def test_and_resolution(self):
tree = (UnknownOperation(Word("a"), Word("b"),
OrOperation(Word("c"), Word("d"))))
expected = (AndOperation(Word("a"), Word("b"),
OrOperation(Word("c"), Word("d"))))
resolver = UnknownOperationResolver(resolve_to=AndOperation)
self.assertEqual(resolver(tree), expected)
def test_repeating_expression(self):
# non regression test
tree = AndOperation(
Group(OrOperation(Word('bar'), Word('foo'))),
Group(OrOperation(Word('bar'), Word('foo'), Word('spam'))),
)
# basic transformer should not change tree
same_tree = TreeTransformer().visit(copy.deepcopy(tree))
self.assertEqual(same_tree, tree)
def test_mro(self):
visitor = self.MROVisitor()
tree = OrOperation(Word('a'), Word('b'))
result = visitor.visit(tree)
self.assertEquals(result, 'a OR b')
tree = OrOperation(AndOperation(Word('a'), Word('b')), Word('c'))
result = visitor.visit(tree)
self.assertEquals(result, 'a BASE_OP b OR c')
def test_groups(self):
tree = (OrOperation(
Word('test', tail=" "),
Group(AndOperation(
SearchField("subject",
FieldGroup(
OrOperation(Word('foo', tail=" "),
Word('bar', head=" "))),
tail=" "), Word('baz', head=" ")),
head=" ")))
parsed = parser.parse('test OR (subject:(foo OR bar) AND baz)')
self.assertEqual(str(parsed), str(tree))
self.assertEqual(parsed, tree)
def test_removal(self):
tree = AndOperation(
OrOperation(Word("spam"), Word("ham")),
AndOperation(Word("foo"), Phrase('"bar"')),
AndOperation(Phrase('"baz"'), Phrase('"biz"')),
)
transformer = self.BasicTransformer()
new_tree = transformer.visit(tree)
self.assertEqual(
new_tree,
AndOperation(OrOperation(Word("lol"), Word("lol")), Word("lol")),
)
def test_lucene_resolution_last_op_with_group(self):
tree = (OrOperation(
Word("a"), Word("b"),
Group(
AndOperation(Word("c"), UnknownOperation(Word("d"),
Word("e")))),
UnknownOperation(Word("f"), Word("g")),
Group(UnknownOperation(Word("h"), Word("i")))))
expected = (OrOperation(
Word("a"), Word("b"),
Group(AndOperation(Word("c"), AndOperation(Word("d"), Word("e")))),
OrOperation(Word("f"), Word("g")),
Group(AndOperation(Word("h"), Word("i")))))
resolver = UnknownOperationResolver(resolve_to=None)
self.assertEqual(resolver(tree), expected)
def test_or_operation(self):
tree = OrOperation(Word("foo"), Phrase('"bar"'), Word("baz"))
all_paths = {(0, ), (1, ), (2, )}
matching = set()
paths_ok, paths_ko = self.propagate_matching(tree, matching,
all_paths - matching)
self.assertEqual(paths_ok, set())
self.assertEqual(paths_ko, {
(),
(0, ),
(1, ),
(2, ),
})
matching = {(2, )}
paths_ok, paths_ko = self.propagate_matching(tree, matching,
all_paths - matching)
self.assertEqual(paths_ok, {(), (2, )})
self.assertEqual(paths_ko, {(0, ), (1, )})
matching = {(0, ), (2, )}
paths_ok, paths_ko = self.propagate_matching(tree, matching,
all_paths - matching)
self.assertEqual(paths_ok, {(), (0, ), (2, )})
self.assertEqual(paths_ko, {(1, )})
matching = {(0, ), (1, ), (2, )}
paths_ok, paths_ko = self.propagate_matching(tree, matching,
all_paths - matching)
self.assertEqual(paths_ok, {(), (0, ), (1, ), (2, )})
self.assertEqual(paths_ko, set())
def test_named_queries_or(self):
tree = OrOperation(SearchField("text", Word("foo")),
SearchField("spam", Word("bar")))
set_name(tree.operands[0], "a")
set_name(tree.operands[1], "b")
result = self.transformer(tree)
self.assertEqual(
result, {
'bool': {
'should': [{
'term': {
'text': {
'_name': 'a',
'value': 'foo'
}
}
}, {
'match': {
'spam': {
'_name': 'b',
'query': 'bar',
'zero_terms_query': 'none'
}
}
}]
}
})
def visit_phrase(self, node, parent):
