Service Data Objects (SDOs) enable PHP applications to work with data from different sources (like a database query, an XML file, and a spreadsheet) using a single interface.
Each different kind of data source requires a Data Access Service (DAS) to provide access to the data in the data source. In your PHP application, you use a DAS to create an SDO instance that represents some data in the data source. You can then set and get values in the SDO instance using the standard SDO interface. Finally, you use a DAS to write the modified data back to a data source, typically the same one.
See the list of Data Access Services for details on those currently available. In addition to the provided DASs, SDO also provides interfaces to enable others to be implemented (see the section on SDO Data Access Services Interface for more details).
This extension is derived from concepts taken from the Service Data Objects specification. It includes a version of the Apache Tuscany SDO 2.0 for C++ project.
A Service Data Object instance is made up of a tree of data objects. The tree is defined by containment relationships between the data objects. For example, a Company data object might consist of a number of Department data objects and therefore the Company would have a containment relationship to the Departments.
An SDO may also have non-containment references between data objects in the tree. For example, one Employee data object might reference another Employee to identify a career mentor.
As well as data objects referencing each other, they can also have primitive properties. For example, the Company data object might have a property called "name" of type string, for holding the name of the company (for example, "Acme").
Each of these properties of a data object - containment relationships, non-containment references, or primitive properties - may be many-valued or single-valued. In the above examples, Departments is many-valued and the Company name is single-valued.
In PHP, each SDO data object is represented as a PHP object. The properties of the data object can be accessed using either object syntax or associative array syntax. We'll see some examples of this later.
The SDO extension requires PHP 5.1 or higher. It also requires the libxml2 library. Normally libxml2 will already be installed, but if not, it can be downloaded from http://www.xmlsoft.org/.
You can download and install the latest stable release of all three SDO components - the SDO core, the XML DAS and the Relational DAS - with the command:
pear install sdo |
This command will build the SDO and XML shared libraries as well as installing the PHP files that make the Relational DAS.
If you want to use the latest beta version, then instead run:
pear install sdo-beta |
The pear command automatically installs the SDO and SDO_DAS_XML modules into your PHP extensions directory. To enable the SDO extensions you must add the following lines to php.ini:
extension=sdo.so extension=sdo_das_xml.so |
For more information about building PECL packages, consult the PECL installation section of the manual.
The latest DLLs for the SDO core and the XML DAS can be downloaded from php_sdo.dll and php_sdo_das_xml.dll respectively.
Note that currently the pecl4win site does not provide these binaries at the current release level; you can only download the latest level.
The pear command automatically installs the SDO and SDO_DAS_XML modules into your PHP extensions directory. To enable the SDO extensions you must add the following lines to php.ini:
extension=php_sdo.dll extension=php_sdo_das_xml.dll |
The Relational DAS can be downloaded and installed with the command:
pear install -B sdo |
The Relational DAS is written in PHP. You may need to update your include_path in php.ini to point to the directory that contains sdo/DAS/Relational.
This section describes how to build the SDO core and XML DAS on Linux. You would only need to know how to do this if you wish to build a recent version that you have checked out of CVS.
Change to the main extension directory: cd < wherever your sdo code is >
Run phpize, which will set up the environment to compile both SDO and the XML Data Access Service.
Next, run ./configure; make; make install. Please note, you may need to login as root to install the extension.
Make sure that these modules are loaded by PHP, by adding extension=sdo.so and extension=sdo_das_xml.so to your php.ini file in the same order.
The table below lists the currently provided SDO Data Access Services:
| DAS Name | Description |
|---|---|
| SDO_DAS_XML | An XML Data Access Service supporting reading/writing SDOs as XML documents. |
| SDO_DAS_Relational | A PDO-based Data Access Service supporting reading/writing SDO to relational databases. Implements an optimistic concurrency policy for updates. |
The following are limitations in the current SDO implementation:
There is no support for multi-byte character sets. This will be considered, depending on community requirements, in the Unicode-enabled version of PHP. See Unicode Functions.
