The input data file should be a CSV-formatted text file. A comma (“,”) will be used to separate each column. A carriage-return+line-feed character combination will be used to terminate each row. If a carriage-return+line-feed character combination must be included as part of the data of a single column, the column’s data should be enclosed in double-quotes. The same is true if a comma must be included as part of the data of a column. If a double-quote character must be included as part of the data of a column, then it should be “escaped” by being immediately proceeded by another doublequote character. (e.g. "Hello world" would be encoded as ""Hello World"" if the double-quotes were required as part of the value for a field.)

The file will consist of a header row followed by one or more data rows. The header row will explicitly state which fields will be updated (and, by inference, which tables). The header row will include fields from a set of five tables. The row does not need to list all the fields in a table, nor does it need to reference each available table; it should only list as many fields as will be imported. The order in which it lists these fields will dictate the order in which the data will appear in the subsequent data rows.

Immediately following the header row there should be one or more data rows. Each data row shall contain the necessary command, key and data values needed to import one row of data into the database table(s). There must be one value column for each field-definition column declared in the header (though the columns can be left blank when appropriate).