mirror of https://github.com/milvus-io/milvus.git
test: Add a new range search test for all indexes and align some index params (#32724)
related issue: https://github.com/milvus-io/milvus/issues/32653 1. align some default index params 2. add a new range search tests for all indexes and float vectors --------- Signed-off-by: yanliang567 <yanliang.qiao@zilliz.com>pull/32743/head
yanliang567
GitHub
parent
c70c21e70b
commit
5bb672d70d
|
|
@ -242,7 +242,8 @@ class TestcaseBase(Base):
|
|||
expected: return collection and raw data, insert ids
|
||||
"""
|
||||
log.info("Test case of search interface: initialize before test case")
|
||||
self._connect()
|
||||
if not self.connection_wrap.has_connection(alias=DefaultConfig.DEFAULT_USING)[0]:
|
||||
self._connect()
|
||||
collection_name = cf.gen_unique_str(prefix)
|
||||
if name is not None:
|
||||
collection_name = name
|
||||
|
|
|
|||
|
|
@ -359,9 +359,9 @@ def gen_collection_schema_all_datatype(description=ct.default_desc,
|
|||
else:
|
||||
multiple_dim_array.insert(0, dim)
|
||||
for i in range(len(multiple_dim_array)):
|
||||
fields.append(gen_float_vec_field(name=f"multiple_vector_{ct.vector_data_type_all[i%3]}",
|
||||
fields.append(gen_float_vec_field(name=f"multiple_vector_{ct.all_float_vector_types[i%3]}",
|
||||
dim=multiple_dim_array[i],
|
||||
vector_data_type=ct.vector_data_type_all[i%3]))
|
||||
vector_data_type=ct.all_float_vector_types[i%3]))
|
||||
|
||||
schema, _ = ApiCollectionSchemaWrapper().init_collection_schema(fields=fields, description=description,
|
||||
primary_field=primary_field, auto_id=auto_id,
|
||||
|
|
@ -485,8 +485,8 @@ def gen_default_dataframe_data(nb=ct.default_nb, dim=ct.default_dim, start=0, wi
|
|||
|
||||
def gen_general_default_list_data(nb=ct.default_nb, dim=ct.default_dim, start=0, with_json=True,
|
||||
random_primary_key=False, multiple_dim_array=[], multiple_vector_field_name=[],
|
||||
vector_data_type="FLOAT_VECTOR", auto_id = False,
|
||||
primary_field = ct.default_int64_field_name):
|
||||
vector_data_type="FLOAT_VECTOR", auto_id=False,
|
||||
primary_field=ct.default_int64_field_name):
|
||||
insert_list = []
|
||||
if not random_primary_key:
|
||||
int_values = pd.Series(data=[i for i in range(start, start + nb)])
|
||||
|
|
@ -496,14 +496,15 @@ def gen_general_default_list_data(nb=ct.default_nb, dim=ct.default_dim, start=0,
|
|||
string_values = pd.Series(data=[str(i) for i in range(start, start + nb)], dtype="string")
|
||||
json_values = [{"number": i, "float": i*1.0} for i in range(start, start + nb)]
|
||||
float_vec_values = gen_vectors(nb, dim, vector_data_type=vector_data_type)
|
||||
insert_list = [int_values, float_values, string_values, json_values, float_vec_values]
|
||||
insert_list = [int_values, float_values, string_values]
|
||||
|
||||
if with_json is True:
|
||||
insert_list.append(json_values)
|
||||
insert_list.append(float_vec_values)
|
||||
|
||||
if with_json is False:
|
||||
index = insert_list.index(json_values)
|
||||
del insert_list[index]
|
||||
if auto_id is True:
|
||||
if primary_field == ct.default_int64_field_name:
|
||||
index = insert_list.index(int_values)
|
||||
index = 0
|
||||
elif primary_field == ct.default_string_field_name:
|
||||
index = 2
|
||||
del insert_list[index]
|
||||
|
|
@ -699,7 +700,7 @@ def gen_dataframe_all_data_type(nb=ct.default_nb, dim=ct.default_dim, start=0, w
|
|||
df[ct.default_float_vec_field_name] = float_vec_values
|
||||
else:
|
||||
for i in range(len(multiple_dim_array)):
|
||||
df[multiple_vector_field_name[i]] = gen_vectors(nb, multiple_dim_array[i], ct.vector_data_type_all[i%3])
|
||||
df[multiple_vector_field_name[i]] = gen_vectors(nb, multiple_dim_array[i], ct.all_float_vector_types[i%3])
|
||||
|
||||
if with_json is False:
|
||||
