Rejected rows and unmapped columns

Guide · Data sources

Updated

Browse documentation
On this page

Earlier product guidance · version not specified

What the Engine refuses, and what it only reports

Missing required columns prevent admission of the affected source. Unmapped headers trigger warnings. Rejected rows identify their source file, row number and reason, and stay out of every number.

A required column is missing.

The affected source cannot be admitted, and the error names the column it needs. A required source blocks the run. An optional failed source may be excluded when another analytical source supplies admitted rows; see run and validation recovery. Nothing is published from a file the contract cannot read.

A column is not mapped.

The run continues and lists the columns it did not read, in a warning called input_columns_unmapped.

A row cannot be read.

The Engine refuses the row and names the file, the row number and the reason. The row stays out of every number.

A header has another name.

Copy the catalogue into a folder beside your configuration and add the name to the field’s column_map list. For a required field, replace the old name in required_fields. Point ADAPTERS_DIR at the folder.

The warnings by name

input_columns_unmapped, invalid_timestamp and end_before_start are the ones you meet first; the full list is under Warnings. The Reader preview uses a separate demonstration result; its Overview · Data quality view shows that result’s evidence. Inspect the downloadable sample’s three rejected rows and their reasons separately.