mirror of
https://gitee.com/milvus-io/milvus.git
synced 2025-12-07 01:28:27 +08:00
7aed88113c
issue: #44320 This change adds deduplication logic to handle duplicate primary keys within a single upsert batch, keeping the last occurrence of each primary key. Key changes: - Add DeduplicateFieldData function to remove duplicate PKs from field data, supporting both Int64 and VarChar primary keys - Refactor fillFieldPropertiesBySchema into two separate functions: validateFieldDataColumns for validation and fillFieldPropertiesOnly for property filling, improving code clarity and reusability - Integrate deduplication logic in upsertTask.PreExecute to automatically deduplicate data before processing - Add comprehensive unit tests for deduplication with various PK types (Int64, VarChar) and field types (scalar, vector) - Add Python integration tests to verify end-to-end behavior --------- Signed-off-by: Wei Liu <wei.liu@zilliz.com>
892 lines
44 KiB
Python
892 lines
44 KiB
Python
import pytest
|
|
import numpy as np
|
|
|
|
from base.client_v2_base import TestMilvusClientV2Base
|
|
from utils.util_log import test_log as log
|
|
from common import common_func as cf
|
|
from common import common_type as ct
|
|
from common.common_type import CaseLabel, CheckTasks
|
|
from utils.util_pymilvus import *
|
|
|
|
prefix = "client_insert"
|
|
epsilon = ct.epsilon
|
|
default_nb = ct.default_nb
|
|
default_nb_medium = ct.default_nb_medium
|
|
default_nq = ct.default_nq
|
|
default_dim = ct.default_dim
|
|
default_limit = ct.default_limit
|
|
default_search_exp = "id >= 0"
|
|
exp_res = "exp_res"
|
|
default_search_string_exp = "varchar >= \"0\""
|
|
default_search_mix_exp = "int64 >= 0 && varchar >= \"0\""
|
|
default_invaild_string_exp = "varchar >= 0"
|
|
default_json_search_exp = "json_field[\"number\"] >= 0"
|
|
perfix_expr = 'varchar like "0%"'
|
|
default_search_field = ct.default_float_vec_field_name
|
|
default_search_params = ct.default_search_params
|
|
default_primary_key_field_name = "id"
|
|
default_vector_field_name = "vector"
|
|
default_dynamic_field_name = "field_new"
|
|
default_float_field_name = ct.default_float_field_name
|
|
default_bool_field_name = ct.default_bool_field_name
|
|
default_string_field_name = ct.default_string_field_name
|
|
default_int32_array_field_name = ct.default_int32_array_field_name
|
|
default_string_array_field_name = ct.default_string_array_field_name
|
|
default_int32_field_name = ct.default_int32_field_name
|
|
default_int32_value = ct.default_int32_value
|
|
|
|
|
|
class TestMilvusClientUpsertInvalid(TestMilvusClientV2Base):
|
|
""" Test case of search interface """
|
|
|
|
@pytest.fixture(scope="function", params=[False, True])
|
|
def auto_id(self, request):
|
|
yield request.param
|
|
|
|
@pytest.fixture(scope="function", params=["COSINE", "L2"])
|
|
def metric_type(self, request):
|
|
yield request.param
|
|
|
|
"""
|
|
******************************************************************
|
|
# The following are invalid base cases
|
|
******************************************************************
|
|
"""
|
|
|
|
@pytest.mark.tags(CaseLabel.L2)
|
|
def test_milvus_client_upsert_column_data(self):
|
|
"""
|
|
target: test insert column data
|
|
method: create connection, collection, insert and search
|
|
expected: raise error
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim)
|
|
# 2. insert
|
|
vectors = [[random.random() for _ in range(default_dim)] for _ in range(default_nb)]
|
|
data = [[i for i in range(default_nb)], vectors]
|
|
error = {ct.err_code: 999,
|
|
ct.err_msg: "The Input data type is inconsistent with defined schema, please check it."}
|
|
self.upsert(client, collection_name, data,
|
|
check_task=CheckTasks.err_res, check_items=error)
|
|
self.drop_collection(client, collection_name)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
def test_milvus_client_upsert_empty_collection_name(self):
|
|
"""
|
|
target: test high level api: client.create_collection
|
|
method: create collection with invalid primary field
|
|
expected: Raise exception
|
|
"""
|
|
client = self._client()
|
|
collection_name = ""
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [{default_primary_key_field_name: i, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: i * 1.0, default_string_field_name: str(i)} for i in range(default_nb)]
|
|
error = {ct.err_code: 1, ct.err_msg: f"`collection_name` value {collection_name} is illegal"}
|
|
self.upsert(client, collection_name, rows,
|
|
check_task=CheckTasks.err_res, check_items=error)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
@pytest.mark.parametrize("collection_name", ["12-s", "12 s", "(mn)", "中文", "%$#"])
|
|
def test_milvus_client_upsert_invalid_collection_name(self, collection_name):