df.drop(ct.default_json_field_name, axis=1, inplace=True)
|
||||
|
|
@ -737,7 +738,7 @@ def gen_general_list_all_data_type(nb=ct.default_nb, dim=ct.default_dim, start=0
|
|||
insert_list.append(float_vec_values)
|
||||
else:
|
||||
for i in range(len(multiple_dim_array)):
|
||||
insert_list.append(gen_vectors(nb, multiple_dim_array[i], ct.vector_data_type_all[i%3]))
|
||||
insert_list.append(gen_vectors(nb, multiple_dim_array[i], ct.all_float_vector_types[i%3]))
|
||||
|
||||
if with_json is False:
|
||||
# index = insert_list.index(json_values)
|
||||
|
|
@ -782,7 +783,7 @@ def gen_default_rows_data_all_data_type(nb=ct.default_nb, dim=ct.default_dim, st
|
|||
else:
|
||||
for i in range(len(multiple_dim_array)):
|
||||
dict[multiple_vector_field_name[i]] = gen_vectors(nb, multiple_dim_array[i],
|
||||
ct.vector_data_type_all[i])[0]
|
||||
ct.all_float_vector_types[i])[0]
|
||||
if len(multiple_dim_array) != 0:
|
||||
with open(ct.rows_all_data_type_file_path + f'_{partition_id}' + f'_dim{dim}.txt', 'wb') as json_file:
|
||||
pickle.dump(array, json_file)
|
||||
|
|
@ -1233,7 +1234,7 @@ def gen_simple_index():
|
|||
elif ct.all_index_types[i] in ct.sparse_support:
|
||||
continue
|
||||
dic = {"index_type": ct.all_index_types[i], "metric_type": "L2"}
|
||||
dic.update({"params": ct.default_index_params[i]})
|
||||
dic.update({"params": ct.default_all_indexes_params[i]})
|
||||
index_params.append(dic)
|
||||
return index_params
|
||||
|
||||
|
|
@ -1671,6 +1672,16 @@ def index_to_dict(index):
|
|||
}
|
||||
|
||||
|
||||
def get_index_params_params(index_type):
|
||||
"""get default params of index params by index type"""
|
||||
return ct.default_all_indexes_params[ct.all_index_types.index(index_type)]
|
||||
|
||||
|
||||
def get_search_params_params(index_type):
|
||||
"""get default params of search params by index type"""
|
||||
return ct.default_all_search_params_params[ct.all_index_types.index(index_type)]
|
||||
|
||||
|
||||
def assert_json_contains(expr, list_data):
|
||||
opposite = False
|
||||
if expr.startswith("not"):
|
||||
|
|
|
|||
|
|
@ -14,14 +14,6 @@ default_nq = 2
|
|||
default_limit = 10
|
||||
default_batch_size = 1000
|
||||
max_limit = 16384
|
||||
default_search_params = {"metric_type": "COSINE", "params": {"nprobe": 10}}
|
||||
default_search_ip_params = {"metric_type": "IP", "params": {"nprobe": 10}}
|
||||
default_search_binary_params = {"metric_type": "JACCARD", "params": {"nprobe": 10}}
|
||||
default_index = {"index_type": "IVF_SQ8", "metric_type": "COSINE", "params": {"nlist": 64}}
|
||||
default_binary_index = {"index_type": "BIN_IVF_FLAT", "params": {"nlist": 128}, "metric_type": "JACCARD"}
|
||||
default_diskann_index = {"index_type": "DISKANN", "metric_type": "COSINE", "params": {}}
|
||||
default_diskann_search_params = {"metric_type": "COSINE", "params": {"search_list": 30}}
|
||||
default_sparse_search_params = {"metric_type": "IP", "params": {"drop_ratio_search": "0.2"}}
|
||||
max_top_k = 16384
|
||||
max_partition_num = 4096
|
||||
max_role_num = 10
|
||||
|
|
@ -52,7 +44,7 @@ default_binary_vec_field_name = "binary_vector"
|
|||
float_type = "FLOAT_VECTOR"
|
||||
float16_type = "FLOAT16_VECTOR"
|
||||
bfloat16_type = "BFLOAT16_VECTOR"
|
||||
vector_data_type_all = [float_type, float16_type, bfloat16_type]
|
||||
all_float_vector_types = [float_type, float16_type, bfloat16_type]
|
||||
default_sparse_vec_field_name = "sparse_vector"
|
||||
default_partition_name = "_default"
|
||||
default_resource_group_name = '__default_resource_group'
|
||||
|
|
@ -108,11 +100,6 @@ code = "code"
|
|||
err_code = "err_code"
|
||||
err_msg = "err_msg"
|
||||
in_cluster_env = "IN_CLUSTER"
|
||||
|
||||
default_flat_index = {"index_type": "FLAT", "params": {}, "metric_type": "COSINE"}
|
||||
