Data Services Developer's Guide
|
 |
 |
|
 |
 |
A first step in creating data services for the BEA Aqualogic Data Services Platform (DSP) is to obtain metadata from physical data needed by your application.
This chapter describes this process, including the following topics:
Metadata is simply information about the structure of a data source. For example, a list of the tables and columns in a relational database is metadata. A list of operations in a Web service is metadata.
In DSP, a physical data service is based almost entirely on the introspection of physical data sources.
Figure 3-1 Data Services Available to the RTL Sample Application
Table 3-2 list the types of sources from which DSP can create metadata.
Table 3-2 Data Sources Available for Creating Data Service Metadata
When information about physical data is developed using the Metadata Import Wizard two things happen:
.ds) is created in your DSP-based project.extension.xsd), is created. This schema describes quite exactly the XML type of the data service. Such schemas are placed in a directory named schemas which is a sub-directory of your newly created data service.Figure 3-3 DSP Application Pane Displaying a Data Service and Its Schema Directory
You can import metadata on the data sources needed by your application using the DSP Metadata Import wizard. This wizard introspects available data sources and identifies data objects that can be rendered as data services and functions. Once created, physical data services become the building-blocks for queries and logical data services.
Data source metadata can be imported as Data Services Platform functions or procedures. For example, the following source resulted from importing a Web service operation:
(::pragma function <f:function xmlns:f="urn:annotations.ld.bea.com" kind="read" nativeName="getCustomerOrderByOrderID" nativeLevel1Container="ElecDBTest" nativeLevel2Container="ElecDBTestSoap" style="docu ment"/>::)
declare function f1:getCustomerOrderByOrderID($x1 as element(t1:getCustomerOrderByOrderID)) as schema-element(t1:getCustomerOrderByOrderIDResponse) external;
Notice that the imported Web service is described as a "read" function in the pragma. "External" refers to the fact that the schema is in a separate file. You can find a detailed description of source code annotations in "Understanding Data Services Platform Annotations" in the XQuery Reference Guide.
For some data sources such as Web services imported metadata represents functions which typically return void (in other words, these functions perform operations rather than returning data). Such routines are classified as side-effecting functions or, more formally, as DSP procedures. You also have the option of marking routines imported from certain data sources as procedures. (See Identifying DSP Procedures.)
The following source resulted from importing Web service metadata that includes an operation that has been identified as a side-effecting procedure:
(::pragma function <f:function xmlns:f="urn:annotations.ld.bea.com" kind="hasSideEffects" nativeName="setCustomerOrder" style="document"/>::)
declare function f1:setCustomerOrder($x1 as element(t3:setCustomerOrder)) as schema-element(t3:setCustomerOrderResponse) external;
In the above pragma the function is identified as "hasSideEffects".
Note: DSP procedures are only associated with physical data services and can only be created through the metadata import process. So, for example, attempting to add procedures to a logical data service through Source View will result in an error condition.
When you import source metadata for Web services, relational stored procedures, or Java functions you have an opportunity to identify the metadata that represents side-effecting routines. A typical example is a Web service that creates a new customer record. From the point of view of the data service such routines are procedures.
Procedures are not standalone; they always are part of a data service from the same data source.
When importing data from such sources the Metadata Import wizard automatically categorizes routines that return void as procedures. The reason for this is simply: if a routine does not return data it cannot inter-operate with other data service functions.
There are, however, routines that both return data and have side-effects; it is these routines which you need to identify as procedures during the metadata import process. Identification of such procedures provides the application developer with two key benefits:
Table 3-4 lists common DSP operations, identifying which operations are available or unavailable for data service procedures.
Table 3-4 Data Services Platform Scope of Procedures
Procedures greatly simplify the process of updating non-relational back-end data sources by providing an invokeProcedure( ) API. This API encapsulates the operational logic necessary to invoke relational stored procedures, Web services, or Java functions. In such cases update logic can be built into a back-end data source routine which, in turn, updates the data.
For information on updating non-relational sources and other special cases see "Enabling SDO Data Source Updates" in the Client Application Developer's Guide.
For an example showing how you can identify side-effecting procedures during the metadata import process see Importing Web Services Metadata.
You can obtain metadata on any relational data source available to the BEA WebLogic Platform. For details see the BEA Platform document entitled How Do I Connect a Database Control to a Database Such as SQL Server or Oracle.
Four types of metadata can be obtained from a relational data source:
Note: When using an XA transaction driver you need to mark your data source's connection pool to allow LocalTransaction in order for single database reads and updates to succeed.
For additional information in XA transaction adaptor settings see "Developing Adaptors" in BEA WebLogic Integration documentation: http://download.oracle.com/docs/cd/E13214_01/wli/docs81/devadapt/dbmssamp.html
To create metadata on relational tables and views follow these steps:
Figure 3-5 Selecting a Relational Source from the Import Metadata Wizard
Figure 3-6 Import Data Source Metadata Selection Dialog Box
For information on creating a new data source see Creating a New Data Source.
If you choose to select from an existing data source, several options are available (Figure 3-6).
