3 Using Oracle XML DB

What is XMLType?

Oracle9i release 1 (9.0.1) introduced a new datatype, XMLType, to facilitate native handling of XML data in the database:

• XMLType can represent an XML document in the database, so it is accessible in SQL.

• XMLType has built-in methods that operate on XML content. For example, you can use XMLType methods to create, extract, and index XML data stored in Oracle Database.

• XMLType functionality is also available through a set of Application Program Interfaces (APIs) provided in PL/SQL and Java.

• XMLType can be used in PL/SQL stored procedures for parameters, return values, and variables

With XMLType, SQL developers can leverage the power of the relational database while working in the context of XML. XML developers can leverage the power of XML standards while working in the context of a relational database.

XMLType can be used as the datatype of columns in tables and views. XMLType variables can be used in PL/SQL stored procedures as parameters and return values. You can also use XMLType in SQL, PL/SQL, C, Java (through JDBC), and Oracle Data Provider for .NET (ODP.NET).

The XMLType API provides a number of useful methods that operate on XML content. For example, method extract() extracts one or more nodes from an XMLType instance. Many of these XMLType methods are also provided as SQL functions. For example, SQL function extract corresponds to XMLType method extract().

Oracle XML DB functionality is based on the Oracle XML Developer's Kit C implementations of the relevant XML standards such as XML Parser, XML DOM, and XML Schema Validator.

Benefits of the XMLType Datatype and API

The XMLType datatype and application programming interface (API) enable SQL operations on XML content and XML operations on SQL content:

• Versatile API. XMLType has a versatile API for application development that includes built-in functions, indexing, and navigation support.

• XMLType and SQL. You can use XMLType in SQL statements, combined with other datatypes. For example, you can query XMLType columns and join the result of the extraction with a relational column. Oracle Database determines an optimal way to run such queries.

• Indexing:

  • Oracle XML DB lets you create B*Tree indexes on the object-relational tables that provide structured storage of XMLType tables and columns.

  • Oracle Text indexing supports text indexing of the content of structured and unstructured XMLType tables and columns.

  • The CTXXPATH domain index type of Oracle Text provides an XML-specific text index with transactional semantics. This index type can speed up certain XPath-based searches on both structured and unstructured content.

  • Function-based indexes can be used to create indexes on explicit XPath expressions for both structured and unstructured XMLType storage.

When to Use XMLType

Use XMLType any time you want to use the database as a persistent storage of XML. XMLType functionality includes the following:

• SQL queries on part of or the whole XML document – SQL functions existsNode and extract provide the necessary SQL query functions over XML documents.

• XPath access using SQL functions existsNode and extract – XMLType uses the built-in C XML parser and processor and hence provides better performance and scalability when used inside the server.

• Strong typing inside SQL statements and PL/SQL functions – The strong typing offered by XMLType ensures that the values passed in are XML values and not any arbitrary text string.

• Indexing on XPath document queries – XMLType has methods that you can use to create function-based indexes that optimize searches.

• Separation of applications from storage models – Using XMLType instead of CLOB values or relational storage allows applications to gracefully move to various storage alternatives later without affecting any of the query or DML statements in the application.

• Support for future optimizations – New XML functionality will support XMLType. Because Oracle Database is natively aware that XMLType can store XML data, better optimizations and indexing techniques can be done. By writing applications to use XMLType, these optimizations and enhancements can be easily achieved and preserved in future releases without your needing to rewrite applications.

Two Ways to Store XMLType Data: LOBs and Structured

XMLType data can be stored in two ways:

• In Large Objects (LOBs) – LOB storage maintains content fidelity, also called document fidelity. The original XML is preserved, including whitespace. An entire XML document is stored as a whole in a LOB. For non-schema-based storage, XMLType offers a Character Large Object (CLOB) storage option.

• In Structured storage (in tables and views) – Structured storage maintains DOM (Document Object Model) fidelity.

Native XMLType instances contain hidden columns that store this extra information that does not fit into the SQL object model. This information can be accessed through APIs in SQL or Java, using functions such as extractNode().

Changing XMLType storage from structured storage to LOB, or vice versa, is possible using database IMPORT and EXPORT. Your application code does not need to change. You can then change XML storage options when tuning your application, because each storage option has its own benefits.

Advantages and Disadvantages of XML Storage Options in Oracle XML DB

Table 3-1 summarizes some advantages and disadvantages to consider when selecting your Oracle XML DB storage option. Storage options are also discussed in Table 1-3, "XML Storage Options: Structured or Unstructured" and Chapter 2, "Getting Started with Oracle XML DB".

When to Use CLOB Storage for XMLType

Use CLOB storage for XMLType in the following cases:

• When you are interested in storing and retrieving the whole document.

• When you do not need to perform piece-wise updates on XML documents.

Loading XML Content Using SQL or PL/SQL

You can use a simple INSERT operation in SQL or PL/SQL to load an XML document into the database. Before the document can be stored as an XMLType column or table, it must be converted into an XMLType instance using one of the XMLType constructors.

XMLType constructors allow an XMLType instance to be created from different sources, including VARCHAR, CLOB, and BFILE values. The constructors accept additional arguments that reduce the amount of processing associated with XMLType creation. For example, if you are sure that a given source XML document is valid, you can provide an argument to the constructor that disables the type-checking that is otherwise performed.

In addition, if the source data is not encoded in the database character set, an XMLType instance can be constructed using a BFILE or BLOB value. The encoding of the source data is specified through the character set id (csid) argument of the constructor.

Create a SQL Directory That Points to the Needed Directory

Example 3-3 shows how to insert XML content into an XMLType table. Before making this insertion, you must create a SQL directory object that points to the directory containing the file to be processed. To do this, you must have the CREATE ANY DIRECTORY privilege.

CREATE DIRECTORY xmldir AS path_to_folder_containing_XML_file';

INSERT INTO mytable2 VALUES (XMLType(bfilename('XMLDIR', 'purchaseOrder.xml'),
                                     nls_charset_id('AL32UTF8')));

1 row created.

Loading XML Content Using Java

public void doInsert(Connection conn, Document doc)
throws Exception
{
   String SQLTEXT = "INSERT INTO purchaseorder VALUES (?)";
   XMLType xml = null;
   xml = XMLType.createXML(conn,doc);
   OraclePreparedStatement sqlStatement = null;
   sqlStatement = (OraclePreparedStatement) conn.prepareStatement(SQLTEXT);
   sqlStatement.setObject(1,xml);
   sqlStatement.execute();
}
 
1 row selected.

The "Simple Bulk Loader Application" available on the Oracle Technology Network (OTN) site at http://www.oracle.com/technology/sample_code/tech/xml/xmldb/content.html demonstrates how to load a directory of XML files into Oracle XML DB using Java Database Connectivity (JDBC). JDBC is a set of Java interfaces to Oracle Database.

Loading XML Content Using C

Example 3-5 shows, in C, how to insert XML content into an XMLType table by creating an XMLType instance given a DOM.

#include "stdio.h"
#include <xml.h>
#include <stdlib.h>
#include <string.h>
#include <ocixmldb.h>
OCIEnv *envhp;
OCIError *errhp;
OCISvcCtx *svchp;
OCIStmt *stmthp;
OCIServer *srvhp;
OCIDuration dur;
OCISession *sesshp;
oratext *username = "QUINE";
oratext *password = "CURRY";
oratext *filename = "AMCEWEN-20021009123336171PDT.xml";
oratext *schemaloc = "http://localhost:8080/source/schemas/poSource/xsd/purchaseOrder.xsd";
 
/*--------------------------------------------------------*/
/* Execute a SQL statement that binds XML data            */
/*--------------------------------------------------------*/
 
sword exec_bind_xml(OCISvcCtx *svchp, OCIError *errhp, OCIStmt *stmthp,
                    void *xml,        OCIType *xmltdo, OraText *sqlstmt)
{
  OCIBind *bndhp1 = (OCIBind *) 0;
  sword  status = 0;
  OCIInd ind = OCI_IND_NOTNULL;
  OCIInd *indp = &ind;
  if(status = OCIStmtPrepare(stmthp, errhp, (OraText *)sqlstmt,
                             (ub4)strlen((const char *)sqlstmt),
                             (ub4) OCI_NTV_SYNTAX, (ub4) OCI_DEFAULT))
    return OCI_ERROR;
  if(status = OCIBindByPos(stmthp, &bndhp1, errhp, (ub4) 1, (dvoid *) 0,
                           (sb4) 0, SQLT_NTY, (dvoid *) 0, (ub2 *)0,
                           (ub2 *)0, (ub4) 0, (ub4 *) 0, (ub4) OCI_DEFAULT))
    return OCI_ERROR;
  if(status = OCIBindObject(bndhp1, errhp, (CONST OCIType *) xmltdo,
                            (dvoid **) &xml, (ub4 *) 0,
                            (dvoid **) &indp, (ub4 *) 0))
    return OCI_ERROR;
  if(status = OCIStmtExecute(svchp, stmthp, errhp, (ub4) 1, (ub4) 0,
                             (CONST OCISnapshot*) 0, (OCISnapshot*) 0,
                             (ub4) OCI_DEFAULT))
    return OCI_ERROR;
  return OCI_SUCCESS;
}
 
/*--------------------------------------------------------*/
/* Initialize OCI handles, and connect                    */
/*--------------------------------------------------------*/
 
sword init_oci_connect()
{
. . .
}
 
/*--------------------------------------------------------*/
/* Free OCI handles, and disconnect                       */
/*--------------------------------------------------------*/
 
void free_oci()
{
. . .
}
 
void main()
{
  OCIType *xmltdo;
  xmldocnode  *doc;
  ocixmldbparam params[1];
  xmlerr       err;
  xmlctx  *xctx;
  oratext *ins_stmt;
  sword    status;
  xmlnode *root;
  oratext buf[10000];
 
  /* Initialize envhp, svchp, errhp, dur, stmthp */
  init_oci_connect();
 
  /* Get an XML context */
  params[0].name_ocixmldbparam = XCTXINIT_OCIDUR;
  params[0].value_ocixmldbparam = &dur;
  xctx = OCIXmlDbInitXmlCtx(envhp, svchp, errhp, params, 1);
  if (!(doc = XmlLoadDom(xctx, &err, "file", filename,
                         "schema_location", schemaloc, NULL)))
    {
      printf("Parse failed.\n");
      return;
    }
  else
    printf("Parse succeeded.\n");
  root = XmlDomGetDocElem(xctx, doc);
  printf("The xml document is :\n");
  XmlSaveDom(xctx, &err, (xmlnode *)doc, "buffer", buf, "buffer_length", 10000, NULL);
  printf("%s\n", buf);
 
  /* Insert the document into my_table */
  ins_stmt = (oratext *)"insert into purchaseorder values (:1)";
  status = OCITypeByName(envhp, errhp, svchp, (const text *) "SYS",
                         (ub4) strlen((const char *)"SYS"), (const text *) "XMLTYPE",
                         (ub4) strlen((const char *)"XMLTYPE"), (CONST text *) 0,
                         (ub4) 0, OCI_DURATION_SESSION, OCI_TYPEGET_HEADER,
                         (OCIType **) &xmltdo);
  if (status == OCI_SUCCESS)
    {
      status = exec_bind_xml(svchp, errhp, stmthp, (void *)doc,
                             xmltdo, ins_stmt);
    }
  if (status == OCI_SUCCESS)
    printf ("Insert successful\n");
  else
    printf ("Insert failed\n");
 
  /* Free XML instances */
  if (doc)
    XmlFreeDocument((xmlctx *)xctx, (xmldocnode *)doc);
  /* Free XML CTX */
  OCIXmlDbFreeXmlCtx(xctx);
  free_oci();
}

Loading Large XML Files That Contain Small XML Documents

When loading large XML files consisting of a collection of smaller XML documents, it is often more efficient to use Simple API for XML (SAX) parsing to break the file into a set of smaller documents, and then insert those documents. SAX is an XML standard interface provided by XML parsers for event-based applications.

You can use SAX to load a database table from very large XML files in the order of 30 Mb or larger, by creating individual documents from a collection of nodes. You can also bulk load XML files.

Loading Large XML Files Using SQL*Loader

Use SQL*Loader to load large amounts of XML data into Oracle Database. SQL*Loader loads in one of two modes, conventional or direct path. Table 3-2 compares these modes.

Loading XML Documents into the Repository Using DBMS_XDB

You can also store XML documents in Oracle XML DB Repository, and access these documents using path-based rather than table-based techniques. To load an XML document into the repository under a given path, use PL/SQL package DBMS_XDB. This is illustrated by the following example.

DECLARE
  res BOOLEAN;
BEGIN
  res := DBMS_XDB.createResource('/home/QUINE/purchaseOrder.xml',
                                 bfilename('XMLDIR', 'purchaseOrder.xml'),
                                 nls_charset_id('AL32UTF8'));
END;/

Many operations for configuring and using Oracle XML DB are based on processing one or more XML documents – for example, registering an XML schema and performing an XSL transformation. The easiest way to make these XML documents available to Oracle Database is to load them into Oracle XML DB Repository.

Loading Documents into the Repository Using Protocols

You can load XML documents from a local file system into Oracle XML DB Repository using protocols such as WebDAV, from Windows Explorer or other tools that support WebDAV. Figure 3-1 shows a simple drag and drop operation for copying the contents of the SCOTT folder from the local hard drive to the poSource folder in the Oracle XML DB Repository.

Figure 3-1 Using Windows Explorer to Load Content into the Repository

Description of "Figure 3-1 Using Windows Explorer to Load Content into the Repository"

The copied folder might contain, for example, an XML schema document, an HTML page, and some XSLT style sheets.

XML Encoding Declaration

Each XML document is composed of units called entities. Each entity in an XML document may use a different encoding for its characters. Entities that are stored in an encoding other than UTF-8 or UTF-16 must begin with a declaration containing an encoding specification indicating the character encoding in use. For example:

<?xml version='1.0' encoding='EUC-JP' ?>

Entities encoded in UTF-16 must begin with the Byte Order Mark (BOM), as described in Appendix F of the XML 1.0 Reference. For example, on big-endian platforms, the BOM required of a UTF-16 data stream is #xFEFF.

In the absence of both the encoding declaration and the BOM, the XML entity is assumed to be encoded in UTF-8. Because ASCII is a subset of UTF-8, ASCII entities do not require an encoding declaration.

In many cases, external sources of information are available, besides the XML data, to provide the character encoding in use. For example, the encoding of the data can be obtained from the charset parameter of the Content-Type field in an HTTP(S) request as follows:

Content-Type: text/xml; charset=ISO-8859-4

Character-Set Determination When Loading XML Documents into the Database

In releases prior to Oracle Database 10g Release 1, all XML documents were assumed to be in the database character set, regardless of the document encoding declaration. With Oracle Database 10g Release 1, the document encoding is detected from the encoding declaration when the document is loaded into the database.