default_bin_flat_index = {"index_type": "BIN_FLAT", "params": {}, "metric_type": "JACCARD"}
|
||||
default_sparse_inverted_index = {"index_type": "SPARSE_INVERTED_INDEX", "metric_type": "IP",
|
||||
"params": {"drop_ratio_build": 0.2}}
|
||||
default_count_output = "count(*)"
|
||||
|
||||
rows_all_data_type_file_path = "/tmp/rows_all_data_type"
|
||||
|
|
@ -250,26 +237,50 @@ get_wrong_format_dict = [
|
|||
]
|
||||
|
||||
""" Specially defined list """
|
||||
all_index_types = ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ", "HNSW", "SCANN", "DISKANN", "BIN_FLAT", "BIN_IVF_FLAT",
|
||||
"SPARSE_INVERTED_INDEX", "SPARSE_WAND", "GPU_IVF_FLAT", "GPU_IVF_PQ"]
|
||||
L0_index_types = ["IVF_SQ8", "HNSW", "DISKANN"]
|
||||
all_index_types = ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ",
|
||||
"HNSW", "SCANN", "DISKANN",
|
||||
"BIN_FLAT", "BIN_IVF_FLAT",
|
||||
"SPARSE_INVERTED_INDEX", "SPARSE_WAND",
|
||||
"GPU_IVF_FLAT", "GPU_IVF_PQ"]
|
||||
|
||||
default_index_params = [{"nlist": 128}, {"nlist": 128}, {"nlist": 128}, {"nlist": 128, "m": 16, "nbits": 8},
|
||||
{"M": 48, "efConstruction": 500}, {"nlist": 128}, {}, {"nlist": 128}, {"nlist": 128},
|
||||
{"drop_ratio_build": 0.2}, {"drop_ratio_build": 0.2},
|
||||
{"nlist": 64}, {"nlist": 64, "m": 16, "nbits": 8}]
|
||||
default_all_indexes_params = [{}, {"nlist": 128}, {"nlist": 128}, {"nlist": 128, "m": 16, "nbits": 8},
|
||||
{"M": 32, "efConstruction": 360}, {"nlist": 128}, {},
|
||||
{}, {"nlist": 64},
|
||||
{"drop_ratio_build": 0.2}, {"drop_ratio_build": 0.2},
|
||||
{"nlist": 64}, {"nlist": 64, "m": 16, "nbits": 8}]
|
||||
|
||||
default_all_search_params_params = [{}, {"nprobe": 32}, {"nprobe": 32}, {"nprobe": 32},
|
||||
{"ef": 100}, {"nprobe": 32, "reorder_k": 100}, {"search_list": 30},
|
||||
{}, {"nprobe": 32},
|
||||
{"drop_ratio_search": "0.2"}, {"drop_ratio_search": "0.2"},
|
||||
{}, {}]
|
||||
|
||||
Handler_type = ["GRPC", "HTTP"]
|
||||
binary_support = ["BIN_FLAT", "BIN_IVF_FLAT"]
|
||||
delete_support = ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ"]
|
||||
ivf = ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ"]
|
||||
skip_pq = ["IVF_PQ"]
|
||||
sparse_support = ["SPARSE_INVERTED_INDEX", "SPARSE_WAND"]
|
||||
default_L0_metric = "COSINE"
|
||||
float_metrics = ["L2", "IP", "COSINE"]
|
||||
binary_metrics = ["JACCARD", "HAMMING", "SUBSTRUCTURE", "SUPERSTRUCTURE"]
|
||||
structure_metrics = ["SUBSTRUCTURE", "SUPERSTRUCTURE"]
|
||||
all_scalar_data_types = ['int8', 'int16', 'int32', 'int64', 'float', 'double', 'bool', 'varchar']
|
||||
|
||||
|
||||
default_flat_index = {"index_type": "FLAT", "params": {}, "metric_type": default_L0_metric}
|
||||
default_bin_flat_index = {"index_type": "BIN_FLAT", "params": {}, "metric_type": "JACCARD"}
|
||||
default_sparse_inverted_index = {"index_type": "SPARSE_INVERTED_INDEX", "metric_type": "IP",
|
||||
"params": {"drop_ratio_build": 0.2}}
|
||||
|
||||
default_search_params = {"params": default_all_search_params_params[2]}
|
||||
default_search_ip_params = {"metric_type": "IP", "params": default_all_search_params_params[2]}
|
||||
default_search_binary_params = {"metric_type": "JACCARD", "params": {"nprobe": 32}}
|
||||
default_index = {"index_type": "IVF_SQ8", "metric_type": default_L0_metric, "params": default_all_indexes_params[2]}
|
||||
default_binary_index = {"index_type": "BIN_IVF_FLAT", "metric_type": "JACCARD", "params": default_all_indexes_params[8]}
|
||||
default_diskann_index = {"index_type": "DISKANN", "metric_type": default_L0_metric, "params": {}}
|
||||
default_diskann_search_params = {"params": {"search_list": 30}}
|
||||
default_sparse_search_params = {"metric_type": "IP", "params": {"drop_ratio_search": "0.2"}}
|
||||
|
||||
|
||||
class CheckTasks:
|
||||
