Added item_frequencies to QuerySet

This commit is contained in:
Harry Marr 2009-12-20 17:17:56 +00:00
parent bb23cdb038
commit 6438bd52b7
2 changed files with 49 additions and 0 deletions

View File

@ -159,6 +159,26 @@ class QuerySet(object):
def __iter__(self): def __iter__(self):
return self return self
def item_frequencies(self, list_field):
"""Returns a dictionary of all items present in a list field across
the whole queried set of documents, and their corresponding frequency.
This is useful for generating tag clouds, or searching documents.
"""
freq_func = """
function(collection, query, listField) {
var frequencies = {};
db[collection].find(query).forEach(function(doc) {
doc[listField].forEach(function(item) {
frequencies[item] = 1 + (frequencies[item] || 0);
});
});
return frequencies;
}
"""
db = _get_db()
collection = self._document._meta['collection']
return db.eval(freq_func, collection, self._query, list_field)
class QuerySetManager(object): class QuerySetManager(object):

View File

@ -185,6 +185,35 @@ class QuerySetTest(unittest.TestCase):
ages = [p.age for p in self.Person.objects.order_by('-name')] ages = [p.age for p in self.Person.objects.order_by('-name')]
self.assertEqual(ages, [30, 40, 20]) self.assertEqual(ages, [30, 40, 20])
def test_item_frequencies(self):
"""Ensure that item frequencies are properly generated from lists.
"""
class BlogPost(Document):
hits = IntField()
tags = ListField(StringField())
BlogPost.drop_collection()
BlogPost(hits=1, tags=['music', 'film', 'actors']).save()
BlogPost(hits=2, tags=['music']).save()
BlogPost(hits=3, tags=['music', 'actors']).save()
f = BlogPost.objects.item_frequencies('tags')
f = dict((key, int(val)) for key, val in f.items())
self.assertEqual(set(['music', 'film', 'actors']), set(f.keys()))
self.assertEqual(f['music'], 3)
self.assertEqual(f['actors'], 2)
self.assertEqual(f['film'], 1)
# Ensure query is taken into account
f = BlogPost.objects(hits__gt=1).item_frequencies('tags')
f = dict((key, int(val)) for key, val in f.items())
self.assertEqual(set(['music', 'actors']), set(f.keys()))
self.assertEqual(f['music'], 2)
self.assertEqual(f['actors'], 1)
BlogPost.drop_collection()
def tearDown(self): def tearDown(self):
self.Person.drop_collection() self.Person.drop_collection()