|
|
"""
|
|
target: test high level api: client.create_collection
|
|
method: create collection with invalid primary field
|
|
expected: Raise exception
|
|
"""
|
|
client = self._client()
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [{default_primary_key_field_name: i, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: i * 1.0, default_string_field_name: str(i)} for i in range(default_nb)]
|
|
error = {ct.err_code: 1100, ct.err_msg: f"Invalid collection name: {collection_name}. the first character of a "
|
|
f"collection name must be an underscore or letter: invalid parameter"}
|
|
self.upsert(client, collection_name, rows,
|
|
check_task=CheckTasks.err_res, check_items=error)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
def test_milvus_client_upsert_collection_name_over_max_length(self):
|
|
"""
|
|
target: test high level api: client.create_collection
|
|
method: create collection with invalid primary field
|
|
expected: Raise exception
|
|
"""
|
|
client = self._client()
|
|
collection_name = "a".join("a" for i in range(256))
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [{default_primary_key_field_name: i, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: i * 1.0, default_string_field_name: str(i)} for i in range(default_nb)]
|
|
error = {ct.err_code: 1100, ct.err_msg: f"the length of a collection name must be less than 255 characters"}
|
|
self.upsert(client, collection_name, rows,
|
|
check_task=CheckTasks.err_res, check_items=error)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
def test_milvus_client_upsert_not_exist_collection_name(self):
|
|
"""
|
|
target: test high level api: client.create_collection
|
|
method: create collection with invalid primary field
|
|
expected: Raise exception
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_unique_str("insert_not_exist")
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [{default_primary_key_field_name: i, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: i * 1.0, default_string_field_name: str(i)} for i in range(default_nb)]
|
|
error = {ct.err_code: 100, ct.err_msg: f"can't find collection[database=default][collection={collection_name}]"}
|
|
self.upsert(client, collection_name, rows,
|
|
check_task=CheckTasks.err_res, check_items=error)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
@pytest.mark.parametrize("data", ["12-s", "12 s", "(mn)", "中文", "%$#", " "])
|
|
def test_milvus_client_upsert_data_invalid_type(self, data):
|
|
"""
|
|
target: test high level api: client.create_collection
|
|
method: create collection with invalid primary field
|
|
expected: Raise exception
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim, consistency_level="Strong")
|
|
# 2. insert
|
|
error = {ct.err_code: 1, ct.err_msg: f"wrong type of argument 'data',expected 'Dict' or list of 'Dict'"}
|
|
self.upsert(client, collection_name, data,
|
|
check_task=CheckTasks.err_res, check_items=error)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
def test_milvus_client_upsert_data_empty(self):
|
|
"""
|
|
target: test high level api: client.create_collection
|
|
method: create collection with invalid primary field
|
|
expected: Raise exception
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim, consistency_level="Strong")
|
|
# 2. insert
|
|
error = {ct.err_code: 1, ct.err_msg: f"wrong type of argument 'data',expected 'Dict' or list of 'Dict'"}
|
|
self.upsert(client, collection_name, data="",
|
|
check_task=CheckTasks.err_res, check_items=error)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
def test_milvus_client_upsert_data_vector_field_missing(self):
|
|
"""
|
|
target: test high level api: client.create_collection
|
|
method: create collection with invalid primary field
|
|
expected: Raise exception
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim, consistency_level="Strong")
|
|
# 2. insert
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [{default_primary_key_field_name: i,
|
|
default_float_field_name: i * 1.0, default_string_field_name: str(i)} for i in range(10)]
|
|
error = {ct.err_code: 1,
|
|
ct.err_msg: "Insert missed an field `vector` to collection without set nullable==true or set default_value"}
|
|
self.upsert(client, collection_name, data=rows,
|
|
check_task=CheckTasks.err_res, check_items=error)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
def test_milvus_client_upsert_data_id_field_missing(self):
|
|
"""
|
|
target: test high level api: client.create_collection
|
|
method: create collection with invalid primary field
|
|
expected: Raise exception
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim, consistency_level="Strong")
|
|
