Delimited parsing is based a preexisting source definitions files and a set of properties. The properties specify the delimiters to use and other rules, such as whether there are column names and before values. The source definitions file determines the valid tables to be processed and the order and data type of the columns in the tables.

The format of the delimited message is:

METACOLSn[,COLNAMES]m[,COLBEFOREVALS]m,{COLVALUES}m\n

Where:

• There can be n metadata columns each followed by a field delimiter such as the comma shown in the format statement.

• There can be m column values. Each of these are preceded by a field delimiter such as a comma.

• The column name and before value are optional.

• Each record is terminated by an end of line delimiter, such as \n.

The message to be parsed must contain at least the header and metadata columns. If the number of columns is fewer than the number of header and meta columns, then the capture process terminates and provides an error message.

The remaining number of columns after the header and metadata columns are the column data for the corresponding table, specified in the order of the columns in the resolved metadata. Ideally, the number of table columns present in the message is exactly the same as the expected number of columns according to the metadata. However, missing columns in the message towards the end of message is allowed and the parser marks those last columns (not present in the rest of the message) as missing column data.

Although missing data is allowed from parser perspective, if the key @ column(s) is/are missing, then the capture process will also terminate.

Oracle GoldenGate primary key updates and unified updates are not supported. The only supported operations are inserts, updates, deletes, and truncates.

• Metadata Columns
  
• Parsing Properties
  
• Parsing Steps