""" The name of the method used to check the result """
|
||||
check_nothing = "check_nothing"
|
||||
|
|
|
|||
|
|
@ -1,94 +1,94 @@
|
|||
import datetime
|
||||
import pytest
|
||||
|
||||
from base.client_base import TestcaseBase
|
||||
from common import common_func as cf
|
||||
from common import common_type as ct
|
||||
from common.common_type import CaseLabel
|
||||
from utils.util_log import test_log as log
|
||||
from pymilvus import utility
|
||||
|
||||
|
||||
rounds = 100
|
||||
per_nb = 100000
|
||||
default_field_name = ct.default_float_vec_field_name
|
||||
default_index_params = {"index_type": "IVF_SQ8", "metric_type": "L2", "params": {"nlist": 64}}
|
||||
|
||||
|
||||
class TestLoad(TestcaseBase):
|
||||
""" Test case of end to end"""
|
||||
@pytest.mark.tags(CaseLabel.L3)
|
||||
def test_load_default(self):
|
||||
name = 'load_test_collection_1'
|
||||
name2 = 'load_test_collection_2'
|
||||
# create
|
||||
# collection_w = self.init_collection_wrap(name=name)
|
||||
# collection_w2 = self.init_collection_wrap(name=name2)
|
||||
# assert collection_w.name == name
|
||||
|
||||
for i in range(50):
|
||||
name = f"load_collection2_{i}"
|
||||
self.init_collection_wrap(name=name)
|
||||
log.debug(f"total collections: {len(utility.list_collections())}")
|
||||
|
||||
# # insert
|
||||
# data = cf.gen_default_list_data(per_nb)
|
||||
# log.debug(f"data len: {len(data[0])}")
|
||||
# for i in range(rounds):
|
||||
# t0 = datetime.datetime.now()
|
||||
# ins_res, res = collection_w.insert(data, timeout=180)
|
||||
# tt = datetime.datetime.now() - t0
|
||||
# log.debug(f"round{i} insert: {len(ins_res.primary_keys)} entities in {tt}s")
|
||||
# assert res # and per_nb == len(ins_res.primary_keys)
|
||||
#
|
||||
# t0 = datetime.datetime.now()
|
||||
# ins_res2, res = collection_w2.insert(data, timeout=180)
|
||||
# tt = datetime.datetime.now() - t0
|
||||
# log.debug(f"round{i} insert2: {len(ins_res2.primary_keys)} entities in {tt}s")
|
||||
# assert res
|
||||
#
|
||||
# # flush
|
||||
# t0 = datetime.datetime.now()
|
||||
# log.debug(f"current collection num_entities: {collection_w.num_entities}")
|
||||
# tt = datetime.datetime.now() - t0
|
||||
# log.debug(f"round{i} flush in {tt}")
|
||||
#
|
||||
# t0 = datetime.datetime.now()
|
||||
# log.debug(f"current collection2 num_entities: {collection_w2.num_entities}")
|
||||
# tt = datetime.datetime.now() - t0
|
||||
# log.debug(f"round{i} flush2 in {tt}")
|
||||
|
||||
# index, res = collection_w.create_index(default_field_name, default_index_params, timeout=60)
|
||||
# assert res
|
||||
|
||||
# # search
|
||||
# collection_w.load()
|
||||
# search_vectors = cf.gen_vectors(1, ct.default_dim)
|
||||
# t0 = datetime.datetime.now()
|
||||
# res_1, _ = collection_w.search(data=search_vectors,
|
||||
# anns_field=ct.default_float_vec_field_name,
|
||||
# param={"nprobe": 16}, limit=1)
|
||||
# tt = datetime.datetime.now() - t0
|
||||
# log.debug(f"assert search: {tt}")
|
||||
# assert len(res_1) == 1
|
||||
# # collection_w.release()
|
||||
#
|
||||
# # index
|
||||
# collection_w.insert(cf.gen_default_dataframe_data(nb=5000))
|
||||
# assert collection_w.num_entities == len(data[0]) + 5000
|
||||
# _index_params = {"index_type": "IVF_SQ8", "metric_type": "L2", "params": {"nlist": 64}}
|
||||
# t0 = datetime.datetime.now()
|
||||
# index, _ = collection_w.create_index(field_name=ct.default_float_vec_field_name,
|
||||
# index_params=_index_params,
|
||||
# name=cf.gen_unique_str())
|
||||
# tt = datetime.datetime.now() - t0
|
||||
# log.debug(f"assert index: {tt}")
|
||||
# assert len(collection_w.indexes) == 1
|
||||
#
|
||||
# # query
|
||||
# term_expr = f'{ct.default_int64_field_name} in [3001,4001,4999,2999]'
|
||||
# t0 = datetime.datetime.now()
|
||||