If you choose to select all, a table will appear containing all the tables, views, and stored procedures in your data source organized by catalog and schema.
Sometimes you know exactly the objects in your data source that you want to turn into data services. Or your data source may be so large that a filter is needed. Or you may be looking for objects with specific naming characteristics (such as %audit2003%, a string which would retrieve all objects containing the enclosed string).
In such cases you can identify the exact parts of your relational source that you want to become data service candidates using standard JDBC wildcards. An underscore (_) creates a wildcard for an individual character. A percentage sign (%) indicates a wildcard for a string. Entries are case-sensitive.
For example, you could search for all tables starting with CUST with the entry: CUST%. Or, if you had a relational schema called ELECTRONICS, you could enter that term in the Schema field and retrieve all the tables, views, and stored procedure that are a part of that schema.
CUST%, PAY%
entered in the Tables/Views field retrieves all tables and views starting with either CUST or PAY.
Note: If no items are entered for a particular field, all matching items are retrieved. For example, if no filtering entry is made for the Procedure field, all stored procedures in the data object will be retrieved.
For relational tables and views you should choose either the Select all option or Selected data source objects.
You can also use wildcards to support importing metadata on internal stored procedures. For example, entering the following string as a stored procedure filter:
%TRIM%
retrieves metadata on the system stored procedure:
STANDARD.TRIM
In such a situation you would also want to make a nonsense entry in the Table/View field to avoid retrieving all tables and views in the database.
For details on stored procedures see Importing Stored Procedure-Based Metadata.
Allows you to enter an SQL statement that is used as the basis for creating a data service. See Using SQL to Import Metadata for details.
Most often you will work with existing data sources. However, if you choose New... the WLS DataSource Viewer appears (Figure 3-7). Using the DataSource Viewer you can create new data pools and sources.
Figure 3-7 BEA WebLogic Data Source Viewer
For details on using the DataSource Viewer see Configuring a Data Source in WebLogic Workshop documentation.
Only data sources that have set up through the BEA WebLogic Administration Console are available to a Data Services Platform application or project. In order for the BEA WebLogic Server used by DSP to access a particular relational data source you need to set up a JDBC connection pool and a JDBC data source.
http://download.oracle.com/docs/cd/E13222_01/wls/docs81/ConsoleHelp/domain_jdbcconnectionpool_config_general.html
http://download.oracle.com/docs/cd/E13222_01/wls/docs81/ConsoleHelp/domain_jdbcdatasource_config.html
Figure 3-8 Selecting a Data Source
Once you have selected a data source, you need to choose how you want to develop your metadata — by selecting all objects in the database, by filtering database objects, or by entering a SQL statement. (see Figure 3-6).
Once you have selected a data source and any optional filters, a list of available database objects appears.
Figure 3-9 Identifying Database Objects to be Used as Data Services
Using standard dialog commands you can add one or several tables to the list of selected data objects. To deselect a table, select that table in the right-hand column and click Remove.
A Search field is also available. This is useful for data sources which have many objects. Enter a search string, then click Search repeatedly to move through your list.
You can edit the file name to clarify the name or to avoid conflicts. Simply click on the name of the file and make any editing changes.
Database vendors variously support database catalogs and schemas. Table 3-11 describes this support for several major vendors.
Table 3-11 Vendor Support for Catalog and Schema Objects
When a source name is encountered that does not fit within XML naming conventions, default generated names are converted according to rules described by the SQLX standard. Generally speaking, an invalid XML name character is replaced by its hexadecimal escape sequence (having the form _xUUUU_).
For additional details see section 9.1 of the W3C draft version of this standard:
http://www.sqlx.org/SQL-XML-documents/5WD-14-XML-2003-12.pdf
Once you have created your data services you are ready to start constructing logical views on your physical data. See Designing Data Services. and Modeling Data Services.
Enterprise databases utilize stored procedures to improve query performance, manage and schedule data operations, enhance security, and so forth. You can import metadata based on stored procedures. Each stored procedure becomes a data service.
Note: Refer to your database documentation for details on managing stored procedures.
Stored procedures are essentially database objects that logically group a set of SQL and native database programming language statements together to perform a specific task.
Table 3-12 defines some commonly used terms as they apply to this discussion of stored procedures.
Table 3-12 Terms Commonly Used When Discussing Stored Procedures
Imported stored procedure metadata is quite similar to imported metadata for relational tables and views. The initial three steps for importing stored procedures are the same as importing any relational metadata (described under Importing Relational Table and View Metadata).
Note: If a stored procedure has only one return value and the value is either simple type or a RowSet which is mapping to an existing schema, no schema file created.
You can select any combination of database tables, views, and stored procedures. If you select one or several stored procedures, the Metadata Import wizard will guide you through the additional steps required to turn a stored procedure into a data service. These steps are:
Figure 3-13 Selecting Stored Procedure Database Objects to Import
Figure 3-14 Configuring a Stored Procedure in Pre-editing Mode
Data objects in the stored procedure that cannot be identified by the Metadata Import wizard will appear in red, without a datatype. In such cases you need to enter Edit mode (click the Edit button) to identify the data type.
