Introduction

The Torque generator can be configured by the following files:
  • The file conf/control.xml tells the generator which files should be generated, and which sources, options and outlets should be used to generate them.
  • For the generation process, outlets (which create String output out of sources) must be defined. In each outlet XML configuration file in the outlets directory, one or more outlets can be defined. The minimum information needed for each outlet is a unique name, its type, and the information to invoke the outlet of this given type. For example, if the outlet has the type "velocityOutlet" i.e. the outlet is a velocity template, then the path to the template is needed as well.
    All outlet properties can be overridden in the controller configuration.
    For each template with a known file extension, an outlet definition is created automatically.
    See below for details on defining outlets.

Some settings can also be passed directly when calling the generator. Typically these settings specify which config directory resp. package is used, which source directory is used, which source files are included and excluded etc. Also, debug output can be added to the output which helps determine which output snippet stems from which outlet.
These settings are typically made available through the build system adapter, e.g.the maven plugin or the ant tasks.

Standard directory layout

The Torque generator uses a standard directory layout for its configuration:
      - torque-gen directory (e.g. src/main/torque-gen for maven 3)
        - conf
          - control.xml
          - options.properties (not always present, name can differ)
        - outlets
          - outlet_config_example.xml (name not fixed)
          - ... (other outlet definition files)
        - templates (only if template generators are used)
          - template_example.vm (name not fixed)
          - ... (other templates)
        - resources (only if xsd schema validation is used for sources)
          - schema_example.xsd (name not fixed)
          - ... (other xsd schemata)
          - binary resources (images etc.)
    
The sources are not considered part of the configuration; thus they do not appear here (If you have no idea where to put the sources, create a src directory in the torque-gen directory and put them there.) The content of the various directories is described below.

The control configuration

The main configuration consists of the file control.xml, and maybe of one or more options file, and is located in the directory conf.

The configuraton file complies to configuration.xsd and the referenced outlet.xsd.

In the file control.xml, the option files are declared, and the output files are defined. A control file may like this:
<?xml version="1.0" encoding="UTF-8"?>
<control loglevel="info"
    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xsi:schemaLocation="http://db.apache.org/torque/4.0/generator/configuration http://db.apache.org/torque/4.0/generator/configuration.xsd"
    xmlns="http://db.apache.org/torque/4.0/generator/configuration">

  <options xsi:type="propertiesOptions" path="options.properties"/>

  <entityReference
      systemId="http://foo.org/bar.xsd"
      resource="bar.xsd" />

  <output name="torque.om.dbObject" existingTargetStrategy="skip" outputDirKey="modifiable">
    <filenameOutlet
        xsi:type="javaOutlet"
        class="org.apache.torque.generator.outlet.java.JavaFilenameOutlet">
      <mergepoint name="package">
        <action
           xsi:type="sourceElementAttributeAction"
           element="."
           attribute="dbObjectPackage"
           acceptNotSet="false"/>
      </mergepoint>
      <mergepoint name="classname">
        <action 
            xsi:type="sourceElementAttributeAction"
            element="."
            attribute="dbObjectClassName"
            acceptNotSet="false"/>
      </mergepoint>
    </filenameOutlet>
    <source xsi:type="fileSource" 
        elements="database/table"
        skipDecider="org.apache.torque.templates.sourcefilter.OmSkipDecider">
      <transformer class="org.apache.torque.templates.transformer.om.OMTransformer"/>
      <include>*schema.xml</include>
      <exclude>id-table-schema.xml</exclude>
    </source>
    <outlet name="torque.om.dbObject"/>
  </output>
</control>
    
The loglevel attribute on the control file defines the amount of logging during generation. Possible values are "debug", "info", "warn" and "error", default is "info".
The "options" elements define which option files should be read (more than one options file is possible). If the same option is defined in different files, the value in the last file takes precedence.
The "entityReference" element tells an XML parser used to read XML source files to resolve the systemId "http://foo.org/bar.xsd" to the file bar.xsd in the resources directory. More than one entityReference is possible. If you do not use XML sources, XSD validation cannot be used and defining entity references makes no sense.
The output tags define which output files are generated. The name attribute of the output is used for debugging purposes and must be set. The existingTargetStrategy attribute of the output tag defines how to handle existing files. Default is "replace" (replace the existing file), other modes are "skip" (skip generation and leave existing file as is), "append" (append generation result to the existing file) and "merge" (merge changes in the generation output with changed generation result, see the existing target modes documentation for details). The outputDirKey attribute of the output tag allows to write the output files to another output directory, the outputDirKey defines a symbolic name for this directory. On generation time, the user needs to assign real directories to the symbolic names. E.g. in the torque templates, the outputDirKey "modifiable" is used for files which are intended to be modifiable by the user; the "modifiable" key defaults to the src/main/generated-java directory. The filename of the output files can either be defined in the "filename" attribute if the filename is fixed, or in the filenameOutlet child element if the filename is generated from the sources. The files to read as input and the source elements for which output files are generated are defined in the source child element of the output tag. The filename outlet can be either referenced by name (if it is defined elsewhere), or can be defined here by using the "type" and "class" attributes.
A XPath-like notation can be used to traverse the source graph (note that only the simplest XPath elements are recognized, i.e. explicit element names and the Tokens "/", ".." and "."). In principle, for each source element which matches the description in the elements attribute in each source file matching the include/exclude pattern, an output file is generated. However, the SkipDecider class which can be configured in the "skipDecider" attribute of the "source" element can determine to skip generation under certain circumstances. Note that this multiple processing also applies for fixed output filenames.
If the source needs to be transformed before generation, transformer tags are included as children of the source tag. For each transformer, the class must be given. If more than one transformer is defined for a given source, the output of the first transformer is used as input for the second transformer, and so on, and the output of the last transformer is fed into the generators.