The following SDO 2.0 concepts are not supported in the current PHP implementation. It is not necessarily the case that these will all be added over time. Their inclusion will depend on community requirements.
Bi-directional relationships.
Type and property alias names.
Read-only properties.
The Helper classes defined in SDO 2.0 are not directly implemented. However equivalent function is provided in a more natural way for PHP. For example the function of CopyHelper::copy() is provided by applying the PHP clone keyword to a data object.
The examples below assume an SDO created with the schema and instance information shown below, using the XML Data Access Service.
The instance document below describes a single company, called 'MegaCorp', which contains a single department, called 'Advanced Technologies'. The Advanced Technologies department contains three employees. The company employeeOfTheMonth is referencing the second employee, 'Jane Doe'.
<?xml version="1.0" encoding="UTF-8" ?>
<company xmlns="companyNS" name="MegaCorp"
employeeOfTheMonth="E0003">
<departments name="Advanced Technologies" location="NY" number="123">
<employees name="John Jones" SN="E0001"/>
<employees name="Jane Doe" SN="E0003"/>
<employees name="Al Smith" SN="E0004" manager="true"/>
</departments>
</company> |
The root element of the schema is a company. The company contains departments, and each department contains employees. Each element has a number of attributes to store things like name, serial number, and so on. Finally, the company also has an IDREF attribute which identifies one of the employees as the 'employeeOfTheMonth'.
<xsd:schema
xmlns:xsd="http://www.w3.org/2001/XMLSchema"
xmlns:sdo="commonj.sdo"
xmlns:sdoxml="commonj.sdo/xml"
xmlns:company="companyNS"
targetNamespace="companyNS">
<xsd:element name="company" type="company:CompanyType"/>
<xsd:complexType name="CompanyType">
<xsd:sequence>
<xsd:element name="departments" type="company:DepartmentType"
maxOccurs="unbounded"/>
</xsd:sequence>
<xsd:attribute name="name" type="xsd:string"/>
<xsd:attribute name="employeeOfTheMonth" type="xsd:IDREF"
sdoxml:propertyType="company:EmployeeType"/>
</xsd:complexType>
<xsd:complexType name="DepartmentType">
<xsd:sequence>
<xsd:element name="employees" type="company:EmployeeType"
maxOccurs="unbounded"/>
</xsd:sequence>
<xsd:attribute name="name" type="xsd:string"/>
<xsd:attribute name="location" type="xsd:string"/>
<xsd:attribute name="number" type="xsd:int"/>
</xsd:complexType>
<xsd:complexType name="EmployeeType">
<xsd:attribute name="name" type="xsd:string"/>
<xsd:attribute name="SN" type="xsd:ID"/>
<xsd:attribute name="manager" type="xsd:boolean"/>
</xsd:complexType>
</xsd:schema> |
The XML Data Access Service maps the schema to an SDO. Attributes such as "name" become primitive properties, the sequence of employees becomes a many-valued containment relationship, and so on. Note that the containment relationships are expressed as one complex type within another, whereas non-containment references are expressed in terms of ID and IDREF, with a special sdoxml:propertyType attribute specifying the type of the non-containment reference.
The following examples assume $company is the root of a tree of data objects created from the schema and instance document shown above.
Example 3. Data Object iteration We can iterate over the properties of a data object using foreach. The following iterates over the properties of the employee of the month.
which will output:
The 'manager' property is not output, because it has not been set. |
Example 6. Many-valued property iteration Many-valued properties can also be iterated over using foreach. The following iterates over the company's departments.
Each iteration will assign the next department in the list to the variable $department. |
Example 7. Chained property access We can chain property references on a single line. The following sets and gets the name of the first department.