Your goal in correcting an "<unknown>" condition associated with a stored procedure (Figure 3-14) is to bring the metadata obtained by the import wizard into conformance with the actual metadata of the stored procedure. In some cases this will be by correcting the location of the return type. In others you will need to adjust the type associated with an element of the procedure or add elements that were not found during the initial introspection of the stored procedure.
Figure 3-15 Stored Procedure in Editing Mode (with Callouts)
Each element in a stored procedure is associated with a type. If the item is a simple type, you can simply choose from the pop-up list of types.
Figure 3-16 Changing the Type of an Element in a Stored Procedure
If the type is complex, you may need to supply an appropriate schema. Click on the schema location button and either enter a schema path name or browse to a schema. The schema must reside in your application.
After selecting a schema, both the path to the schema file and the URI appear. For example:
http://temp.openuri.org/schemas/Customer.xsd}CUSTOMER
Not all databases support rowsets. In addition, JDBC does not report information related to defined rowsets. In order to create data services from stored procedures that use rowset information, supply the correct ordinal (matching number) and a schema. If the schema has multiple global elements, you can select the one you want from the Type column. Otherwise the type will be the first global element in your schema file.
The order of rowset information is significant; it must match the order in your data source. Use the Move Up / Move Down commands to adjust the ordinal number assigned to the rowset.
Complete the importation of your procedures by reviewing and accepting items in the Summary screen (see step 4. in Importing Relational Table and View Metadata for details).
Note: XML types in data services generated from stored procedures do not display native types. However, you can view the native type in the Source View pragma (see Working with XQuery Source).
Handling Stored Procedure Rowsets
A rowset type is a complex type. The name of the rowset type can be:
The rowset type contains a sequence of a repeatable elements (for example called CUSTOMER) with the fields of the rowset.
Note: All rowset-type definitions must conform to this structure.
In some cases the Metadata Import wizard can automatically detect the structure of a rowset and create an element structure. However, if the structure is unknown, you will need to provide it through the wizard.
It is often convenient to leverage independent routines as part of managing enterprise information through a data service. An obvious example would be to leverage standalone update or security functions through data services. Such functions have noXML type; in fact they typically return nothing (or void). Instead the data service knows that they have side-effects and are associated as procedures with a data service of the same data source.
Stored procedures are very often side-effecting from the perspective of the data service, since they perform internal operations on data. In such cases all you need to do is identify the stored procedures as a DSP procedure during the metadata import process.
After you have identified the stored procedures that you want to add to your data service or XML file library (XFL), you also have an opportunity to identify which of these should be identified as DSP procedures.
Figure 3-17 Identifying Stored Procedures Having Side Effects
Note: DSP procedures based around atomic (simple) types are collected in an identified XML function library (XFL) file. Other procedures need to be associated with a data service that is local to your DSP-enabled project.
You can import metadata for an internal stored procedures. See Filter Data Source Objects for details.
Only the most recent version of a stored procedure can be imported into DSP. For this reason you cannot identify a version number when importing a stored procedure through the Metadata Import wizard. Similarly, adding a version number to DSP source will result in a query exception.
Each database vendor approaches stored procedures differently. XQuery support limitations are, in general, due to JDBC driver limitations.
DSP does not support rowset as an input parameter.
Table 3-18 summarizes DSP support for Oracle database procedures.
Table 3-18 Support for Oracle Store Procedures
|
Any Oracle PL/SQL data type except those listed below: Note: When defining function signatures, note that the Oracle %TYPE and %ROWTYPE types must be translated to XQuery types that match the true types underlying the stored procedure's %TYPE and %ROWTYPE declarations. %TYPE declarations map to simple types; %ROWTYPE declarations map to rowset types. For a list of database types supported by DSP see Relational Data Types-to-Metadata Conversion. |
|
|
Oracle supports returning PL/SQL data types such as NUMBER, VARCHAR, %TYPE, and %ROWTYPE as parameters. |
|
|
The following identifies limitations associated with importing Oracle database procedure metadata.
|
Table 3-19 summarizes DSP support for Sybase SQL Server database procedures.
Table 3-19 Support for Sybase Stored Procedures
|
For the complete list of database types supported by DSP see Relational Data Types-to-Metadata Conversion. |
|
|
Sybase functions supports returning a single value or a table. Procedures return data in the following ways:
|
|
|
The following identifies limitations associated with importing Sybase database procedure metadata:
|
Table 3-20 summarizes DSP support for IBM DB2 database procedures.
Table 3-20 Support for IBM Store Procedures
|
Each function is also categorized as a scalar, column, row, or table function. |
|
|
For the complete list of database types supported by DSP see Relational Data Types-to-Metadata Conversion. |
|
|
DB2 supports returning a single value, a row of values, or a table. |
|
|
The following identifies limitations associated with importing DB2 database procedure metadata: |
Table 3-21 summarizes DSP support for Informix database stored procedures.
