Script to populate uid_catalog and reference_catalog in order to make link integrity work

populate_uid_and_reference_catalogs.py

# coding=utf8
u"""Script to populate uid_catalog and reference_catalog in order to make link integrity work.

Only content items that implements the IReferenceable behavior (plone.app.referenceablebehavior) are
added to the catalogs.

ATTENTION: Be sure to adjust the global variables after the imports according to your environment.
"""

from Acquisition import aq_base
from Products.Archetypes import config
from plone import api
from plone.app.referenceablebehavior.referenceable import IReferenceable
from zope.component.hooks import setSite

DRY_RUN = True
PLONE_SITE_ID = 'Plone'
ADMIN_USER_ID = 'admin'


def main(app):
    portal = app[PLONE_SITE_ID]
    setSite(portal)

    with api.env.adopt_user(username=ADMIN_USER_ID):
        portal_catalog = api.portal.get_tool('portal_catalog')
        brains = portal_catalog(object_provides=IReferenceable.__identifier__)
        brains = sorted(brains, key=lambda b: b.getPath())
        for b in brains:
            obj = b.getObject()
            add_to_uid_catalog(obj)
            for child_obj in get_references(obj):
                add_to_uid_catalog(child_obj)
                add_to_reference_catalog(child_obj)
    
    if not DRY_RUN:
        # Commit transaction
        import transaction; transaction.commit()
        # Perform ZEO client synchronization (if running in clustered mode)
        app._p_jar.sync()


def add_to_catalog(obj, catalog):
    path = get_path(obj)
    print 'Adding to {}: {} ...'.format(catalog.getId(), path)
    catalog.catalog_object(obj, path)


def add_to_uid_catalog(obj):
    add_to_catalog(obj=obj, catalog=api.portal.get_tool(config.UID_CATALOG))


def add_to_reference_catalog(obj):
    add_to_catalog(obj=obj, catalog=api.portal.get_tool(config.REFERENCE_CATALOG))


def get_references(obj):
    if not hasattr(aq_base(obj), config.REFERENCE_ANNOTATION):
        return ()
    at_references = getattr(obj, config.REFERENCE_ANNOTATION)
    return () if (not at_references) else at_references.values()


def get_path(obj):
    physical_path = obj.getPhysicalPath()
    if config.REFERENCE_ANNOTATION in physical_path:

        physical_path = physical_path[len(api.portal.get().getPhysicalPath()):]
    path = '/'.join(physical_path)
    assert ('/' in path), 'Invalid path.'  # To avoid regressions.
    return path


main(app)