However, if the XML data is obtained from a CLOB or VARCHAR value, then the encoding declaration is ignored, because these two data types are always encoded in the database character set.

In addition, when loading data into Oracle XML DB, either through programmatic APIs or transfer protocols, you can provide external encoding to override the document encoding declaration. An error is raised if you try to load a schema-based XML document that contains characters that are not legal in the determined encoding.

The following examples show different ways to specify external encoding:

• Using PL/SQL function DBMS_XDB.createResource to create a file resource from a BFILE, you can specify the file encoding with the CSID argument. If a zero CSID is specified then the file encoding is auto-detected from the document encoding declaration.

  CREATE DIRECTORY xmldir AS '/private/xmldir';
  CREATE OR REPLACE PROCEDURE loadXML(filename VARCHAR2, file_csid NUMBER) IS
     xbfile  BFILE;
     RET     BOOLEAN;
  BEGIN
     xbfile := bfilename('XMLDIR', filename);
     ret := DBMS_XDB.createResource('/public/mypurchaseorder.xml', 
                                    xbfile,
                                    file_csid);
  END;/
  
  

• Use the FTP protocol to load documents into Oracle XML DB. Use the quote set_charset FTP command to indicate the encoding of the files to be loaded.

  FTP> quote set_charset Shift_JIS  
  FTP> put mypurchaseorder.xml
  
  

• Use the HTTP(S) protocol to load documents into Oracle XML DB. Specify the encoding of the data to be transmitted to Oracle XML DB in the request header.

  Content-Type: text/xml; charset= EUC-JP
  
  

Character-Set Determination When Retrieving XML Documents from the Database

XML documents stored in Oracle XML DB can be retrieved using a SQL client, programmatic APIs, or transfer protocols. You can specify the encoding of the retrieved data (except in Oracle Database releases prior to 10g, where XML data is retrieved only in the database character set).The character set for an XML document retrieved from the database is determined in the following ways:

• SQL Client – If a SQL client (such as SQL*Plus) is used to retrieve XML data, then the character set is determined by the client-side environment variable NLS_LANG. In particular, this setting overrides any explicit character-set declarations in the XML data itself.

  For example, if you set the client side NLS_LANG variable to AMERICAN_AMERICA.AL32UTF8 and then retrieve an XML document with encoding EUC_JP provided by declaration <?xml version="1.0" encoding="EUC-JP"?>, the character set of the retrieved document is AL32UTF8, not EUC_JP.

  
  

  See Also: Oracle Database Globalization Support Guide for information on NLS_LANG

  
  

• PL/SQL and APIs – Using PL/SQL or programmatic APIs, you can retrieve XML data into VARCHAR, CLOB, or XMLType datatypes. As for SQL clients, you can control the encoding of the retrieved data by setting NLS_LANG.

  You can also retrieve XML data into a BLOB value using XMLType and URIType methods. These methods let you specify the character set of the returned BLOB value. Here is an example:

  CREATE OR REPLACE FUNCTION getXML(pathname VARCHAR2, charset VARCHAR2) 
                       RETURN BLOB IS
      xblob  BLOB;
  BEGIN
      SELECT e.RES.getBlobVal(nls_charset_id(charset)) INTO xblob
        FROM RESOURCE_VIEW e WHERE ANY_PATH = pathname;
     RETURN xblob;
  END;/
  
  

• FTP – You can use the FTP quote set_nls_locale command to set the character set:

  FTP> quote set_nls_locale EUC-JP
  FTP> get mypurchaseorder.xml
  
  

  
  

  See Also: FTP Quote Methods

  
  

• HTTP(S) – You can use the Accept-Charset parameter in an HTTP(S) request:

  /httptest/mypurchaseorder.xml  1.1 HTTP/Host: localhost:2345
  Accept: text/*
  Accept-Charset:  iso-8859-1, utf-8
  
  

  
  

  See Also: Controlling Character Sets for HTTP(S)

  
  

XML Instance Documents

Documents conforming to a given XML schema can be considered as members or instances of the class defined by that XML schema. Consequently the term instance document is often used to describe an XML document that conforms to a given XML schema. The most common use of an XML schema is to validate that a given instance document conforms to the rules defined by the XML schema.

XML Schema for Schemas

The W3C Schema working group publishes an XML schema, often referred to as the "Schema for Schemas". This XML schema provides the definition, or vocabulary, of the XML Schema language. All valid XML schemas can be considered as members of the class defined by this XML schema. This means that an XML schema is an XML document that conforms to the class defined by the XML schema published at http://www.w3.org/2001/XMLSchema.

Editing XML Schemas

XML schemas can be authored and edited using any of the following:

• A simple text editor, such as emacs or vi

• An XML schema-aware editor, such as the XML editor included with Oracle JDeveloper

• An explicit XML schema-authoring tool, such as XMLSpy from Altova Corporation

XML Schema Features

The XML Schema language defines 47 scalar datatypes. This provides for strong typing of elements and attributes. The W3C XML Schema Recommendation also supports object-oriented techniques such as inheritance and extension, hence you can design XML schema with complex objects from base data types defined by the XML Schema language. The vocabulary includes constructs for defining and ordering, default values, mandatory content, nesting, repeated sets, and redefines. Oracle XML DB supports all the constructs, except for redefines.

<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema" version="1.0">
  <xs:element name="PurchaseOrder" type="PurchaseOrderType"/>
  <xs:complexType name="PurchaseOrderType">
    <xs:sequence>
      <xs:element name="Reference" type="ReferenceType"/>
      <xs:element name="Actions" type="ActionsType"/>
      <xs:element name="Reject" type="RejectionType" minOccurs="0"/>
      <xs:element name="Requestor" type="RequestorType"/>
      <xs:element name="User" type="UserType"/>
      <xs:element name="CostCenter" type="CostCenterType"/>
      <xs:element name="ShippingInstructions" type="ShippingInstructionsType"/>
      <xs:element name="SpecialInstructions" type="SpecialInstructionsType"/>
      <xs:element name="LineItems" type="LineItemsType"/>
    </xs:sequence>
  </xs:complexType>
  <xs:complexType name="LineItemsType">
    <xs:sequence>
      <xs:element name="LineItem" type="LineItemType" maxOccurs="unbounded"/>
    </xs:sequence>
  </xs:complexType>
  <xs:complexType name="LineItemType">
    <xs:sequence>
      <xs:element name="Description" type="DescriptionType"/>
      <xs:element name="Part" type="PartType"/>
    </xs:sequence>
    <xs:attribute name="ItemNumber" type="xs:integer"/>
  </xs:complexType>
  <xs:complexType name="PartType">
    <xs:attribute name="Id">
      <xs:simpleType>
        <xs:restriction base="xs:string">
          <xs:minLength value="10"/>
          <xs:maxLength value="14"/>
        </xs:restriction>
      </xs:simpleType>
    </xs:attribute>
    <xs:attribute name="Quantity" type="moneyType"/>
    <xs:attribute name="UnitPrice" type="quantityType"/>
  </xs:complexType>
  <xs:simpleType name="ReferenceType">
    <xs:restriction base="xs:string">
      <xs:minLength value="18"/>
      <xs:maxLength value="30"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:complexType name="ActionsType">
    <xs:sequence>
      <xs:element name="Action" maxOccurs="4">
        <xs:complexType>
          <xs:sequence>
            <xs:element name="User" type="UserType"/>
            <xs:element name="Date" type="DateType" minOccurs="0"/>
          </xs:sequence>
        </xs:complexType>
      </xs:element>
    </xs:sequence>
  </xs:complexType>
  <xs:complexType name="RejectionType">
    <xs:all>
      <xs:element name="User" type="UserType" minOccurs="0"/>
      <xs:element name="Date" type="DateType" minOccurs="0"/>
      <xs:element name="Comments" type="CommentsType" minOccurs="0"/>
    </xs:all>
  </xs:complexType>
  <xs:complexType name="ShippingInstructionsType">
    <xs:sequence>
      <xs:element name="name" type="NameType" minOccurs="0"/>
      <xs:element name="address" type="AddressType" minOccurs="0"/>
      <xs:element name="telephone" type="TelephoneType" minOccurs="0"/>
    </xs:sequence>
  </xs:complexType>
  <xs:simpleType name="moneyType">
    <xs:restriction base="xs:decimal">
      <xs:fractionDigits value="2"/>
      <xs:totalDigits value="12"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="quantityType">
    <xs:restriction base="xs:decimal">
      <xs:fractionDigits value="4"/>
      <xs:totalDigits value="8"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="UserType">
    <xs:restriction base="xs:string">
      <xs:minLength value="0"/>
      <xs:maxLength value="10"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="RequestorType">
    <xs:restriction base="xs:string">
      <xs:minLength value="0"/>
      <xs:maxLength value="128"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="CostCenterType">
    <xs:restriction base="xs:string">
      <xs:minLength value="1"/>
      <xs:maxLength value="4"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="VendorType">
    <xs:restriction base="xs:string">
      <xs:minLength value="0"/>
      <xs:maxLength value="20"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="PurchaseOrderNumberType">
    <xs:restriction base="xs:integer"/>
  </xs:simpleType>
  <xs:simpleType name="SpecialInstructionsType">
    <xs:restriction base="xs:string">
      <xs:minLength value="0"/>
      <xs:maxLength value="2048"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="NameType">
    <xs:restriction base="xs:string">
      <xs:minLength value="1"/>
      <xs:maxLength value="20"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="AddressType">
    <xs:restriction base="xs:string">
      <xs:minLength value="1"/>
      <xs:maxLength value="256"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="TelephoneType">
    <xs:restriction base="xs:string">
      <xs:minLength value="1"/>
      <xs:maxLength value="24"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="DateType">
    <xs:restriction base="xs:date"/>
  </xs:simpleType>
  <xs:simpleType name="CommentsType">
    <xs:restriction base="xs:string">
      <xs:minLength value="1"/>
      <xs:maxLength value="2048"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="DescriptionType">
    <xs:restriction base="xs:string">
      <xs:minLength value="1"/>
      <xs:maxLength value="256"/>
    </xs:restriction>
  </xs:simpleType>
</xs:schema> 

Graphical Representation of the Purchase-Order XML Schema

Figure 3-2 shows the purchase-order XML schema displayed using XMLSpy. XMLSpy is a graphical and user-friendly tool from Altova Corporation for creating and editing XML schema and XML documents. See http://www.altova.com for details. XMLSpy also supports WebDAV and FTP protocols hence can directly access and edit content stored in Oracle XML DB Repository.

Figure 3-2 XMLSpy Graphical Representation of the PurchaseOrder XML Schema

Description of "Figure 3-2 XMLSpy Graphical Representation of the PurchaseOrder XML Schema"

The PurchaseOrder XML schema is a simple XML schema that demonstrates key features of a typical XML document:

• Global element PurchaseOrder is an instance of the complexType PurchaseOrderType

• PurchaseOrderType defines the set of nodes that make up a PurchaseOrder element

• LineItems element consists of a collection of LineItem elements

• Each LineItem element consists of two elements: Description and Part

• Part element has attributes Id, Quantity, and UnitPrice

Why Use XML Schema With Oracle XML DB?

The following paragraphs describe the main reasons for using XML schema with Oracle XML DB.

Structured Storage of XML Documents

Structured storage of XML documents is based on decomposing the content of the document into a set of SQL objects. These SQL objects are based on the SQL 1999 Type framework. When an XML schema is registered with Oracle XML DB, the required SQL type definitions are automatically generated from the XML schema.

A SQL type definition is generated from each complexType defined by the XML schema. Each element or attribute defined by the complexType becomes a SQL attribute in the corresponding SQL type. Oracle XML DB automatically maps the 47 scalar data types defined by the XML Schema Recommendation to the 19 scalar datatypes supported by SQL. A varray type is generated for each element and this can occur multiple times.

The generated SQL types allow XML content, compliant with the XML schema, to be decomposed and stored in the database as a set of objects without any loss of information. When the document is ingested the constructs defined by the XML schema are mapped directly to the equivalent SQL types. This allows Oracle XML DB to leverage the full power of Oracle Database when managing XML and can lead to significant reductions in the amount of space required to store the document. It can also reduce the amount of memory required to query and update XML content.

Annotating an XML Schema to Control Naming, Mapping, and Storage

The W3C XML Schema Recommendation defines an annotation mechanism that allows vendor-specific information to be added to an XML schema. Oracle XML DB uses this to control the mapping between the XML schema and the SQL object model.

Annotating an XML schema allows control over the naming of the SQL objects and attributes created. Annotations can also be used to override the default mapping between the XML schema data types and SQL data types and to specify which table should be used to store the data.

Controlling How XML Collections are Stored in the Database

Annotations are also used to control how collections in an XML document are stored in the database. Currently there are four options:

• Character Large Object (CLOB). The entire set of elements is persisted as XML text stored in a CLOB column.

• Varray in a LOB. Each element in the collection is converted into a SQL object. The collection of SQL objects is serialized and stored in a LOB column.

• Varray as a nested table. Each element in the collection is converted into a SQL object. The collection of SQL objects is stored as a set of rows in an Index Organized Nested Table (IOT).

• Varray as a set of XMLType values. Each element in the collection is treated as a separate XMLType value. The collection of XMLType values is stored as a set of rows in an XMLType table.

These storage options allow you to tune the performance of applications that use XMLType datatypes to store XML in the database.

However, there is no requirement to annotate an XML schema before using it with Oracle XML DB. Oracle XML DB uses a set of default assumptions when processing an XML schema that contains no annotations.

Declaring the Oracle XML DB Namespace

Before annotating an XML schema you must first declare the Oracle XML DB namespace. The Oracle XML DB namespace is defined as:

http://xmlns.oracle.com/xdb

The namespace is declared in the XML schema by adding a namespace declaration such as the following to the root element of the XML schema:

xmlns:xdb="http://xmlns.oracle.com/xdb"

Note the use of a namespace prefix (xdb). This makes it possible to abbreviate the namespace to xdb when adding annotations.

Example 3-8 shows the beginning of the PurchaseOrder XML schema with annotations. See Example D-1 for the complete schema listing.

<xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema"
           xmlns:xdb="http://xmlns.oracle.com/xdb"
           version="1.0"
           xdb:storeVarrayAsTable="true">
  <xs:element name="PurchaseOrder" type="PurchaseOrderType" xdb:defaultTable="PURCHASEORDER"/>
  <xs:complexType name="PurchaseOrderType" xdb:SQLType="PURCHASEORDER_T">
    <xs:sequence>
      <xs:element name="Reference" type="ReferenceType" minOccurs="1" xdb:SQLName="REFERENCE"/>
      <xs:element name="Actions" type="ActionsType" xdb:SQLName="ACTIONS"/>
      <xs:element name="Reject" type="RejectionType" minOccurs="0" xdb:SQLName="REJECTION"/>
      <xs:element name="Requestor" type="RequestorType" xdb:SQLName="REQUESTOR"/>
      <xs:element name="User" type="UserType" minOccurs="1" xdb:SQLName="USERID"/>
      <xs:element name="CostCenter" type="CostCenterType" xdb:SQLName="COST_CENTER"/>
      <xs:element name="ShippingInstructions" type="ShippingInstructionsType" 
                  xdb:SQLName="SHIPPING_INSTRUCTIONS"/>
      <xs:element name="SpecialInstructions" type="SpecialInstructionsType" 
                  xdb:SQLName="SPECIAL_INSTRUCTIONS"/>
      <xs:element name="LineItems" type="LineItemsType" xdb:SQLName="LINEITEMS"/>
    </xs:sequence>
  </xs:complexType>
  <xs:complexType name="LineItemsType" xdb:SQLType="LINEITEMS_T">
    <xs:sequence>
      <xs:element name="LineItem" type="LineItemType" maxOccurs="unbounded" 
                  xdb:SQLName="LINEITEM" xdb:SQLCollType="LINEITEM_V"/>
    </xs:sequence>
  </xs:complexType>
  <xs:complexType name="LineItemType" xdb:SQLType="LINEITEM_T">
    <xs:sequence>
      <xs:element name="Description" type="DescriptionType" 
                  xdb:SQLName="DESCRIPTION"/>
      <xs:element name="Part" type="PartType" xdb:SQLName="PART"/>
    </xs:sequence>
    <xs:attribute name="ItemNumber" type="xs:integer" xdb:SQLName="ITEMNUMBER" 
                  xdb:SQLType="NUMBER"/>
  </xs:complexType>
  <xs:complexType name="PartType" xdb:SQLType="PART_T">
    <xs:attribute name="Id" xdb:SQLName="PART_NUMBER" xdb:SQLType="VARCHAR2">
      <xs:simpleType>
        <xs:restriction base="xs:string">
          <xs:minLength value="10"/>
          <xs:maxLength value="14"/>
        </xs:restriction>
      </xs:simpleType>
    </xs:attribute>
    <xs:attribute name="Quantity" type="moneyType" xdb:SQLName="QUANTITY"/>
    <xs:attribute name="UnitPrice" type="quantityType" xdb:SQLName="UNITPRICE"/>
  </xs:complexType>
  <xs:simpleType name="ReferenceType">
    <xs:restriction base="xs:string">
      <xs:minLength value="18"/>
      <xs:maxLength value="30"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:complexType name="ActionsType" xdb:SQLType="ACTIONS_T">
    <xs:sequence>
      <xs:element name="Action" maxOccurs="4" xdb:SQLName="ACTION" xdb:SQLCollType="ACTION_V">
        <xs:complexType xdb:SQLType="ACTION_T">
          <xs:sequence>
            <xs:element name="User" type="UserType" xdb:SQLName="ACTIONED_BY"/>
            <xs:element name="Date" type="DateType" minOccurs="0" xdb:SQLName="DATE_ACTIONED"/>
          </xs:sequence>
        </xs:complexType>
      </xs:element>
    </xs:sequence>
  </xs:complexType>
  <xs:complexType name="RejectionType" xdb:SQLType="REJECTION_T">
    <xs:all>
      <xs:element name="User" type="UserType" minOccurs="0" xdb:SQLName="REJECTED_BY"/>
      <xs:element name="Date" type="DateType" minOccurs="0" xdb:SQLName="DATE_REJECTED"/>
      <xs:element name="Comments" type="CommentsType" minOccurs="0" xdb:SQLName="REASON_REJECTED"/>
    </xs:all>
  </xs:complexType>
  <xs:complexType name="ShippingInstructionsType" xdb:SQLType="SHIPPING_INSTRUCTIONS_T">
    <xs:sequence>
      <xs:element name="name" type="NameType" minOccurs="0" xdb:SQLName="SHIP_TO_NAME"/>
      <xs:element name="address" type="AddressType" minOccurs="0" xdb:SQLName="SHIP_TO_ADDRESS"/>
      <xs:element name="telephone" type="TelephoneType" minOccurs="0" xdb:SQLName="SHIP_TO_PHONE"/>
    </xs:sequence>
  </xs:complexType>
  <xs:simpleType name="moneyType">
    <xs:restriction base="xs:decimal">
      <xs:fractionDigits value="2"/>
      <xs:totalDigits value="12"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="quantityType">
    <xs:restriction base="xs:decimal">
      <xs:fractionDigits value="4"/>
      <xs:totalDigits value="8"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="UserType">
    <xs:restriction base="xs:string">
      <xs:minLength value="0"/>
      <xs:maxLength value="10"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="RequestorType">
    <xs:restriction base="xs:string">
      <xs:minLength value="0"/>
      <xs:maxLength value="128"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="CostCenterType">
    <xs:restriction base="xs:string">
      <xs:minLength value="1"/>
      <xs:maxLength value="4"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="VendorType">
    <xs:restriction base="xs:string">
      <xs:minLength value="0"/>
      <xs:maxLength value="20"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="PurchaseOrderNumberType">
    <xs:restriction base="xs:integer"/>
  </xs:simpleType>
  <xs:simpleType name="SpecialInstructionsType">
    <xs:restriction base="xs:string">
      <xs:minLength value="0"/>
      <xs:maxLength value="2048"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="NameType">
    <xs:restriction base="xs:string">
      <xs:minLength value="1"/>
      <xs:maxLength value="20"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="AddressType">
    <xs:restriction base="xs:string">
      <xs:minLength value="1"/>
      <xs:maxLength value="256"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="TelephoneType">
    <xs:restriction base="xs:string">
      <xs:minLength value="1"/>
      <xs:maxLength value="24"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="DateType">
    <xs:restriction base="xs:date"/>
  </xs:simpleType>
  <xs:simpleType name="CommentsType">
    <xs:restriction base="xs:string">
      <xs:minLength value="1"/>
      <xs:maxLength value="2048"/>
    </xs:restriction>
  </xs:simpleType>
  <xs:simpleType name="DescriptionType">
    <xs:restriction base="xs:string">
      <xs:minLength value="1"/>
      <xs:maxLength value="256"/>
    </xs:restriction>
  </xs:simpleType>
</xs:schema>

The PurchaseOrder XML schema defines the following two namespaces:

• http://www.w3c.org/2001/XMLSchema. This is reserved by W3C for the Schema for Schemas.

• http://xmlns.oracle.com/xdb. This is reserved by Oracle for the Oracle XML DB schema annotations.

The PurchaseOrder schema uses several annotations, including the following:

• defaultTable annotation in the PurchaseOrder element. This specifies that XML documents, compliant with this XML schema are stored in a database table called purchaseorder.

• SQLType annotation.

  The first occurrence of SQLType specifies that the name of the SQL type generated from complexType PurchaseOrderType is purchaseorder_t.

  The second occurrence of SQLType specifies that the name of the SQL type generated from the complexType LineItemType is lineitem_t and the SQL type that manages the collection of LineItem elements is lineitem_v.

• SQLName annotation. This provides an explicit name for each SQL attribute of purchaseorder_t.

Figure 3-3 shows the XMLSpy Oracle tab, which facilitates adding Oracle XML DB schema annotations to an XML schema while working in the graphical editor.

Figure 3-3 XMLSpy Showing Support for Oracle XML DB Schema Annotations

Description of "Figure 3-3 XMLSpy Showing Support for Oracle XML DB Schema Annotations"

Registering an XML Schema with Oracle XML DB

For an XML schema to be useful to Oracle XML DB you must first register it with Oracle XML DB. After it has been registered, it can be used for validating XML documents and for creating XMLType tables and columns bound to the XML schema.

Two items are required to register an XML schema with Oracle XML DB:

• The XML schema document

• A string that can be used as a unique identifier for the XML schema, after it is registered with Oracle Database. Instance documents use this unique identifier to identify themselves as members of the class defined by the XML schema. The identifier is typically in the form of a URL, and is often referred to as the schema location hint.

You register an XML schema with PL/SQL procedure DBMS_XMLSCHEMA.registerschema. See Example 3-9. By default, when an XML schema is registered, Oracle XML DB automatically generates all of the SQL object types and XMLType tables required to manage the instance documents.

XML schemas can be registered as global or local.

BEGIN
  DBMS_XMLSCHEMA.registerSchema(
    'http://localhost:8080/source/schemas/poSource/xsd/purchaseOrder.xsd',
    XDBURIType('/source/schemas/poSource/xsd/purchaseOrder.xsd').getClob(),
    TRUE,
    TRUE,
    FALSE,
    TRUE);
END;
/

In this example, the unique identifier for the XML schema is:

http://localhost:8080/source/schemas/poSource/xsd/purchaseOrder.xsd

The XML schema document was previously loaded into Oracle XML DB Repository at this path: /source/schemas/poSource/xsd/purchaseOrder.xsd.

During XML schema registration, an XDBURIType accesses the content of the XML schema document, based on its location in the repository. Flags passed to procedure registerSchema specify that the XML schema must be registered as a local schema, and that SQL objects and tables must be generated by the registration process.

Procedure DBMS_XMLSCHEMA.registerSchema performs the following operations:

• Parses and validates the XML schema

• Creates a set of entries in Oracle Data Dictionary that describe the XML schema

• Creates a set of SQL object definitions, based on complexTypes defined in the XML schema

• Creates an XMLType table for each global element defined by the XML schema

DESCRIBE purchaseorder_t
 purchaseorder_t is NOT FINAL
 Name                                      Null?    Type
 ----------------------------------------- -------- ----------------------------
 SYS_XDBPD$                                         XDB.XDB$RAW_LIST_T
 REFERENCE                                          VARCHAR2(30 CHAR)
 ACTIONS                                            ACTIONS_T
 REJECTION                                          REJECTION_T
 REQUESTOR                                          VARCHAR2(128 CHAR)
 USERID                                             VARCHAR2(10 CHAR)
 COST_CENTER                                        VARCHAR2(4 CHAR)
 SHIPPING_INSTRUCTIONS                              SHIPPING_INSTRUCTIONS_T
 SPECIAL_INSTRUCTIONS                               VARCHAR2(2048 CHAR)
 LINEITEMS                                          LINEITEMS_T
 
DESCRIBE lineitems_t
 lineitems_t is NOT FINAL
 Name                                      Null?    Type
 ----------------------------------------- -------- ----------------------------
 SYS_XDBPD$                                         XDB.XDB$RAW_LIST_T
 LINEITEM                                           LINEITEM_V
 
DESCRIBE lineitem_v
 lineitem_v VARRAY(2147483647) OF LINEITEM_T
 LINEITEM_T is NOT FINAL
 Name                                      Null?    Type
 ----------------------------------------- -------- ----------------------------
 SYS_XDBPD$                                         XDB.XDB$RAW_LIST_T
 ITEMNUMBER                                         NUMBER(38)
 DESCRIPTION                                        VARCHAR2(256 CHAR)
 PART                                               PART_T

Creating XML Schema-Based XMLType Columns and Tables

After an XML schema has been registered with Oracle XML DB, it can be referenced when defining tables that contain XMLType columns or creating XMLType tables.

Example 3-11 shows how to manually create the PurchaseOrder table, the default table for PurchaseOrder elements.

CREATE TABLE purchaseorder OF XMLType
  XMLSCHEMA "http://localhost:8080/source/schemas/poSource/xsd/purchaseOrder.xsd"
  ELEMENT "PurchaseOrder"
  VARRAY "XMLDATA"."ACTIONS"."ACTION"
    STORE AS TABLE action_table 
                   ((PRIMARY KEY (NESTED_TABLE_ID, SYS_NC_ARRAY_INDEX$))
                   ORGANIZATION INDEX OVERFLOW)
  VARRAY "XMLDATA"."LINEITEMS"."LINEITEM"
    STORE AS TABLE lineitem_table 
                   ((PRIMARY KEY (NESTED_TABLE_ID, SYS_NC_ARRAY_INDEX$))
                   ORGANIZATION INDEX OVERFLOW);

DESCRIBE purchaseorder
 Name                                      Null?    Type
 ----------------------------------------- -------- ----------------------------
TABLE of SYS.XMLTYPE(XMLSchema
"http://localhost:8080/source/schemas/poSource/xsd/purchaseOrder.xsd"
Element "PurchaseOrder") STORAGE Object-relational TYPE "PURCHASEORDER_T"

The output of the DESCRIBE statement shows the following information about the purchaseorder table:

• The table is an XMLType table

• The table is constrained to storing PurchaseOrder documents as defined by the PurchaseOrder XML schema

• Rows in this table are stored as a set of objects in the database

• SQL type purchaseorder_t is the base object for this table

Default Tables

The XML schema in Example 3-11 specifies that the PurchaseOrder table is the default table for PurchaseOrder elements. When an XML document compliant with the XML schema is inserted into Oracle XML DB Repository using protocols or PL/SQL, the content of the XML document is stored as a row in the purchaseorder table.

When an XML schema is registered as a global schema, you must grant the appropriate access rights on the default table to all other users of the database before they can work with instance documents that conform to the globally registered XML schema.

Attributes noNamespaceSchemaLocation and schemaLocation

If the target XML schema does not declare a target namespace, the noNamespaceSchemaLocation attribute is used to identify the XML schema. The value of the attribute is the schema location hint. This is the unique identifier passed to PL/SQL procedure DBMS_XMLSCHEMA.registerSchema when the schema is registered with the database.

For the purchaseOrder.xsd XML schema, the correct definition of the root element of the instance document would read as follows:

<PurchaseOrder
  xmlns:xsi=http://www.w3.org/2001/XMLSchema-instance
  xsi:noNamespaceSchemaLocation=
    "http://localhost:8080/source/schemas/poSource/xsd/purchaseOrder.xsd">

If the target XML schema declares a target namespace, then the schemaLocation attribute is used to identify the XML schema. The value of this attribute is a pair of values separated by a space:

• the value of the target namespace declared in the XML schema

• the schema location hint, the unique identifier passed to procedure DBMS_XMLSCHEMA.registerSchema when the schema is registered with the database

For example, assume that the PurchaseOrder XML schema includes a target namespace declaration. The root element of the schema would look like this:

<xs:schema targetNamespace="http://demo.oracle.com/xdb/purchaseOrder"
           xmlns:xs="http://www.w3.org/2001/XMLSchema"
           xmlns:xdb="http://xmlns.oracle.com/xdb"
           version="1.0" xdb:storeVarrayAsTable="true">
   <xs:element name="PurchaseOrder" type="PurchaseOrderType"
               xdb:defaultTable="PURCHASEORDER"/>

In this case, the correct form of the root element of the instance document would read as follows:

<PurchaseOrder
    xnlns="http://demo.oracle.com/xdb/purchaseOrder"
    xmlns:xsi=http://www.w3.org/2001/XMLSchema-instance
    xsi:schemaLocation=
      "http://demo.oracle.com/xdb/purchaseOrder
       http://mdrake-lap:8080/source/schemas/poSource/xsd/purchaseOrder.xsd">

Dealing with Multiple Namespaces

When the XML schema includes elements defined in multiple namespaces, an entry must occur in the schemaLocation attribute for each of the XML schemas. Each entry consists of the namespace declaration and the schema location hint. The entries are separated from each other by one or more whitespace characters. If the primary XML schema does not declare a target namespace, then the instance document also needs to include a noNamespaceSchemaLocation attribute that provides the schema location hint for the primary XML schema.