Configuring the outlets

The outlets can be defined and wired together in the outlets subdirectory (recommended) or in the control.xml file (should only be used when overriding settings in existing template sets). Each file in this directory defines one or more outlets. Each outlet must have an unique name; it is possible to use dot(.)-separated namespaces for avoiding name conflicts if more than one template set is used (e.g. the outlet name org.apache.torque.someOutlet has the namespace org.apache.torque and the unqualified name someOutlet). For template-based outlets, the name attribute can be omitted; the name is then calculated from the path of the template relative to the templates directory (e.g. the template with the relative path someDir/someTemplate.groovy will get the name someDir.someTemplate if no explicit name is set). Also, the type of the outlet must be given (currently supported are "velocity", "groovy", and "java"). Depending on the type, further informations need to be given: The velocity and groovy outlets need the path to the template (relative to the templates subdirectory), the java outlets need the class of the outlet class. For example, this defines a java and a velocity outlet:

<?xml version="1.0" encoding="UTF-8"?>
<outlets xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xsi:schemaLocation="http://db.apache.org/torque/4.0/generator/configuration http://db.apache.org/torque/4.0/generator/configuration.xsd"
    xmlns="http://db.apache.org/torque/4.0/generator/configuration">
  <outlet name="org.apache.torque.generator.javaOutlet"
      xsi:type="javaOutlet" 
      class="org.apache.torque.generator.JavaGenerator">
  </generator>
  <outlet name="torque.om.bean.baseBean" 
      xsi:type="velocityOutlet"
      path="bean/base/baseBean.vm"/>
</outlets>

mergepoints

In each outlet, mergepoints can be defined. Into each mergepoint, external content may be inserted (e.g. the value of an option or a variable or the output of another template). To keep your templates flexible, it makes sense to define mergepoints where you do not need them now but may need them in the future (you do not have to assign content to a mergepoint, it can also stay empty). The mergepoint inside the outlet is referenced by its name. For example, the following configuration defines three outlets, and the output of the last two outlet is used in mergepoints of the first outlet:

<?xml version="1.0" encoding="UTF-8"?>
<outlets xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xsi:schemaLocation="http://db.apache.org/torque/4.0/generator/configuration http://db.apache.org/torque/4.0/generator/configuration.xsd"
    xmlns="http://db.apache.org/torque/4.0/generator/configuration">
  <outlet name="torque.om.bean.base.getterSetter"
      xsi:type="velocityOutlet"
      path="general/getterSetter.vm">
    <mergepoint name="getter">
      <action xsi:type="applyAction" outlet="torque.om.bean.base.getter"/>
    </mergepoint>
    <mergepoint name="setter">
      <action xsi:type="applyAction" outlet="torque.om.bean.base.setter"/>
    </mergepoint>
  </outlet>

  <outlet name="torque.om.bean.base.getter"
      xsi:type="velocityOutlet"
      path="bean/base/getter.vm">
  </outlet>

  <outlet name="torque.om.bean.base.setter"
      xsi:type="velocityOutlet"
      path="bean/base/setter.vm">
  </outlet>
</outlets>

A more exotic case is the definition of mergepoints outside an outlet tag. This is useful when overriding a mergepoint definition in a child project. The mergepoint can be accessed by ${outletName}.${mergepointName}. For example, if we want to override the setter mergepoint in the torque.om.bean.base.getterSetter to use a custom velocity template named setter2.vm, we could use the following configuration in the child project:

<?xml version="1.0" encoding="UTF-8"?>
<outlets xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xsi:schemaLocation="http://db.apache.org/torque/4.0/generator/configuration http://db.apache.org/torque/4.0/generator/configuration.xsd"
    xmlns="http://db.apache.org/torque/4.0/generator/configuration">
  <outlet name="torque.om.bean.base.setter2"
      xsi:type="velocityOutlet"
      path="bean/base/setter2.vm">
  </outlet>

  <mergepoint name="torque.om.bean.base.getterSetter.setter">
    <action xsi:type="applyAction" outlet="torque.om.bean.base.setter2"/>
  </mergepoint>
</outlets>

Note that by virtue of the standalone mergepoint definition, we need not override the whole definition for the outlet torque.om.bean.base.getterSetter, but can override a single mergepoint definition.

automatic outlet definition for templates

For each template with a known file extension (currently .vm for velocity templates and .groovy for groovy scripts), the generator automatically creates an outlet. The name of these outlets is the path of the file in the template directory, with path separators replaced by dots and the file extension removed. For example, for a file templates/peer/impl/peerImpl.vm, an outlet with the name peer.impl.peerImpl will be created. No outlet will be defined automatically if an outlet definition with the given name already exists.

Note that the outlet definitions which are created automatically do not have mergepoint mappings, so you will still have to map the mergepoint mappings explicitly (or define an extra outlets with the mergepoint mappings). Also, the encoding of the templates will be assumed to be the vm's default encoding.

If a template exists in two known languages (e.g. myTemplate.vm and myTemplate.groovy), it is currently undefined which template of the two will be used for automatic outlet definition. So if automatic outlet definition is used for defining outlets, make sure all templates have a different filename even when omitting the suffix.