Query for content sorted by path and position in parent

pbauer · September 1, 2021, 6:48am

How can I iterate over content in a portal in exactly the order it appears to Users?

portal_catalog(sort_on=["path", "getObjPositionInParent"]) does not help because getObjPositionInParent is never used because the id of the object is in the path.

The only ideas I have right now:

Add a custom index parent_path and do portal_catalog(sort_on=["parent_path", "getObjPositionInParent"]).
Instead of doing one query do a query for each container somehow using the index is_folderish and path:
Walk the tree instead of using a catalog-query.

dieter · September 1, 2021, 7:07am

If you need to process the complete content, walk it (not using the catalog at all).
Otherwise, locate the content via a catalog search and perform the sorting externally (based on parent_path and getObjPositionInParent).

yurj · May 23, 2024, 10:51am

Hi! did you find a solution? Maybe we could use the same walk procedure from Catalog Rebuild:

https://github.com/zopefoundation/Products.ZCatalog/blob/4066d07968c85f4f54c4470ce476ac54379fbb33/src/Products/ZCatalog/ZCatalog.py#L656

instead of passing indexObject, you can pass the function to export the content or get the object position and use the position while creating the content or whatever.

djay · May 25, 2024, 6:21am

Plone out of the box walks the content.

github.com

eea/plone.restapi/blob/90c32570e19cd02a56490091df0b49f504aeb3a9/src/plone/restapi/services/navigation/get.py#L107


      
          @property
          def current_language(self):
              return (
                  self.request.get("LANGUAGE", None)
                  or (self.context and aq_inner(self.context).Language())
                  or self.default_language
              )
          
          @property
          @memoize
          def navtree(self):
              ret = defaultdict(list)
              navtree_path = self.navtree_path
              for tab in self.portal_tabs:
                  entry = {}
                  entry.update(
                      {
                          "path": "/".join((navtree_path, tab["id"])),
                          "description": tab["description"],
                          "@id": tab["url"],
                      }

you can use the restapi.

A new index with parent_path and getObjPositionInParent wouldn't solve it.

But a path index of positions in parents would. like /00001/00023/00024.

I think the reason we don't have this and walk the tree is that waking up all parents when you want a list of all content seems a reasonable performance hit?