milvus-io /
milvus
| Total Complexity | 71 |
| Total Lines | 457 |
| Duplicated Lines | 10.07 % |
| Changes | 0 | ||
Duplicate code is one of the most pungent code smells. A rule that is often used is to re-structure code once it is duplicated in three or more places.
Common duplication problems, and corresponding solutions are:
Complex classes like test_insert often do a lot of different things. To break such a class down, we need to identify a cohesive component within that class. A common approach to find such a component is to look for fields/methods that share the same prefixes, or suffixes.
Once you have determined the fields that belong together, you can apply the Extract Class refactoring. If the component makes sense as a sub-class, Extract Subclass is also a candidate, and is often faster.
| 1 | import logging |
||
| 2 | import time |
||
| 3 | import pdb |
||
| 4 | import copy |
||
| 5 | import threading |
||
| 6 | from multiprocessing import Pool, Process |
||
| 7 | import pytest |
||
| 8 | from milvus import DataType |
||
| 9 | from utils import * |
||
| 10 | from constants import * |
||
| 11 | |||
| 12 | ADD_TIMEOUT = 60 |
||
| 13 | uid = "test_insert" |
||
| 14 | field_name = default_float_vec_field_name |
||
|
Comprehensibility
Best Practice
introduced
by
|
|||
| 15 | binary_field_name = default_binary_vec_field_name |
||
| 16 | default_single_query = { |
||
| 17 | "bool": { |
||
| 18 | "must": [ |
||
| 19 | {"vector": {field_name: {"topk": 10, "query": gen_vectors(1, default_dim), "metric_type": "L2", |
||
| 20 | "params": {"nprobe": 10}}}} |
||
| 21 | ] |
||
| 22 | } |
||
| 23 | } |
||
| 24 | |||
| 25 | |||
| 26 | class TestInsertBase: |
||
| 27 | """ |
||
| 28 | ****************************************************************** |
||
| 29 | The following cases are used to test `insert` function |
||
| 30 | ****************************************************************** |
||
| 31 | """ |
||
| 32 | |||
| 33 | @pytest.fixture( |
||
| 34 | scope="function", |
||
| 35 | params=gen_simple_index() |
||
| 36 | ) |
||
| 37 | def get_simple_index(self, request, connect): |
||
| 38 | if str(connect._cmd("mode")) == "CPU": |
||
| 39 | if request.param["index_type"] in index_cpu_not_support(): |
||
| 40 | pytest.skip("CPU not support index_type: ivf_sq8h") |
||
| 41 | return request.param |
||
| 42 | |||
| 43 | @pytest.fixture( |
||
| 44 | scope="function", |
||
| 45 | params=gen_single_filter_fields() |
||
| 46 | ) |
||
| 47 | def get_filter_field(self, request): |
||
| 48 | yield request.param |
||
| 49 | |||
| 50 | @pytest.fixture( |
||
| 51 | scope="function", |
||
| 52 | params=gen_single_vector_fields() |
||
| 53 | ) |
||
| 54 | def get_vector_field(self, request): |
||
| 55 | yield request.param |
||
| 56 | |||
| 57 | def test_add_vector_with_empty_vector(self, connect, collection): |
||
| 58 | ''' |
||
| 59 | target: test add vectors with empty vectors list |
||
| 60 | method: set empty vectors list as add method params |
||
| 61 | expected: raises a Exception |
||
| 62 | ''' |
||
| 63 | vector = [] |
||
| 64 | with pytest.raises(Exception) as e: |
||
| 65 | status, ids = connect.insert(collection, vector) |
||
| 66 | |||
| 67 | def test_add_vector_with_None(self, connect, collection): |
||
| 68 | ''' |
||
| 69 | target: test add vectors with None |
||
| 70 | method: set None as add method params |
||
| 71 | expected: raises a Exception |
||
| 72 | ''' |
||
| 73 | vector = None |
||
| 74 | with pytest.raises(Exception) as e: |
||
| 75 | status, ids = connect.insert(collection, vector) |
||
| 76 | |||
| 77 | @pytest.mark.timeout(ADD_TIMEOUT) |
||
| 78 | def test_insert_collection_not_existed(self, connect): |
||
| 79 | ''' |
||
| 80 | target: test insert, with collection not existed |
||
| 81 | method: insert entity into a random named collection |
||
| 82 | expected: error raised |
||
| 83 | ''' |
||
| 84 | collection_name = gen_unique_str(uid) |
||
| 85 | with pytest.raises(Exception) as e: |
||
| 86 | connect.insert(collection_name, default_entities_rows) |
||
| 87 | |||
| 88 | @pytest.mark.timeout(ADD_TIMEOUT) |
||
| 89 | def test_insert_drop_collection(self, connect, collection): |
||
| 90 | ''' |
||
| 91 | target: test delete collection after insert vector |
||