if (len(parent) > 0 and (isinstance(parent[-1], luqum.tree.SearchField)
or isinstance(parent[-1], luqum.tree.Range))):
return node
else:
search_list = [SearchField(f, node) for f in self.fields]
return Group(OrOperation(*search_list))
def test_zealous_or_not(self):
query = (OrOperation(Not(Word("foo")), Word("bar")))
check_zealous = LuceneCheck(zeal=1)
self.assertFalse(check_zealous(query))
self.assertIn("inconsistent", check_zealous.errors(query)[0])
check_easy_going = LuceneCheck()
self.assertTrue(check_easy_going(query))
def test_mro(self):
visitor = self.MROVisitor()
tree = OrOperation(Word('a'), Word('b'))
result = visitor.visit(tree)
self.assertEquals(list(result), ['a OR b', 'a', 'b'])
tree = AndOperation(Word('a'), Word('b'))
result = visitor.visit(tree)
self.assertEquals(list(result), ['a BASE_OP b', 'a', 'b'])
def test_auto_name_nested(self):
tree = AndOperation(
OrOperation(
SearchField("bar", Word("test")),
AndOperation(
Proximity(Phrase('"test"'), 2),
SearchField("baz", Word("test")),
),
),
Group(UnknownOperation(
Fuzzy(Word("test")),
Phrase('"test"'),
), ),
)
names = auto_name(tree)
self.assertEqual(sorted(names.keys()), list("abcdefgh"))
# and
and1 = tree
self.assertEqual(get_name(and1), None)
# - or
or1 = and1.children[0]
self.assertEqual(get_name(or1), "a")
self.assertEqual(names["a"], (0, ))
# -- search field
sfield1 = or1.children[0]
self.assertEqual(get_name(sfield1), "c")
self.assertEqual(names["c"], (0, 0))
self.assertEqual(get_name(sfield1.expr), None)
# -- and
and2 = or1.children[1]
self.assertEqual(get_name(and2), "d")
self.assertEqual(names["d"], (0, 1))
# --- proximity phrase
self.assertEqual(get_name(and2.children[0]), "e")
self.assertEqual(names["e"], (0, 1, 0))
self.assertEqual(get_name(and2.children[0].term), None)
# --- search field
sfield2 = and2.children[1]
self.assertEqual(get_name(sfield2), "f")
self.assertEqual(names["f"], (0, 1, 1))
self.assertEqual(get_name(sfield2.expr), None)
# - group
group1 = and1.children[1]
self.assertEqual(get_name(group1), "b")
self.assertEqual(names["b"], (1, ))
# -- unknown op
unknownop1 = group1.children[0]
self.assertEqual(get_name(unknownop1), None)
# --- fuzzy word
self.assertEqual(get_name(unknownop1.children[0]), "g")
self.assertEqual(names["g"], (1, 0, 0))
self.assertEqual(get_name(unknownop1.children[0].term), None)
# --- phrase
self.assertEqual(get_name(unknownop1.children[1]), "h")
self.assertEqual(names["h"], (1, 0, 1))
def test_visit_complex(self):
tree = AndOperation(
Group(
OrOperation(Word("foo"), Word("bar"),
Boost(Fuzzy(Word("baz")), force=2))),
Proximity(Phrase('"spam ham"')),
SearchField("fizz", Regex("/fuzz/")),
)
transformed = self.transform(tree)
expected = AndOperation(
Group(
OrOperation(
Word("foo@0-0-0"),
Word("bar@0-0-1"),
Boost(Fuzzy(Word("baz@0-0-2-0-0")), force=2),
)),
Proximity(Phrase('"spam ham@1-0"')),
SearchField("fizz", Regex("/fuzz@2-0/")),
)
self.assertEqual(transformed, expected)
def test_with_unknown_op_nested(self):
prettify = Prettifier(indent=8, max_len=20)
tree = OrOperation(
UnknownOperation(Word("baaaaaaaaaar"), Word("baaaaaaaaaaaaaz")),
Word("fooooooooooo"))
self.assertEqual(
"\n" + prettify(tree), """
baaaaaaaaaar
baaaaaaaaaaaaaz
OR
fooooooooooo""")
def test_bad_fieldgroup(self):
check = LuceneCheck()
query = FieldGroup(Word("foo"))
self.assertFalse(check(query))
self.assertEqual(len(check.errors(query)), 1)
self.assertIn("FieldGroup misuse", check.errors(query)[0])
query = OrOperation(FieldGroup(Word("bar")), Word("foo"))
