XML Schema is a W3C recommendation to specify "schema"s for XML data, used to put constraints on otherwise extremely flexible XML data. The constraints include basic data types, structures, occurrences, uniqueness, and referential integrity, etc. DB2 9 for z/OS pureXML provides XML schema validation for XML data through a user-defined function (UDF) called DSN_XMLVALIDATE. The following provides some guidelines in using XML schema validation.
- An XML schema is not required for an XML column. An XML column can store any well-formed XML documents, and it contains untyped XML data, even if a document is validated using the DSN_XMLVALIDATE function. In other words, type annotations are not kept after schema validation.
- XML Schema validation can be performed in an application using any schema validation tools, or in DB2 by invoking DSN_XMLVALIDATE. To invoke schema validation in DB2, an XML schema has to be registered in DB2 XML Schema Repository (XSR). XSR consists of a set of DB2 tables storing XML schema info, and a set of stored procedures to manage the schemas. A special JDBC interface is also provided for schema registration.
- Setting up XSR requires Java V5, JDBC and Java stored procedure setup, which requires WLM setup. See the "XSR setup and troubleshooting" article in the DB2 for z/OS pureXML wiki page for references to the setup information and troubleshooting details.
- XSR tables and stored procedures are created during the installation or migration. Please customize XSR related settings by changing the installation and migration jobs DSNTIJSG and DSNTIJNX.
- XML Schema documents (XSDs) have to be provided physically to DB2 for schema registration. During registration, an SQL ID in the form of "SYSXSR.<name>" can be provided to reference the schema. The schema can also be referenced by the target namespace with optional schema location. Schema documents are compiled into a binary format for runtime use. The original schema documents and the binary representations generated at XSR_COMPLETE are stored in XSR table SYSIBM.XSROBJECTS and other XSR tables. Note that string IDs are used for XML target namespaces and schema locations in the tables.
- To validate a document against a schema, invoke DSN_XMLVALIDATE in one of the two forms: two-argument form and three-argument form. The two-argument form takes an instance document and the schema name (SQL ID), while the three-argument form takes an instance document and schema target namespace, with optional schema location. As of this writing, DSN_XMLVALIDATE function has to be wrapped in the XMLPARSE function, and has a 50MB size limit on a document. For example, the following INSERT statement inserts an XML document after validation against a schema named SYSXSR.PURCHASEORDER:
INSERT INTO MYTABLE VALUES(10, XMLPARSE(DOCUMENT DSN_XMLVALIDATE(xmllob, 'SYSXSR.PURCHASEORDER') ) );Validation may cause XML data to change if there are default values and normalization specified in the schema that are applicable to the instance document.
- DB2 invokes XLXP-C high-performance validating parser. However, validation is still 2 to 3 times as expensive as simple parsing in terms of CPU time. Avoid validation in DB2 to save CPU and improve performance if performance is a concern.
- To support XML schema versioning, either use a totally different target namespace for a new schema version (usually for major changes), or use the same target namespace with a different schema location (usually for minor changes). You have to register schema versions with different SQL names in the XSR. Usually target namespace and schema location are also used in instance documents to identify the schema version. DB2 does not have the capability to use the target namespace and schema location to identify a registered schema for validation yet.
- Use Command Line Processor (CLP) or IBM Data Studio to register and manage XML schemas. Either way invokes the underlying XSR stored procedures, and is more convenient than using the native XSR stored procedures. See XML QuickDemo for a simple example of using CLP to register a schema.
Contact us if you have any questions related to XML schema validation.
-Guogen (Gene) Zhang (GGZ)[Read More]