Comparing Partial to Full XML Schema Validation

This section describes the differences between partial and full XML schema validation used when inserting XML documents into the database.

INSERT INTO purchaseorder
  VALUES(XMLType(bfilename('XMLDIR', 'InvalidElement.xml'),
                 nls_charset_id('AL32UTF8')));
         XMLType
         *
ERROR at line 4:
ORA-30937: No schema definition for 'UserName' (namespace '##local') in parent
 'PurchaseOrder'
ORA-06512: at "SYS.XMLTYPE", line 259
ORA-06512: at "SYS.XMLTYPE", line 284
ORA-06512: at line 1

ALTER TABLE purchaseorder
  ADD CONSTRAINT validate_purchaseorder
  CHECK (XMLIsValid(OBJECT_VALUE) = 1);
 
Table altered.
 
INSERT INTO purchaseorder
  VALUES (XMLType(bfilename('XMLDIR', 'InvalidReference.xml'),
                  nls_charset_id('AL32UTF8')));

INSERT INTO purchaseorder
*
 
ERROR at line 1:
ORA-02290: check constraint (QUINE.VALIDATE_PURCHASEORDER) violated

CREATE OR REPLACE TRIGGER validate_purchaseorder
   BEFORE INSERT ON purchaseorder
   FOR EACH ROW
BEGIN
  IF (:new.OBJECT_VALUE IS NOT NULL) THEN :new.OBJECT_VALUE.schemavalidate();
  END IF;
END;
/
 
Trigger created.

INSERT INTO purchaseorder  VALUES (XMLType(bfilename('XMLDIR', 'InvalidReference.xml'),
                  nls_charset_id('AL32UTF8')));
           *
ERROR at line 2:
ORA-31154: invalid XML document
ORA-19202: Error occurred in XML processing
LSX-00221: "SBELL-20021009" is too short (minimum length is 18)
ORA-06512: at "SYS.XMLTYPE", line 333
ORA-06512: at "QUINE.VALIDATE_PURCHASEORDER", line 3
ORA-04088: error during execution of trigger 'QUINE.VALIDATE_PURCHASEORDER'

Using SQL Constraints to Enforce Referential Integrity

The W3C XML Schema Recommendation defines a powerful language for defining the contents of an XML document. However, there are a number of simple data management concepts not currently addressed by the W3C XML Schema Recommendation. These include the ability to ensure that the value of an element or attribute:

• Is unique across a set of XML documents (a UNIQUE constraint)

• Exists in a particular data source outside the current document (FOREIGN KEY constraint)

The mechanisms used to enforce integrity on XML are the same mechanisms used to enforce integrity on conventional relational data. Simple rules such as uniqueness and foreign-key relationships, are enforced by specifying constraints. More complex rules are enforced by specifying database triggers. Example 3-17 and Example 3-18 illustrate how you can use SQL constraints to enforce referential integrity.

Oracle XML DB makes it possible to use the database to enforce business rules on XML content, in addition to rules that can be specified using the XML schema constructs. The database enforces these business rules regardless of whether XML is inserted directly into a table or uploaded using one of the protocols supported by Oracle XML DB Repository.

ALTER TABLE purchaseorder
  ADD CONSTRAINT reference_is_unique
  UNIQUE (XMLDATA."REFERENCE");
 
Table altered.
 
ALTER TABLE purchaseorder
  ADD CONSTRAINT user_is_valid
  FOREIGN KEY (XMLDATA."USERID") REFERENCES hr.employees(email);
 
Table altered.
 
INSERT INTO purchaseorder
  VALUES (XMLType(bfilename('XMLDIR', 'purchaseOrder.xml'),
                  nls_charset_id('AL32UTF8')));
 
1 row created.
 
INSERT INTO purchaseorder
  VALUES (XMLType(bfilename('XMLDIR', 'DuplicateReference.xml'),
                  nls_charset_id('AL32UTF8')));

INSERT INTO purchaseorder
*
 
ERROR at line 1:
ORA-00001: unique constraint (QUINE.REFERENCE_IS_UNIQUE) violated
 
INSERT INTO purchaseorder
  VALUES (XMLType(bfilename('XMLDIR', 'InvalidUser.xml'),
                  nls_charset_id('AL32UTF8')));

INSERT INTO purchaseorder
*
 
ERROR at line 1:
ORA-02291: integrity constraint (QUINE.USER_IS_VALID) violated - parent key not
 found

$ ftp localhost 2100
Connected to localhost.
220 mdrake-sun FTP Server (Oracle XML DB/Oracle Database 10g Enterprise Edition
Release 10.1.0.0.0 - Beta) ready.
Name (localhost:oracle10): QUINE
331 pass required for QUINE
Password:
230 QUINE logged in
ftp> cd /source/schemas
250 CWD Command successful
ftp> put InvalidReference.xml
200 PORT Command successful
150 ASCII Data Connection
550- Error Response
ORA-00604: error occurred at recursive SQL level 1
ORA-31154: invalid XML document
ORA-19202: Error occurred in XML processing
LSX-00221: "SBELL-20021009" is too short (minimum length is 18)
ORA-06512: at "SYS.XMLTYPE", line 333
ORA-06512: at "QUINE.VALIDATE_PURCHASEORDER", line 3
ORA-04088: error during execution of trigger 'QUINE.VALIDATE_PURCHASEORDER'
550 End Error Response
ftp> put InvalidElement.xml
200 PORT Command successful
150 ASCII Data Connection
550- Error Response
ORA-30937: No schema definition for 'UserName' (namespace '##local') in parent
'PurchaseOrder'
550 End Error Response
ftp> put DuplicateReference.xml
200 PORT Command successful
150 ASCII Data Connection
550- Error Response
ORA-00604: error occurred at recursive SQL level 1
ORA-00001: unique constraint (QUINE.REFERENCE_IS_UNIQUE) violated
550 End Error Response
ftp> put InvalidUser.xml
200 PORT Command successful
150 ASCII Data Connection
550- Error Response
ORA-00604: error occurred at recursive SQL level 1
ORA-02291: integrity constraint (QUINE.USER_IS_VALID) violated - parent key not
 found
550 End Error Response

Full SQL Error Trace

When an error occurs while a document is being uploaded with a protocol, Oracle XML DB provides the client with the full SQL error trace. How the error is interpreted and reported to you is determined by the error-handling built into the client application. Some clients, such as the command line FTP tool, reports the error returned by Oracle XML DB, while others, such as Microsoft Windows Explorer, simply report a generic error message.

XPath and Oracle XML

Oracle XML DB includes new XMLType methods and XML-specific SQL functions. WIth these you can query and update XML content stored in Oracle Database. They use the W3C XPath Recommendation to identify the required node or nodes. Every node in an XML document can be uniquely identified by an XPath expression. An XPath expression consists of a slash-separated list of element names, attributes names, and XPath functions. XPath expressions may contain indexes and conditions that determine which branch of the tree is traversed in determining the target nodes.

By supporting XPath-based methods and functions, Oracle XML DB makes it possible for XML programmers to query and update XML documents in a familiar, standards-compliant manner.

PurchaseOrder XML Document

Examples in this section are based on the following PurchaseOrder XML document:

<PurchaseOrder 
  xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
  xsi:noNamespaceSchemaLocation=
    "http://localhost:8080/source/schemas/poSource/xsd/purchaseOrder.xsd">
  <Reference>SBELL-2002100912333601PDT</Reference>
  <Actions>
    <Action>
      <User>SVOLLMAN</User>
    </Action>
  </Actions>
  <Reject/>
  <Requestor>Sarah J. Bell</Requestor>
  <User>SBELL</User>
  <CostCenter>S30</CostCenter>
  <ShippingInstructions>
    <name>Sarah J. Bell</name>
    <address>400 Oracle Parkway
      Redwood Shores
      CA
      94065
      USA</address>
    <telephone>650 506 7400</telephone>
  </ShippingInstructions>
  <SpecialInstructions>Air Mail</SpecialInstructions>
  <LineItems>
    <LineItem ItemNumber="1">
      <Description>A Night to Remember</Description>
      <Part Id="715515009058" UnitPrice="39.95" Quantity="2"/>
    </LineItem>
    <LineItem ItemNumber="2">
      <Description>The Unbearable Lightness Of Being</Description>
      <Part Id="37429140222" UnitPrice="29.95" Quantity="2"/>
    </LineItem>
    <LineItem ItemNumber="3">
      <Description>Sisters</Description>
      <Part Id="715515011020" UnitPrice="29.95" Quantity="4"/>
    </LineItem>
  </LineItems>
</PurchaseOrder>

Retrieving the Content of an XML Document Using Pseudocolumn OBJECT_VALUE

The OBJECT_VALUE pseudocolumn can be used as an alias for the value of an object table. For an XMLType table that consists of a single column of XMLType, the entire XML document is retrieved. (OBJECT_VALUE replaces the value(x) and SYS_NC_ROWINFO$ aliases used in releases prior to Oracle Database10g Release 1.)

SET LONG 10000
SET PAGESIZE 100

SELECT OBJECT_VALUE FROM purchaseorder;
 
OBJECT_VALUE
-----------------------------------------------------------------------
<PurchaseOrder xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
 xsi:noNamespaceSchemaLocation="http://localhost:8080/source/schemas
/poSource/xsd/purchaseOrder.xsd">
  <Reference>SBELL-2002100912333601PDT</Reference>
  <Actions>
    <Action>
      <User>SVOLLMAN</User>
    </Action>
  </Actions>
  <Reject/>
  <Requestor>Sarah J. Bell</Requestor>
  <User>SBELL</User>
  <CostCenter>S30</CostCenter>
  <ShippingInstructions>
    <name>Sarah J. Bell</name>
    <address>400 Oracle Parkway
Redwood Shores
CA
94065
USA</address>
    <telephone>650 506 7400</telephone>
  </ShippingInstructions>
  <SpecialInstructions>Air Mail</SpecialInstructions>
  <LineItems>
    <LineItem ItemNumber="1">
      <Description>A Night to Remember</Description>
      <Part Id="715515009058" UnitPrice="39.95" Quantity="2"/>
    </LineItem>
    <LineItem ItemNumber="2">
      <Description>The Unbearable Lightness Of Being</Description>
      <Part Id="37429140222" UnitPrice="29.95" Quantity="2"/>
    </LineItem>
    <LineItem ItemNumber="3">
      <Description>Sisters</Description>
      <Part Id="715515011020" UnitPrice="29.95" Quantity="4"/>
    </LineItem>
  </LineItems>
</PurchaseOrder>
 
1 row selected.

Accessing Fragments or Nodes of an XML Document Using EXTRACT

SQL function extract returns the nodes that match an XPath expression. Nodes are returned as an instance of XMLType. The result of extract can be either a complete document or an XML fragment. The functionality of SQL function extract is also available through XMLType method extract().

SELECT extract(OBJECT_VALUE, '/PurchaseOrder/Reference')
  FROM purchaseorder;

EXTRACT(OBJECT_VALUE, '/PURCHASEORDER/REFERENCE')
---------------------------------------------------
<Reference>SBELL-2002100912333601PDT</Reference>
 
1 row selected.

SELECT extract(OBJECT_VALUE, '/PurchaseOrder/LineItems/LineItem[1]')
  FROM purchaseorder;
 
EXTRACT(OBJECT_VALUE, '/PURCHASEORDER/LINEITEMS/LINEITEM[1]')
--------------------------------------------------------------------------------
<LineItem ItemNumber="1">
  <Description>A Night to Remember</Description>
  <Part Id="715515009058" UnitPrice="39.95" Quantity="2"/>
</LineItem>
 
1 row selected.

SELECT extract(OBJECT_VALUE, '/PurchaseOrder/LineItems/LineItem/Description')
  FROM purchaseorder;
 
EXTRACT(OBJECT_VALUE, '/PURCHASEORDER/LINEITEMS/LINEITEM/DESCRIPTION')
--------------------------------------------------------------------------------
<Description>A Night to Remember</Description>
<Description>The Unbearable Lightness Of Being</Description>
<Description>Sisters</Description>
 
1 row selected.

Accessing Text Nodes and Attribute Values Using EXTRACTVALUE

The SQL function extractValue returns the value of the text node or attribute value that matches the supplied XPath expression. The value is returned as a SQL scalar value. The XPath expression passed to extractValue must uniquely identify a single text node or attribute value within the document.

SELECT extractValue(OBJECT_VALUE, '/PurchaseOrder/Reference')
FROM purchaseorder;
 
EXTRACTVALUE(OBJECT_VALUE, '/PURCHASEORDER/REFERENCE)
----------------------------------------------------------------------
SBELL-2002100912333601PDT
 
1 row selected.

SELECT extractValue(OBJECT_VALUE,
                    '/PurchaseOrder/LineItems/LineItem[1]/Description')
  FROM purchaseorder;
 
EXTRACTVALUE(OBJECT_VALUE, '/PURCHASEORDER/LINEITEMS/LINEITEM[1]/DESCRIPTION')
------------------------------------------------------------------------------
A Night to Remember
 
1 row selected.

SELECT extractValue(
         OBJECT_VALUE,
         '/PurchaseOrder/LineItems/LineItem[Part/@Id="715515011020"]/Description')
FROM purchaseorder;
 
EXTRACTVALUE(OBJECT
_VALUE, '/PURCHASEORDER/LINEITEMS/LINEITEM[PART/@ID="715515011020"]/DESCRIPTION')
---------------------------------------------------------------------------------
Sisters
 
1 row selected.

SELECT extractValue(OBJECT_VALUE,
                    '/PurchaseOrder/LineItems/LineItem/Description')
  FROM purchaseorder;
SELECT extractValue(OBJECT_VALUE,
                    '/PurchaseOrder/LineItems/LineItem/Description')
 *
ERROR at line 1:
ORA-01427: single-row subquery returns more than one row

 
SELECT extractValue(OBJECT_VALUE,'/PurchaseOrder/LineItems/LineItem[1]')
  FROM purchaseorder;
  FROM purchaseorder
     *
ERROR at line 3:
ORA-19026: EXTRACTVALUE can only retrieve value of leaf node
 

SELECT extractValue(OBJECT_VALUE,
                    '/PurchaseOrder/LineItems/LineItem/Description/text()')
  FROM purchaseorder;
SELECT extractValue(OBJECT_VALUE,
                    '/PurchaseOrder/LineItems/LineItem/Description/text()')
 *
ERROR at line 1:
ORA-01427: single-row subquery returns more than one row

ORA-01427: single-row subquery returns more than one row

Searching the Content of an XML Document Using EXISTSNODE

The SQL function existsNode evaluates whether or not a given document contains a node that matches a W3C XPath expression. Function existsNode returns true (1) if the document contains the node specified by the XPath expression supplied to the function and false (0) if it does not. Since XPath expressions can contain predicates, existsNode can determine whether or not a given node exists in the document, and whether or not a node with the specified value exists in the document. The functionality provided by SQL function existsNode is also available through XMLType method existsNode.