self.assertFalse(check(query))
self.assertEqual(len(check.errors(query)), 1)
self.assertIn("FieldGroup misuse", check.errors(query)[0])
def test_mro(self):
visitor = self.MROVisitor()
tree = OrOperation(Word('a'), Word('b'))
result = visitor.visit(tree)
self.assertEquals(list(result), ['a OR b', 'a', 'b'])
# AndOperation has no specific method,
# but inherists BaseOperation, hence uses visit_base_operation
tree = AndOperation(Word('a'), Word('b'))
result = visitor.visit(tree)
self.assertEquals(list(result), ['a BASE_OP b', 'a', 'b'])
def test_named_queries_complex(self):
tree = (AndOperation(
SearchField("text", Phrase('"foo bar"')),
Group(OrOperation(
Word("bar"),
SearchField("spam", Word("baz")),
), ),
))
and_op = tree
search_text = and_op.operands[0]
or_op = and_op.operands[1].children[0]
bar = or_op.operands[0]
search_spam = or_op.operands[1]
set_name(search_text, "foo_bar")
set_name(bar, "bar")
set_name(search_spam, "baz")
expected = {
'bool': {
'must': [{
'term': {
'text': {
'_name': 'foo_bar',
'value': 'foo bar'
}
}
}, {
'bool': {
'should': [{
'term': {
'text': {
'_name': 'bar',
'value': 'bar'
}
}
}, {
'match': {
'spam': {
'_name': 'baz',
'query': 'baz',
'zero_terms_query': 'none'
}
}
}]
}
}]
}
}
result = self.transformer(tree)
self.assertEqual(result, expected)
def test_check_ok(self):
query = (AndOperation(
SearchField(
"f",
FieldGroup(
AndOperation(
Boost(Proximity(Phrase('"foo bar"'), 4), "4.2"),
Prohibit(Range("100", "200"))))),
Group(OrOperation(Fuzzy(Word("baz"), ".8"), Plus(Word("fizz"))))))
check = LuceneCheck()
self.assertTrue(check(query))
self.assertEqual(check.errors(query), [])
check = LuceneCheck(zeal=1)
self.assertTrue(check(query))
self.assertEqual(check.errors(query), [])
def test_set_children(self):
test = self._test_set_children
test(Word("foo"), [])
test(Phrase('"foo"'), [])
test(Regex("/foo/"), [])
test(SearchField("foo", Word("bar")), [Word("baz")])
test(Group(Word("foo")), [Word("foo")])
test(FieldGroup(Word("foo")), [Word("foo")])
test(Range(Word("20"), Word("30")), [Word("40"), Word("50")])
test(Proximity(Word("foo")), [Word("foo")])
test(Fuzzy(Word("foo")), [Word("foo")])
test(Boost(Word("foo"), force=1), [Word("foo")])
many_terms = tuple(Word(f"foo_{i}") for i in range(5))
test(UnknownOperation(Word("foo"), Word("bar")),
(Word("foo"), Word("bar")))
test(UnknownOperation(*many_terms), many_terms)
test(AndOperation(Word("foo"), Word("bar")),
(Word("foo"), Word("bar")))
test(AndOperation(*many_terms), many_terms)
test(OrOperation(Word("foo"), Word("bar")), (Word("foo"), Word("bar")))
test(OrOperation(*many_terms), many_terms)
test(Plus(Word("foo")), [Word("foo")])
test(Not(Word("foo")), [Word("foo")])
test(Prohibit(Word("foo")), [Word("foo")])
def test_visit_complex(self):
tree = AndOperation(
Group(
OrOperation(Word("foo"), Word("bar"),
Boost(Fuzzy(Word("baz")), force=2))),
Proximity(Phrase('"spam ham"')),
SearchField("fizz", Regex("/fuzz/")),
)
paths = self.visit(tree)
self.assertEqual(sorted(paths), [
((0, 0, 0), "foo"),
((0, 0, 1), "bar"),
((0, 0, 2, 0, 0), "baz"),
((1, 0), '"spam ham"'),
((2, 0), '/fuzz/'),
])
def test_auto_name_nested(self):
tree = AndOperation(
OrOperation(
SearchField("bar", Word("test")),
AndOperation(
Proximity(Phrase('"test"'), 2),
SearchField("baz", Word("test")),
),
),
Group(
UnknownOperation(
Fuzzy(Word("test")),
Phrase('"test"'),
),
),
)
auto_name(tree)
# and
and1 = tree
self.assertEqual(get_name(and1), "0")
# - or
or1 = and1.children[0]
self.assertEqual(get_name(or1), "0_0")