# res, _ = collection_w.query(term_expr)
|
||||
# tt = datetime.datetime.now() - t0
|
||||
# log.debug(f"assert query: {tt}")
|
||||
# assert len(res) == 4
|
||||
# import datetime
|
||||
# import pytest
|
||||
#
|
||||
# from base.client_base import TestcaseBase
|
||||
# from common import common_func as cf
|
||||
# from common import common_type as ct
|
||||
# from common.common_type import CaseLabel
|
||||
# from utils.util_log import test_log as log
|
||||
# from pymilvus import utility
|
||||
#
|
||||
#
|
||||
# rounds = 100
|
||||
# per_nb = 100000
|
||||
# default_field_name = ct.default_float_vec_field_name
|
||||
# default_index_params = {"index_type": "IVF_SQ8", "metric_type": "L2", "params": {"nlist": 64}}
|
||||
#
|
||||
#
|
||||
# class TestLoad(TestcaseBase):
|
||||
# """ Test case of end to end"""
|
||||
# @pytest.mark.tags(CaseLabel.L3)
|
||||
# def test_load_default(self):
|
||||
# name = 'load_test_collection_1'
|
||||
# name2 = 'load_test_collection_2'
|
||||
# # create
|
||||
# # collection_w = self.init_collection_wrap(name=name)
|
||||
# # collection_w2 = self.init_collection_wrap(name=name2)
|
||||
# # assert collection_w.name == name
|
||||
#
|
||||
# for i in range(50):
|
||||
# name = f"load_collection2_{i}"
|
||||
# self.init_collection_wrap(name=name)
|
||||
# log.debug(f"total collections: {len(utility.list_collections())}")
|
||||
#
|
||||
# # # insert
|
||||
# # data = cf.gen_default_list_data(per_nb)
|
||||
# # log.debug(f"data len: {len(data[0])}")
|
||||
# # for i in range(rounds):
|
||||
# # t0 = datetime.datetime.now()
|
||||
# # ins_res, res = collection_w.insert(data, timeout=180)
|
||||
# # tt = datetime.datetime.now() - t0
|
||||
# # log.debug(f"round{i} insert: {len(ins_res.primary_keys)} entities in {tt}s")
|
||||
# # assert res # and per_nb == len(ins_res.primary_keys)
|
||||
# #
|
||||
# # t0 = datetime.datetime.now()
|
||||
# # ins_res2, res = collection_w2.insert(data, timeout=180)
|
||||
# # tt = datetime.datetime.now() - t0
|
||||
# # log.debug(f"round{i} insert2: {len(ins_res2.primary_keys)} entities in {tt}s")
|
||||
# # assert res
|
||||
# #
|
||||
# # # flush
|
||||
# # t0 = datetime.datetime.now()
|
||||
# # log.debug(f"current collection num_entities: {collection_w.num_entities}")
|
||||
# # tt = datetime.datetime.now() - t0
|
||||
# # log.debug(f"round{i} flush in {tt}")
|
||||
# #
|
||||
# # t0 = datetime.datetime.now()
|
||||
# # log.debug(f"current collection2 num_entities: {collection_w2.num_entities}")
|
||||
# # tt = datetime.datetime.now() - t0
|
||||
# # log.debug(f"round{i} flush2 in {tt}")
|
||||
#
|
||||
# # index, res = collection_w.create_index(default_field_name, default_all_indexes_params, timeout=60)
|
||||
# # assert res
|
||||
#
|
||||
# # # search
|
||||
# # collection_w.load()
|
||||
# # search_vectors = cf.gen_vectors(1, ct.default_dim)
|
||||
# # t0 = datetime.datetime.now()
|
||||
# # res_1, _ = collection_w.search(data=search_vectors,
|
||||
# # anns_field=ct.default_float_vec_field_name,
|
||||
# # param={"nprobe": 16}, limit=1)
|
||||
# # tt = datetime.datetime.now() - t0
|
||||
# # log.debug(f"assert search: {tt}")
|
||||
# # assert len(res_1) == 1
|
||||
# # # collection_w.release()
|
||||
# #
|
||||
# # # index
|
||||
# # collection_w.insert(cf.gen_default_dataframe_data(nb=5000))
|
||||
# # assert collection_w.num_entities == len(data[0]) + 5000
|
||||
# # _index_params = {"index_type": "IVF_SQ8", "metric_type": "L2", "params": {"nlist": 64}}
|
||||
# # t0 = datetime.datetime.now()
|
||||
# # index, _ = collection_w.create_index(field_name=ct.default_float_vec_field_name,
|
||||
# # index_params=_index_params,
|
||||
# # name=cf.gen_unique_str())
|
||||
# # tt = datetime.datetime.now() - t0
|
||||
# # log.debug(f"assert index: {tt}")
|
||||