Table 3-21 Support for Informix Stored Procedures
|
For the complete list of database types supported by DSP see Relational Data Types-to-Metadata Conversion. |
|
|
Informix supports returning single value, multiple values, and rowsets. |
|
|
Informix treats return value(s) from functions or procedures as a rowset. For this reason a rowset needs to be defined for the return value(s). The following limitations have been identified: Informix Native Driver Limitations
BEA WebLogic Driver Limitations
Due to the limitations described above, the following approach is suggested for importing Informix stored procedure metadata: 2. Define a schema that matches the return value structure (using the same approach as external schemas for other databases). |
Table 3-22 summarizes DSP support for Microsoft SQL Server database procedures.
Table 3-22 DSP Support for Microsoft SQL Server Stored Procedures
One of the relational import metadata options (see Figure 3-6) is to use an SQL statement to customize introspection of a data source. If you select this option the SQL Statement dialog appears.
Figure 3-23 SQL Statement Dialog Box
You can type or paste your SELECT statement into the statement box (Figure 3-23), indicating parameters with a "?" question-mark symbol. Using one of the DSP data samples, the following SELECT statement can be used:
SELECT * FROM RTLCUSTOMER.CUSTOMER WHERE CUSTOMER_ID = ?
RTLCUSTOMER is a schema in the data source, CUSTOMER is, in this case, a table.
For the parameter field, you would need to select a data type. In this case, CHAR or VARCHAR.
The next step is to assign a data service name.
When you run your query under Test View, you will need to supply the parameter in order for the query to run successfully.
Once you have entered your SQL statement and any required parameters click Next to change or verify the name and location of your new data service.
Figure 3-24 Relational SQL Statement Imported Data Summary Screen
The imported data summary screen identifies a proposed name for your new data service.
The final steps are no different than you used to create a data service from a table or view.
The following table shows how data types provided by various relational databases are converted into XQuery data types. Types are listed in alphabetical order.
Table 3-25 Relational Data Types and Their XQuery Counterparts
A Web service is a self-contained, platform-independent unit of business logic that is accessible through application adaptors, as well as standards-based Internet protocols such as HTTP or SOAP.
Web services greatly facilitate application-to-application communication. As such they are increasingly central to enterprise data resources. A familiar example of an externalized Web service is a frequent-update weather portlet or stock quotes portlet that can easily be integrated into a Web application. Similarly, a Web service can be effectively used to track a drop shipment order from a seller to a manufacturer.
Note: Multi-dimensional arrays in RPC mode are not supported.
Creating a data service based on a Web service definition (schema) is similar to importing relational data source metadata (see Importing Relational Table and View Metadata).
Here are the Web service-specific steps involved:
Note: For the purpose of showing how to import Web service metadata a WSDL file from the RTLApp sample is used for the remaining steps. If you are following these instructions enter the following into the URI field to access the WSDL included with RTLApp:
http://localhost:7001/ElecWS/controls/ElecDBTestContract.wsdl
Note: Imported operations returning void are automatically imported as DSP procedures. You can identify other operations as procedures using the Select Side Effect Procedures dialog (Figure 3-27).
It is often convenient to leverage side-effecting operations as part of managing enterprise information through a data service. An obvious example would be to manage standalone update or security functions through data services. The data service registers that such operations have side-effects.
Procedures are not standalone; they always are part of a data service from the same data source.
Web services are side-effecting from the perspective of the data service even when they do return data. In such cases, you need to associate the Web service operation with a data service during the metadata import process.
Figure 3-27 Marking Imported Operations DSP Procedures
Procedures must be associated with a data service that is local to a DSP-enabled project.
Figure 3-28 Identifying Web Service Operations to be Used as Data Services
Using standard dialog editing commands you can select one or several operations to be added to the list of selected Web service operations. To deselect an operation, click on it, then click Remove. Or choose Remove All to return to the initial state.
Figure 3-29 Web Services Imported Data Summary Screen
The summary screen shown in Figure 3-29:
Even if there are no name conflicts you may want to change a data service name for clarity. Simply click on the name of the data service and enter the new name.
Note: Web Service functions identified as side-effecting procedures must be associated with a data service based on the same WSDL.
Note: When importing a Web service operation that itself has one or more dependent (or referenced) schemas, the Metadata Import wizard creates second-level schemas according to internal naming conventions. If several operations reference the same secondary schemas, the generated name for the secondary schema may change if you re-import or synchronize with the Web service.
If you are interested in trying the Metadata Import wizard with an internet Web service URI, the following page (available as of this writing) provides sample URIs:
http://www.strikeiron.com/BrowseMarketplace.aspx?c=14&m=1
Simply select a topic and navigate to a page showing the sample WSDL address such as:
http://ws.strikeiron.com/SwanandMokashi/StockQuotes?WSDL
Copy the string into the Web service URI field and click Next to select the operations want to turn into sample data services or procedures.
Another external Web service that can be used to test metadata import can be located at:
http://www.whitemesa.net/wsdl/std/echoheadersvc.wsdl
You can create metadata based on custom Java functions. When you use the Metadata Import wizard to introspect a .class file, metadata is created around both complex and simple types. Complex types become data services while simple Java routines are converted into XQueries and placed in an XQuery function library (XFL). In Source View (see Working with XQuery Source) a pragma is created that defines the function signature and relevant schema type for complex types such as Java classes and elements.