# 2. insert
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [{default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: i * 1.0, default_string_field_name: str(i)} for i in range(20)]
|
|
error = {ct.err_code: 1,
|
|
ct.err_msg: f"Insert missed an field `id` to collection without set nullable==true or set default_value"}
|
|
self.upsert(client, collection_name, data=rows,
|
|
check_task=CheckTasks.err_res, check_items=error)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
def test_milvus_client_upsert_data_extra_field(self):
|
|
"""
|
|
target: test milvus client: insert extra field than schema
|
|
method: insert extra field than schema when enable_dynamic_field is False
|
|
expected: Raise exception
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
# 1. create collection
|
|
dim = 32
|
|
self.create_collection(client, collection_name, dim, enable_dynamic_field=False)
|
|
# 2. insert
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [{default_primary_key_field_name: i, default_vector_field_name: list(rng.random((1, dim))[0]),
|
|
default_float_field_name: i * 1.0, default_string_field_name: str(i)} for i in range(10)]
|
|
error = {ct.err_code: 1,
|
|
ct.err_msg: f"Attempt to insert an unexpected field `float` to collection without enabling dynamic field"}
|
|
self.upsert(client, collection_name, data=rows,
|
|
check_task=CheckTasks.err_res, check_items=error)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
def test_milvus_client_upsert_data_dim_not_match(self):
|
|
"""
|
|
target: test milvus client: insert extra field than schema
|
|
method: insert extra field than schema when enable_dynamic_field is False
|
|
expected: Raise exception
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim)
|
|
# 2. insert
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [
|
|
{default_primary_key_field_name: i, default_vector_field_name: list(rng.random((1, default_dim + 1))[0]),
|
|
default_float_field_name: i * 1.0, default_string_field_name: str(i)} for i in range(default_nb)]
|
|
error = {ct.err_code: 65536, ct.err_msg: f"of float data should divide the dim({default_dim})"}
|
|
self.upsert(client, collection_name, data=rows,
|
|
check_task=CheckTasks.err_res, check_items=error)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
def test_milvus_client_upsert_not_matched_data(self):
|
|
"""
|
|
target: test milvus client: insert not matched data then defined
|
|
method: insert string to int primary field
|
|
expected: Raise exception
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim)
|
|
# 2. insert
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [
|
|
{default_primary_key_field_name: str(i), default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: i * 1.0, default_string_field_name: str(i)} for i in range(default_nb)]
|
|
error = {ct.err_code: 1,
|
|
ct.err_msg: "The Input data type is inconsistent with defined schema, {id} field should be a int64"}
|
|
self.upsert(client, collection_name, data=rows,
|
|
check_task=CheckTasks.err_res, check_items=error)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
@pytest.mark.parametrize("partition_name", ["12 s", "(mn)", "中文", "%$#", " "])
|
|
def test_milvus_client_upsert_invalid_partition_name(self, partition_name):
|
|
"""
|
|
target: test milvus client: insert extra field than schema
|
|
method: insert extra field than schema when enable_dynamic_field is False
|
|
expected: Raise exception
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim)
|
|
# 2. insert
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [{default_primary_key_field_name: i, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: i * 1.0, default_string_field_name: str(i)} for i in range(default_nb)]
|
|
error = {ct.err_code: 65535, ct.err_msg: f"Invalid partition name: {partition_name}"}
|
|
if partition_name == " ":
|
|
error = {ct.err_code: 1, ct.err_msg: f"Invalid partition name: . Partition name should not be empty."}
|
|
self.upsert(client, collection_name, data=rows, partition_name=partition_name,
|
|
check_task=CheckTasks.err_res, check_items=error)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
def test_milvus_client_upsert_not_exist_partition_name(self):
|
|
"""
|
|
target: test milvus client: insert extra field than schema
|
|
method: insert extra field than schema when enable_dynamic_field is False
|
|
expected: Raise exception
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim)
|
|
# 2. insert
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [{default_primary_key_field_name: i, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: i * 1.0, default_string_field_name: str(i)} for i in range(default_nb)]
|
|
partition_name = cf.gen_unique_str("partition_not_exist")
|
|
error = {ct.err_code: 200, ct.err_msg: f"partition not found[partition={partition_name}]"}
