CSV Format Options

​

Overview

By default, the platform auto-detects the CSV delimiter and reads column names from the first row of the file. For most CSV files, no additional configuration is needed. However, some data providers deliver CSV files without a header row — every row contains data, and there is no first row defining column names. In these cases, you can specify explicit column names so the platform knows how to label each field.

​

CSV Options

The following options are available when the detected (or manually set) format is CSV:

​

When to Use csvColumns

Use csvColumns when your CSV file has no header row. Without it, the platform reads the first row as column names, which means:

• The first row of actual data is consumed as headers instead of being ingested.
• Column names change unpredictably each time the file is refreshed, breaking the Field Mapping.

​

Example

Consider a stock feed delivered without headers: Without csvColumns, the platform treats ABC123, STORE01, and 42 as column names. The second row becomes the only data row, and column names change on the next delivery. With csvColumns set to ["id", "store_code", "quantite"], both rows are ingested as data with stable, meaningful column names:

​

Configuring csvColumns

To set explicit column names, include csvColumns in the format options of the source configuration when calling the configure_sources step via the Datasources API: After the Datasource is configured, verify that the Field Mapping displays the expected column names and that all rows are ingested as data.

​

Applicable Source Types

The csvColumns option is available on all file-based connectors that support CSV:

• URL
• File Upload
• FTP / SFTP
• Google Cloud Storage
• Amazon S3