In the RTLApp DataServices/Demo directory there is a sample that can be used to illustrate Java function metadata import.
Your Java file can contains two types of functions:
Before you can create metadata on a custom Java function you must create a Java class containing both schema and function information. A detailed example is described in Creating XMLBean Support for Java Functions.
Importing Java function metadata is similar to importing relational data source metadata (see Importing Relational Table and View Metadata). Here are the Java function-specific steps involved:
.class file from your .java function and place it in your application's library.Figure 3-30 Selecting a Java Function as the Data Source
.class file must be in your BEA WebLogic application. You can browse to your file or enter a fully-qualified path name starting from the root directory of your DSP-based project.Figure 3-31 Specifying a Java Class File for Metadata Import
Figure 3-32 Selecting Java Functions to Become Either Data Services or XFL Functions
It is often convenient to leverage independent routines as part of managing enterprise information through a data service. An obvious example would be to leverage standalone update or security functions through data services. Such functions have noXML type; in fact they typically return nothing (or void). Instead the data service knows that the routine has side-effects, but those effects are not transparent to the service. DSP procedures can also be thought of as side-effecting functions.
Java functions are "side-effecting" from the perspective of the data service when they perform internal operations on data.
After you have identified the Java functions that you want to add to your project, you can also identify which, if any, of these should be treated as DSP procedures (Figure 3-33). In the case of main(), the Metadata Import wizard detects that it returns void so it is already marked as a procedure.
Figure 3-33 Marking Java Functions as DSP Procedures
Functions based around atomic (simple) types are collected in an identified XML function library (XFL) file.
Note: Side-effecting procedures must to be associated with a data service that is from the same data source. In this case, the source is your Java file. In other words, in order to specify a Java function as a procedure, a function in the same file that returns a complex element must either be created at the same time or already exist in your project.
Figure 3-34 Java Function Imported Data Summary Screen
You can edit the proposed data service name either for clarity or to avoid conflicts with other existing or planned data services. All functions returning complex data types will be in the same data service. Click on the proposed data service name to change it.
Procedures must be associated with a data service that draws data from the same data source (Java file). In the sample shown in Figure 3-34, the only available data service is PRODUCTS (or whatever name you choose).
If there are existing XFL files in your project you have the option of adding atomic functions to that library or creating a new library for them. All the Java file atomic functions are located in the same library.
Before you can import Java function metadata, you need to create a .class file that contains XMLBean classes based on global elements and compiled versions of your Java functions. To do this, you first create XMLBean classes based on a schema of your data. There are several ways to accomplish this. In the example in this section you create a WebLogic Workshop project of type Schema.
Generally speaking, the process involves:
.xsd file) representing the shape of the global elements invoked by your function..class file, if under a DSP-based project, or you can add the JAR file from a Java project to the Library folder of your application..class file.In the following example there are a number of custom functions in a .java file called FuncData.java. In the RTLApp this file can be found at:
ld:DataServices/Demo/Java/FuncData.java
Some functions in this file return primitive data types, while others return a complex element. The complex element representing the data to be introspected is in a schema file called FuncData.xsd.
|
Contains Java functions to be converted into data service query functions. Also contains as small data sample. |
|
|
Contains a schema for the complex element identified in |
The schema file can be found at:
ld:DataServices/Demo/Java/schema/FuncData.xsd
To simplify the example a small data set is included in the .java file as a string.
The following steps will create a data service from the Java functions in FuncData.java:
Importing a schema file into a schema project automatically starts the project build process.
When successful, XMLBean classes are created for each function in your Java file and placed in a JAR file called JavaFunctSchema.jar
The JAR file is located in the Libraries section of your application.
ld:DataServices/Demo/Java folder in the RTLApp and select FuncData.java for import. Click Import.The JAR file named for your DSP-based project is updated to include a.class file named FuncData.class; It is this file that can be introspected by the Metadata Import wizard. The file is located in a folder named JavaFuncMetadata in the Library section of your application.
Figure 3-35 Class File Generated Java Function XML Beans
The .java file used in this example contains both functions and data. More typically, your routine will access data through a data import function.
The first function in Listing 3-1 simply retrieves the first element in an array of PRODUCTS. The second returns the entire array.
Listing 3-1 JavaFunc.java getFirstPRODUCT( ) and getAllPRODUCTS( ) Functions
public class JavaFunc {
...
public static noNamespace.PRODUCTSDocument.PRODUCTS getFirstProduct(){
noNamespace.PRODUCTSDocument.PRODUCTS products = null;
try{
noNamespace.DbDocument dbDoc = noNamespace.DbDocument.Factory.parse(testCustomer);
products = dbDoc.getDb().getPRODUCTSArray(1);
//return products;
}catch(Exception e){
e.printStackTrace();
}
return products;
}
public static noNamespace.PRODUCTSDocument.PRODUCTS[] getAllProducts(){
noNamespace.PRODUCTSDocument.PRODUCTS[] products = null;
try{
noNamespace.DbDocument dbDoc = noNamespace.DbDocument.Factory.parse(testCustomer);
products = dbDoc.getDb().getPRODUCTSArray();
//return products;
}catch(Exception e){
e.printStackTrace();
}
return products;
}
}
The schema used to create XMLBeans is shown in Listing 3-2. It simply models the structure of the complex element; it could have been obtained by first introspecting the data directly.
