Independent of the style of XML, the parsing process needs to determine:

• Transaction boundaries

• Operation entries and metadata including:

  • Table name

  • Operation type

  • Timestamp

• Column entries and metadata including:

  • Either the column name or index; if both are specified the system will check to see if the column with the specified data has the specified name.

  • Column before or after values, sometimes both.

This is done through a set of interrelated rules. For each type of XML message that is to be processed you name a rule that will be used to obtain the required data. For each of these named rules you add properties to:

• Specify the rule as a transaction, operation, or column rule type. Rules of any type are required to have a specified name and type.

• Specify the XPath expression to match to see if the rule is active for the document being processed. This is optional; if not defined the parser will match the node of the parent rule or the whole document if this is the first rule.

• List detailed rules (subrules) that are to be processed in the order listed. Which subrules are valid is determined by the rule type. Subrules are optional.

In the following example the top-level rule is defined as genericrule. It is a transaction type rule. Its subrules are defined in oprule and they are of the type operation.

xmlparser.rules=genericrule
xmlparser.rules.genericrule.type=tx
xmlparser.rules.genericrule.subrules=oprule
xmlparser.rules.oprule.type=op