SELECT COUNT(*)
  FROM purchaseorder
WHERE existsNode(OBJECT_VALUE, '/PurchaseOrder/Reference') = 1;

  COUNT(*)
----------
       132


1 row selected.

SELECT count(*) 
  FROM purchaseorder
  WHERE existsNode(OBJECT_VALUE,      
                   '/PurchaseOrder[Reference="SBELL-2002100912333601PDT"]')
      = 1;
  COUNT(*)
----------
         1
1 row selected.

SELECT count(*)
  FROM purchaseorder
  WHERE existsNode(
          OBJECT_VALUE,
          '/PurchaseOrder/Reference[Reference="SBELL-XXXXXXXXXXXXXXXXXX"]')
        = 1;
 
  COUNT(*)
----------
         0
 
1 row selected.

SELECT count(*)
  FROM purchaseorder
  WHERE existsNode(OBJECT_VALUE, '/PurchaseOrder/LineItems/LineItem/Part/@Id')
        = 1;
 
  COUNT(*)
----------
       132
 
1 row selected.

SELECT count(*)
  FROM purchaseorder
  WHERE existsNode(
          OBJECT_VALUE,
          '/PurchaseOrder/LineItems/LineItem/Part[@Id="715515009058"]')
        = 1;
 
  COUNT(*)
----------
        21

SELECT count(*)
  FROM purchaseorder
  WHERE existsNode(
          OBJECT_VALUE,
          '/PurchaseOrder/LineItems/LineItem[3]/Part[@Id="715515009058"]')
        = 1;

  COUNT(*)
----------
         1
1 row selected.

SELECT extractValue(OBJECT_VALUE, '/PurchaseOrder/Reference') "Reference"
  FROM purchaseorder
  WHERE extractValue(OBJECT_VALUE, '/PurchaseOrder/User') LIKE 'S%';
 
Reference
------------------------------
SBELL-20021009123336231PDT
SBELL-20021009123336331PDT
SKING-20021009123336321PDT
...
36 rows selected.

SELECT extractValue(OBJECT_VALUE, '/PurchaseOrder/Reference') "Reference"
  FROM purchaseorder, hr.employees e
  WHERE extractValue(OBJECT_VALUE, '/PurchaseOrder/User') = e.email
    AND e.employee_id = 100;
 
Reference
------------------------------
SKING-20021009123336321PDT
SKING-20021009123337153PDT
SKING-20021009123335560PDT
SKING-20021009123336952PDT
SKING-20021009123336622PDT
SKING-20021009123336822PDT
SKING-20021009123336131PDT
SKING-20021009123336392PDT
SKING-20021009123337974PDT
SKING-20021009123338294PDT
SKING-20021009123337703PDT
SKING-20021009123337383PDT
SKING-20021009123337503PDT
 
13 rows selected.

Using EXTRACTVALUE and EXISTSNODE in a WHERE Clause

The examples in the preceding section demonstrate how SQL function extractValue can be used in a SELECT list to return information contained in an XML document. You can also use these functions in a WHERE clause to determine whether or not a document must be included in the result set of a SELECT, UPDATE, or DELETE statement.

You can use SQL function existsNode to restrict the result set to documents containing nodes that match an XPath expression. You can use SQL function extractValue when joining across multiple tables based on the value of one or more nodes in the XML document. Also, you can use use extractValue whenever specifying a condition is easier with SQL (for example, using keyword LIKE for pattern matching) than with XPath.

SELECT extractValue(OBJECT_VALUE, '/PurchaseOrder/Reference') "Reference"
  FROM purchaseorder
  WHERE existsNode(OBJECT_VALUE, '/PurchaseOrder[User="SBELL"]') = 1;
 
Reference
------------------------------
SBELL-20021009123336231PDT
SBELL-20021009123336331PDT
SBELL-20021009123337353PDT
SBELL-20021009123338304PDT
SBELL-20021009123338505PDT
SBELL-20021009123335771PDT
SBELL-20021009123335280PDT
SBELL-2002100912333763PDT
SBELL-2002100912333601PDT
SBELL-20021009123336362PDT
SBELL-20021009123336532PDT
SBELL-20021009123338204PDT
SBELL-20021009123337673PDT
 
13 rows selected.

SELECT extractValue(OBJECT_VALUE, '/PurchaseOrder/Reference') "Reference"
  FROM purchaseorder
  WHERE existsNode(
          OBJECT_VALUE,
          '/PurchaseOrder/LineItems/LineItem[1]/Part[@Id="715515009058"]')
        = 1;

Reference
------------------------------
SBELL-2002100912333601PDT
 
1 row selected.

Using XMLSEQUENCE to Perform SQL Operations on XMLType Fragments

Example 3-21 demonstrates how the SQL function extract returns an XMLType value containing the node or nodes that matched the supplied XPath expression. When the document contains multiple nodes that match the supplied XPath expression, extract returns an XML fragment containing all of the matching nodes. A fragment differs from a document in that it has no single root element.

This kind of result is common when extract is used to retrieve the set of elements contained in a collection (in this case each node in the fragment will be of the same type), or when the XPath expression terminates in a wildcard (where the nodes in the fragment can be of different types).

SQL function XMLSequence can perform SQL operations on an XMLType value that contains a fragment. It generates a collection of XMLType objects from an XMLType containing a fragment. The collection contains one XMLType value for each of the top-level elements in the fragment. This collection of XMLType objects can then be converted into a virtual table using the SQL table function. Converting the fragment into a virtual table makes it easier to use SQL to process the results of an extract function call that returns multiple nodes.

SELECT extractValue(p.OBJECT_VALUE, 
                    '/PurchaseOrder/LineItems/LineItem/Description')
  FROM purchaseorder p
  WHERE existsNode(p.OBJECT_VALUE,
                   '/PurchaseOrder[Reference="SBELL-2002100912333601PDT"]')
        = 1;
SELECT extractValue(p.OBJECT_VALUE,
                    '/PurchaseOrder/LineItems/LineItem/Description')
 *
ERROR at line 1:
ORA-01427: single-row subquery returns more than one row

SELECT extract(p.OBJECT_VALUE, '/PurchaseOrder/LineItems/LineItem/Description')
  FROM purchaseorder p
  WHERE existsNode(p.OBJECT_VALUE,
                   '/PurchaseOrder[Reference="SBELL-2002100912333601PDT"]')
        = 1;
 
EXTRACT(P.OBJECT_VALUE,'/PURCHASEORDER/LINEITEMS/LINEITEM/DESCRIPTION')
--------------------------------------------------------------------------------
<Description>A Night to Remember</Description>
<Description>The Unbearable Lightness Of Being</Description>
<Description>Sisters</Description>
 
1 row selected.

SELECT value(des)
  FROM purchaseorder p,
       table(XMLSequence(
               extract(p.OBJECT_VALUE,
                       '/PurchaseOrder/LineItems/LineItem/Description'))) des
  WHERE existsNode(p.OBJECT_VALUE,
                   '/PurchaseOrder[Reference="SBELL-2002100912333601PDT"]')
        = 1;
 
VALUE(DES)
--------------------------------------------------------------------------------
<Description>A Night to Remember</Description>
<Description>The Unbearable Lightness Of Being</Description>
<Description>Sisters</Description>

3 rows selected.

SELECT extractValue(value(des), '/Description')
  FROM purchaseorder p,
       table(XMLSequence(
               extract(p.OBJECT_VALUE,
                       '/PurchaseOrder/LineItems/LineItem/Description'))) des
  WHERE existsNode(p.OBJECT_VALUE,
                   '/PurchaseOrder[Reference="SBELL-2002100912333601PDT"]')
        = 1;

EXTRACTVALUE(VALUE(DES),'/DESCRIPTION')
--------------------------------------------------------------------------------
A Night to Remember
The Unbearable Lightness Of Being
Sisters
 
3 rows selected.

SELECT extractValue(p.OBJECT_VALUE, '/PurchaseOrder/Reference'),
       count(*)
  FROM purchaseorder p,
       table(XMLSequence(
               extract(p.OBJECT_VALUE,
                       '/PurchaseOrder/LineItems/LineItem'))) d
  WHERE existsNode(p.OBJECT_VALUE, '/PurchaseOrder[User="SBELL"]') = 1
  GROUP BY extractValue(p.OBJECT_VALUE, '/PurchaseOrder/Reference')
  ORDER BY extractValue(p.OBJECT_VALUE, '/PurchaseOrder/Reference');

EXTRACTVALUE(P.OBJECT_VALUE,'/   COUNT(*)
------------------------------ ----------
SBELL-20021009123335280PDT             20
SBELL-20021009123335771PDT             21
SBELL-2002100912333601PDT               3
SBELL-20021009123336231PDT             25
SBELL-20021009123336331PDT             10
SBELL-20021009123336362PDT             15
SBELL-20021009123336532PDT             14
SBELL-20021009123337353PDT             10
SBELL-2002100912333763PDT              21
SBELL-20021009123337673PDT             10
SBELL-20021009123338204PDT             14
SBELL-20021009123338304PDT             24
SBELL-20021009123338505PDT             20

13 rows selected.

SELECT count(*)
  FROM purchaseorder p,
       table(XMLSequence(extract(p.OBJECT_VALUE, '/PurchaseOrder/*'))) n
  WHERE existsNode(p.OBJECT_VALUE,
                   '/PurchaseOrder[Reference="SBELL-2002100912333601PDT"]')
        = 1;
 
  COUNT(*)
----------
         9

1 row selected.

Updating XML Schema-Based and Non-Schema-Based XML Documents

The way SQL functions like updateXML modify an XML document is determined mainly by whether or not the XML document is based on an XML schema, and how the XML document is stored:

• XML documents stored in CLOB values. When a SQL function like updateXML modifies an XML document stored as a CLOB (whether schema-based or not), Oracle XML DB performs the update by creating a Document Object Model (DOM) from the XML document and using DOM API methods to modify the appropriate XML data. After modification, the updated DOM is returned back to the underlying CLOB object.

• XML documents stored object-relationally. When a SQL function like updateXML modifies a schema-based XML document that is stored object-relationally, Oracle XML DB can use XPath rewrite to modify the underlying object in place. This is a partial update. Partial updates translate the XPath argument to the SQL function into an equivalent SQL operation. The SQL operation then directly modifies the attributes of underlying objects. Such a partial update can be much quicker than a DOM-based update. This can make a significant difference when executing a SQL statement that applies a SQL function like updateXML to a large number of documents.

Using EXPLAIN PLAN to Tune XPath Rewrites

XPath rewrite on its own cannot guarantee scalable and performant applications. The performance of SQL statements generated by XPath rewrite is ultimately determined by the available indexes and the way data is stored on disk. Also, as with any other SQL application, a DBA must monitor the database and optimize storage and indexes if the application is to perform well.

The good news, from a DBA perspective, is that this information is nothing new. The same skills are required to tune an XML application as for any other database application. All of the tools that DBAs typically use with SQL-based applications can be applied to XML-based applications using Oracle XML DB functions.

EXPLAIN PLAN FOR
  SELECT extractValue(OBJECT_VALUE, '/PurchaseOrder/Reference') "Reference"
    FROM purchaseorder
    WHERE existsNode(OBJECT_VALUE, '/PurchaseOrder[User="SBELL"]') = 1;
 
Explained.
 
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------
Plan hash value: 841749721
 
----------------------------------------------------------------------------------
| Id  | Operation         | Name          | Rows | Bytes | Cost (%CPU)| Time    |
----------------------------------------------------------------------------------
|   0 | SELECT STATEMENT  |               |    1 |    24 |     5   (0)| 00:00:01|
|*  1 |  TABLE ACCESS FULL| PURCHASEORDER |    1 |    24 |     5   (0)| 00:00:01|
----------------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------
 
   1 - filter("PURCHASEORDER"."SYS_NC00022$"='SBELL')

Note-----   - dynamic sampling used for this statement 
17 rows selected.

Using Indexes to Improve Performance of XPath-Based Functions

Oracle XML DB supports the creation of three kinds of index on XML content:

• Text-based indexes – These can be created on any XMLType table or column.

• Function-based indexes – These can be created on any XMLType table or column.

• B-Tree indexes – When the XMLType table or column is based on structured storage techniques, conventional B-Tree indexes can be created on underlying SQL types.

Indexes are typically created by using SQL function extractValue, although it is also possible to create indexes based on other functions such as existsNode. During the index creation process Oracle XML DB uses XPath rewrite to determine whether it is possible to map between the nodes referenced in the XPath expression used in the CREATE INDEX statement and the attributes of the underlying SQL types. If the nodes in the XPath expression can be mapped to attributes of the SQL types, then the index is created as a conventional B-Tree index on the underlying SQL objects. If the XPath expression cannot be restated using object-relational SQL then a function-based index is created.

CREATE INDEX purchaseorder_user_index
  ON purchaseorder(extractValue(OBJECT_VALUE, '/PurchaseOrder/User'));

EXPLAIN PLAN FOR
  SELECT extractValue(OBJECT_VALUE, '/PurchaseOrder/Reference') "Reference"
    FROM purchaseorder
    WHERE existsNode(OBJECT_VALUE, '/PurchaseOrder[User="SBELL"]') = 1;
 
Explained.
 
PLAN_TABLE_OUTPUT
--------------------------------------------------------------------------------------------------------
Plan hash value: 713050960
 
--------------------------------------------------------------------------------------------------------
| Id  | Operation                   | Name                     | Rows  | Bytes | Cost (%CPU)| Time     |
--------------------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT            |                          |     1 |    24 |     3   (0)| 00:00:01 |
|   1 |  TABLE ACCESS BY INDEX ROWID| PURCHASEORDER            |     1 |    24 |     3   (0)| 00:00:01 |
|*  2 |   INDEX RANGE SCAN          | PURCHASEORDER_USER_INDEX |     1 |       |     1   (0)| 00:00:01 |
--------------------------------------------------------------------------------------------------------
 
Predicate Information (identified by operation id):
---------------------------------------------------
 
   2 - access("PURCHASEORDER"."SYS_NC00022$"='SBELL')
 
18 rows selected.