Listing 3-2 B-PTest.xsd Model Complex Element Parsed by Java Function
<xs:schema elementFormDefault="qualified" xmlns:xs="http://www.w3.org/2001/XMLSchema">
<xs:element name="db">
<xs:complexType>
<xs:sequence>
<xs:element ref="PRODUCTS" maxOccurs="unbounded"/>
</xs:sequence>
</xs:complexType>
</xs:element>
<xs:element name="AVERAGE_SERVICE_COST" type="xs:decimal"/>
<xs:element name="LIST_PRICE" type="xs:decimal"/>
<xs:element name="MANUFACTURER" type="xs:string"/>
<xs:element name="PRODUCTS">
<xs:complexType>
<xs:sequence>
<xs:element ref="PRODUCT_NAME"/>
<xs:element ref="MANUFACTURER"/>
<xs:element ref="LIST_PRICE"/>
<xs:element ref="PRODUCT_DESCRIPTION"/>
<xs:element ref="AVERAGE_SERVICE_COST"/>
</xs:sequence>
</xs:complexType>
</xs:element>
<xs:element name="PRODUCT_DESCRIPTION" type="xs:string"/>
<xs:element name="PRODUCT_NAME" type="xs:string"/>
</xs:schema>
Java functions require that an element returned (as specified in the return signature) come from a valid XML document. A valid XML document has a single root element with zero or more children, and its content matches the schema referred.
Listing 3-3 Approach When Data is Retrieved Through a Document
public static noNamespace.PRODUCTSDocument.PRODUCTS getNextProduct(){
// create the dbDocument (the root)
noNamespace.DbDocument dbDoc = noNamespace.DbDocument.Factory.newInstance();
// the db element from it
noNamespace.DbDocument.Db db = dbDoc.addNewDb();
// get the PRODUCTS element
PRODUCTS product = db.addNewPRODUCTS();
//.. create the children
product.setPRODUCTNAME("productName");
product.setMANUFACTURER("Manufacturer");
product.setLISTPRICE(BigDecimal.valueOf((long)12.22));
product.setPRODUCTDESCRIPTION("Product Description");
product.setAVERAGESERVICECOST(BigDecimal.valueOf((long)122.22));
// .. update children of db
db.setPRODUCTSArray(0,product);
// .. update the document with db
dbDoc.setDb(db);
//.. now dbDoc is a valid document with db and is children.
// we are interested in PRODUCTS which is a child of db.
// Hence always create a valid document before processing the
children.
// Just creating the child element and returning it, is not
// enough, since it does not mean the document is valid.
// The child needs to come from a valid document, which is created
// for the global element only.
return dbDoc.getDb().getPRODUCTSArray(0);
}
In DSP, user-defined functions are typically Java classes. The following are supported:
In order to support this functionality, the Metadata Import wizard supports marshalling and unmarshalling so that token iterators in Java are converted to XML and vice-versa.
Functions you create should be defined as static Java functions. The Java method name when used in an XQuery will be the XQuery function name qualified with a namespace.
Table 3-36 shows the casting algorithms for simple Java types, schema types and XQuery types.
Table 3-36 Simple Java Types and XQuery Counterparts
Java functions can also consume variables of XMLBean type that are generated by processing a schema via XMLBeans. The classes generated by XMLBeans can be referred in a Java function as parameters or return types.
The elements or types referred to in the schema should be global elements because these are the only types in XMLBeans that have static parse methods defined.
The next section provides additional code samples that illustrate how Java functions are used by the Metadata Import wizard to create data services.
In order to create data services or members of an XQuery function library, you would first start with a Java function.
