How it works

You give the node a file in Source: type a File path, or bind it to a column of the Loop's Current item so each iteration reads a different file. The column types are sniffed from the file; once the source has been previewed, Column types lists every column with its detected type, and you can override one with VARCHAR, BIGINT, DOUBLE, DATE, TIMESTAMP, TIME or BOOLEAN. Read all as text switches the sniffing off entirely, so every column arrives as text and you re-type what you need downstream with Cast Types. Send to AI as decides what an AI node downstream sees of this source: a few rows (5 by default, the first, a random pick or the last), a description you write, the schema, or nothing.

Inputs and outputs

Inputs

None — this node starts a workflow.

Outputs

  • data — Table

Settings

The Attributes panel of CSV Reader, as in the app: the controls work, but nothing runs — there is no data behind it.

Example

A customer export where postal codes lose their leading zeros. Keep the sniffed types and set the one column that matters:

ColumnDetectedOverride
customer_idBIGINTauto
postal_codeBIGINTVARCHAR
joinedDATEauto