repository.dtd

一个Struts写的网上购物系统,是一个不错的实例.

	A class-descriptor may contain user defined custom attribute elements.

    The insert-procedure, update-procedure and delete-procedure elements
    identify the procedure/function that is defined in the database which
    will handle the insertion, update or deletion of an instance of this
    class.  These elements are all optional.  If they are absent then basic
    sql statements (INSERT INTO xxx..., UPDATE xxx, DELETE FROM xxx) will
    be utilized.

  -->
<!ELEMENT class-descriptor
	((documentation?, extent-class+, attribute*) |
	(documentation?, object-cache?, extent-class*, field-descriptor+,
	 reference-descriptor*, collection-descriptor*,
     index-descriptor*, attribute*,
     insert-procedure?, update-procedure?, delete-procedure?))>


<!--
	The class attribute contains the full qualified name of the specified class.
	As this attribute is of type ID there can only be one class-descriptor per
	class.

	The isolation-level attribute specifies the transactional isolation to be
	used for this class on ODMG api level (this does NOT touch the jdbc-level
    of the connection).

	If the proxy attribute is set, proxies are used for all loading operations
	of instances of this class. If set to "dynamic", dynamic proxies are used.
	If set to another value this value is interpreted as the full-qualified
	name of the proxy class to use.

    If the proxy-prefetching-limit attribute (used with the proxy attribute)
    is set to the value > 0, the collections of objects of this class are materialized
    by groups of the specified size, say when user tries to access the first
    object of the collection, next proxy-prefetching-limit objects are loaded
    by one database query.
    Set this parameter to 0 if you want to turn this feature off.

	The schema attribute may contain the database schema owning the table
	mapped to this class.

	The table attribute specifies the table name this class is mapped to.

	The row-reader attribute may contain a full qualified class name.
	This class will be used as the RowReader implementation used to
	materialize instances of the persistent class.

	The accept-locks attribute specifies whether implicit locking should
	propagate to this class.  Currently relevant for the ODMG layer only.

	The optional initialization-method specifies a no-argument instance
	method that is invoked after reading an instance from a database row.
	It can be used to do initialization and validations.

	The optional factory-class specifies a factory-class that
	that is to be used instead of a no argument constructor.
	If the factory-class is specified the factory-method
	also must be defined.
	factory-method refers to a static no-argument method
	of the factory-class class.

	The refresh attribute can be set to true to force OJB to refresh
	instances when loaded from cache. It's set to false by default.

  -->
<!ATTLIST class-descriptor
	class ID #REQUIRED
	isolation-level (read-uncommitted | read-committed | repeatable-read |
	                 serializable | optimistic) "read-uncommitted"
	proxy CDATA #IMPLIED
	proxy-prefetching-limit CDATA #IMPLIED
	schema CDATA #IMPLIED
	table CDATA #IMPLIED
	row-reader CDATA #IMPLIED
    extends IDREF #IMPLIED
	accept-locks (true | false) "true"
	initialization-method CDATA #IMPLIED
	factory-class CDATA #IMPLIED
	factory-method CDATA #IMPLIED
	refresh (true | false) "false"
>


<!--
	An extent-class element is used to specify an implementing class or a
	derived class that belongs to the extent of all instances of the interface
	or base class.
  -->
<!ELEMENT extent-class (documentation?)>
<!--
	The class-ref attribute must contain a fully qualified classname.
	The repository file must contain a class-descriptor for this class.
  -->
<!ATTLIST extent-class
	class-ref IDREF #REQUIRED
>

<!--
  A field descriptor contains mapping info for a primitive typed
  attribute of a persistent class.

  A field descriptor may contain custom attribute elements.
  -->
<!ELEMENT field-descriptor (documentation?, attribute*)>
<!--
	<b>The id attribute is optional.</b> If not specified, OJB internally
	sorts field-descriptors according to their order of appearance in the
	repository file.
	If a different sort order is intended the id attribute may be used to
	hold a unique number identifying the decriptors position in the sequence of
	field-descriptors.
	The order of the numbers for the field-descriptors must correspond to
	the order of columns in the mapped table.

	The name attribute holds the name of the persistent classes attribute.
	If the PersistentFieldDefaultImpl is used there must be an attribute
	in the persistent class with this name.
	If the PersistentFieldPropertyImpl is used there must be a JavaBeans
	compliant property of this name.

	The table attribute may specify a table different from the mapped
	table for the persistent class. (currently not implemented).

	The column attribute specifies the column the persistent classes field
	is mapped to.

	The optional jdbc-type attribute specifies the JDBC type of the column.
	If not specified OJB tries to identify the JDBC type by inspecting the
	Java attribute by reflection.

	The primarykey specifies if the column is a primary key column.

	The nullable attribute specifies if the column may contain null values.

	The indexed attribute specifies if there is an index on this column

	The autoincrement attribute specifies if the values for the persistent
	attribute are automatically generated by OJB.

    The sequence-name attribute can be used to state explicitly a sequence
    name used by the sequence manager implementations. Check the docs/javadocs
    of the used sequence manager implementation to get information if this
    is a mandatory attribute. OJB standard sequence manager implementations
    build a sequence name by its own, if the attribute was not set.

	The locking attribute is set to true if the persistent attribute is
	used for optimistic locking. can only be set for TIMESTAMP and INTEGER
	columns.

	The updatelock attribute is set to false if the persistent attribute is
	used for optimistic locking AND the dbms should update the lock column
    itself. The default is true which means that when locking is true then
    OJB will update the locking fields. Can only be set for TIMESTAMP and INTEGER
	columns.