# # assert len(collection_w.indexes) == 1
|
||||
# #
|
||||
# # # query
|
||||
# # term_expr = f'{ct.default_int64_field_name} in [3001,4001,4999,2999]'
|
||||
# # t0 = datetime.datetime.now()
|
||||
# # res, _ = collection_w.query(term_expr)
|
||||
# # tt = datetime.datetime.now() - t0
|
||||
# # log.debug(f"assert query: {tt}")
|
||||
# # assert len(res) == 4
|
||||
|
|
|
|||
|
|
@ -274,7 +274,7 @@ class TestMilvusClientIndexValid(TestcaseBase):
|
|||
@pytest.mark.skip("https://github.com/milvus-io/pymilvus/issues/1886")
|
||||
@pytest.mark.parametrize("index, params",
|
||||
zip(ct.all_index_types[:7],
|
||||
ct.default_index_params[:7]))
|
||||
ct.default_all_indexes_params[:7]))
|
||||
def test_milvus_client_index_default(self, index, params, metric_type):
|
||||
"""
|
||||
target: test search (high level api) normal case
|
||||
|
|
@ -324,7 +324,7 @@ class TestMilvusClientIndexValid(TestcaseBase):
|
|||
@pytest.mark.skip(reason="pymilvus issue 1884")
|
||||
@pytest.mark.parametrize("index, params",
|
||||
zip(ct.all_index_types[:7],
|
||||
ct.default_index_params[:7]))
|
||||
ct.default_all_indexes_params[:7]))
|
||||
def test_milvus_client_index_with_params(self, index, params, metric_type):
|
||||
"""
|
||||
target: test search (high level api) normal case
|
||||
|
|
@ -372,7 +372,7 @@ class TestMilvusClientIndexValid(TestcaseBase):
|
|||
@pytest.mark.skip("wait for modification")
|
||||
@pytest.mark.parametrize("index, params",
|
||||
zip(ct.all_index_types[:7],
|
||||
ct.default_index_params[:7]))
|
||||
ct.default_all_indexes_params[:7]))
|
||||
def test_milvus_client_index_after_insert(self, index, params, metric_type):
|
||||
"""
|
||||
target: test search (high level api) normal case
|
||||
|
|
@ -518,7 +518,7 @@ class TestMilvusClientIndexValid(TestcaseBase):
|
|||
@pytest.mark.skip("wait for modification")
|
||||
@pytest.mark.parametrize("index, params",
|
||||
zip(ct.all_index_types[:7],
|
||||
ct.default_index_params[:7]))
|
||||
ct.default_all_indexes_params[:7]))
|
||||
def test_milvus_client_index_drop_create_same_index(self, index, params, metric_type):
|
||||
"""
|
||||
target: test search (high level api) normal case
|
||||
|
|
@ -570,7 +570,7 @@ class TestMilvusClientIndexValid(TestcaseBase):
|
|||
@pytest.mark.skip("wait for modification")
|
||||
@pytest.mark.parametrize("index, params",
|
||||
zip(ct.all_index_types[:7],
|
||||
ct.default_index_params[:7]))
|
||||
ct.default_all_indexes_params[:7]))
|
||||
def test_milvus_client_index_drop_create_different_index(self, index, params, metric_type):
|
||||
"""
|
||||
target: test search (high level api) normal case
|
||||
|
|
|
|||
|
|
@ -189,12 +189,12 @@ class TestRestartBase:
|
|||
# # logging.getLogger().info(file)
|
||||
# if file["field"] == field_name and file["name"] != "_raw":
|
||||
# assert file["data_size"] > 0
|
||||
# if file["index_type"] != default_index["index_type"]:
|
||||
# if file["index_type"] != default_ivf_flat_index["index_type"]:
|
||||
# continue
|
||||
# for file in stats["partitions"][0]["segments"][0]["files"]:
|
||||
# if file["field"] == field_name and file["name"] != "_raw":
|
||||
# assert file["data_size"] > 0
|
||||
# if file["index_type"] != default_index["index_type"]:
|
||||
# if file["index_type"] != default_ivf_flat_index["index_type"]:
|
||||
# assert False
|
||||
# else:
|
||||
# assert True
|
||||
|
|
|
|||
|
|
@ -19,7 +19,7 @@ from pymilvus.exceptions import MilvusException
|
|||
prefix = "index"
|
||||
default_schema = cf.gen_default_collection_schema()
|
||||
default_field_name = ct.default_float_vec_field_name
|
||||
default_index_params = {"index_type": "IVF_SQ8", "metric_type": "COSINE", "params": {"nlist": 64}}
|
||||
default_index_params = ct.default_index
|
||||