Optimizing Operations on Collections

The majority of XML documents contain collections of repeating elements. For Oracle XML DB to be able to efficiently process the collection members, it is important that the storage model for managing the collection provide an efficient way of accessing the individual members of the collection. Selecting the correct storage structure makes it possible to index elements within the collection and perform direct operations on individual elements within the collection.

Oracle XML DB offers the following ways to manage the members of a collection:

• When a collection is stored as a CLOB value, you cannot directly access its members.

• When a varray is stored as a LOB, you cannot directly access members of the collection.

  Storing the members as XML text in a CLOB value means that any operation on the collection requires parsing the contents of the CLOB and then using functional evaluation to perform the required operation.

  Converting the collection into a set of SQL objects that are serialized into a LOB removes the need to parse the documents. However any operations on the members of the collection still require that the collection be loaded from disk into memory before the necessary processing can take place.

• When a varray is stored as a nested table, you can directly access members of the collection.

• When a varray is stored as an XMLType value, you can directly access members of the collection.

In the latter two cases (nested table and XMLType), each member of the varray becomes a row in a table, so you can access it directly though SQL.

Using Indexes to Tune Queries on Collections Stored as Nested Tables

Example 3-41 shows the execution plan for a query to find the Reference element from any document that contains an order for part number 717951002372 (Part element with an Id attribute of value 717951002372).

EXPLAIN PLAN FOR
  SELECT extractValue(OBJECT_VALUE, '/PurchaseOrder/Reference') "Reference"
    FROM purchaseorder
    WHERE existsNode(OBJECT_VALUE, '/PurchaseOrder/LineItems/LineItem/Part[@Id="717951002372"]') = 1;
 
Explained.
 
PLAN_TABLE_OUTPUT
---------------------------------------------------------------------------------------------
Plan hash value: 47905112
 
-------------------------------------------------------------------------------------------------------
| Id  | Operation                    | Name                   | Rows  | Bytes | Cost (%CPU)| Time     |
-------------------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT             |                        |    22 | 14300 |    10  (10)| 00:00:01 |
|   1 |  NESTED LOOPS                |                        |    22 | 14300 |    10  (10)| 00:00:01 |
|   2 |   SORT UNIQUE                |                        |    22 |  2640 |     8   (0)| 00:00:01 |
|*  3 |    INDEX UNIQUE SCAN         | LINEITEM_TABLE_DATA    |    22 |  2640 |     8   (0)| 00:00:01 |
|*  4 |     INDEX RANGE SCAN         | LINEITEM_PART_INDEX    |     9 |       |     2   (0)| 00:00:01 |
|   5 |   TABLE ACCESS BY INDEX ROWID| PURCHASEORDER          |     1 |   530 |     1   (0)| 00:00:01 |
|*  6 |    INDEX UNIQUE SCAN         | LINEITEM_TABLE_MEMBERS |     1 |       |     0   (0)| 00:00:01 |
-------------------------------------------------------------------------------------------------------
 
Predicate Information (identified by operation id):
---------------------------------------------------
 
   3 - access("SYS_NC00011$"='717951002372')
   4 - access("SYS_NC00011$"='717951002372')
   6 - access("NESTED_TABLE_ID"="PURCHASEORDER"."SYS_NC0003400035$")
 
20 rows selected.

CREATE INDEX lineitem_part_index
  ON lineitem_table l(l.part.part_number, l.NESTED_TABLE_ID);

Index created.

EXPLAIN PLAN FOR
  SELECT extractValue(OBJECT_VALUE, '/PurchaseOrder/Reference') "Reference"
    FROM purchaseorder
    WHERE existsNode(OBJECT_VALUE, '/PurchaseOrder/LineItems/LineItem/Part[@Id="717951002372"]') = 1;
 
Explained.

PLAN_TABLE_OUTPUT
-------------------------------------------------------------------------------------------------------
Plan hash value: 497281434
 
-------------------------------------------------------------------------------------------------------
| Id  | Operation                    | Name                   | Rows  | Bytes | Cost (%CPU)| Time     |
-------------------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT             |                        |    22 |  1012 |     4  (25)| 00:00:01 |
|   1 |  NESTED LOOPS                |                        |    22 |  1012 |     4  (25)| 00:00:01 |
|   2 |   SORT UNIQUE                |                        |    22 |   418 |     2   (0)| 00:00:01 |
|*  3 |    INDEX RANGE SCAN          | LINEITEM_PART_INDEX    |    22 |   418 |     2   (0)| 00:00:01 |
|   4 |   TABLE ACCESS BY INDEX ROWID| PURCHASEORDER          |     1 |    27 |     1   (0)| 00:00:01 |
|*  5 |    INDEX UNIQUE SCAN         | LINEITEM_TABLE_MEMBERS |     1 |       |     0   (0)| 00:00:01 |
-------------------------------------------------------------------------------------------------------
 
Predicate Information (identified by operation id):
---------------------------------------------------
 
   3 - access("SYS_NC00011$"='717951002372')
   5 - access("NESTED_TABLE_ID"="PURCHASEORDER"."SYS_NC0003400035$")
 
18 rows selected.

3 - filter(SYS_CHECKACL("ACLOID","OWNERID",xmltype(''<privilege
              xmlns="http://xmlns.oracle.com/xdb/acl.xsd"
              xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
              xsi:schemaLocation="http://xmlns.oracle.com/xdb/acl.xsd
              http://xmlns.oracle.com/xdb/acl.xsd
              DAV:http://xmlns.oracle.com/xdb/dav.xsd">
                        <read-properties/><read-contents/></privilege>''))=1)

Predicate Information (identified by operation id):
---------------------------------------------------
   1 - access("NESTED_TABLE_ID"=:B1)
   2 - access("NESTED_TABLE_ID"=:B1)
   3 - filter(EXISTSNODE(SYS_MAKEXML('C0A5497E8DCF110BE034080020E5CF39', 
                                     3044, "SYS_ALIAS_4". "XMLEXTRA", 
                                     "SYS_ALIAS_4"."XMLDATA"),
                         '/PurchaseOrder[User="SBELL"]')
              =1)
   5 - access("NESTED_TABLE_ID"=:B1)
   6 - access("NESTED_TABLE_ID"=:B1)

Generating XML From Relational Tables Using DBURIType

Another way to generate XML from relational data is with SQL function DBURIType. Function DBURIType exposes one or more rows in a given table or view as a single XML document. The name of the root element is derived from the name of the table or view. The root element contains a set of ROW elements. There is one ROW element for each row in the table or view. The children of each ROW element are derived from the columns in the table or view. Each child element contains a text node with the value of the column for the given row.

SELECT DBURIType('/HR/DEPARTMENTS').getXML() FROM DUAL;
 
DBURITYPE('/HR/DEPARTMENTS').GETXML()
------------------------------------------------------
<?xml version="1.0"?>
<DEPARTMENTS>
 <ROW>
  <DEPARTMENT_ID>10</DEPARTMENT_ID>
  <DEPARTMENT_NAME>Administration</DEPARTMENT_NAME>
  <MANAGER_ID>200</MANAGER_ID>
  <LOCATION_ID>1700</LOCATION_ID>
 </ROW>
...
 <ROW>
  <DEPARTMENT_ID>20</DEPARTMENT_ID>
  <DEPARTMENT_NAME>Marketing</DEPARTMENT_NAME>
  <MANAGER_ID>201</MANAGER_ID>
  <LOCATION_ID>1800</LOCATION_ID>
 </ROW>
</DEPARTMENTS>

SELECT DBURIType('/HR/DEPARTMENTS/ROW[DEPARTMENT_ID="10"]').getXML()
  FROM DUAL;
 
DBURITYPE('/HR/DEPARTMENTS/ROW[DEPARTMENT_ID="10"]').GETXML()
------------------------------------------------------------------
<?xml version="1.0"?>
 <ROW>
  <DEPARTMENT_ID>10</DEPARTMENT_ID>
  <DEPARTMENT_NAME>Administration</DEPARTMENT_NAME>
  <MANAGER_ID>200</MANAGER_ID>
  <LOCATION_ID>1700</LOCATION_ID>
 </ROW>

1 row selected.

SELECT DBURIType(
         '/HR/DEPARTMENTS/ROW[DEPARTMENT_ID="10"]/DEPARTMENT_NAME').getXML()
  FROM DUAL;
 
DBURITYPE('/HR/DEPARTMENTS/ROW[DEPARTMENT_ID="10"]/DEPARTMENT_NAME').GETXML()
-----------------------------------------------------------------------------
<?xml version="1.0"?>
 <DEPARTMENT_NAME>Administration</DEPARTMENT_NAME>
 
1 row selected.

Installing and Uninstalling Oracle XML DB Repository

All of the metadata and content managed by Oracle XML DB Repository is stored using a set of tables in the database schema owned by database user XDB. User XDB is a locked account installed with DBCA or by running the script catqm.sql. Script catqm.sql is located in the directory ORACLE_HOME/rdbms/admin. The repository can be uninstalled using DBCA or by running the script catnoqm.sql. Great care should be taken when running catnoqm.sql as this will drop all content stored in Oracle XML DB Repository and invalidate any XMLType tables or columns associated with registered XML schemas.

Oracle XML DB Provides Name-Level Locking

When using a relational database to maintain hierarchical folder structures, ensuring a high degree of concurrency when adding and removing items in a folder is a challenge. In conventional file system there is no concept of a transaction. Each operation (add a file, create a subfolder, rename a file, delete a file, and so on) is treated as an atomic transaction. Once the operation has completed the change is immediately available to all other users of the file system.

One key advantage of Oracle XML DB Repository is the ability to use SQL for repository operations in the context of a logical transaction. Applications can create long-running transactions that include updates to one or more folders. In this situation a conventional locking strategy that takes an exclusive lock on each updated folder or directory tree would quickly result in significant concurrency problems.

Queued Folder Modifications are Locked Until Committed

Oracle XML DB solves this by providing for name-level locking rather than folder-level locking. Repository operations such as creating, renaming, moving, or deleting a sub-folder or file do not require that your operation be granted an exclusive write lock on the target folder. The repository manages concurrent folder operations by locking the name within the folder rather than the folder itself. The name and the modification type are put on a queue.Only when the transaction is committed is the folder locked and its contents modified. Hence Oracle XML DB allows multiple applications to perform concurrent updates on the contents of a folder. The queue is also used to manage folder concurrency by preventing two applications from creating objects with the same name.Queuing folder modifications until commit time also minimizes I/O when a number of changes are made to a single folder in the same transaction.This is useful when several applications generate files quickly in the same directory, for example when generating trace or log files, or when maintaining a spool directory for printing or email delivery.

Use Protocols or SQL to Access and Process Repository Content

There are two ways to work with content stored in Oracle XML DB Repository:

• Using industry standard protocols such as HTTP(S), WebDAV, or FTP to perform document level operations such as insert, update and delete.

• By directly accessing Oracle XML DB Repository content at the table or row level using SQL.

Using Standard Protocols to Store and Retrieve Content

Oracle XML DB supports industry-standard internet protocols such as HTTP(S), WebDav, and FTP. The combination of protocol support and URL-based access makes it possible to insert, retrieve, update, and delete content stored in Oracle Database from standard desktop applications such as Windows Explorer, Microsoft Word, and XMLSpy.

Figure 3-4 shows Windows Explorer used to insert a folder from the local hard drive into Oracle Database. Windows Explorer includes support for the WebDAV protocol. WebDAV extends the HTTP standard, adding additional verbs that allow an HTTP server to act as a file server.

When a Windows Explorer copy operation or FTP input command is used to transfer a number of documents into Oracle XML DB Repository, each put or post command is treated as a separate atomic operation. This ensures that the client does not get confused if one of the file transfers fails. It also means that changes made to a document through a protocol are visible to other users as soon as the request has been processed.

Figure 3-4 Copying Files into Oracle XML DB Repository

Description of "Figure 3-4 Copying Files into Oracle XML DB Repository"

Uploading Content Into Oracle XML DB Using FTP

The following example shows commands issued and output generated when a standard command line FTP tool loads documents into Oracle XML DB Repository:

$ ftp mdrake-sun 2100
Connected to mdrake-sun.
220 mdrake-sun FTP Server (Oracle XML DB/Oracle Database 10g Enterprise Edition
Release 10.1.0.1.0 - Beta) ready.
Name (mdrake-sun:oracle10): QUINE
331 pass required for QUINE
Password:
230 QUINE logged in
ftp> cd /source/schemas
250 CWD Command successful
ftp> mkdir PurchaseOrders
257 MKD Command successful
ftp> cd PurchaseOrders
250 CWD Command successful
ftp> mkdir 2002
257 MKD Command successful
ftp> cd 2002
250 CWD Command successful
ftp> mkdir "Apr"
257 MKD Command successful
ftp> put "Apr/AMCEWEN-20021009123336171PDT.xml"
"Apr/AMCEWEN-20021009123336171PDT.xml"
200 PORT Command successful
150 ASCII Data Connection
226 ASCII Transfer Complete
local: Apr/AMCEWEN-20021009123336171PDT.xml remote:
Apr/AMCEWEN-20021009123336171PDT.xml
4718 bytes sent in 0.0017 seconds (2683.41 Kbytes/s)
ftp> put "Apr/AMCEWEN-20021009123336271PDT.xml"
"Apr/AMCEWEN-20021009123336271PDT.xml"
200 PORT Command successful
150 ASCII Data Connection
226 ASCII Transfer Complete
local: Apr/AMCEWEN-20021009123336271PDT.xml remote:
Apr/AMCEWEN-20021009123336271PDT.xml
4800 bytes sent in 0.0014 seconds (3357.81 Kbytes/s)
.....
ftp> cd "Apr"
250 CWD Command successful
ftp> ls -l
200 PORT Command successful
150 ASCII Data Connection
-rw-r--r1 QUINE oracle 0 JUN 24 15:41 AMCEWEN-20021009123336171PDT.xml
-rw-r--r1 QUINE oracle 0 JUN 24 15:41 AMCEWEN-20021009123336271PDT.xml
-rw-r--r1 QUINE oracle 0 JUN 24 15:41 EABEL-20021009123336251PDT.xml
-rw-r--r1 QUINE oracle 0 JUN 24 15:41 PTUCKER-20021009123336191PDT.xml
-rw-r--r1 QUINE oracle 0 JUN 24 15:41 PTUCKER-20021009123336291PDT.xml
-rw-r--r1 QUINE oracle 0 JUN 24 15:41 SBELL-20021009123336231PDT.xml
-rw-r--r1 QUINE oracle 0 JUN 24 15:41 SBELL-20021009123336331PDT.xml
-rw-r--r1 QUINE oracle 0 JUN 24 15:41 SKING-20021009123336321PDT.xml
-rw-r--r1 QUINE oracle 0 JUN 24 15:41 SMCCAIN-20021009123336151PDT.xml
-rw-r--r1 QUINE oracle 0 JUN 24 15:41 SMCCAIN-20021009123336341PDT.xml
-rw-r--r1 QUINE oracle 0 JUN 24 15:41 VJONES-20021009123336301PDT.xml
226 ASCII Transfer Complete
remote: -l
959 bytes received in 0.0027 seconds (349.45 Kbytes/s)
ftp> cd ".."
250 CWD Command successful
....
ftp> quit
221 QUIT Goodbye.
$

The key point demonstrated by both of these examples is that neither Windows Explorer nor the FTP tool is aware that it is working with Oracle XML DB. Since the tools and Oracle XML DB both support open Internet protocols they simply work with each other out of the box.