# --- search field word
sfield1 = or1.children[0]
self.assertFalse(get_name(sfield1))
self.assertEqual(get_name(sfield1.expr), "0_0_0")
# --- and
and2 = or1.children[1]
self.assertEqual(get_name(and2), "0_0_1")
# ----- proximity phrase
self.assertEqual(get_name(and2.children[0].term), "0_0_1_0")
# ----- search field word
sfield2 = and2.children[1]
self.assertFalse(get_name(sfield2))
self.assertEqual(get_name(sfield2.expr), "0_0_1_1")
# - group
group1 = and1.children[1]
self.assertEqual(get_name(group1), None)
# --- unknown op
unknownop1 = group1.children[0]
self.assertEqual(get_name(unknownop1), "0_1")
# ----- fuzzy word
self.assertEqual(get_name(unknownop1.children[0].term), "0_1_0")
# ----- phrase
self.assertEqual(get_name(unknownop1.children[1]), "0_1_1")
def test_lucene_resolution_last_op(self):
tree = (OrOperation(
Word("a"), Word("b"), UnknownOperation(Word("c"), Word("d")),
AndOperation(Word("e"), UnknownOperation(Word("f"), Word("g"))),
UnknownOperation(Word("i"), Word("j")),
OrOperation(Word("k"), UnknownOperation(Word("l"), Word("m"))),
UnknownOperation(Word("n"), Word("o"))))
expected = (OrOperation(
Word("a"), Word("b"), OrOperation(Word("c"), Word("d")),
AndOperation(Word("e"), AndOperation(Word("f"), Word("g"))),
AndOperation(Word("i"), Word("j")),
OrOperation(Word("k"), OrOperation(Word("l"), Word("m"))),
OrOperation(Word("n"), Word("o"))))
resolver = UnknownOperationResolver(resolve_to=None)
self.assertEqual(resolver(tree), expected)
def test_combinations(self):
# self.assertEqual(parser.parse("subject:test desc:(house OR car)").pval, "")
tree = (UnknownOperation(
SearchField("subject", Word("test"), tail=" "),
AndOperation(
SearchField("desc",
FieldGroup(
OrOperation(Word("house", tail=" "),
Word("car", head=" "))),
tail=" "),
Not(Proximity(Phrase('"approximatly this"'), 3, head=" "),
head=" "))))
parsed = parser.parse(
'subject:test desc:(house OR car) AND NOT "approximatly this"~3')
self.assertEqual(str(parsed), str(tree))
self.assertEqual(parsed, tree)
def test_auto_name_integration(self):
tree = (AndOperation(
SearchField("text", Phrase('"foo bar"')),
Group(OrOperation(
Word("bar"),
SearchField("spam", Word("baz")),
), ),
))
auto_name(tree)
expected = {
'bool': {
'must': [{
'term': {
'text': {
'_name': 'a',
'value': 'foo bar'
}
}
}, {
'bool': {
'should': [{
'term': {
'text': {
'_name': 'c',
'value': 'bar'
}
}
}, {
'match': {
'spam': {
'_name': 'd',
'query': 'baz',
'zero_terms_query': 'none'
}
}
}]
}
}]
}
}
result = self.transformer(tree)
self.assertEqual(result, expected)
def test_nested_negation(self):
for NegClass in (Prohibit, Not):
with self.subTest("%r" % NegClass):
tree = AndOperation(
NegClass(
OrOperation(
NegClass(
AndOperation(
NegClass(Word("a")),
Word("b"),
)),
Word("c"),
)),
Word("d"),
)
a, b, c, d = (0, 0, 0, 0, 0, 0), (0, 0, 0, 0, 1), (0, 0,
1), (1, )
not_a, ab, not_ab = (0, 0, 0, 0, 0), (0, 0, 0, 0), (0, 0, 0)
abc, not_abc, abcd = (0, 0), (0, ), ()
paths_ok, paths_ko = self.propagate_matching(
tree, set(), {a, b, c, d})
self.assertEqual(paths_ok, {not_a, not_ab, abc})
self.assertEqual(paths_ko, {a, b, ab, c, not_abc, d, abcd})
paths_ok, paths_ko = self.propagate_matching(
tree, {b, d}, {a, c})
self.assertEqual(paths_ok, {not_a, b, ab, not_abc, d, abcd})
self.assertEqual(paths_ko, {a, not_ab, c, abc})
paths_ok, paths_ko = self.propagate_matching(
tree, {a, b, c}, {d})
self.assertEqual(paths_ok, {a, b, not_ab, c, abc})
self.assertEqual(paths_ko, {not_a, ab, not_abc, d, abcd})
paths_ok, paths_ko = self.propagate_matching(