As an example, the Java function getListGivenMixed( ) can be defined as:
public static float[] getListGivenMixed(float[] fpList, int size) {
int listLen = ((fpList.length > size) ? size : fpList.length);
float fpListop = new float[listLen];
for (int i =0; i < listLen; i++)
fpListop[i]=fpList[i];
return fpListop;
}
After the function is processed through the wizard the following metadata information is created:
xquery version "1.0" encoding "WINDOWS-1252";
(::pragma xfl <x:xfl xmlns:x="urn:annotations.ld.bea.com">
<creationDate>2005-06-01T14:25:50</creationDate>
<javaFunction class="DocTest"/>
</x:xfl>::)
declare namespace f1 = "lib:testdoc/library";
(::pragma function <f:function xmlns:f="urn:annotations.ld.bea.com" nativeName="getListGivenMixed">
<params>
<param nativeType="[F"/>
<param nativeType="int"/>
</params>
</f:function>::)
declare function f1:getListGivenMixed($x1 as xsd:float*, $x2 as xsd:int) as xsd:float* external;
Here is the corresponding XQuery for executing the above function:
declare namespace f1 = "ld:javaFunc/float";
let $y := (2.0, 4.0, 6.0, 8.0, 10.0)
let $x := f1:getListGivenMixed($y, 2)
return $x
Consider that you have a schema called Customer (customer.xsd), as shown below:
<?xml version="1.0" encoding="UTF-8" ?>
<xs:schema targetNamespace="ld:xml/cust:/BEA_BB10000" xmlns:xs="http://www.w3.org/2001/XMLSchema">
<xs:element name="CUSTOMER">
<xs:complexType>
<xs:sequence>
<xs:element name="FIRST_NAME" type="xs:string" minOccurs="1"/>
<xs:element name="LAST_NAME" type="xs:string" minOccurs="1"/>
</xs:sequence>
</xs:complexType>
</xs:element>
</xs:schema>
If you want to generate a list conforming to the CUSTOMER element you could process the schema via XMLBeans and obtain xml.cust.beaBB10000.CUSTOMERDocument.CUSTOMER. Now you can use the CUSTOMER element as shown:
public static xml.cust.beaBB10000.CUSTOMERDocument.CUSTOMER[]
getCustomerListGivenCustomerList(
xml.cust.beaBB10000.CUSTOMERDocument.CUSTOMER[] ipListOfCust)
throws XmlException {
xml.cust.beaBB10000.CUSTOMERDocument.CUSTOMER [] mylocalver =
ipListOfCust;
return mylocalver;
}
Then the metadata information produced by the wizard will be:
(::pragma function <f:function xmlns:f="urn:annotations.ld.bea.com" kind="datasource" access="public">
<params>
<param nativeType="[Lxml.cust.beaBB10000.CUSTOMERDocument$CUSTOMER;"/>
</params>
</f:function>::)
declare function f1:getCustomerListGivenCustomerList($x1 as element(t1:CUSTOMER)*) as element(t1:CUSTOMER)* external;
The corresponding XQuery for executing the above function is:
declare namespace f1 = "ld:javaFunc/CUSTOMER";
let $z := (
validate(<n:CUSTOMER xmlns:n="ld:xml/cust:/BEA_BB10000"><FIRST_NAME>John2</FIRST_NAME><LAST_NAME>Smith2</LAST_NAME>
</n:CUSTOMER>),
validate(<n:CUSTOMER xmlns:n="ld:xml/cust:/BEA_BB10000"><FIRST_NAME>John2</FIRST_NAME><LAST_NAME>Smith2</LAST_NAME>
</n:CUSTOMER>),
validate(<n:CUSTOMER xmlns:n="ld:xml/cust:/BEA_BB10000"><FIRST_NAME>John2</FIRST_NAME><LAST_NAME>Smith2</LAST_NAME>
</n:CUSTOMER>),
validate(<n:CUSTOMER xmlns:n="ld:xml/cust:/BEA_BB10000"><FIRST_NAME>John2</FIRST_NAME><LAST_NAME>Smith2</LAST_NAME>
</n:CUSTOMER>))
for $zz in $z
return
f1:getCustomerListGivenCustomerList($z)
The following restrictions apply to Java functions:
Spreadsheets offer a highly adaptable means of storing and manipulating information, especially information which needs to be changed quickly. You can easily turn such spreadsheet data in a data services.
Spreadsheet documents are often referred to as CSV files, standing for comma-separated values. Although CSV is not a typical native format for spreadsheets, the capability to save spreadsheets as CSV files is nearly universal.
Although the separator field is often a comma, the Metadata Import wizard supports any ASCII character as a separator, as well as fixed-length fields.
Note: Delimited files in a single server must share the same encoding format. This encoding can be specified through the system property ld.csv.encoding and set through the JVM command-line directly or via a script such as startWebLogic.cmd (Windows) or startWebLogic.sh (UNIX).
Here is the format for this command:
-Dld.csv.encoding=<encoding format>
In the RTLApp DataServices/Demo directory there is a sample that can be used to illustrate delimited file metadata import.
There are several approaches to developing metadata around delimited information, depending on your needs and the nature of the source.
Note: The generated schema takes the name of the source file.
Importing XML file information is similar to importing a relational data source metadata (see Importing Relational Table and View Metadata). Here are the steps that are involved:
Figure 3-37 Selecting a Delimited Source from the Import Metadata Wizard
file:///
For example, on Windows systems you can access an XML file such as Orders.xml from the root C: directory using the following URI:
file:///<c:/home>/Orders.csv
On a UNIX system, you would access such a file with the URI:
file:///<home>/Orders.csv
,).Figure 3-38 Specifying Import Delimited Metadata Characteristics
Figure 3-39 Delimited Document Imported Data Summary Screen
Note: When importing CSV-type data there are several things to keep in mind:
XML files are a convenient means of handling hierarchical data. XML files and associated schemas are easily turned into data services.
Importing XML file information is similar to importing a relational data source metadata (see Importing Relational Table and View Metadata).
