def test_nested_delete():
schema = fields.Schema(kind=fields.ID,
name=fields.KEYWORD(scorable=True, stored=True))
ix = RamStorage().create_index(schema)
with ix.writer() as w:
with w.group():
w.add_document(kind=u("class"), name=u("Index"))
w.add_document(kind=u("method"), name=u("add document"))
w.add_document(kind=u("method"), name=u("add reader"))
w.add_document(kind=u("method"), name=u("close"))
with w.group():
w.add_document(kind=u("class"), name=u("Accumulator"))
w.add_document(kind=u("method"), name=u("add"))
w.add_document(kind=u("method"), name=u("get result"))
with w.group():
w.add_document(kind=u("class"), name=u("Calculator"))
w.add_document(kind=u("method"), name=u("add"))
w.add_document(kind=u("method"), name=u("add all"))
w.add_document(kind=u("method"), name=u("add some"))
w.add_document(kind=u("method"), name=u("multiply"))
w.add_document(kind=u("method"), name=u("close"))
with w.group():
w.add_document(kind=u("class"), name=u("Deleter"))
w.add_document(kind=u("method"), name=u("add"))
w.add_document(kind=u("method"), name=u("delete"))
# Delete "Accumulator" class
with ix.writer() as w:
q = query.NestedParent(query.Term("kind", "class"),
query.Term("name", "Accumulator"))
w.delete_by_query(q)
# Check that Accumulator AND ITS METHODS are deleted
with ix.searcher() as s:
r = s.search(query.Term("kind", "class"))
assert sorted(hit["name"]
for hit in r) == ["Calculator", "Deleter", "Index"]
names = [fs["name"] for _, fs in s.iter_docs()]
assert names == [
"Index", "add document", "add reader", "close", "Calculator",
"add", "add all", "add some", "multiply", "close", "Deleter",
"add", "delete"
]
# Delete any class with a close method
with ix.writer() as w:
q = query.NestedParent(query.Term("kind", "class"),
query.Term("name", "close"))
w.delete_by_query(q)
# Check the CLASSES AND METHODS are gone
with ix.searcher() as s:
names = [fs["name"] for _, fs in s.iter_docs()]
assert names == ["Deleter", "add", "delete"]
def test_nested_parent():
schema = fields.Schema(name=fields.ID(stored=True),
type=fields.ID,
part=fields.ID,
price=fields.NUMERIC)
ix = RamStorage().create_index(schema)
with ix.writer() as w:
with w.group():
w.add_document(name=u("iPad"), type=u("product"))
w.add_document(part=u("screen"), price=100)
w.add_document(part=u("battery"), price=50)
w.add_document(part=u("case"), price=20)
with w.group():
w.add_document(name=u("iPhone"), type=u("product"))
w.add_document(part=u("screen"), price=60)
w.add_document(part=u("battery"), price=30)
w.add_document(part=u("case"), price=10)
with w.group():
w.add_document(name=u("Mac mini"), type=u("product"))
w.add_document(part=u("hard drive"), price=50)
w.add_document(part=u("case"), price=50)
with ix.searcher() as s:
price = s.schema["price"]
pq = query.Term("type", "product")
cq = query.Term("price", 50)
q = query.NestedParent(pq, cq)
r = s.search(q)
assert sorted([hit["name"] for hit in r]) == ["Mac mini", "iPad"]
def test_scoring():
schema = fields.Schema(kind=fields.ID,
name=fields.KEYWORD(scorable=True, stored=True))
ix = RamStorage().create_index(schema)
with ix.writer() as w:
with w.group():
w.add_document(kind=u("class"), name=u("Index"))
w.add_document(kind=u("method"), name=u("add document"))
w.add_document(kind=u("method"), name=u("add reader"))
w.add_document(kind=u("method"), name=u("close"))
with w.group():
w.add_document(kind=u("class"), name=u("Accumulator"))
