admin/milvus
1
0
Fork 0
mirror of https://gitee.com/milvus-io/milvus.git synced 2025-12-06 09:08:43 +08:00
Code Issues Packages Projects Releases Wiki Activity

test: [cp2.6] cp hybrid search tests from master (#46018)

Browse Source 
related issue: https://github.com/milvus-io/milvus/issues/45326
pr: #46003

---------

Signed-off-by: yanliang567 <yanliang.qiao@zilliz.com>
This commit is contained in:
yanliang567 2025-12-05 15:47:12 +08:00 committed by GitHub
parent f412ab8892
commit 6ce05b7d45
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
6 changed files with 647 additions and 901 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

51
tests/python_client/check/func_check.py
View File

@ -436,14 +436,13 @@ class ResponseChecker:
original_entities = pandas.DataFrame(original_entities)
pc.output_field_value_check(search_res, original_entities, pk_name=pk_name)
if len(search_res) != check_items["nq"]:
log.error("search_results_check: Numbers of query searched (%d) "
log.error("search_results_check: Numbers of query searched(nq) (%d) "
"is not equal with expected (%d)"
% (len(search_res), check_items["nq"]))
assert len(search_res) == check_items["nq"]
else:
log.info("search_results_check: Numbers of query searched is correct")
# log.debug(search_res)
nq_i = 0
for hits in search_res:
ids = []
distances = []
@ -461,25 +460,24 @@ class ResponseChecker:
% (len(hits), check_items["limit"]))
assert len(hits) == check_items["limit"]
assert len(ids) == check_items["limit"]
else:
if check_items.get("ids", None) is not None:
ids_match = pc.list_contain_check(ids, list(check_items["ids"]))
if not ids_match:
log.error("search_results_check: ids searched not match")
assert ids_match
elif check_items.get("metric", None) is not None:
# verify the distances are already sorted
if check_items.get("metric").upper() in ["IP", "COSINE", "BM25"]:
assert pc.compare_lists_with_epsilon_ignore_dict_order(distances, sorted(distances, reverse=True))
else:
assert pc.compare_lists_with_epsilon_ignore_dict_order(distances, sorted(distances, reverse=False))
if check_items.get("vector_nq") is None or check_items.get("original_vectors") is None:
log.debug("skip distance check for knowhere does not return the precise distances")
else:
pass
if check_items.get("ids", None) is not None:
ids_match = pc.list_contain_check(ids, list(check_items["ids"]))
if not ids_match:
log.error("search_results_check: ids searched not match")
assert ids_match
if check_items.get("metric", None) is not None:
# verify the distances are already sorted
num_to_check = min(100, len(distances)) # check 100 items if more than that
if check_items.get("metric").upper() in ["IP", "COSINE", "BM25"]:
assert np.allclose(distances[:num_to_check], sorted(distances[:num_to_check], reverse=True))
else:
pass # just check nq and topk, not specific ids need check
nq_i += 1
assert np.allclose(distances[:num_to_check], sorted(distances[:num_to_check], reverse=False))
if check_items.get("vector_nq") is None or check_items.get("original_vectors") is None:
log.debug("skip distance check for knowhere does not return the precise distances")
else:
pass
else:
pass # just check nq and topk, not specific ids need check
log.info("search_results_check: limit (topK) and "
"ids searched for %d queries are correct" % len(search_res))
@ -586,12 +584,13 @@ class ResponseChecker:
for single_query_result in query_res:
single_query_result[vector_field] = np.frombuffer(single_query_result[vector_field][0], dtype=np.int8).tolist()
if isinstance(query_res, list):
result = pc.compare_lists_with_epsilon_ignore_dict_order(a=query_res, b=exp_res)
if result is False:
# Only for debug, compare the result with deepdiff
pc.compare_lists_with_epsilon_ignore_dict_order_deepdiff(a=query_res, b=exp_res)
assert result
return result
debug_mode = check_items.get("debug_mode", False)
if debug_mode is True:
assert pc.compare_lists_with_epsilon_ignore_dict_order_deepdiff(a=query_res, b=exp_res)
else:
assert pc.compare_lists_with_epsilon_ignore_dict_order(a=query_res, b=exp_res), \
f"there exists different values between query_results and expected_results, " \
f"use debug_mode in check_items to print the difference entity by entity(but it is slow)"
else:
log.error(f"Query result {query_res} is not list")
return False