default_autoindex_params = {"index_type": "AUTOINDEX", "metric_type": "IP"}
|
||||
|
||||
# copied from pymilvus
|
||||
|
|
@ -33,9 +33,9 @@ index_name2 = cf.gen_unique_str("varhar")
|
|||
index_name3 = cf.gen_unique_str("binary")
|
||||
default_string_index_params = {}
|
||||
default_binary_schema = cf.gen_default_binary_collection_schema()
|
||||
default_binary_index_params = {"index_type": "BIN_IVF_FLAT", "metric_type": "JACCARD", "params": {"nlist": 64}}
|
||||
default_binary_index_params = ct.default_binary_index
|
||||
# query = gen_search_vectors_params(field_name, default_entities, default_top_k, 1)
|
||||
default_index = {"index_type": "IVF_FLAT", "params": {"nlist": 128}, "metric_type": "L2"}
|
||||
default_ivf_flat_index = {"index_type": "IVF_FLAT", "params": {"nlist": 128}, "metric_type": "L2"}
|
||||
default_ip_index_params = {"index_type": "IVF_FLAT", "metric_type": "IP", "params": {"nlist": 64}}
|
||||
default_nq = ct.default_nq
|
||||
default_limit = ct.default_limit
|
||||
|
|
@ -215,7 +215,7 @@ class TestIndexOperation(TestcaseBase):
|
|||
self.index_wrap.init_index(collection_w.collection, default_field_name, default_index_params)
|
||||
error = {ct.err_code: 65535, ct.err_msg: "CreateIndex failed: at most one "
|
||||
"distinct index is allowed per field"}
|
||||
self.index_wrap.init_index(collection_w.collection, default_field_name, default_index,
|
||||
self.index_wrap.init_index(collection_w.collection, default_field_name, default_ivf_flat_index,
|
||||
check_task=CheckTasks.err_res, check_items=error)
|
||||
|
||||
assert len(collection_w.indexes) == 1
|
||||
|
|
@ -611,7 +611,7 @@ class TestNewIndexBase(TestcaseBase):
|
|||
self.connection_wrap.remove_connection(ct.default_alias)
|
||||
res_list, _ = self.connection_wrap.list_connections()
|
||||
assert ct.default_alias not in res_list
|
||||
collection_w.create_index(ct.default_float_vec_field_name, ct.default_index_params,
|
||||
collection_w.create_index(ct.default_float_vec_field_name, ct.default_all_indexes_params,
|
||||
check_task=CheckTasks.err_res,
|
||||
check_items={ct.err_code: 1, ct.err_msg: "should create connect first"})
|
||||
|
||||
|
|
@ -716,7 +716,7 @@ class TestNewIndexBase(TestcaseBase):
|
|||
collection_w = self.init_collection_wrap(name=c_name)
|
||||
data = cf.gen_default_list_data()
|
||||
collection_w.insert(data=data)
|
||||
index_prams = [default_index, {"metric_type": "L2", "index_type": "IVF_SQ8", "params": {"nlist": 1024}}]
|
||||
index_prams = [default_ivf_flat_index, {"metric_type": "L2", "index_type": "IVF_SQ8", "params": {"nlist": 1024}}]
|
||||
for index in index_prams:
|
||||
index_name = cf.gen_unique_str("name")
|
||||
collection_w.create_index(default_float_vec_field_name, index, index_name=index_name)
|
||||
|
|
@ -1122,7 +1122,7 @@ class TestNewIndexBase(TestcaseBase):
|
|||
"limit": default_limit})
|
||||
|
||||
@pytest.mark.tags(CaseLabel.L2)
|
||||
@pytest.mark.parametrize("index, params", zip(ct.all_index_types[:6], ct.default_index_params[:6]))
|
||||
@pytest.mark.parametrize("index, params", zip(ct.all_index_types[:6], ct.default_all_indexes_params[:6]))
|
||||
def test_drop_mmap_index(self, index, params):
|
||||
"""
|
||||
target: disabling and re-enabling mmap for index
|
||||
|
|
@ -1384,7 +1384,7 @@ class TestIndexInvalid(TestcaseBase):
|
|||
"""
|
||||
collection_name = get_collection_name
|
||||
with pytest.raises(Exception) as e:
|
||||
connect.create_index(collection_name, field_name, default_index)
|
||||
connect.create_index(collection_name, field_name, default_ivf_flat_index)
|
||||
|
||||
@pytest.mark.tags(CaseLabel.L2)
|
||||
def test_drop_index_with_invalid_collection_name(self, connect, get_collection_name):
|
||||
|
|
|
|||
|
|
@ -25,8 +25,7 @@ default_schema = cf.gen_default_collection_schema()
|
|||
