Skip to main content

sql_column

Purpose​

sql_column extracts one cell of a Postgres or MySQL result set, so transforms such as date or rand_string can fix up timestamps and generated ids in mock responses the same way they do in JSON bodies.

Usage​

"extractor": {
  "type": "sql_column",
  "config": {
    "row": "1",
    "name": "created_at"
  }
}
ParameterRequiredDescription
rowNoThe row, counting from 1 across every result set in the response. Defaults to 1.
columnOne of column or nameThe column position, counting from 1.
nameOne of column or nameThe column name.

It applies to responses only. NULL extracts as an empty value and stays NULL when a transform passes it through unchanged.

Limits​

  • Only text-format cells can be read and changed. Postgres drivers such as pgx often ask for binary results on prepared statements, and the extractor skips those cells.
  • Column names are only available where the response carries them: a Postgres simple query or any MySQL result. A Postgres prepared statement describes its columns in a separate message, so select its columns by position.
  • A MySQL prepared statement returns binary rows, where each cell keeps its column's type. A new value that does not fit the type, such as text in an integer column, is refused and the cell keeps its recorded value.
  • A name selects the first column with that name. When a join returns two columns with the same name, select the later ones by position.

Example​

Shift a recorded timestamp to the time of the replay:

sql_column(row=1,name=created_at) -> date(layout=auto)