| 92 | method: insert vector and delete collection |
||
| 93 | expected: no error raised |
||
| 94 | ''' |
||
| 95 | ids = connect.insert(collection, default_entity_row) |
||
| 96 | assert len(ids) == 1 |
||
| 97 | connect.drop_collection(collection) |
||
| 98 | |||
| 99 | @pytest.mark.timeout(ADD_TIMEOUT) |
||
| 100 | def test_insert_sleep_drop_collection(self, connect, collection): |
||
| 101 | ''' |
||
| 102 | target: test delete collection after insert vector for a while |
||
| 103 | method: insert vector, sleep, and delete collection |
||
| 104 | expected: no error raised |
||
| 105 | ''' |
||
| 106 | ids = connect.insert(collection, default_entity_row) |
||
| 107 | assert len(ids) == 1 |
||
| 108 | connect.flush([collection]) |
||
| 109 | connect.drop_collection(collection) |
||
| 110 | |||
| 111 | @pytest.mark.timeout(ADD_TIMEOUT) |
||
| 112 | def test_insert_create_index(self, connect, collection, get_simple_index): |
||
| 113 | ''' |
||
| 114 | target: test build index insert after vector |
||
| 115 | method: insert vector and build index |
||
| 116 | expected: no error raised |
||
| 117 | ''' |
||
| 118 | ids = connect.insert(collection, default_entities_rows) |
||
| 119 | assert len(ids) == default_nb |
||
| 120 | connect.flush([collection]) |
||
| 121 | connect.create_index(collection, field_name, get_simple_index) |
||
| 122 | info = connect.get_collection_info(collection) |
||
| 123 | fields = info["fields"] |
||
| 124 | for field in fields: |
||
| 125 | if field["name"] == field_name: |
||
| 126 | assert field["indexes"][0] == get_simple_index |
||
| 127 | |||
| 128 | View Code Duplication | @pytest.mark.timeout(ADD_TIMEOUT) |
|
| 129 | def test_insert_after_create_index(self, connect, collection, get_simple_index): |
||
| 130 | ''' |
||
| 131 | target: test build index insert after vector |
||
| 132 | method: insert vector and build index |
||
| 133 | expected: no error raised |
||
| 134 | ''' |
||
| 135 | connect.create_index(collection, field_name, get_simple_index) |
||
| 136 | ids = connect.insert(collection, default_entities_rows) |
||
| 137 | assert len(ids) == default_nb |
||
| 138 | info = connect.get_collection_info(collection) |
||
| 139 | fields = info["fields"] |
||
| 140 | for field in fields: |
||
| 141 | if field["name"] == field_name: |
||
| 142 | assert field["indexes"][0] == get_simple_index |
||
| 143 | |||
| 144 | @pytest.mark.timeout(ADD_TIMEOUT) |
||
| 145 | def test_insert_search(self, connect, collection): |
||
| 146 | ''' |
||
| 147 | target: test search vector after insert vector after a while |
||
| 148 | method: insert vector, sleep, and search collection |
||
| 149 | expected: no error raised |
||
| 150 | ''' |
||
| 151 | ids = connect.insert(collection, default_entities_rows) |
||
| 152 | connect.flush([collection]) |
||
| 153 | res = connect.search(collection, default_single_query) |
||
| 154 | logging.getLogger().debug(res) |
||
| 155 | assert res |
||
| 156 | |||
| 157 | def test_insert_segment_row_count(self, connect, collection): |
||
| 158 | nb = default_segment_row_limit + 1 |
||
| 159 | res_ids = connect.insert(collection, gen_entities_rows(nb)) |
||
| 160 | connect.flush([collection]) |
||
| 161 | assert len(res_ids) == nb |
||
| 162 | stats = connect.get_collection_stats(collection) |
||
| 163 | assert len(stats['partitions'][0]['segments']) == 2 |
||
| 164 | for segment in stats['partitions'][0]['segments']: |
||
| 165 | assert segment['row_count'] in [default_segment_row_limit, 1] |
||
| 166 | |||
| 167 | @pytest.fixture( |
||
| 168 | scope="function", |
||
| 169 | params=[ |
||
| 170 | 1, |
||
| 171 | 2000 |
||
| 172 | ], |
||
| 173 | ) |
||
| 174 | def insert_count(self, request): |
||
| 175 | yield request.param |
||
| 176 | |||
| 177 | @pytest.mark.timeout(ADD_TIMEOUT) |
||
| 178 | def test_insert_ids_not_match(self, connect, id_collection, insert_count): |
||
| 179 | ''' |
||
| 180 | target: test insert vectors in collection, use customize ids |
||
| 181 | method: create collection and insert vectors in it, check the ids returned and the collection length after vectors inserted |
||
| 182 | expected: the length of ids and the collection row count |
||
| 183 | ''' |
||
| 184 | nb = insert_count |
||
| 185 | with pytest.raises(Exception) as e: |
