Add unittest for delete and retrieve (#14373)

Signed-off-by: zhenshan.cao <zhenshan.cao@zilliz.com>
This commit is contained in:
zhenshan.cao 2021-12-28 22:18:47 +08:00 committed by GitHub
parent 2c0b89c6c9
commit f729853be7
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

View File

@ -257,3 +257,99 @@ TEST(GetEntityByIds, PrimaryKey) {
// }
// }
}
TEST(GetEntityByIds, delete_retrieve) {
auto schema = std::make_shared<Schema>();
auto fid_64 = schema->AddDebugField("i64", DataType::INT64);
auto DIM = 16;
auto fid_vec = schema->AddDebugField("vector_64", DataType::VECTOR_FLOAT, DIM, MetricType::METRIC_L2);
schema->set_primary_key(FieldOffset(0));
int64_t N = 10;
int64_t req_size = 10;
auto choose = [=](int i) { return i; };
auto dataset = DataGen(schema, N);
auto segment = CreateSealedSegment(schema);
SealedLoader(dataset, *segment);
auto i64_col = dataset.get_col<int64_t>(0);
auto plan = std::make_unique<query::RetrievePlan>(*schema);
auto term_expr = std::make_unique<query::TermExprImpl<int64_t>>();
term_expr->field_offset_ = FieldOffset(0);
term_expr->data_type_ = DataType::INT64;
for (int i = 0; i < req_size; ++i) {
term_expr->terms_.emplace_back(i64_col[choose(i)]);
}
plan->plan_node_ = std::make_unique<query::RetrievePlanNode>();
plan->plan_node_->predicate_ = std::move(term_expr);
std::vector<FieldOffset> target_offsets{FieldOffset(0), FieldOffset(1)};
plan->field_offsets_ = target_offsets;
{
auto retrieve_results = segment->Retrieve(plan.get(), 100);
Assert(retrieve_results->fields_data_size() == target_offsets.size());
FieldOffset field_offset(0);
auto field0 = retrieve_results->fields_data(0);
Assert(field0.has_scalars());
auto field0_data = field0.scalars().long_data();
for (int i = 0; i < req_size; ++i) {
auto index = choose(i);
auto data = field0_data.data(i);
}
for (int i = 0; i < req_size; ++i) {
auto index = choose(i);
auto data = field0_data.data(i);
ASSERT_EQ(data, i64_col[index]);
}
auto field1 = retrieve_results->fields_data(1);
Assert(field1.has_vectors());
auto field1_data = field1.vectors().float_vector();
ASSERT_EQ(field1_data.data_size(), DIM * req_size);
}
int64_t row_count = 0;
// strange, when enable load_delete_record, this test failed
auto load_delete_record = false;
if (load_delete_record) {
std::vector<idx_t> pks{1, 2, 3, 4, 5};
std::vector<Timestamp> timestamps{10, 10, 10, 10, 10};
LoadDeletedRecordInfo info = {timestamps.data(), pks.data(), row_count};
segment->LoadDeletedRecord(info);
row_count = 5;
}
int64_t new_count = 6;
std::vector<idx_t> new_pks{0,1,2,3,4,5};
std::vector<idx_t> new_timestamps{10, 10, 10,10,10, 10};
auto reserved_offset = segment->PreDelete(new_count);
ASSERT_EQ(reserved_offset, row_count);
segment->Delete(reserved_offset, new_count, reinterpret_cast<const int64_t*>(new_pks.data()),
reinterpret_cast<const Timestamp*>(new_timestamps.data()));
{
auto retrieve_results = segment->Retrieve(plan.get(), 100);
Assert(retrieve_results->fields_data_size() == target_offsets.size());
FieldOffset field_offset(0);
auto field0 = retrieve_results->fields_data(0);
Assert(field0.has_scalars());
auto field0_data = field0.scalars().long_data();
auto size = req_size - new_count;
for (int i = 0; i < size; ++i) {
auto index = choose(i);
auto data = field0_data.data(i);
ASSERT_EQ(data, i64_col[index+new_count]);
}
auto field1 = retrieve_results->fields_data(1);
Assert(field1.has_vectors());
auto field1_data = field1.vectors().float_vector();
ASSERT_EQ(field1_data.data_size(), DIM * size);
}
}