default_binary_schema = cf.gen_default_binary_collection_schema()
|
||||
default_index_params = {"index_type": "IVF_SQ8",
|
||||
"metric_type": "L2", "params": {"nlist": 64}}
|
||||
default_binary_index_params = {
|
||||
"index_type": "BIN_IVF_FLAT", "metric_type": "JACCARD", "params": {"nlist": 64}}
|
||||
default_binary_index_params = ct.default_binary_index
|
||||
default_search_exp = "int64 >= 0"
|
||||
|
||||
|
||||
|
|
|
|||
|
|
@ -27,8 +27,8 @@ default_mix_expr = "int64 >= 0 && varchar >= \"0\""
|
|||
default_expr = f'{ct.default_int64_field_name} >= 0'
|
||||
default_invalid_expr = "varchar >= 0"
|
||||
default_string_term_expr = f'{ct.default_string_field_name} in [\"0\", \"1\"]'
|
||||
default_index_params = {"index_type": "IVF_SQ8", "metric_type": "L2", "params": {"nlist": 64}}
|
||||
binary_index_params = {"index_type": "BIN_IVF_FLAT", "metric_type": "JACCARD", "params": {"nlist": 64}}
|
||||
default_index_params = ct.default_index
|
||||
binary_index_params = ct.default_binary_index
|
||||
|
||||
default_entities = ut.gen_entities(ut.default_nb, is_normal=True)
|
||||
default_pos = 5
|
||||
|
|
|
|||
File diff suppressed because it is too large
Load Diff
|
|
@ -62,18 +62,6 @@ def binary_support():
|
|||
return ["BIN_FLAT", "BIN_IVF_FLAT"]
|
||||
|
||||
|
||||
def delete_support():
|
||||
return ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ"]
|
||||
|
||||
|
||||
def ivf():
|
||||
return ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ"]
|
||||
|
||||
|
||||
def skip_pq():
|
||||
return ["IVF_PQ"]
|
||||
|
||||
|
||||
def binary_metrics():
|
||||
return ["JACCARD", "HAMMING", "SUBSTRUCTURE", "SUPERSTRUCTURE"]
|
||||
|
||||
|
|
@ -721,30 +709,6 @@ def gen_invalid_vectors():
|
|||
return invalid_vectors
|
||||
|
||||
|
||||
def gen_invaild_search_params():
|
||||
invalid_search_key = 100
|
||||
search_params = []
|
||||
for index_type in all_index_types:
|
||||
if index_type == "FLAT":
|
||||
continue
|
||||
search_params.append({"index_type": index_type, "search_params": {"invalid_key": invalid_search_key}})
|
||||
if index_type in delete_support():
|
||||
for nprobe in gen_invalid_params():
|
||||
ivf_search_params = {"index_type": index_type, "search_params": {"nprobe": nprobe}}
|
||||
search_params.append(ivf_search_params)
|
||||
elif index_type in ["HNSW"]:
|
||||
for ef in gen_invalid_params():
|
||||
hnsw_search_param = {"index_type": index_type, "search_params": {"ef": ef}}
|
||||
search_params.append(hnsw_search_param)
|
||||
elif index_type == "ANNOY":
|
||||
for search_k in gen_invalid_params():
|
||||
if isinstance(search_k, int):
|
||||
continue
|
||||
annoy_search_param = {"index_type": index_type, "search_params": {"search_k": search_k}}
|
||||
search_params.append(annoy_search_param)
|
||||
return search_params
|
||||
|
||||
|
||||
def gen_invalid_index():
|
||||
index_params = []
|
||||
for index_type in gen_invalid_strs():
|
||||
|
|
@ -825,23 +789,6 @@ def gen_normal_expressions():
|
|||
return expressions
|
||||
|
||||
|
||||
def get_search_param(index_type, metric_type="L2"):
|
||||
search_params = {"metric_type": metric_type}
|
||||
if index_type in ivf() or index_type in binary_support():
|
||||
nprobe64 = {"nprobe": 64}
|
||||
search_params.update({"params": nprobe64})
|
||||
elif index_type in ["HNSW"]:
|
||||
ef64 = {"ef": 64}
|
||||
search_params.update({"params": ef64})
|
||||
elif index_type == "ANNOY":
|
||||
search_k = {"search_k": 1000}
|
||||
search_params.update({"params": search_k})
|
||||
else:
|
||||
log.error("Invalid index_type.")
|
||||
raise Exception("Invalid index_type.")
|
||||
return search_params
|
||||
|
||||
|
||||
def assert_equal_vector(v1, v2):
|
||||
if len(v1) != len(v2):
|
||||
assert False
|
||||
|
|
|
|||
Loading…
Reference in New Issue