|
|
self.upsert(client, collection_name, data=rows, partition_name=partition_name,
|
|
check_task=CheckTasks.err_res, check_items=error)
|
|
|
|
@pytest.mark.tags(CaseLabel.L2)
|
|
def test_milvus_client_upsert_collection_partition_not_match(self):
|
|
"""
|
|
target: test milvus client: insert extra field than schema
|
|
method: insert extra field than schema when enable_dynamic_field is False
|
|
expected: Raise exception
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
another_collection_name = cf.gen_unique_str(prefix + "another")
|
|
partition_name = cf.gen_unique_str("partition")
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim)
|
|
self.create_collection(client, another_collection_name, default_dim)
|
|
self.create_partition(client, another_collection_name, partition_name)
|
|
# 2. insert
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [{default_primary_key_field_name: i, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: i * 1.0, default_string_field_name: str(i)} for i in range(default_nb)]
|
|
error = {ct.err_code: 200, ct.err_msg: f"partition not found[partition={partition_name}]"}
|
|
self.upsert(client, collection_name, data=rows, partition_name=partition_name,
|
|
check_task=CheckTasks.err_res, check_items=error)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
@pytest.mark.parametrize("nullable", [True, False])
|
|
def test_milvus_client_insert_array_element_null(self, nullable):
|
|
"""
|
|
target: test search with null expression on each key of json
|
|
method: create connection, collection, insert and search
|
|
expected: raise exception
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
dim = 5
|
|
# 1. create collection
|
|
nullable_field_name = "nullable_field"
|
|
schema = self.create_schema(client, enable_dynamic_field=False)[0]
|
|
schema.add_field(default_primary_key_field_name, DataType.VARCHAR, max_length=64, is_primary=True,
|
|
auto_id=False)
|
|
schema.add_field(default_vector_field_name, DataType.FLOAT_VECTOR, dim=dim)
|
|
schema.add_field(nullable_field_name, DataType.ARRAY, element_type=DataType.INT64, max_capacity=12,
|
|
max_length=64, nullable=nullable)
|
|
index_params = self.prepare_index_params(client)[0]
|
|
index_params.add_index(default_vector_field_name, metric_type="COSINE")
|
|
self.create_collection(client, collection_name, dimension=dim, schema=schema, index_params=index_params)
|
|
# 2. insert
|
|
vectors = cf.gen_vectors(default_nb, dim)
|
|
rows = [{default_primary_key_field_name: str(i), default_vector_field_name: vectors[i],
|
|
nullable_field_name: [None, 2, 3]} for i in range(default_nb)]
|
|
error = {ct.err_code: 1,
|
|
ct.err_msg: "The Input data type is inconsistent with defined schema, {nullable_field} field "
|
|
"should be a array, but got a {<class 'list'>} instead."}
|
|
self.insert(client, collection_name, rows,
|
|
check_task=CheckTasks.err_res,
|
|
check_items=error)
|
|
|
|
|
|
class TestMilvusClientUpsertValid(TestMilvusClientV2Base):
|
|
""" Test case of search interface """
|
|
|
|
@pytest.fixture(scope="function", params=[False, True])
|
|
def auto_id(self, request):
|
|
yield request.param
|
|
|
|
@pytest.fixture(scope="function", params=["COSINE", "L2"])
|
|
def metric_type(self, request):
|
|
yield request.param
|
|
|
|
"""
|
|
******************************************************************
|
|
# The following are valid base cases
|
|
******************************************************************
|
|
"""
|
|
|
|
@pytest.mark.tags(CaseLabel.L0)
|
|
def test_milvus_client_upsert_default(self):
|
|
"""
|
|
target: test search (high level api) normal case
|
|
method: create connection, collection, insert and search
|
|
expected: search/query successfully
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim, consistency_level="Strong")
|
|
collections = self.list_collections(client)[0]
|
|
assert collection_name in collections
|
|
self.describe_collection(client, collection_name,
|
|
check_task=CheckTasks.check_describe_collection_property,
|
|
check_items={"collection_name": collection_name,
|
|
"dim": default_dim,
|
|
"consistency_level": 0})
|
|
# 2. insert
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [{default_primary_key_field_name: i, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: i * 1.0, default_string_field_name: str(i)} for i in range(default_nb)]
|
|
results = self.upsert(client, collection_name, rows)[0]
|
|
assert results['upsert_count'] == default_nb
|
|
# 3. search
|
|
vectors_to_search = rng.random((1, default_dim))
|
|
insert_ids = [i for i in range(default_nb)]
|
|
self.search(client, collection_name, vectors_to_search,
|
|
check_task=CheckTasks.check_search_results,
|
|