w.add_document(kind=u("method"), name=u("add"))
w.add_document(kind=u("method"), name=u("get result"))
with w.group():
w.add_document(kind=u("class"), name=u("Calculator"))
w.add_document(kind=u("method"), name=u("add"))
w.add_document(kind=u("method"), name=u("add all"))
w.add_document(kind=u("method"), name=u("add some"))
w.add_document(kind=u("method"), name=u("multiply"))
w.add_document(kind=u("method"), name=u("close"))
with ix.searcher() as s:
q = query.NestedParent(query.Term("kind", "class"),
query.Term("name", "add"))
r = s.search(q)
assert [hit["name"]
for hit in r] == ["Calculator", "Index", "Accumulator"]
def test_no_parents():
schema = fields.Schema(id=fields.STORED,
kind=fields.ID,
name=fields.ID(stored=True))
k = u("alfa")
ix = RamStorage().create_index(schema)
with ix.writer() as w:
w.add_document(id=0, kind=k, name=u("one"))
w.add_document(id=1, kind=k, name=u("two"))
w.add_document(id=2, kind=k, name=u("three"))
w.add_document(id=3, kind=k, name=u("four"))
w.add_document(id=4, kind=k, name=u("one"))
w.add_document(id=5, kind=k, name=u("two"))
w.add_document(id=6, kind=k, name=u("three"))
w.add_document(id=7, kind=k, name=u("four"))
w.add_document(id=8, kind=k, name=u("one"))
w.add_document(id=9, kind=k, name=u("two"))
w.add_document(id=10, kind=k, name=u("three"))
w.add_document(id=11, kind=k, name=u("four"))
with ix.searcher() as s:
pq = query.Term("kind", "bravo")
cq = query.Or([query.Term("name", "two"), query.Term("name", "four")])
q = query.NestedParent(pq, cq)
r = s.search(q)
assert r.is_empty()
def test_everything_is_a_parent():
schema = fields.Schema(id=fields.STORED,
kind=fields.ID,
name=fields.ID(stored=True))
k = u("alfa")
ix = RamStorage().create_index(schema)
with ix.writer() as w:
w.add_document(id=0, kind=k, name=u("one"))
w.add_document(id=1, kind=k, name=u("two"))
w.add_document(id=2, kind=k, name=u("three"))
w.add_document(id=3, kind=k, name=u("four"))
w.add_document(id=4, kind=k, name=u("one"))
w.add_document(id=5, kind=k, name=u("two"))
w.add_document(id=6, kind=k, name=u("three"))
w.add_document(id=7, kind=k, name=u("four"))
w.add_document(id=8, kind=k, name=u("one"))
w.add_document(id=9, kind=k, name=u("two"))
w.add_document(id=10, kind=k, name=u("three"))
w.add_document(id=11, kind=k, name=u("four"))
with ix.searcher() as s:
pq = query.Term("kind", k)
cq = query.Or([query.Term("name", "two"), query.Term("name", "four")])
q = query.NestedParent(pq, cq)
r = s.search(q)
assert [hit["id"] for hit in r] == [1, 3, 5, 7, 9, 11]
def test_missing():
schema = fields.Schema(kind=fields.ID,
name=fields.KEYWORD(scorable=True, stored=True))
ix = RamStorage().create_index(schema)
with ix.writer() as w:
with w.group():
w.add_document(kind=u("class"), name=u("Index"))
w.add_document(kind=u("method"), name=u("add document"))
w.add_document(kind=u("method"), name=u("add reader"))
w.add_document(kind=u("method"), name=u("close"))
with w.group():
w.add_document(kind=u("class"), name=u("Accumulator"))
w.add_document(kind=u("method"), name=u("add"))
w.add_document(kind=u("method"), name=u("get result"))
with w.group():
w.add_document(kind=u("class"), name=u("Calculator"))
w.add_document(kind=u("method"), name=u("add"))
w.add_document(kind=u("method"), name=u("add all"))
w.add_document(kind=u("method"), name=u("add some"))
w.add_document(kind=u("method"), name=u("multiply"))
