Merge pull request #1999 from strikaco/db_tag_improve

Typeclass get_by_tag scalability improvements and weighted tag querysets
This commit is contained in:
Griatch 2019-12-16 19:02:13 +01:00 committed by GitHub
commit 809e07b4e3

View file

@ -5,7 +5,7 @@ all Attributes and TypedObjects).
"""
import shlex
from django.db.models import Q
from django.db.models import Q, Count
from evennia.utils import idmapper
from evennia.utils.utils import make_iter, variable_from_module
from evennia.typeclasses.attributes import Attribute
@ -236,7 +236,7 @@ class TypedObjectManager(idmapper.manager.SharedMemoryManager):
"""
return self.get_tag(key=key, category=category, obj=obj, tagtype="alias")
def get_by_tag(self, key=None, category=None, tagtype=None):
def get_by_tag(self, key=None, category=None, tagtype=None, **kwargs):
"""
Return objects having tags with a given key or category or combination of the two.
Also accepts multiple tags/category/tagtype
@ -250,6 +250,12 @@ class TypedObjectManager(idmapper.manager.SharedMemoryManager):
tagtype (str, optional): 'type' of Tag, by default
this is either `None` (a normal Tag), `alias` or
`permission`. This always apply to all queried tags.
Kwargs:
match (str): ALL or ANY, determines whether the target object must match
ALL of the provided tags/categories or ANY single one. ANY will perform
a weighted sort, so objects with more tag matches will outrank those
with fewer tag matches.
Returns:
objects (list): Objects with matching tag.
@ -262,6 +268,12 @@ class TypedObjectManager(idmapper.manager.SharedMemoryManager):
if not (key or category):
return []
global _Tag
if not _Tag:
from evennia.typeclasses.models import Tag as _Tag
match = kwargs.get('match', 'all').lower().strip()
keys = make_iter(key) if key else []
categories = make_iter(category) if category else []
n_keys = len(keys)
@ -286,14 +298,27 @@ class TypedObjectManager(idmapper.manager.SharedMemoryManager):
"get_by_tag needs a single category or a list of categories "
"the same length as the list of tags."
)
clauses = Q()
for ikey, key in enumerate(keys):
query = query.filter(
db_tags__db_key__iexact=key, db_tags__db_category__iexact=categories[ikey]
)
# Keep each key and category together, grouped by AND
clauses |= Q(db_key__iexact=key, db_category__iexact=categories[ikey])
else:
# only one or more categories given
for category in categories:
query = query.filter(db_tags__db_category__iexact=category)
clauses = Q()
uniq_categories = sorted(set(categories))
for category in uniq_categories:
clauses |= Q(db_category__iexact=category,)
tags = _Tag.objects.filter(clauses)
query = query.filter(db_tags__in=tags).annotate(matches=Count('db_tags__pk', filter=Q(db_tags__in=tags), distinct=True))
# Default ALL: Match all of the tags and optionally more
if match == 'all':
query = query.filter(matches__gte=tags.count())
# ANY: Match any single tag, ordered by weight
elif match == 'any':
query = query.order_by('-matches')
return query