Using the associative array syntax, this is equivalent to
In either case, the dept_name variable is set to 'Emerging Technologies'. |
Example 8. XPath navigation The associative array index can be an XPath-like expression. Valid expressions are defined by an augmented sub-set of XPath. Two forms of indexing into many-valued properties are supported. The first is the standard XPath array syntax with the indexing starting at one, the second is an SDO extension to XPath with an index starting at zero. The standard syntax is:
and the SDO XPath extension syntax is:
Both these examples get the second employee from the first department. |
Example 10. Creating child data objects A data object can be a factory for its child data objects. A child data object is automatically part of the data graph. The following add a new employee to the 'Advanced Technologies' department.
|
Example 13. Unset a referenced data object The following removes the 'employeeOfTheMonth' from the company. If this were a containment relationship then the employee would be removed from the company (probably not a good idea to sack your best employee each month!), but since this is a non-containment reference, the employee being referenced will remain in the department in the company, but will no longer be accessible via the employeeOfTheMonth property.
|
Example 14. Access via property index Data object properties can be accessed via their property index using array syntax. The property index is the position at which the property's definition appears in the model (in this case the xml schema). We can see from the schema listing above that the company name attribute is the second company property (the SDO interface makes no distinction between XML attributes and elements). The following sets the company name to 'Acme', with the same result as Access via property name
Using the index directly in this way is likely to be fragile. Normally the property name syntax should be preferred, but the property index may be required in special cases. |
Sequenced data objects are SDOs which can track property ordering across the properties of a data object. They can also contain unstructured text elements (text element which do not belong to any of the SDO's properties). Sequenced data objects are useful for working with XML documents which allow unstructured text (i.e. mixed=true) or if the elements can be interleaved ( <A/><B/><A/>). This can occur for example when the schema defines maxOccurs>1 on a element which is a complexType with a choice order indicator.
The examples below assume an SDO created with the following schema and instance information, using the XML Data Access Service.
The schema below describes the format of a letter. The letter can optionally contain three properties; date, firstName, and lastName. The schema states mixed="true" which means that unstructured text can be interspersed between the three properties.
<xsd:schema xmlns:xsd="http://www.w3.org/2001/XMLSchema"
xmlns:letter="http://letterSchema"
targetNamespace="http://letterSchema">
<xsd:element name="letters" type="letter:FormLetter"/>
<xsd:complexType name="FormLetter" mixed="true">
<xsd:sequence>
<xsd:element name="date" minOccurs="0" type="xsd:string"/>
<xsd:element name="firstName" minOccurs="0" type="xsd:string"/>
<xsd:element name="lastName" minOccurs="0" type="xsd:string"/>
</xsd:sequence>
</xsd:complexType>
</xsd:schema> |
The following is an instance letter document. It contains the three letter properties; date, firstName and lastName, and has unstructured text elements for the address and letter body.
<letter:letters xmlns:letter="http://letterSchema"> <date>March 1, 2005</date> Mutual of Omaha Wild Kingdom, USA Dear <firstName>Casy</firstName> <lastName>Crocodile</lastName> Please buy more shark repellent. Your premium is past due. </letter:letters> |
When loaded, the letter data object will have the sequence and property indices shown in the table below:
| Sequence Index | Property Index:Name | Value |
|---|---|---|
| 0 | 0:date | March 1, 2005 |
| 1 | - | Mutual of Omaha |
| 2 | - | Wild Kingdom, USA |
| 3 | - | Dear |
| 4 | 1:firstName | Casy |
| 5 | 2:lastName | Crocodile |
| 6 | - | Please buy more shark repellent. |
| 7 | - | Your premium is past due. |
To ensure sequence indices are maintained, sequenced data objects should be manipulated through the SDO_Sequence interface. This allows the data object's instance data to be manipulated in terms of the sequence index as opposed to the property index (shown in the table above). The following examples assume the letter instance has been loaded into a data object referenced by the variable $letter.
All subsequent examples assume that the $letter_seq variable has been assigned the sequence for the letter data object.