tree, {a, b, c, d}, set())
self.assertEqual(paths_ok, {a, b, not_ab, c, abc, d})
self.assertEqual(paths_ko, {not_a, ab, not_abc, abcd})
def test_unknown_operation(self):
tree = UnknownOperation(Word("foo"), Phrase('"bar"'), Word("baz"))
tree_or = OrOperation(Word("foo"), Phrase('"bar"'), Word("baz"))
tree_and = AndOperation(Word("foo"), Phrase('"bar"'), Word("baz"))
propagate_or = self.propagate_matching
propagate_and = MatchingPropagator(default_operation=AndOperation)
all_paths = {(0, ), (1, ), (2, )}
for matching in [
set(), {(2, )}, {(0, ), (2, )}, {(0, ), (1, ), (2, )}
]:
self.assertEqual(
propagate_or(tree, matching),
self.propagate_matching(tree_or, matching,
matching - all_paths),
)
self.assertEqual(
propagate_and(tree, matching),
self.propagate_matching(tree_and, matching,
matching - all_paths),
)
def test_complex(self):
tree = Group(
OrOperation(
SearchField(
"foo",
FieldGroup(
UnknownOperation(Word("bar"),
Range(Word("baz"), Word("spam")))),
),
Not(Proximity(Phrase('"ham ham"'), 2)),
Plus(Fuzzy(Word("hammer"), 3)),
))
self.assertEqual(str(tree),
'(foo:(bar[bazTOspam])ORNOT"ham ham"~2OR+hammer~3)')
self.assertEqual(
str(auto_head_tail(tree)),
'(foo:(bar [baz TO spam]) OR NOT "ham ham"~2 OR +hammer~3)',
)
# idem potent
self.assertEqual(
str(auto_head_tail(auto_head_tail(tree))),
'(foo:(bar [baz TO spam]) OR NOT "ham ham"~2 OR +hammer~3)',
)
def test_element_from_path(self):
tree = AndOperation(
OrOperation(
SearchField("bar", Word("test")),
Group(
AndOperation(
Proximity(Phrase('"test"'), 2),
SearchField("baz", Word("test")),
Fuzzy(Word("test")),
Phrase('"test"'),
), ),
), )
names = {
"a": (),
"b": (0, 1),
"c": (0, 1, 0, 2),
"d": (0, 1, 0, 2, 0),
"e": (0, 1, 0, 3)
}
self.assertEqual(element_from_path(tree, ()), tree)
self.assertEqual(element_from_name(tree, "a", names), tree)
self.assertEqual(element_from_path(tree, (0, 1)),
tree.children[0].children[1])
self.assertEqual(element_from_name(tree, "b", names),
tree.children[0].children[1])
self.assertEqual(element_from_path(tree, (0, 1, 0, 2)),
Fuzzy(Word("test")))
self.assertEqual(element_from_name(tree, "c", names),
Fuzzy(Word("test")))
self.assertEqual(element_from_path(tree, (0, 1, 0, 2, 0)),
Word("test"))
self.assertEqual(element_from_name(tree, "d", names), Word("test"))
self.assertEqual(element_from_path(tree, (0, 1, 0, 3)),
Phrase('"test"'))
self.assertEqual(element_from_name(tree, "e", names), Phrase('"test"'))
with self.assertRaises(IndexError):
element_from_path(tree, (1, ))
def test_or_operator(self):
tree = parser.parse("\tfoo\nOR bar\rOR\t\nbaz\r\r")
self.assertEqual(tree,
OrOperation(Word("foo"), Word("bar"), Word("baz")))
self.assertEqual(tree.head, "")
self.assertEqual(tree.tail, "")
self.assertEqual(tree.pos, 0)
self.assertEqual(tree.size, 22)
foo, bar, baz = tree.children
self.assertEqual(foo.head, "\t")
self.assertEqual(foo.tail, "\n")
self.assertEqual(foo.pos, 1)
self.assertEqual(foo.size, 3)
self.assertEqual(bar.head, " ")
self.assertEqual(bar.tail, "\r")
self.assertEqual(bar.pos, 9)
self.assertEqual(bar.size, 3)
self.assertEqual(baz.head, "\t\n")
self.assertEqual(baz.tail, "\r\r")
self.assertEqual(baz.pos, 17)
self.assertEqual(baz.size, 3)
self.assertEqual(str(tree), "\tfoo\nOR bar\rOR\t\nbaz\r\r")
self.assertEqual(tree.__str__(head_tail=True),
"\tfoo\nOR bar\rOR\t\nbaz\r\r")
def test_not_equality(self):
# non regression test, adding a child should trigger non equality
tree1 = OrOperation(Word("bar"))
tree2 = OrOperation(Word("bar"), Word("foo"))
self.assertNotEqual(tree1, tree2)
