Question

Assuming I have entries in a datastore that reference other datastore entities:

id    link    val_1    val_2
-----------------------------
1      None      10    None
2       1       None     4
3       2       None     5
4       3       None     2
etc...  etc...  etc...   etc...

I need to recursively progress through the datastore doing the following:

1. Get entry 1 via get_by_id(1);
2. Get val_1 from entry 1
3. Query db for any entries that have '1' as the link.
4. With the results, add entry 1's val_1 to their val_2 and update their val_1
   with the result
5. For each result, query db for other entries that have their 'link' set to the
   results id, add the result's val_1 to their val_2 and update their val_1
6. Query db for any entries that have the result's id as their link
7. Repeat from step 5

I’ve tried to translate this into python, but I’m getting nowhere. Does anyone have some ideas on how I can accomplish this?

edit: I got the following code to do what I need. I feel that it could be refactored significantly. Any suggestions?

def updateLinked(cid):
    c = Info.get_by_id(int(cid))
    q = Info.all()
    q.filter('link =',int(cid))
    result = q.fetch(1000)
    for item in result:
        linked = 1

        new_link_id = item.key().id()
        val_2 = item.val_2
        val_1 = c.val_1 + val_2

        item.val_1 = val_1
        item.put()

        while linked == 1:
            q2 = Info.all()
            q2.filter('link=',new_link_lid)
            result2 = q2.fetch(1000)

            if len(result2):    
                linked = 1
                for item2 in result2:
                    c = Info.get_by_id(new_link_id)
                    new_link_id = item2.key().id()

                    val_2 = item2.val_2
                    val_1 = c.val_1 + val_2

                    item2.val_1 = val_1
                    item2.put()
            else:
                linked = 0

Answer 1

Looks like you are essentially pushing an update out across a hierarchical dataset. Consider using either GAE’s built in parent/ancestor functionality or rolling your own by adding a db.ListProperty(db.Key) to your model that includes a list of all ‘link’ record keys. This would save you considerable time by allowing you to query all the data that needs to be updated in one go.

class Info(db.Model):
    link = db.SelfReferenceProperty()
    val_1 = db.IntegerProperty()
    val_2 = db.IntegerProperty()
    ancestors = db.ListProperty(db.Key) #list of ancestor links

def updateLinked(cid):
    c = Info.get_by_id(int(cid))
    result = Info.all().filter('ancestors = ', c.key())
    for item in result:
        item.val_1 = c.val_1 + item.val_2
        item.put()