Example 16. Get/set sequence values We can get and set individual values (including unstructured text) using the sequence index. The following sets the firstName to 'Snappy' and gets the last sequence values (the unstructured text, 'Your premium is past due.').
|
Example 18. Sequence versus Data Object Setting values through the data object interface may result in the value not being part of the sequence. A value set through the data object will only be accessible through the sequence if the property was already part of the sequence. The following example sets the lastName through the data object and gets it through the sequence. This is fine because lastName already exists in the sequence. If it had not previously been set, then lastName would be set to 'Smith', but would not be part of the sequence.
|
Example 19. Adding to a sequence We can add new values to a sequence using the SDO_Sequence::insert() method. The following examples assume that the 'firstName' and 'lastName' properties are initially unset.
|
Example 20. Removing from a sequence We can use the isset() and unset() functions to test and remove items from the sequence (Note: unset() currently leaves the values in the data object, but this behaviour is likely to change to also remove the data from the data object). A sequence behaves like a contiguous list; therefore, removing items from the middle will shift entries at higher indices down. The following example tests to see if the first sequence element is set and unsets it if is.
|
SDOs have a knowledge of the structure they have been created to represent (the model). For example, a Company SDO created using the Company XML schema above would only be permitted to contain DepartmentType data objects which in turn could only contain EmployeeType data objects.
Sometimes it is useful to be able to access this model information at runtime. For example, this could be used to automatically generate a user interface for populating a data object. The model information is accessed using reflection.
Example 21. Reflecting on a Data Object The following example shows how we can reflect on an empty Employee data object.
The above example will output:
Using print on the SDO_Model_ReflectionDataObject writes out the data object's model. We can see from the output how the type companyNS:EmployeeType has three properties and we can see the names of the properties along with their types. Note, the primitive types are listed as SDO types (e.g. commonj.sdo namespace, String type). It is worth noting that this is the SDO model and when these are surfaced to an application they can be treated as the PHP equivalent types (e.g. string and boolean). |
Example 22. Accessing the type information We can query the type information of a data object using reflection. The following example checks the type corresponds to a data object rather than a primitive and then iterates through the properties of the type, writing out the name of each property ($type and $property are SDO_Model_Type and SDO_Model_Property objects, respectively).
The above example will output:
|
SDO consists of three sets of interfaces. The first set covers those interfaces for use by typical SDO applications. These are identified by the package prefix 'SDO_'. The second set is those used to reflect on, and work with, the model of a data object. These are identified by the package prefix 'SDO_Model_'. Finally, the third set are those use by Data Access Service implementations and are identified by the package prefix 'SDO_DAS_'. The majority of SDO users will not need to use or understand the 'SDO_Model_' and 'SDO_DAS_' interfaces.
The main interface through which data objects are manipulated. In addition to the methods below, SDO_DataObject extends the ArrayAccess, SDO_PropertyAccess (defines __get() / __set() methods for property access overloading), Iterator, and Countable interfaces.
getSequence - get the sequence for the data object
createDataObject - create a child data object
clear - unset the properties of a data object
getContainer - get the container (also known as 'parent') of this data object
getTypeName - get the name of the type for this data object
getTypeNamespaceURI - get the namespace URI of the type for this data object
The interface through which sequenced data objects can be accessed to preserve ordering across a data object's properties and to allow unstructured text. SDO_Sequence preserves contiguous indices and therefore inserting or removing elements may shift other elements up or down. In addition to the methods below, SDO_Sequence extends the ArrayAccess, Iterator and Countable interface.
getProperty - get the property for a given sequence index
move - move an element from one property index to another
insert - insert a new value into the sequence
The interface through which many-valued properties are manipulated. In addition to the method defined below, SDO_List extends ArrayAccess, Iterator and Countable. SDO_List preserves contiguous indices and therefore inserting or removing elements may shift other elements up or down.
The interface through which data objects can be created. A Data Access Service is responsible for populating the model (i.e. configuring the data factory with the type and structure information for the data objects it can create.) for the factory and can then optionally return an instance of, or implement, the SDO_DataFactory interface.