w.add_document(kind=u("method"), name=u("close"))
with w.group():
w.add_document(kind=u("class"), name=u("Deleter"))
w.add_document(kind=u("method"), name=u("add"))
w.add_document(kind=u("method"), name=u("delete"))
with ix.searcher() as s:
q = query.NestedParent(query.Term("kind", "class"),
query.Term("name", "add"))
r = s.search(q)
assert_equal([hit["name"] for hit in r],
["Calculator", "Index", "Accumulator", "Deleter"])
with ix.writer() as w:
w.delete_by_term("name", "Accumulator")
w.delete_by_term("name", "Calculator")
with ix.searcher() as s:
pq = query.Term("kind", "class")
assert_equal(len(list(pq.docs(s))), 2)
q = query.NestedParent(pq, query.Term("name", "add"))
r = s.search(q)
assert_equal([hit["name"] for hit in r], ["Index", "Deleter"])
def test_parent_score_fn():
from whoosh.scoring import Frequency
schema = fields.Schema(name=fields.ID(unique=True, stored=True),
keys=fields.TEXT,
type=fields.ID)
with TempIndex(schema) as ix:
with ix.writer() as w:
w.add_document(name=u"p1", type=u"parent")
w.add_document(name=u"c1.1", type=u"child", keys=u"key key")
w.add_document(name=u"c1.2", type=u"child", keys=u"key key key")
w.add_document(name=u"c1.3", type=u"child", keys=u"key key")
w.add_document(name=u"p2", type=u"parent")
w.add_document(name=u"c2.1", type=u"child", keys=u"")
w.add_document(name=u"c2.2",
type=u"child",
keys=u"key key key key")
w.add_document(name=u"c2.3", type=u"child", keys=u"key")
with ix.searcher(weighting=Frequency()) as s:
parents = query.Term("type", u"parent")
children = query.Term("keys", u"key")
q = query.NestedParent(parents, children, score_fn=max)
print(list(q.docs(s)))
r = s.search(q)
assert r.scored_length() == 2
assert r[0]["name"] == u"p2"
assert r[0].score == 4
assert r[1]["name"] == u"p1"
assert r[1].score == 3
q = query.NestedParent(parents, children, score_fn=min)
r = s.search(q)
assert r.scored_length() == 2
assert r[0]["name"] == u"p1"
assert r[0].score == 2
assert r[1]["name"] == u"p2"
assert r[1].score == 1
def test_all_parents_deleted():
schema = fields.Schema(kind=fields.ID,
name=fields.KEYWORD(scorable=True, stored=True))
ix = RamStorage().create_index(schema)
with ix.writer() as w:
with w.group():
w.add_document(kind=u("class"), name=u("Index"))
w.add_document(kind=u("method"), name=u("add document"))
w.add_document(kind=u("method"), name=u("add reader"))
w.add_document(kind=u("method"), name=u("close"))
with w.group():
w.add_document(kind=u("class"), name=u("Accumulator"))
w.add_document(kind=u("method"), name=u("add"))
w.add_document(kind=u("method"), name=u("get result"))
with w.group():
w.add_document(kind=u("class"), name=u("Calculator"))
w.add_document(kind=u("method"), name=u("add"))
w.add_document(kind=u("method"), name=u("add all"))
w.add_document(kind=u("method"), name=u("add some"))
w.add_document(kind=u("method"), name=u("multiply"))
w.add_document(kind=u("method"), name=u("close"))
with w.group():
w.add_document(kind=u("class"), name=u("Deleter"))
w.add_document(kind=u("method"), name=u("add"))
w.add_document(kind=u("method"), name=u("delete"))
with ix.writer() as w:
w.delete_by_term("name", "Index")
w.delete_by_term("name", "Accumulator")
w.delete_by_term("name", "Calculator")
w.delete_by_term("name", "Deleter")
with ix.searcher() as s:
q = query.NestedParent(query.Term("kind", "class"),
query.Term("name", "add"))
r = s.search(q)
assert r.is_empty()
