Reindexing Context's Page Container on Change/Insert/Delete

[drfho]

The latest changes address the different cases:

[1] CASE-1: ZMS_INSERT

• is page-like class and filterMatch of current node (eg. new ZMSFolder)
• is not page-like and filterMatch of container-page (eg. new ZMSTextarea into ZMSFolder)
• is not page-like and filterMatch of current node (e.g. new ZMSFile)

[2] CASE-2: CHANGE

• container-page is not current node and filterMatch of container-page
• filterMatch of current node

ZMS/Products/zms/ZMSZCatalogAdapter.py

Lines 201 to 222 in 1b4ec16

	# Reindex node's content by each connector.
	for connector in connectors:
	if self.matches_ids_filter(node) or self.matches_ids_filter(page):
	# CASE-1: ZMS_INSERT
	# Reindex the current node if page or its container page.
	if self.REQUEST.get('ZMS_INSERT', None):
	if node.isPage() and self.matches_ids_filter(node):
	self.reindex(connector, node, recursive=False, fileparsing=fileparsing)
	elif not node.isPage() and self.matches_ids_filter(page):
	self.reindex(connector, page, recursive=False, fileparsing=fileparsing)
	if not node.isPage() and self.matches_ids_filter(node):
	# After reindexing the page add the node if filter-match.
	self.reindex(connector, node, recursive=False, fileparsing=fileparsing)
	# CASE-2: CHANGE
	# Reindex container page if node!=page and the node if filter-match.
	else:
	if node != page and self.matches_ids_filter(page):
	self.reindex(connector, page, recursive=False, fileparsing=fileparsing)
	if self.matches_ids_filter(node):
	self.reindex(connector, node, recursive=False, fileparsing=fileparsing)

The incremental reindex still shows a surprising logical (?) problem: on ZMS_INSERT the content of the container page is aggregated only of the content-objects that have the same meta_id like the objects that is inserted. Exp: if a ZMSTextarea is added to a document, the indexed page content only concats the ZMSTextarea-content but ignores any other object class e.g. the content of ZMSTable-object.
It seems that the context that is needed for the attribute rendering is in case of standard_html not referring to the iterated object's standard_html, LINE 355 child_node.getBodyContent(request):

ZMS/Products/zms/ZMSZCatalogAdapter.py

Lines 349 to 357 in 1b4ec16

	try:
	# ZMSFile.standard_html will work in get_file.
	if not (node.meta_id == 'ZMSFile' and attr_id == 'standard_html'):
	if attr_id == 'standard_html' and request.get('ZMS_INSERT', None) and node.isPage():
	# ZMS_INSERT Page: attr('standard_html') does not work.
	for child_node in node.getObjChildren('e',request,self.PAGEELEMENTS):
	value += child_node.getBodyContent(request)
	else:
	value = node.attr(attr_id)

[1] On INSERTING an new ZMSTextarea prevents Reindexing Page-Content not derived from ZMSTextarea

[2] On CHANGE a PAGEELEMENT or the PAGE-Container renders/reindexes the full content

Debugging the get_attr_data()-iteration

ZMS/Products/zms/ZMSZCatalogAdapter.py

Lines 353 to 355 in 1b4ec16

	# ZMS_INSERT Page: attr('standard_html') does not work.
	for child_node in node.getObjChildren('e',request,self.PAGEELEMENTS):
	value += child_node.getBodyContent(request)

reveals that standard_html-attr of the inserted obj-class is applied and not the one of the iterated item. In case of ZMSFile this results in empty data for any ZMSTextarea, because the standard_html of ZMSFile cannot render ZMSTextarea appropiately.

The other way round: if inserting a ZMSTextarea the ZMSFile cannot be renderd

child_node.meta_id
'ZMSFile'
child_node.getBodyContent(request)
'<!-- ZMSTextarea.standard_html -->\n\n\n\n<!-- /ZMSTextarea.standard_html -->'
child_node.attr('standard_html')
'<!-- ZMSTextarea.standard_html -->\n\n\n\n<!-- /ZMSTextarea.standard_html -->'
child_node.attr('title')
'antrag.pdf'

NOTE: Fullindex works correctly and thus corrects any missing page content.

Ref:
[1] Ensure Reindexing Context's Page Container on Change/Insert #277
[2] https://github.com/idasm-unibe-ch/unibe-cms-opensearch/issues/50

[drfho]

Addressing the iteration-item's class' content-modell standard_html explictly does the job:

self.getMetaobjAttr(child_node.meta_id, 'standard_html')['ob'](zmscontext=child_node)

ZMS/Products/zms/ZMSZCatalogAdapter.py

Lines 349 to 353 in 936b091

	# ZMS_INSERT Page: attr('standard_html') does not work.
	for child_node in node.getObjChildren('e',request,self.PAGEELEMENTS):
	# value += child_node.getBodyContent(request) # This would not work.
	# Explicitly call ob-method to get the correct standard_html:
	value += self.getMetaobjAttr(child_node.meta_id, 'standard_html')['ob'](zmscontext=child_node)

[drfho]

Unindexing

If a ZMS-node may be a page or a page-element: if latter is deleted (moved to trashcan) the containg page mus be reindexed If a page is deleted this node must be removed from index in all it's languages.

ZMS/Products/zms/ZMSZCatalogAdapter.py

Lines 228 to 261 in de70f0f

	# --------------------------------------------------------------------------
	# ZMSZCatalogAdapter.unindex_node
	# --------------------------------------------------------------------------
	def unindex_nodes(self, nodes=[], forced=False):
	# Is triggered by zmscontainerobject.moveObjsToTrashcan().
	standard.writeBlock(self, "[unindex_nodes]")
	# Get closest catalog-connectors.
	path_nodes = self.breadcrumbs_obj_path()
	path_nodes.reverse()
	for path_node in path_nodes:
	if path_node.getCatalogAdapter():
	connectors = path_node.getCatalogAdapter().get_connectors()
	break
	if not connectors:
	root = self.getRootElement()
	connectors = root.getCatalogAdapter().get_connectors()
	try:
	if self.getConfProperty('ZMS.CatalogAwareness.active', 1) or forced:
	# [1] Reindex page-container nodes of deleted page-elements.
	pageelement_nodes = [node for node in nodes if not node.isPage()]
	if pageelement_nodes:
	for pageelement_node in pageelement_nodes:
	# Todo: Avoid redundant reindexing of page-container.
	self.reindex_node(node=pageelement_node)
	# [2] Unindex deleted page-nodes if filter-match.
	nodes = [node for node in nodes if self.matches_ids_filter(node)]
	for connector in connectors:
	connector.manage_objects_remove(nodes)
	return True
	except:
	standard.writeError( self, "unindex_nodes not successful")
	return False

The possible languages of a deleted node (not the catalog-connector node, which may have a different language set) is set as global var in the main interface function manage_opensearch_objects_remove() ...

ZMS/Products/zms/conf/metaobj_manager/com.zms.catalog.opensearch/opensearch_connector/manage_opensearch_objects_remove.py

Lines 63 to 67 in de70f0f

	for node in nodes:
	# Set node's language list as global variable.
	global langs
	langs = node.getLangIds()
	break

... and then used in the bulk-deleting bulk_opensearch_delete():

ZMS/Products/zms/conf/metaobj_manager/com.zms.catalog.opensearch/opensearch_connector/manage_opensearch_objects_remove.py

Lines 52 to 56 in de70f0f

	# Create language specific opensearch id
	for lang in globals()['langs']:
	_id = "%s:%s"%(x['uid'],lang)
	d = {"_op_type":"delete", "_index":index_name, "_id":_id}
	actions.append(d)