check_items={"enable_milvus_client_api": True,
|
|
"nq": len(vectors_to_search),
|
|
"ids": insert_ids,
|
|
"limit": default_limit,
|
|
"pk_name": default_primary_key_field_name})
|
|
# 4. query
|
|
self.query(client, collection_name, filter=default_search_exp,
|
|
check_task=CheckTasks.check_query_results,
|
|
check_items={exp_res: rows,
|
|
"with_vec": True,
|
|
"pk_name": default_primary_key_field_name})
|
|
self.release_collection(client, collection_name)
|
|
self.drop_collection(client, collection_name)
|
|
|
|
@pytest.mark.tags(CaseLabel.L2)
|
|
def test_milvus_client_upsert_empty_data(self):
|
|
"""
|
|
target: test search (high level api) normal case
|
|
method: create connection, collection, insert and search
|
|
expected: search/query successfully
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim, consistency_level="Strong")
|
|
# 2. insert
|
|
rows = []
|
|
results = self.upsert(client, collection_name, rows)[0]
|
|
assert results['upsert_count'] == 0
|
|
# 3. search
|
|
rng = np.random.default_rng(seed=19530)
|
|
vectors_to_search = rng.random((1, default_dim))
|
|
self.search(client, collection_name, vectors_to_search,
|
|
check_task=CheckTasks.check_search_results,
|
|
check_items={"enable_milvus_client_api": True,
|
|
"nq": len(vectors_to_search),
|
|
"ids": [],
|
|
"pk_name": default_primary_key_field_name,
|
|
"limit": 0})
|
|
self.drop_collection(client, collection_name)
|
|
|
|
@pytest.mark.tags(CaseLabel.L2)
|
|
def test_milvus_client_upsert_partition(self):
|
|
"""
|
|
target: test fast create collection normal case
|
|
method: create collection
|
|
expected: create collection with default schema, index, and load successfully
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
partition_name = cf.gen_unique_str(prefix)
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim, consistency_level="Strong")
|
|
# 2. create partition
|
|
self.create_partition(client, collection_name, partition_name)
|
|
partitions = self.list_partitions(client, collection_name)[0]
|
|
assert partition_name in partitions
|
|
index = self.list_indexes(client, collection_name)[0]
|
|
assert index == ['vector']
|
|
# load_state = self.get_load_state(collection_name)[0]
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [{default_primary_key_field_name: i, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: i * 1.0, default_string_field_name: str(i)} for i in range(default_nb)]
|
|
# 3. upsert to default partition
|
|
results = self.upsert(client, collection_name, rows, partition_name=partitions[0])[0]
|
|
assert results['upsert_count'] == default_nb
|
|
# 4. upsert to non-default partition
|
|
results = self.upsert(client, collection_name, rows, partition_name=partition_name)[0]
|
|
assert results['upsert_count'] == default_nb
|
|
# 5. search
|
|
vectors_to_search = rng.random((1, default_dim))
|
|
insert_ids = [i for i in range(default_nb)]
|
|
self.search(client, collection_name, vectors_to_search,
|
|
check_task=CheckTasks.check_search_results,
|
|
check_items={"enable_milvus_client_api": True,
|
|
"nq": len(vectors_to_search),
|
|
"ids": insert_ids,
|
|
"limit": default_limit,
|
|
"pk_name": default_primary_key_field_name})
|
|
# partition_number = self.get_partition_stats(client, collection_name, "_default")[0]
|
|
# assert partition_number == default_nb
|
|
# partition_number = self.get_partition_stats(client, collection_name, partition_name)[0]
|
|
# assert partition_number[0]['value'] == 0
|
|
if self.has_partition(client, collection_name, partition_name)[0]:
|
|
self.release_partitions(client, collection_name, partition_name)
|
|
self.drop_partition(client, collection_name, partition_name)
|
|
if self.has_collection(client, collection_name)[0]:
|
|
self.drop_collection(client, collection_name)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
def test_milvus_client_insert_upsert(self):
|
|
"""
|
|
target: test fast create collection normal case
|
|
method: create collection
|
|
expected: create collection with default schema, index, and load successfully
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
partition_name = cf.gen_unique_str(prefix)
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim, consistency_level="Strong")
|
|
# 2. create partition
|
|
self.create_partition(client, collection_name, partition_name)
|
|
partitions = self.list_partitions(client, collection_name)[0]
|
|
assert partition_name in partitions
|
|
index = self.list_indexes(client, collection_name)[0]
|
|
assert index == ['vector']
|
|
# load_state = self.get_load_state(collection_name)[0]
|
|
# 3. insert and upsert
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [{default_primary_key_field_name: i, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: i * 1.0, default_string_field_name: str(i)} for i in range(default_nb)]