An SDO_Exception is thrown when the caller's request cannot be completed. The subclasses of SDO_Exception are:
SDO_PropertyNotSetException - the property specified exists but has not been set or does not have a default value
SDO_PropertyNotFoundException - the property specified is not part of the data object's type
SDO_TypeNotFoundException - the specified namespace URI or type name is unknown
SDO_InvalidConversionException - conversion between the types of the assignment is not possible
SDO_IndexOutOfBoundsException - the numeric index into a data object, sequence or list is not in the valid range
SDO_UnsupportedOperationException - the request cannot be completed because it is not allowed, for example an attempt to set a read-only property.
The main interface used to reflect on a data object instance to obtain its model type and property information. It is designed to follow the reflection pattern introduced in PHP 5.
__construct - construct a new SDO_Model_ReflectionDataObject.
export - get a string describing the data object.
getType - get the SDO_Model_Type for the data object.
getInstanceProperties - get the instance properties of the data object.
getContainmentProperty - get the property which defines the containment relationship to the data object.
The interface through which a data object's type information can be retrieved. This interface can be used to find out the type name and namespace URI of the type, whether the type allow open content, and so on.
getName - get the name of the type.
getNamespaceURI - get the namespace URI of the type.
isInstance - test for a data object being an instance of the type.
getProperties - get the properties of the type.
getProperty - get a property of the type.
isDataType - test to see if this type is a primitive scalar type.
isSequencedType - test to see if this is a sequenced type.
isOpenType - test to see if this is an open type.
isAbstractType - test to see if this is an abstract type.
getBaseType - get the base type of this type (if one exists).
The interface through which a data object's property information can be retrieved. This interface can be used to find out the type of a property, whether a property has a default value, whether the property is contained or reference by its parent, its cardinality, and so on.
getName - get the name of the property.
getType - get the type of the property.
isMany - test to see if the property is many-valued.
isContainment - test to see if the property describes a containment relationship.
getContainingType - get the type which contains this property.
getDefault - get the default value for a property.
The interface through which a Data Access Service can access a data object's SDO_DAS_ChangeSummary. The change summary is used by the Data Access Service to check for conflicts when applying changes back to a data source.
getChangeSummary - get the change summary for a data object
The interface through which the change history of a data object is accessed. The change summary holds information for any modifications on a data object which occurred since logging was activated. In the case of deletions and modifications, the old values are also held in the change summary.
If logging is no longer active then the change summary only holds changes made up to the point when logging was deactivated. Reactivating logging clears the change summary. This is useful when a set of changes have been written out by a DAS and the data object is to be reused.
beginLogging - begin logging changes made to a data object
endLogging - end logging changes made to a data object
isLogging - test to see if change logging is on
getChangedDataObjects - get a list of the data objects which have been changed
getChangeType - get the type of change which has been made to a data object
getOldValues - get a list of old values for a data object
getOldContainer - get the old container data object for a deleted data object
The interface through which the old value for a property is accessed. A list of settings is returned by the change summary method getOldValues() .
getPropertyIndex - get the property index for the changed property
getPropertyName - get the property name for the changed property
getValue - get the old value for the changed property
getListIndex - get the list index for the old value if it was part of a many-valued property
isSet - test to see if the property was set prior to being modified
The interface for constructing the model for an SDO_DataObject. The SDO_DAS_DataFactory is an abstract class providing a static method which returns a concrete data factory implementation. The implementation is used by Data Access Services to create an SDO model from their model. For example, a Relational Data Access Service might create and populate an SDO_DAS_DataFactory model based on a schema for a relational database.
getDataFactory - static methods for getting a concrete data factory instance
addType - add a new type to the SDO model
addPropertyToType - add a new property to a type definition in the SDO model
The constants below are defined by this extension, and will only be available when the extension has either been compiled into PHP or dynamically loaded at runtime.
Represents a change type of 'none'.
Represents a change type of 'modification'.
Represents a change type of 'addition'.
Represents a change type of 'deletion'.