Any tool that understands the WebDAV or FTP protocol can be used to create content managed by Oracle XML DB Repository. No additional software has to installed on the client or the mid-tier.

When the contents of the folders are viewed using a tool such as Windows Explorer or FTP, the length of any schema-based XML documents contained in the folder is shown as zero (0) bytes. This was designed as such for two reasons:

• It is not clear what the size of the document should be. Is it the size of the CLOB instance generated by printing the document, or the number of bytes required to store the objects used to persist the document inside the database?

• Regardless of which definition is chosen, calculating and maintaining this information is costly.

Figure 3-5 shows Internet Explorer using a URL and the HTTP protocol to view an XML document stored in the database.

Figure 3-5 Path-Based Access Using HTTP and a URL

Description of "Figure 3-5 Path-Based Access Using HTTP and a URL"

Accessing Oracle XML DB Repository Programmatically

Oracle XML DB Repository can be accessed and updated directly from SQL. This means that any application or programming language that can use SQL to interact with Oracle Database can also access and update content stored in the repository. Oracle XML DB includes PL/SQL package DBMS_XDB, which provides methods that allow resources to be created, modified, and deleted programmatically.

DECLARE
  res BOOLEAN;
BEGIN
  res := DBMS_XDB.createResource('/home/QUINE/NurseryRhyme.txt',
                                 bfilename('XMLDIR', 'tdadxdb-03-01.txt'),
                                 nls_charset_id('AL32UTF8'));
END;
/

Accessing and Updating XML Content in the Repository

This section describes features for accessing and updating Oracle XML DB Repository content.

Access XML Documents Using SQL

Content stored in the repository can be accessed and updated from SQL and PL/SQL. You can interrogate the structure of the repository in complex ways. For example, you can query to determine how many files with extension .xsl are under a location other than /home/mystylesheetdir.

You can also mix path-based repository access with content-based access. You can, for example, ask "How many documents not under /home/purchaseOrders have a node identified by the XPath /PurchaseOrder/User/text() with a value of KING?"

All of the metadata for managing the repository is stored in a database schema owned by the database user XDB. This user is created during Oracle XML DB installation. The primary table in this schema is an XMLType table called XDB$RESOURCE. This contains one row for each resource (file or folder) in the repository. Documents in this table are referred to as resource documents . The XML schema that defines the structure of an Oracle XML DB resource document is registered under URL, "http://xmlns.oralce.com/xdb/XDBResource.xsd.

Repository Content is Exposed Through RESOURCE_VIEW and PATH_VIEW

Table XDB$RESOURCE is not directly exposed to SQL programmers. Instead, the contents of the repository are exposed through two public views, RESOURCE_VIEW and PATH_VIEW. Through these views, you can access and update both the metadata and the content of documents stored in the repository. Both views contain a virtual column, RES. Use RES to access and update resource documents with SQL statements using a path notation. Operations on the views use underlying tables in the repository.

Use EXISTS_PATH and UNDER_PATH to Include Path-Based Predicates in the WHERE Clause

Oracle XML DB includes two repository-specific SQL functions: exists_path and under_path. Use these functions to include path-based predicates in the WHERE clause of a SQL statement. SQL operations can select repository content based on the location of the content in the repository folder hierarchy. The hierarchical index ensures that path-based queries are executed efficiently.

When XML schema-based XML documents are stored in the repository, the document content is stored as an object in the default table identified by the XML schema. The repository contains only metadata about the document and a pointer (REF of XMLType) that identifies the row in the default table that contains the content.

Documents Other Than XML Can Be Stored In the Repository

It is also possible to store other kinds of documents in the repository. When a document that is not XML or is not schema-based XML is stored in the repository, the document content is stored in a LOB along with the metadata about the document.

PL/SQL Packages to Create, Delete, Rename, Move, ... Folders and Documents

Since Oracle XML DB repository can be accessed and updated using SQL, any application capable of calling a PL/SQL procedure can use the repository. All SQL and PL/SQL repository operations are transactional, and access to the repository and its contents is subject to database security, as well as the repository Access Control Lists (ACLs).

With supplied PL/SQL packages DBMS_XDB, DBMS_XDBZ, and DBMS_XDB_VERSION, you can create, delete, and rename documents and folders, move a file or folder within the folder hierarchy, set and change the access permissions on a file or folder, and initiate and manage versioning.

DECLARE
  RESULT BOOLEAN;
BEGIN
  IF (NOT DBMS_XDB.existsResource('/public/mysource')) THEN
     result := DBMS_XDB.createFolder('/public/mysource');
  END IF;
  IF (NOT DBMS_XDB.existsResource('/public/mysource/schemas')) THEN
     result := DBMS_XDB.createFolder('/public/mysource/schemas');
  END IF;
  IF (NOT DBMS_XDB.existsResource('/public/mysource/schemas/poSource')) THEN
     result := DBMS_XDB.createFolder('/public/mysource/schemas/poSource');
  END IF;
  IF (NOT DBMS_XDB.existsResource('/public/mysource/schemas/poSource/xsd')) THEN
     result := DBMS_XDB.createFolder('/public/mysource/schemas/poSource/xsd');
  END IF;
  IF (NOT DBMS_XDB.existsResource('/public/mysource/schemas/poSource/xsl')) THEN
     result := DBMS_XDB.createFolder('/public/mysource/schemas/poSource/xsl');
  END IF;
END;
/

Accessing the Content of Documents Using SQL

You can access the content of documents stored in Oracle XML DB Repository in several ways. The easiest way is to use XDBURIType. XDBURIType uses a URL to specify which resource to access. The URL passed to the XDBURIType is assumed to start at the root of the repository. Datatype XDBURIType provides methods getBLOB(), getCLOB(), and getXML() to access the different kinds of content that can be associated with a resource.

SELECT XDBURIType('/home/QUINE/NurseryRhyme.txt').getClob()
  FROM DUAL;
 
XDBURITYPE('/HOME/QUINE/NURSERYRHYME.TXT').GETCLOB()
----------------------------------------------------
Mary had a little lamb
Its fleece was white as snow
and everywhere that Mary went
that lamb was sure to go
 
1 row selected.

SELECT
  DBMS_XMLGEN.convert(
    extract(RES,
            '/Resource/Contents/text/text()',
            'xmlns="http://xmlns.oracle.com/xdb/XDBResource.xsd"').getClobVal(),
    1)
  FROM RESOURCE_VIEW r
  WHERE equals_path(RES, '/home/QUINE/NurseryRhyme.txt') = 1;

DBMS_XMLGEN.CONVERT(EXTRACT(RES,'/RESOURCE/CONTENTS/TEXT/TEXT()','XMLNS="HTTP://
--------------------------------------------------------------------------------
Mary had a little lamb
Its fleece was white as snow
and everywhere that Mary went
that lamb was sure to go
 
1 row selected.

SELECT extractValue(value(des), '/Description')
  FROM RESOURCE_VIEW r,
       table(
         XMLSequence(
           extract(RES,
                   '/r:Resource/r:Contents/PurchaseOrder/LineItems/LineItem/Description',
                   'xmlns:r="http://xmlns.oracle.com/xdb/XDBResource.xsd"'))) des
  WHERE 
    equals_path(RES, '/home/QUINE/PurchaseOrders/2002/Mar/SBELL-2002100912333601PDT.xml') = 1;

EXTRACTVALUE(VALUE(L),'/DESCRIPTION')
--------------------------------------------
A Night to Remember
The Unbearable Lightness Of Being
The Wizard of Oz
 
3 rows selected.

Accessing the Content of XML Schema-Based Documents

The content of a schema-based XML document can be accessed in two ways.

• In the same manner as for non-schema-based XML documents, by using the resource document. This allows the RESOURCE_VIEW to be used to query different types of schema-based XML documents with a single SQL statement.

• As a row in the default table that was defined when the XML schema was registered with Oracle XML DB.

SELECT extractValue(value(des), '/Description')
  FROM RESOURCE_VIEW r, 
       purchaseorder p,
       table(XMLSequence(extract(OBJECT_VALUE,
                                 '/PurchaseOrder/LineItems/LineItem/Description'))) des
 WHERE equals_path(res, '/home/QUINE/PurchaseOrders/2002/Mar/SBELL-2002100912333601PDT.xml') = 1
   AND ref(p) = extractValue(res, '/Resource/XMLRef');
 
EXTRACTVALUE(VALUE(L),'/DESCRIPTION')
-------------------------------------
A Night to Remember
The Unbearable Lightness Of Being
The Wizard of Oz
 
3 rows selected.

SELECT extractValue(OBJECT_VALUE, '/PurchaseOrder/Reference')
  FROM RESOURCE_VIEW r, purchaseorder p
  WHERE under_path(res, '/home/QUINE/PurchaseOrders/2002/Mar') = 1
    AND ref(p) = extractValue(res, '/Resource/XMLRef')
    AND existsNode(OBJECT_VALUE,
                   '/PurchaseOrder/LineItems/LineItem/Part[@Id="715515009058"]')
        = 1;
 
EXTRACTVALUE(OBJECT_VALUE,'/PU
------------------------------
CJOHNSON-20021009123335851PDT
LSMITH-2002100912333661PDT
SBELL-2002100912333601PDT
 
3 rows selected.

Updating the Content of Documents Stored in the Repository

You can also update the content of documents stored in Oracle XML DB Repository using protocols or SQL.

Updating Repository Content Using Protocols

The most popular content authoring tools now support HTTP, FTP, and WebDAV protocols. These tools can use a URL and the HTTP verb get to access the content of a document, and the HTTP verb put to save the contents of a document. Hence, given the appropriate access permissions, a simple URL is all you need to access and edit content stored in Oracle XML DB Repository.

Figure 3-6 shows how, with the WebDAV support included in Microsoft Word, you can use Microsoft Word to update and edit a document stored in Oracle XML DB Repository.

Figure 3-6 Using Microsoft Word to Update and Edit Content Stored in Oracle XML DB

Description of "Figure 3-6 Using Microsoft Word to Update and Edit Content Stored in Oracle XML DB"

When an editor like Microsoft Word updates an XML document stored in Oracle XML DB the database receives an input stream containing the new content of the document. Unfortunately products such as Word do not provide Oracle XML DB with any way of identifying what changes have taken place in the document. This means that partial updates are not possible and it is necessary to re-parse the entire document, replacing all the objects derived from the original document with objects derived from the new content.

DECLARE
  file         BFILE;
  contents     CLOB;
  dest_offset  NUMBER := 1;
  src_offset   NUMBER := 1;
  lang_context NUMBER := 0;
  conv_warning NUMBER := 0;
BEGIN
  file := bfilename('XMLDIR', 'tdadxdb-03-02.txt');
  DBMS_LOB.createTemporary(contents, true, DBMS_LOB.SESSION);
  DBMS_LOB.fileopen(file, DBMS_LOB.file_readonly);
  DBMS_LOB.loadClobfromFile(contents,
                            file,
                            DBMS_LOB.getLength(file),
                            dest_offset,
                            src_offset,
                            nls_charset_id('AL32UTF8'),
                            lang_context,
                            conv_warning);
  DBMS_LOB.fileclose(file);
  UPDATE RESOURCE_VIEW
    SET res = updateXML(res,
                        '/Resource/Contents/text/text()',
                        contents,
                        'xmlns="http://xmlns.oracle.com/xdb/XDBResource.xsd"')
      WHERE equals_path(res, '/home/QUINE/NurseryRhyme.txt') = 1;
  DBMS_LOB.freeTemporary(contents);
END;
/

UPDATE RESOURCE_VIEW
  SET res = updateXML(res,
                      '/r:Resource/r:Contents/PurchaseOrder/User/text()',
                      'SKING',
                      'xmlns:r="http://xmlns.oracle.com/xdb/XDBResource.xsd"')
  WHERE equals_path(
          res,
          '/home/QUINE/PurchaseOrders/2002/Mar/SBELL-2002100912333601PDT.xml')
        = 1;
 
1 row updated.

SELECT extractValue(res,
                    '/r:Resource/r:Contents/PurchaseOrder/User/text()',
                    'xmlns:r="http://xmlns.oracle.com/xdb/XDBResource.xsd"')
  FROM RESOURCE_VIEW
  WHERE equals_path(
          res,
          '/home/QUINE/PurchaseOrders/2002/Mar/SBELL-2002100912333601PDT.xml')
        = 1;

EXTRACTVALUE(RES, '/R:RESOURCE/R:CONTENTS/PURCHASEORDER/USER/TEXT()', 
             'XMLNS:R="HTTP://XMLNS.ORACLE.COM/XDB/XDBRESOURCE.XSD"')
---------------------------------------------------------------------
SKING

1 row selected.

UPDATE purchaseorder p
  SET OBJECT_VALUE = updateXML(OBJECT_VALUE, '/PurchaseOrder/User/text()', 'SBELL')
    WHERE ref(p) =
      (SELECT extractValue(res,'/Resource/XMLRef')
         FROM RESOURCE_VIEW
         WHERE equals_path(res, 
                           '/home/QUINE/PurchaseOrders/2002/Mar/SBELL-2002100912333601PDT.xml') 
               = 1);

1 row updated.

SELECT extractValue(OBJECT_VALUE, '/PurchaseOrder/User/text()')
  FROM purchaseorder p, RESOURCE_VIEW
  WHERE ref(p) = extractValue(res, '/Resource/XMLRef')
    AND equals_path(res, '/home/QUINE/PurchaseOrders/2002/Mar/SBELL-2002100912333601PDT.xml')
        = 1;
 
EXTRACTVAL
----------
SBELL
 
1 row selected.

Controlling Access to Repository Data

You can control access to the resources in Oracle XML DB Repository by using Access Control Lists (ACLs). An ACL is a list of access control entries, each of which grants or denies a set of privileges to a specific principal. The principal can be a database user, a database role, an LDAP user, an LDAP group or the special principal dav:owner, which refers to the owner of the resource. Each resource in the repository is protected by an ACL. The ACL determines what privileges, such as read-properties and update, a user has on the resource. Each repository operation includes a check of the ACL to determine if the current user is allowed to perform the operation. By default, a new resource inherits the ACL of its parent folder. But you can set the ACL of a resource using procedure DBMS_XDB.setACL(). For more details on Oracle XML DB resource security, see Chapter 24, "Repository Resource Security".