|
|
results = self.insert(client, collection_name, rows, partition_name=partition_name)[0]
|
|
assert results['insert_count'] == default_nb
|
|
rows = [{default_primary_key_field_name: i, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: i * 1.0, "new_diff_str_field": str(i)} for i in range(default_nb)]
|
|
results = self.upsert(client, collection_name, rows, partition_name=partition_name)[0]
|
|
assert results['upsert_count'] == default_nb
|
|
# 3. search
|
|
vectors_to_search = rng.random((1, default_dim))
|
|
insert_ids = [i for i in range(default_nb)]
|
|
self.search(client, collection_name, vectors_to_search,
|
|
check_task=CheckTasks.check_search_results,
|
|
check_items={"enable_milvus_client_api": True,
|
|
"nq": len(vectors_to_search),
|
|
"ids": insert_ids,
|
|
"limit": default_limit,
|
|
"pk_name": default_primary_key_field_name})
|
|
if self.has_partition(client, collection_name, partition_name)[0]:
|
|
self.release_partitions(client, collection_name, partition_name)
|
|
self.drop_partition(client, collection_name, partition_name)
|
|
if self.has_collection(client, collection_name)[0]:
|
|
self.drop_collection(client, collection_name)
|
|
|
|
|
|
class TestMilvusClientUpsertDedup(TestMilvusClientV2Base):
|
|
"""Test case for upsert deduplication functionality"""
|
|
|
|
@pytest.fixture(scope="function", params=["COSINE", "L2"])
|
|
def metric_type(self, request):
|
|
yield request.param
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
def test_milvus_client_upsert_dedup_int64_pk(self):
|
|
"""
|
|
target: test upsert with duplicate int64 primary keys in same batch
|
|
method:
|
|
1. create collection with int64 primary key
|
|
2. upsert data with duplicate primary keys [1, 2, 3, 2, 1]
|
|
3. query to verify only last occurrence is kept
|
|
expected: only 3 unique records exist, with data from last occurrence
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim, consistency_level="Strong")
|
|
|
|
# 2. upsert data with duplicate PKs: [1, 2, 3, 2, 1]
|
|
# Expected: keep last occurrence -> [3, 2, 1] at indices [2, 3, 4]
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [
|
|
{default_primary_key_field_name: 1, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: 1.0, default_string_field_name: "str_1_first"},
|
|
{default_primary_key_field_name: 2, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: 2.0, default_string_field_name: "str_2_first"},
|
|
{default_primary_key_field_name: 3, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: 3.0, default_string_field_name: "str_3"},
|
|
{default_primary_key_field_name: 2, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: 2.1, default_string_field_name: "str_2_last"},
|
|
{default_primary_key_field_name: 1, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: 1.1, default_string_field_name: "str_1_last"},
|
|
]
|
|
|
|
results = self.upsert(client, collection_name, rows)[0]
|
|
# After deduplication, should only have 3 records
|
|
assert results['upsert_count'] == 3
|
|
|
|
# 3. query to verify deduplication - should have only 3 unique records
|
|
query_results = self.query(client, collection_name, filter="id >= 0")[0]
|
|
assert len(query_results) == 3
|
|
|
|
# Verify that last occurrence data is kept
|
|
id_to_data = {item['id']: item for item in query_results}
|
|
assert 1 in id_to_data
|
|
assert 2 in id_to_data
|
|
assert 3 in id_to_data
|
|
|
|
# Check that data from last occurrence is preserved
|
|
assert id_to_data[1]['float'] == 1.1
|
|
assert id_to_data[1]['varchar'] == "str_1_last"
|
|
assert id_to_data[2]['float'] == 2.1
|
|
assert id_to_data[2]['varchar'] == "str_2_last"
|
|
assert id_to_data[3]['float'] == 3.0
|
|
assert id_to_data[3]['varchar'] == "str_3"
|
|
|
|
self.drop_collection(client, collection_name)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
def test_milvus_client_upsert_dedup_varchar_pk(self):
|
|
"""
|
|
target: test upsert with duplicate varchar primary keys in same batch
|
|
method:
|
|
1. create collection with varchar primary key
|
|
2. upsert data with duplicate primary keys ["a", "b", "c", "b", "a"]
|
|
3. query to verify only last occurrence is kept
|
|
expected: only 3 unique records exist, with data from last occurrence
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
|
|
# 1. create collection with varchar primary key
|
|
schema = self.create_schema(client, enable_dynamic_field=True)[0]
|
|
schema.add_field("id", DataType.VARCHAR, max_length=64, is_primary=True, auto_id=False)
|
|
schema.add_field(default_vector_field_name, DataType.FLOAT_VECTOR, dim=default_dim)
|
|
schema.add_field("age", DataType.INT64)
|
|
index_params = self.prepare_index_params(client)[0]
|
|