	The default-fetch attribute specifies whether the persistent attribute
	belongs to the JDO default fetch group.

	The conversion attribute contains a fully qualified class name.
	This class must implement the interface
	org.apache.ojb.accesslayer.conversions.FieldConversion.
	A FieldConversion can be used to implement conversions between Java-
	attributes and database columns.

	The length attribute can be used to specify a length setting, if
	required by the jdbc-type of the underlying database column.

	The precision attribute can be used to specify a precision setting, if
	required by the jdbc-type of the underlying database column.

	The scale attribute can be used to specify a scale setting, if
	required by the jdbc-type of the underlying database column.

	The access attribute specifies the accessibility of the field.
	"readonly" marks fields that are not to modified. "readwrite" marks
	fields that may be read and written to. "anonymous" marks anonymous fields.
	An anonymous field has a database representation (column) but no
	corresponding Java attribute. Hence the name of such a field does not
	refer to a Java attribute of the class, but is used as a unique
	identifier only.

  -->
<!ATTLIST field-descriptor
	id CDATA #IMPLIED
	name CDATA #REQUIRED
	table CDATA #IMPLIED
	column CDATA #REQUIRED
	jdbc-type (BIT | TINYINT | SMALLINT | INTEGER | BIGINT | DOUBLE |
	           FLOAT | REAL | NUMERIC | DECIMAL | CHAR | VARCHAR |
	           LONGVARCHAR | DATE | TIME | TIMESTAMP | BINARY |
	           VARBINARY | LONGVARBINARY | CLOB | BLOB | STRUCT |
               ARRAY | REF | BOOLEAN | DATALINK) #IMPLIED
	primarykey (true | false) "false"
	nullable (true | false) "true"
	indexed (true | false) "false"
	autoincrement (true | false) "false"
    sequence-name CDATA #IMPLIED
	locking (true | false) "false"
	update-lock (true | false) "true"
	default-fetch (true | false) "false"
	conversion CDATA #IMPLIED
	length CDATA #IMPLIED
	precision CDATA #IMPLIED
	scale CDATA #IMPLIED
	access (readonly | readwrite | anonymous) "readwrite"
>


<!--
  An attribute element allows arbitrary name/value pairs to
  be represented in the repository.
  -->
<!ELEMENT attribute (documentation?)>

<!--
	The attribute-name identifies the name of the attribute.
	The attribute-value identifies the value of the attribute.
-->
<!ATTLIST attribute
	attribute-name CDATA #REQUIRED
	attribute-value CDATA #REQUIRED
>

<!--
    A reference-descriptor contains mapping info for an attribute of a
    persistent class that is not primitive but references another
    persistent entity Object.

    A foreignkey element contains information on foreign key columns that
    implement the association on the database level.

    A reference-decriptor may contain user defined attribute elements.
  -->
<!ELEMENT reference-descriptor (documentation?, foreignkey+, attribute*)>
<!--
	The name attribute holds the name of the persistent classes attribute.
	If the PersistentFieldDefaultImpl is used there must be an attribute
	in the persistent class with this name.
	If the PersistentFieldPropertyImpl is used there must be a JavaBeans
	compliant property of this name.

	The class-ref attribute contains a fully qualified class name.
	This class is the Object type of the persistent reference attribute.
	As this is an IDREF there must be a class-descriptor for this class
	in the repository too.

	The proxy attribute can be set to true to specify that proxy based
	lazy loading should be used for this attribute.

    If the proxy-prefetching-limit attribute (used with the proxy attribute)
    is set to the value > 0, then loading of the reference for the first
    object of some collection causes loading of the references for
    the next proxy-prefetching-limit objects.
    Set this parameter to 0 if you want to turn this feature off.

	The refresh attribute can be set to true to force OJB to refresh
	object references on instance loading.

	The auto-retrieve attribute specifies whether OJB automatically retrieves
	this reference attribute on loading the persistent object.
	If set to false the reference attribute is set to null. In this case the
	user is responsible to fill the reference attribute.

	The auto-update attribute specifies whether OJB automatically stores
	this reference attribute on storing the persistent object.
	This attribute must be set to false if using the OTM, ODMG or JDO layer.

	The auto-delete attribute specifies whether OJB automatically deletes
	this reference attribute on deleting the persistent object.
	This attribute must be set to false if using the OTM, ODMG or JDO layer.

	The otm-dependent attribute specifies whether the OTM layer automatically
    creates the referred object or deletes it if the reference field is set to null.
    Also otm-dependent references behave as if auto-update and auto-delete
    were set to true, but the auto-update and auto-delete attributes themself
    must be always set to false for use with OTM layer.

  -->
<!ATTLIST reference-descriptor
	name CDATA #REQUIRED
	class-ref IDREF #REQUIRED

	proxy (true | false) "false"
	proxy-prefetching-limit CDATA #IMPLIED
	refresh (true | false) "false"

	auto-retrieve (true | false) "true"
	auto-update (true | false) "false"
	auto-delete (true | false) "false"
	otm-dependent (true | false) "false"
>

<!--
    A foreignkey element contains information on a foreign-key persistent
    attribute that implement the association on the database level.
  -->
<!ELEMENT foreignkey (documentation?)>
<!--
	The field-id-ref contains the id attribute of the field-descriptor
	used as a foreign key.
  -->
<!ATTLIST foreignkey
	field-id-ref CDATA #IMPLIED
	field-ref CDATA #IMPLIED
>