archetypes.schemaextender 2.0.2

Dynamically extend Archetypes schemas with named adapters.

Introduction

This package allows you to modify an Archetypes schema, using simple adapters. This can be used to add new fields, reorder fields and fieldsets or make other changes.

The most common use of schema extension is to allow add-on products to enhance standard Plone content types, for example by adding an option that can be set to toggle special behaviour.

schemaextender hooks into the Archetypes framework by registering an ISchema adapter for BaseContent and BaseFolder, making it responsible for providing the schema for all types derived from those classes. This includes all standard Plone content types. Since only one ISchema adapter can be active schemaextender provides its own mechanism to modify schemas using named adapters. Named adapters are allowing to register more than one schemaextender per adapted interface.

There are three types of adapters available:

• ISchemaExtender: using this adapter you can add new fields to a schema.

• IOrderableSchemaExtender: this adapters makes it possible to both add new fields and reorder fields. This is more costly than just adding new fields.

• IBrowserLayerAwareExtender: this adapters are making use of plone.browserlayer, so that the extender is only available if a layer is registered.

• ISchemaModifier: this is a low-level hook that allows direct manipulation of the schema. This can be very dangerous and should never be used if one does not know exactly what she/he is doing!

The adapter types are documented in the ‘’interfaces.py’’ file in archetypes.schemaextender.

Simple example

As an example we will add a simple boolean field to the standard Plone document type. First we need to create a field class:

from Products.Archetypes.public import BooleanField
from archetypes.schemaextender.field import ExtensionField

class MyBooleanField(ExtensionField, BooleanField):
    """A trivial field."""

schemaextender can not use the standard Archetypes fields directly since those rely on the class generation logic generating accessors and mutator methods. By using the ExtensionField mix-in class we can still use them. Make sure the ExtensionField mix-in comes first, so it properly overwrites the standard methods.

Next we have to create an adapter that will add this field:

from zope.component import adapts
from zope.interface import implements
from archetypes.schemaextender.interfaces import ISchemaExtender
from Products.Archetypes.public import BooleanWidget
from Products.ATContentTypes.interface import IATDocument

class PageExtender(object):
    adapts(IATDocument)
    implements(ISchemaExtender)


    fields = [
        MyBooleanField("super_power",
        widget = BooleanWidget(
            label="This page has super powers")),
            ]

    def __init__(self, context):
        self.context = context

    def getFields(self):
        return self.fields

Try to store the fields on the class, that way they aren’t created each time the getFields method gets called. Generally you should make sure getFields does as few things as possible, because it’s called very often.

The final step is registering this adapter with the Zope component architecture. Since we already declared the interface we provide and which type of object we adapt this can be done very quickly in configure.zcml (assuming you put the code above in a file extender.py):

<configure xmlns="http://namespaces.zope.org/zope"
           xmlns:five="http://namespaces.zope.org/five">

    <include package="archetypes.schemaextender" />
    <adapter factory=".extender.PageExtender" />
</configure>

Custom fields

If you want you can make more complicated field types as well. The only requirement is that you need to have ExtensionField as the first parent class for your field type. As an example here is a field that toggles a marker interface on an object:

from zope.interface import Interface
from zope.interface import alsoProvides
from zope.interface import noLongerProvides
from Products.Archetypes.public import BooleanField
from archetypes.schemaextender.field import ExtensionField

def addMarkerInterface(obj, *ifaces):
    for iface in ifaces:
        if not iface.providedBy(obj):
            alsoProvides(obj, iface)


def removeMarkerInterface(obj, *ifaces):
    for iface in ifaces:
        if iface.providedBy(obj):
            noLongerProvides(obj, iface)


class ISuperPower(Interface):
    """Marker interface for classes that can do amazing things."""


class InterfaceMarkerField(ExtensionField, BooleanField):
    def get(self, instance, **kwargs):
        return ISuperPower.providedBy(instance)

    def getRaw(self, instance, **kwargs):
        return ISuperPower.providedBy(instance)

    def set(self, instance, value, **kwargs):
        if value:
            addMarkerInterface(instance, ISuperPower)
        else:
            removeMarkerInterface(instance, ISuperPower)

Changelog

2.0.2 - 2010-06-13

• Changed the schema cache again, to use the id() only as a fallback when there is not yet an UID assigned. The id() is unstable with Acquisition wrappers and ZODB level ghosting of objects. We provide an explicit API for disabling the cache instead. [hannosch]

• Avoid deprecation warnings under Zope 2.13. [hannosch]

• Change the schema cache key to avoid an issue during content migration when there might be two objects that have the same UID but different schemata. Now concatenating the Python id() with the UID (id() alone isn’t good enough, as two objects with non-overlapping lifetimes may have the same id() value). Fixes http://dev.plone.org/plone/ticket/10637. [davisagli]

2.0.1 - 2010-05-23

• Disable the schema cache during test runs. [hannosch]

2.0 - 2010-05-23

• Removed the caching.zcml and enabled caching by default. You can use a different caching implementation by using an overrides.zcml file. [hannosch]

• Added z3c.autoinclude entry point to mark this as a Plone plugin. [hannosch]

1.6 - 2010-03-22

• Fix index accessor to support setting custom accessor methods. [witsch]

1.5 - 2009-11-18

• Fixed test failure in usage.txt. [hannosch]

• Standardized package metadata layout. [hannosch]

1.4 - 2009-11-05

• Fix schema copying to also include properties and layers. [maerteijn]

1.3 - 2009-10-20

• Refactored the TranslatableExtensionField getMutator to directly reuse the generatedMutatorWrapper from LinguaPlone itself. This avoids duplicating the logic and lets schemaextender fields use the special reference field handling introduced in LinguaPlone. This change introduces a version requirement for LinguaPlone of at least 3.0b6. [hannosch]

1.2 - 2009-10-10

• Add ISchema adapter using simple caching on the request in order to avoid redundant calculation of the (extended) schema. The adapter is not enabled by default and can be activated by loading caching.zcml. [witsch]

• Avoid using the overridden + operator when copying the original schema as this will needlessly validate all fields again. [witsch]

• Added missing changelog entry. [hannosch, woutervh]

1.1 - 2009-06-03

• Added support for LinguaPlone language independent fields, by seamlessly using a new TranslatableExtensionField when LP is installed. [hannosch]

• Added a proper interface to the IExtensionField. [hannosch]

• Adjusted tests for Plone 3.3. [hannosch]

• Minor adjustment in documentation: a) don’t adapt the class in the example, b) explain why named adapters are used. [jensens]

• Schema modifiers now also browserlayer-aware. [jessesnyder]

1.0 - 2008-07-17

• No changes since 1.0rc1.

1.0rc1 - 2008-04-07

• Added optional plone.browserlayer support. Extenders implementing IBrowserLayerAwareExtender need to have a layer attribute. Those extenders are taken into account only if the specified layer is active. [jensens]

1.0b1 - 2007-12-07

• Schema modifiers implementing ISchemaModifier are now responsible for copying fields they modify. See README and the doc strings. [fschulze]

• Added a simple benchmark and made some optimizations by avoiding a lot of field copying. [fschulze, wiggy]

• Use a marker interface instead of overrides.zcml - this means you don’t need to muck with overrides in dependent products. [optilude]

• Added code to allow addition of new schemata. We need an ordered dictionary to not bork the order of the schemata. [jensens]

• Add a small benchmark utility. [wichert]

• Replace the high-level test with unit-tests and extend the test coverage. [wichert]

• Rewrite the README to be more human readable. [wichert]

1.0a1 - 2007-10-15

• First public release.