index_params.add_index(default_vector_field_name, metric_type="COSINE")
|
|
self.create_collection(client, collection_name, default_dim, schema=schema,
|
|
index_params=index_params, consistency_level="Strong")
|
|
|
|
# 2. upsert data with duplicate PKs: ["a", "b", "c", "b", "a"]
|
|
# Expected: keep last occurrence -> ["c", "b", "a"] at indices [2, 3, 4]
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [
|
|
{"id": "a", default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
"age": 10},
|
|
{"id": "b", default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
"age": 20},
|
|
{"id": "c", default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
"age": 30},
|
|
{"id": "b", default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
"age": 21},
|
|
{"id": "a", default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
"age": 11},
|
|
]
|
|
|
|
results = self.upsert(client, collection_name, rows)[0]
|
|
# After deduplication, should only have 3 records
|
|
assert results['upsert_count'] == 3
|
|
|
|
# 3. query to verify deduplication
|
|
query_results = self.query(client, collection_name, filter='id in ["a", "b", "c"]')[0]
|
|
assert len(query_results) == 3
|
|
|
|
# Verify that last occurrence data is kept
|
|
id_to_data = {item['id']: item for item in query_results}
|
|
assert "a" in id_to_data
|
|
assert "b" in id_to_data
|
|
assert "c" in id_to_data
|
|
|
|
# Check that data from last occurrence is preserved
|
|
assert id_to_data["a"]["age"] == 11
|
|
assert id_to_data["b"]["age"] == 21
|
|
assert id_to_data["c"]["age"] == 30
|
|
|
|
self.drop_collection(client, collection_name)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
def test_milvus_client_upsert_dedup_all_duplicates(self):
|
|
"""
|
|
target: test upsert when all records have same primary key
|
|
method:
|
|
1. create collection
|
|
2. upsert 5 records with same primary key
|
|
3. query to verify only 1 record exists
|
|
expected: only 1 record exists with data from last occurrence
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim, consistency_level="Strong")
|
|
|
|
# 2. upsert data where all have same PK (id=1)
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [
|
|
{default_primary_key_field_name: 1, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: i * 1.0, default_string_field_name: f"version_{i}"}
|
|
for i in range(5)
|
|
]
|
|
|
|
results = self.upsert(client, collection_name, rows)[0]
|
|
# After deduplication, should only have 1 record
|
|
assert results['upsert_count'] == 1
|
|
|
|
# 3. query to verify only 1 record exists
|
|
query_results = self.query(client, collection_name, filter="id == 1")[0]
|
|
assert len(query_results) == 1
|
|
|
|
# Verify it's the last occurrence (i=4)
|
|
assert query_results[0]['float'] == 4.0
|
|
assert query_results[0]['varchar'] == "version_4"
|
|
|
|
self.drop_collection(client, collection_name)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
def test_milvus_client_upsert_dedup_no_duplicates(self):
|
|
"""
|
|
target: test upsert with no duplicate primary keys
|
|
method:
|
|
1. create collection
|
|
2. upsert data with unique primary keys
|
|
3. query to verify all records exist
|
|
expected: all records exist as-is
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim, consistency_level="Strong")
|
|
|
|
# 2. upsert data with unique PKs
|
|
rng = np.random.default_rng(seed=19530)
|
|
nb = 10
|
|
rows = [
|
|
{default_primary_key_field_name: i, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: i * 1.0, default_string_field_name: str(i)}
|
|
for i in range(nb)
|
|
]
|
|
|
|
results = self.upsert(client, collection_name, rows)[0]
|
|
# No deduplication should occur
|
|
assert results['upsert_count'] == nb
|
|
|
|
# 3. query to verify all records exist
|
|
query_results = self.query(client, collection_name, filter=f"id >= 0")[0]
|
|
assert len(query_results) == nb
|
|
|
|
self.drop_collection(client, collection_name)
|
|
|
|
@pytest.mark.tags(CaseLabel.L2)
|
|
def test_milvus_client_upsert_dedup_large_batch(self):
|
|
"""
|
|
target: test upsert deduplication with large batch
|
|
method:
|
|
1. create collection
|
|
2. upsert large batch with 50% duplicate primary keys
|
|
3. query to verify correct number of records
|
|
expected: only unique records exist
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim, consistency_level="Strong")
|
|
|
|
# 2. upsert large batch where each ID appears twice
|
|
rng = np.random.default_rng(seed=19530)
|
|
nb = 500
|
|
unique_ids = nb // 2 # 250 unique IDs
|
|
|
|
rows = []
|
|
for i in range(nb):
|
|
pk = i % unique_ids # This creates duplicates: 0,1,2...249,0,1,2...249
|
|
rows.append({
|
|
default_primary_key_field_name: pk,
|
|