The Metadata Import wizard allows you to browse for an XML file anywhere in your application. You can also import data from any XML file on your system using an absolute path prepended with the following:
file:///
For example, on Windows systems you can access an XML file such as Orders.xml from the root C: directory using the following URI:
file:///c:/Orders.xml
On a UNIX system, you would access such a file with the URI:
file:///home/Orders.xml
In the RTLApp DataServices/Demo directory there is a sample that can be used to illustrate XML file metadata import.
Figure 3-40 Selecting an XML File from the Import Metadata Wizard
Figure 3-41 Specify an XML File Schema for XML Metadata Import
Figure 3-42 XML File Imported Data Summary Screen
You can edit the data service name either to clarify the name or to avoid conflicts with other existing or planned data services. Conflicts are shown in red. Simply click on the name of the data service to change its name. Then click Next.
Figure 3-43 A Selecting a Global Element When Importing XML Metadata
When you create metadata for an XML data source but do not supply a data source name, you will need to identify the URI of your data source as a parameter when you execute the data service's read function (various methods of accessing data service functions are described in detail in the Client Application Developer's Guide).
The identification takes the form of:
<uri>/path/filename.xml
where uri is representative of a path or path alias, path represents the directory and filename.xml represents the filename. The .xml extension is needed.
You can access files using an absolute path prepended with the following:
file:///
For example, on Windows systems you can access an XML file such as Orders.xml from the root C: directory using the following URI:
file:///c:/Orders.xml
On a UNIX system, you would access such a file with the URI:
file:///home/Orders.xml
Figure 3-44 shows how the XML source file is referenced.
Figure 3-44 Specifying an XML Source URI in Test View
When you first create a physical data service its underlying metadata is, by definition, consistent with its data source. Over time, however, your metadata may become "out of sync" for several reasons:
You can use the Update Source Metadata right-click menu option to identify differences between your source metadata files and the structure of the source data including:
In the case of Source Unavailable, the issue likely relates to connectivity or permissions. In the case of the other types of reports, you can determine when and if to update data source metadata to conform with the underlying data sources.
If there are no differences between your metadata and the underlying source, the Update Source Metadata wizard will report up-to-date for each data service tested.
Source metadata should be updated with care since the operation can have both direct and indirect consequences. For example, if you have added a relationship between two physical data services, updating your source metadata can potentially remove the relationship from both data services. If the relationship appears in a model diagram, the relationship line will appear in red, indicating that the relationship is no longer described by the respective data services.
In many cases the Update Source Metadata Wizard can automatically merge user changes with the updated metadata. See Using the Update Source Metadata Wizard, for details.
Direct effects apply to physical data services. Indirect effects occur to logical data services, since such services are themselves ultimately based — at least in part — on physical data service. For example, if you have created a new relationship between a physical and a logical data service, updating the physical data service can invalidate the relationship. In the case of the physical data service, there will be no relationship reference. The logical data service will retain the code describing the relationship but it will be invalid if the opposite relationship notations is no longer be present.
Thus updating source metadata should be done carefully. Several safeguards are in place to protect your development effort while preserving your ability to keep your metadata up-to-date. See Archival of Source Metadata for information of how your current metadata is preserved as part of the source update.
The Update Source Metadata wizard allows you to update your source metadata.
Note: Before attempting to update source metadata you should make sure that your build project has no errors.
Figure 3-45 Updating Source Metadata for Several Data Services
You can verify that your data structure is up-to-date by performing a metadata update on one or multiple physical data services in your DSP-based project. For example, in Figure 3-45 all the physical data services in the project will be updated.
After you select your target(s), the wizard identifies the metadata that will be verified and any differences between your metadata and the underlying source.
You can select/deselect any data service or XFL file listed in the dialog using the checkbox to the left of the name (Figure 3-46).
Figure 3-46 Data Services Metadata to be Updated
Next, an analysis is performed on your metadata by the wizard. The following types of synchronization mismatches are identified:
A update preview screen report (Figure 3-47) is prepared describing these differences both generally and for field-level data.
Figure 3-47 Metadata Update Plan for RTLApp's DataServices Project
The Metadata Update Preview screen identifies:
Icons differentiate elements as to be added, removed, or changed. Table 3-48 describes the update source metadata message types and color legends.
Table 3-48 Source Metadata Update Targets and Color Legend
Under some circumstances the Update Source Metadata wizard flags data service artifacts as changed locally when, in fact, no change was made.
For example, in the case of importing a Web service operation, a schema that is dependent (or referenced) by another schema will be assigned an internally-generated filename. If a second imported Web service operation in your project references the same dependent schema, upon synchronization the wizard may note that the name of the imported secondary schema file has changed. Simply proceed with synchronization; the old second-level schema will automatically be removed.
When you update source metadata two files are created and placed in a special directory in your application:
ld:/updateMetadataHistory/metadatadiff<timestamp>.xml
ld:/updateMetadataHistory/sourceBackUp<timestamp>.zip
An update metadata source operations assigns the same timestamp to both generated files.
Figure 3-49 UpdateMetadataHistory Directory Sample Content
Working with a particular update operations report and source, you can often quickly restore relationships and other changes that were made to your metadata while being assured that your metadata is up-to-date.
|
|
|