1474
tests/python_client/milvus_client_v2/test_milvus_client_hybrid_search_v2.py
View File

File diff suppressed because it is too large Load Diff

4
tests/python_client/milvus_client_v2/test_milvus_client_search_v2_new.py
View File

@ -344,8 +344,8 @@ class TestMilvusClientSearchBasicV2(TestMilvusClientV2Base):
)
@pytest.mark.tags(CaseLabel.L2)
# @pytest.mark.parametrize("limit, nq", zip([1, 1000, ct.max_limit], [ct.max_nq, 10, 1]))
@pytest.mark.parametrize("limit, nq", zip([ct.max_limit], [1]))
@pytest.mark.parametrize("limit, nq", zip([1, 1000, ct.max_limit], [2000, 10, 1]))
# @pytest.mark.parametrize("limit, nq", zip([ct.max_limit], [1]))
def test_search_with_different_nq_limits(self, limit, nq):
"""
target: test search with different nq and limit values

4
tests/python_client/requirements.txt
View File

@ -28,8 +28,8 @@ pytest-parallel
pytest-random-order
# pymilvus
pymilvus==2.6.5rc2
pymilvus[bulk_writer]==2.6.5rc2
pymilvus==2.6.5rc9
pymilvus[bulk_writer]==2.6.5rc9
# for protobuf
protobuf>=5.29.5

7
tests/python_client/testcases/async_milvus_client/test_e2e_async.py
View File

@ -252,10 +252,7 @@ class TestAsyncMilvusClient(TestMilvusClientV2Base):
assert r[0]['insert_count'] == step
# flush
# TODO: call async flush() as https://github.com/milvus-io/pymilvus/issues/3060 fixed
# await self.async_milvus_client_wrap.flush(c_name)
milvus_client = self._client()
self.flush(milvus_client, c_name)
await self.async_milvus_client_wrap.flush(c_name)
stats, _ = await self.async_milvus_client_wrap.get_collection_stats(c_name)
assert stats["row_count"] == async_default_nb
@ -273,7 +270,7 @@ class TestAsyncMilvusClient(TestMilvusClientV2Base):
assert _index["indexed_rows"] == async_default_nb
assert _index["state"] == "Finished"
_load, _ = await self.async_milvus_client_wrap.get_load_state(c_name)
assert _load == LoadState.Loaded
assert _load["state"] == LoadState.Loaded
# dql tasks
tasks = []

8
tests/python_client/testcases/async_milvus_client/test_index_async.py
View File

@ -59,8 +59,7 @@ class TestAsyncMilvusClientIndexInvalid(TestMilvusClientV2Base):
index_params = async_client.prepare_index_params()[0]
index_params.add_index(field_name="vector")
# 3. create index
error = {ct.err_code: 1100, ct.err_msg: f"Invalid collection name: {name}. the first character of a collection "
f"name must be an underscore or letter: invalid parameter"}
error = {ct.err_code: 1100, ct.err_msg: f"collection not found[database=default][collection={name}]"}
await async_client.create_index(name, index_params,
check_task=CheckTasks.err_res,
check_items=error)
@ -87,8 +86,7 @@ class TestAsyncMilvusClientIndexInvalid(TestMilvusClientV2Base):
index_params = async_client.prepare_index_params()[0]
index_params.add_index(field_name="vector")
# 3. create index
error = {ct.err_code: 1100, ct.err_msg: f"Invalid collection name: {name}. the length of a collection name "
f"must be less than 255 characters: invalid parameter"}
error = {ct.err_code: 1100, ct.err_msg: f"collection not found[database=default][collection={name}]"}
await async_client.create_index(name, index_params,
check_task=CheckTasks.err_res,
check_items=error)
@ -116,7 +114,7 @@ class TestAsyncMilvusClientIndexInvalid(TestMilvusClientV2Base):
index_params.add_index(field_name="vector")
# 3. create index
error = {ct.err_code: 100,
ct.err_msg: f"can't find collection[database=default][collection={not_existed_collection_name}]"}
ct.err_msg: f"collection not found[database=default][collection={not_existed_collection_name}]"}
await async_client.create_index(not_existed_collection_name, index_params,
check_task=CheckTasks.err_res,
check_items=error)