default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: float(i), # Different value for each row
|
|
default_string_field_name: f"batch_{i}"
|
|
})
|
|
|
|
results = self.upsert(client, collection_name, rows)[0]
|
|
# After deduplication, should only have unique_ids records
|
|
assert results['upsert_count'] == unique_ids
|
|
|
|
# 3. query to verify correct number of records
|
|
query_results = self.query(client, collection_name, filter=f"id >= 0", limit=1000)[0]
|
|
assert len(query_results) == unique_ids
|
|
|
|
# Verify that last occurrence is kept (should have higher float values)
|
|
for item in query_results:
|
|
pk = item['id']
|
|
# Last occurrence of pk is at index (pk + unique_ids)
|
|
expected_float = float(pk + unique_ids)
|
|
assert item['float'] == expected_float
|
|
assert item['varchar'] == f"batch_{pk + unique_ids}"
|
|
|
|
self.drop_collection(client, collection_name)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
def test_milvus_client_upsert_dedup_with_partition(self):
|
|
"""
|
|
target: test upsert deduplication works correctly with partitions
|
|
method:
|
|
1. create collection with partition
|
|
2. upsert data with duplicates to specific partition
|
|
3. query to verify deduplication in partition
|
|
expected: deduplication works within partition
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
partition_name = cf.gen_unique_str("partition")
|
|
|
|
# 1. create collection and partition
|
|
self.create_collection(client, collection_name, default_dim, consistency_level="Strong")
|
|
self.create_partition(client, collection_name, partition_name)
|
|
|
|
# 2. upsert data with duplicates to partition
|
|
rng = np.random.default_rng(seed=19530)
|
|
rows = [
|
|
{default_primary_key_field_name: 1, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: 1.0, default_string_field_name: "first"},
|
|
{default_primary_key_field_name: 2, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: 2.0, default_string_field_name: "unique"},
|
|
{default_primary_key_field_name: 1, default_vector_field_name: list(rng.random((1, default_dim))[0]),
|
|
default_float_field_name: 1.1, default_string_field_name: "last"},
|
|
]
|
|
|
|
results = self.upsert(client, collection_name, rows, partition_name=partition_name)[0]
|
|
assert results['upsert_count'] == 2
|
|
|
|
# 3. query partition to verify deduplication
|
|
query_results = self.query(client, collection_name, filter="id >= 0",
|
|
partition_names=[partition_name])[0]
|
|
assert len(query_results) == 2
|
|
|
|
# Verify correct data
|
|
id_to_data = {item['id']: item for item in query_results}
|
|
assert id_to_data[1]['float'] == 1.1
|
|
assert id_to_data[1]['varchar'] == "last"
|
|
assert id_to_data[2]['float'] == 2.0
|
|
assert id_to_data[2]['varchar'] == "unique"
|
|
|
|
self.drop_collection(client, collection_name)
|
|
|
|
@pytest.mark.tags(CaseLabel.L1)
|
|
def test_milvus_client_upsert_dedup_with_vectors(self):
|
|
"""
|
|
target: test upsert deduplication preserves correct vector data
|
|
method:
|
|
1. create collection
|
|
2. upsert data with duplicate PKs but different vectors
|
|
3. search to verify correct vector is preserved
|
|
expected: vector from last occurrence is preserved
|
|
"""
|
|
client = self._client()
|
|
collection_name = cf.gen_collection_name_by_testcase_name()
|
|
|
|
# 1. create collection
|
|
self.create_collection(client, collection_name, default_dim, consistency_level="Strong")
|
|
|
|
# 2. upsert data with duplicate PK=1 but different vectors
|
|
# Create distinctly different vectors for easy verification
|
|
first_vector = [1.0] * default_dim # All 1.0
|
|
last_vector = [2.0] * default_dim # All 2.0
|
|
|
|
rows = [
|
|
{default_primary_key_field_name: 1, default_vector_field_name: first_vector,
|
|
default_float_field_name: 1.0, default_string_field_name: "first"},
|
|
{default_primary_key_field_name: 2, default_vector_field_name: [0.5] * default_dim,
|
|
default_float_field_name: 2.0, default_string_field_name: "unique"},
|
|
{default_primary_key_field_name: 1, default_vector_field_name: last_vector,
|
|
default_float_field_name: 1.1, default_string_field_name: "last"},
|
|
]
|
|
|
|
results = self.upsert(client, collection_name, rows)[0]
|
|
assert results['upsert_count'] == 2
|
|
|
|
# 3. query to get vector data
|
|
query_results = self.query(client, collection_name, filter="id == 1",
|
|
output_fields=["id", "vector", "float", "varchar"])[0]
|
|
assert len(query_results) == 1
|
|
|
|
# Verify it's the last occurrence with last_vector
|
|
result = query_results[0]
|
|
assert result['float'] == 1.1
|
|
assert result['varchar'] == "last"
|
|
# Vector should be last_vector (all 2.0)
|
|
assert all(abs(v - 2.0) < 0.001 for v in result['vector'])
|
|
|
|
self.drop_collection(client, collection_name) |