In the following example, the current user is QUINE. The query gives the number of resources in the folder /public. Assume that there are only two resources in this folder: f1 and f2. Also assume that the ACL on f1 grants the read-properties privilege to QUINE while the ACL on f2 does not grant QUINE any privileges. A user needs the read-properties privilege on a resource for it to be visible to the user. The result of the query is 1, because only f1 is visible to QUINE.

SELECT count(*) FROM RESOURCE_VIEW r 
  WHERE under_path(r.res, '/public') = 1;
 
  COUNT(*)
----------
         1

Oracle XML DB Transactional Semantics

When working from SQL, normal transactional behavior is enforced. Multiple calls to SQL functions such as updateXML can be used within a single logical unit of work. Changes made through functions like updateXML are not visible to other database users until the transaction is committed. At any point, ROLLBACK can be used to back out the set of changes made since the last commit.

Querying Metadata and the Folder Hierarchy

In Oracle XML DB, the system-defined metadata for each resource is preserved as an XML document. The structure of these resource documents is defined by the XDBResource.xsd XML schema. This schema is registered as a global XML schema at URL http://xmlns.oracle.com/xdb/XDBResource.xsd.

Oracle XML DB allows you access to metadata and information about the folder hierarchy using two public views, RESOURCE_VIEW and PATH_VIEW.

DESCRIBE RESOURCE_VIEW

Name      Null?    Type
-------------------------------------------------------------
RES                SYS.XMLTYPE(XMLSchema 
                               "http://xmlns.oracle.com/xdb/XDBResource.xsd" 
                               Element 
                               "Resource")
ANY_PATH           VARCHAR2(4000)
RESID              RAW(16)


DESCRIBE PATH_VIEW

Name      Null?    Type
-------------------------------------------------------------
PATH               VARCHAR2(1024)
RES                SYS.XMLTYPE(XMLSchema
                               "http://xmlns.oracle.com/xdb/XDBResource.xsd" 
                               Element 
                               "Resource")
LINK               SYS.XMLTYPE
RESID              RAW(16)

SELECT r.RES.getClobVal()
  FROM RESOURCE_VIEW r
  WHERE equals_path(res, '/home/QUINE/NurseryRhyme.txt') = 1;
 
R.RES.GETCLOBVAL()
--------------------------------------------------------------------------------
<Resource xmlns="http://xmlns.oracle.com/xdb/XDBResource.xsd" 
          Hidden="false" 
          Invalid="false" 
          Container="false" 
          CustomRslv="false" 
          VersionHistory="false" 
          StickyRef="true">
  <CreationDate>2004-08-06T12:12:48.022251</CreationDate>
  <ModificationDate>2004-08-06T12:12:53.215519</ModificationDate>
  <DisplayName>NurseryRhyme.txt</DisplayName>
  <Language>en-US</Language>
  <CharacterSet>UTF-8</CharacterSet>
  <ContentType>text/plain</ContentType>
  <RefCount>1</RefCount>
  <ACL>
    <acl description=
           "Private:All privileges to OWNER only and not accessible to others"
         xmlns="http://xmlns.oracle.com/xdb/acl.xsd" xmlns:dav="DAV:"
         xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" 
         xsi:schemaLocation="http://xmlns.oracle.com/xdb/acl.xsd
         http://xmlns.oracle.com/xdb/acl.xsd">
      <ace>
        <principal>dav:owner</principal>
        <grant>true</grant>
        <privilege>
          <all/>
        </privilege>
      </ace>
    </acl>
  </ACL>
  <Owner>QUINE</Owner>
  <Creator>QUINE</Creator>
  <LastModifier>QUINE</LastModifier>
  <SchemaElement>http://xmlns.oracle.com/xdb/XDBSchema.xsd#text</SchemaElement>
  <Contents>
    <text>Hickory Dickory Dock
The Mouse ran up the clock
The clock struck one
The Mouse ran down
Hickory Dickory Dock
    </text>
  </Contents>
</Resource>
 
1 row selected.

SELECT ANY_PATH FROM RESOURCE_VIEW
  WHERE extractValue(RES, '/Resource/DisplayName') LIKE '%.xsl';
 
ANY_PATH
-------------------------------------------
/source/schemas/poSource/xsl/empdept.xsl
/source/schemas/poSource/xsl/purchaseOrder.xsl
 
2 rows selected.

SELECT count(*)
   FROM RESOURCE_VIEW
   WHERE under_path(RES, '/home/QUINE/PurchaseOrders') = 1;
 
  COUNT(*)
----------
       145

1 row selected.

SELECT PATH
  FROM PATH_VIEW
  WHERE under_path(RES, '/home/QUINE/PurchaseOrders/2002/Apr') = 1;
 
PATH
----------------------------------------------------------------------
/home/QUINE/PurchaseOrders/2002/Apr/AMCEWEN-20021009123336171PDT.xml
/home/QUINE/PurchaseOrders/2002/Apr/AMCEWEN-20021009123336271PDT.xml
/home/QUINE/PurchaseOrders/2002/Apr/EABEL-20021009123336251PDT.xml
/home/QUINE/PurchaseOrders/2002/Apr/PTUCKER-20021009123336191PDT.xml
/home/QUINE/PurchaseOrders/2002/Apr/PTUCKER-20021009123336291PDT.xml
/home/QUINE/PurchaseOrders/2002/Apr/SBELL-20021009123336231PDT.xml
/home/QUINE/PurchaseOrders/2002/Apr/SBELL-20021009123336331PDT.xml
/home/QUINE/PurchaseOrders/2002/Apr/SKING-20021009123336321PDT.xml
/home/QUINE/PurchaseOrders/2002/Apr/SMCCAIN-20021009123336151PDT.xml
/home/QUINE/PurchaseOrders/2002/Apr/SMCCAIN-20021009123336341PDT.xml
/home/QUINE/PurchaseOrders/2002/Apr/VJONES-20021009123336301PDT.xml
 
11 rows selected.

SELECT PATH
  FROM PATH_VIEW
  WHERE extractValue(RES, '/Resource/DisplayName') like 'S%'
    AND under_path(RES, '/home/QUINE/PurchaseOrders/2002/Apr') = 1;
 
PATH
----------------------------------------------------------------------
/home/QUINE/PurchaseOrders/2002/Apr/SBELL-20021009123336231PDT.xml
/home/QUINE/PurchaseOrders/2002/Apr/SBELL-20021009123336331PDT.xml
/home/QUINE/PurchaseOrders/2002/Apr/SKING-20021009123336321PDT.xml
/home/QUINE/PurchaseOrders/2002/Apr/SMCCAIN-20021009123336151PDT.xml
/home/QUINE/PurchaseOrders/2002/Apr/SMCCAIN-20021009123336341PDT.xml
 
5 rows selected.

SELECT ANY_PATH
  FROM RESOURCE_VIEW
  WHERE existsNode(RES,
                   '/Resource[SchemaElement=
                      "http://localhost:8080/source/schemas/poSource/xsd/purchaseOrder.xsd#PurchaseOrder"]')
        = 1;

ANY_PATH
-----------------------------------------------------------------------
/home/QUINE/PurchaseOrders/2002/Apr/AMCEWEN-20021009123336171PDT.xml
/home/QUINE/PurchaseOrders/2002/Apr/AMCEWEN-20021009123336271PDT.xml
/home/QUINE/PurchaseOrders/2002/Apr/EABEL-20021009123336251PDT.xml
/home/QUINE/PurchaseOrders/2002/Apr/PTUCKER-20021009123336191PDT.xml

...

132 rows selected.

Oracle XML DB Hierarchical Index

In a conventional relational database, path-based access and folder-restricted queries would have to be implemented using CONNECT BY operations. Such queries are expensive, so path-based access and folder-restricted queries would become inefficient as the number of documents and depth of the folder hierarchy increase.

To address this issue, Oracle XML DB introduces a new index type, the hierarchical index. A hierarchical index allows the database to resolve folder-restricted queries without relying on a CONNECT BY operation. Hence Oracle XML DB can execute path-based and folder-restricted queries efficiently. The hierarchical index is implemented as an Oracle domain index. This is the same technique used to add Oracle Text indexing support and many other advanced index types to the database.

EXPLAIN PLAN FOR
  SELECT PATH
    FROM PATH_VIEW
    WHERE extractValue(RES, '/Resource/DisplayName') LIKE 'S%'
      AND under_path(RES, '/home/QUINE/PurchaseOrders/2002/Apr') = 1;
 
Explained.
 
PLAN_TABLE_OUTPUT
------------------------------------------------------------------------------------------------------
Plan hash value: 2568289845

------------------------------------------------------------------------------------------------------
| Id  | Operation                            | Name          | Rows  | Bytes | Cost (%CPU)| Time     |
------------------------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT                     |               |   127 | 22606 |    35   (9)| 00:00:01 |
|   1 |  NESTED LOOPS                        |               |   127 | 22606 |    35   (9)| 00:00:01 |
|   2 |   NESTED LOOPS                       |               |   127 | 20447 |    34   (6)| 00:00:01 |
|   3 |    NESTED LOOPS                      |               |   127 | 16891 |    34   (6)| 00:00:01 |
|*  4 |     TABLE ACCESS BY INDEX ROWID      | XDB$RESOURCE  |     1 |   131 |     3   (0)| 00:00:01 |
|*  5 |      DOMAIN INDEX                    | XDBHI_IDX     |       |       |            |          |
|   6 |     COLLECTION ITERATOR PICKLER FETCH|               |       |       |            |          |
|*  7 |    INDEX UNIQUE SCAN                 | XDB_PK_H_LINK |     1 |    28 |     0   (0)| 00:00:01 |
|*  8 |   INDEX UNIQUE SCAN                  | SYS_C003013   |     1 |    17 |     0   (0)| 00:00:01 |
------------------------------------------------------------------------------------------------------
 
Predicate Information (identified by operation id):
---------------------------------------------------
 
   4 - filter("P"."SYS_NC00011$" LIKE 'S%')
   5 - access("XDB"."UNDER_PATH"(SYS_MAKEXML('8758D485E6004793E034080020B242C6',734,"XMLEXTRA"
              ,"XMLDATA"),'/home/QUINE/PurchaseOrders/2002/Apr',9999)=1)
   7 - access("H"."PARENT_OID"=SYS_OP_ATG(VALUE(KOKBF$),3,4,2) AND
              "H"."NAME"=SYS_OP_ATG(VALUE(KOKBF$),2,3,2))
   8 - access("R2"."SYS_NC_OID$"=SYS_OP_ATG(VALUE(KOKBF$),3,4,2))
 
25 rows selected.

How Documents are Stored in the Repository

Oracle XML DB provides special handling for XML documents. The rules for storing the contents of schema-based XML document are defined by the XML schema. The content of the document is stored in the default table associated with the global element definition.

Oracle XML DB Repository also stores files that do not contain XML data, such as JPEG images or Word documents. The XML schema for each resource defines which elements are allowed, and specifies whether the content of these files is to be stored as BLOB or CLOB instances. The content of a non-schema-based XML document is stored as a CLOB instance in the repository.

There is one resource and one link-properties document for every file or folder in the repository. If there are multiple access paths to a given document there will be a link-properties document for each possible link. Both the resource document and the link-properties are stored as XML documents. All these documents are stored in tables in the repository.

When an XML file is loaded into the repository, the following sequence of events that takes place:

1. Oracle XML DB examines the root element of the XML document to see if it is associated with a known (registered) XML schema. This involves looking to see if the document includes a namespace declaration for the XMLSchema-instance namespace, and then looking for a schemaLocation or noNamespaceSchemaLocation attribute that identifies which XML schema the document is associated with.

2. If the document is based on a known XML schema, then the metadata for the XML schema is loaded from the XML schema cache.

3. The XML document is parsed and decomposed into a set the SQL objects derived from the XML schema.

4. The SQL objects created from the XML file are stored in the default table defined when the XML schema was registered with the database.

5. A resource document is created for each document processed. This allows the content of the document to be accessed using the repository. The resource document for a schema-based XMLType includes an element XMLRef. This contents of this element is a REF of XMLType that can be used to locate the row in the default table containing the content associated with the resource.

Using DBUri Servlet to Access Any Table or View From a Browser

Oracle XML DB includes the DBUri servlet that makes it possible to access the content of any table or view directly from a browser. DBUri servlet uses the facilities of the DBURIType to generate a simple XML document from the contents of the table. The servlet is C- based and installed in the Oracle XML DB HTTP server. By default the servlet is installed under the virtual directory /oradb.

The URL passed to the DBUri Servlet is an extension of the URL passed to the DBURIType. The URL is simply extended with the address and port number of the Oracle XML DB HTTP server and the virtual root that directs HTTP(S) requests to the DBUri servlet. The default configuration for this is /oradb.

This means that the URL: http://localhost:8080/oradb/HR/DEPTARTMENTS,

would return an XML document containing the contents of the DEPARTMENTS table in the HR database schema, assuming that the Oracle XML DB HTTP server is running on port 8080, the virtual root for the DBUri servlet is /oradb, and that the user making the request has access to the HR database schema.

DBUri servlet accepts parameters that allow you to specify the name of the ROW tag and MIME-type of the document that is returned to the client.

Content in XMLType table or view can also be accessed through the DBUri servlet. When the URL passed to the DBUri servlet references an XMLType table or XMLType view the URL can be extended with an XPath expression that can determine which documents in the table or row are returned. The XPath expression appended to the URL can reference any node in the document.

XML generated by DBUri servlet can be transformed using the XSLT processor built into Oracle XML DB. This allows XML generated by DBUri servlet to be presented in a more legible format such as HTML.

Style sheet processing is initiated by specifying a transform parameter as part of the URL passed to DBUri servlet. The style sheet is specified using a URI that references the location of the style sheet within database. The URI can either be a DBURIType value that identifies a XMLType column in a table or view, or a path to a document stored in Oracle XML DB Repository. The style sheet is applied directly to the generated XML before it is returned to the client. When using DBUri servlet for XSLT processing it is good practice to use the contenttype parameter to explicitly specify the MIME type of the generated output.

If the XML document being transformed is stored as schema-based XMLType, then Oracle XML DB can reduce the overhead associated with XSL transformation by leveraging the capabilities of the lazily loaded virtual DOM.

Figure 3-7 shows how DBUri can access a row in the purchaseorder table.

Figure 3-7 Using DBUri Servlet to Access XML Content

Description of "Figure 3-7 Using DBUri Servlet to Access XML Content"

The root of the URL is /oradb, so the URL is passed to the DBUri servlet that accesses the purchaseorder table in the SCOTT database schema, rather than as a resource in Oracle XML DB Repository. The URL includes an XPath expression that restricts the result set to those documents where node /PurchaseOrder/Reference/text() contains the value specified in the predicate. The contenttype parameter sets the MIME type of the generated document to text/xml.