||
| 186 | res_ids = connect.insert(id_collection, gen_entities_rows(nb)) |
||
| 187 | |||
| 188 | @pytest.mark.timeout(ADD_TIMEOUT) |
||
| 189 | def test_insert_twice_ids_no_ids(self, connect, collection): |
||
| 190 | ''' |
||
| 191 | target: check the result of insert, with params ids and no ids |
||
| 192 | method: test insert vectors twice, use customize ids first, and then use no ids |
||
| 193 | expected: error raised |
||
| 194 | ''' |
||
| 195 | with pytest.raises(Exception) as e: |
||
| 196 | res_ids = connect.insert(collection, gen_entities_rows(default_nb, _id=False)) |
||
| 197 | |||
| 198 | @pytest.mark.timeout(ADD_TIMEOUT) |
||
| 199 | def test_insert_tag(self, connect, collection): |
||
| 200 | ''' |
||
| 201 | target: test insert entities in collection created before |
||
| 202 | method: create collection and insert entities in it, with the partition_tag param |
||
| 203 | expected: the collection row count equals to nq |
||
| 204 | ''' |
||
| 205 | connect.create_partition(collection, default_tag) |
||
| 206 | ids = connect.insert(collection, default_entities_rows, partition_tag=default_tag) |
||
| 207 | assert len(ids) == default_nb |
||
| 208 | assert connect.has_partition(collection, default_tag) |
||
| 209 | |||
| 210 | @pytest.mark.timeout(ADD_TIMEOUT) |
||
| 211 | def test_insert_tag_with_ids(self, connect, id_collection): |
||
| 212 | ''' |
||
| 213 | target: test insert entities in collection created before, insert with ids |
||
| 214 | method: create collection and insert entities in it, with the partition_tag param |
||
| 215 | expected: the collection row count equals to nq |
||
| 216 | ''' |
||
| 217 | connect.create_partition(id_collection, default_tag) |
||
| 218 | ids = [i for i in range(default_nb)] |
||
| 219 | res_ids = connect.insert(id_collection, gen_entities_rows(default_nb, _id=False), partition_tag=default_tag) |
||
| 220 | assert res_ids == ids |
||
| 221 | |||
| 222 | @pytest.mark.timeout(ADD_TIMEOUT) |
||
| 223 | def test_insert_tag_not_existed(self, connect, collection): |
||
| 224 | ''' |
||
| 225 | target: test insert entities in collection created before |
||
| 226 | method: create collection and insert entities in it, with the not existed partition_tag param |
||
| 227 | expected: error raised |
||
| 228 | ''' |
||
| 229 | tag = gen_unique_str() |
||
| 230 | with pytest.raises(Exception) as e: |
||
| 231 | ids = connect.insert(collection, default_entities_rows, partition_tag=tag) |
||
| 232 | |||
| 233 | @pytest.mark.timeout(ADD_TIMEOUT) |
||
| 234 | def test_insert_tag_existed(self, connect, collection): |
||
| 235 | ''' |
||
| 236 | target: test insert entities in collection created before |
||
| 237 | method: create collection and insert entities in it repeatly, with the partition_tag param |
||
| 238 | expected: the collection row count equals to nq |
||
| 239 | ''' |
||
| 240 | connect.create_partition(collection, default_tag) |
||
| 241 | ids = connect.insert(collection, default_entities_rows, partition_tag=default_tag) |
||
| 242 | ids = connect.insert(collection, default_entities_rows, partition_tag=default_tag) |
||
| 243 | connect.flush([collection]) |
||
| 244 | res_count = connect.count_entities(collection) |
||
| 245 | assert res_count == 2 * default_nb |
||
| 246 | |||
| 247 | @pytest.mark.level(2) |
||
| 248 | def test_insert_collection_not_existed(self, connect): |
||
| 249 | ''' |
||
| 250 | target: test insert entities in collection, which not existed before |
||
| 251 | method: insert entities collection not existed, check the status |
||
| 252 | expected: error raised |
||
| 253 | ''' |
||
| 254 | with pytest.raises(Exception) as e: |
||
| 255 | ids = connect.insert(gen_unique_str("not_exist_collection"), default_entities_rows) |
||
| 256 | |||
| 257 | def test_insert_dim_not_matched(self, connect, collection): |
||
| 258 | ''' |
||
| 259 | target: test insert entities, the vector dimension is not equal to the collection dimension |
||
| 260 | method: the entities dimension is half of the collection dimension, check the status |
||
| 261 | expected: error raised |
||
| 262 | ''' |
||
| 263 | vectors = gen_vectors(default_nb, int(default_dim) // 2) |
||
| 264 | insert_entities = copy.deepcopy(default_entities_rows) |
||
| 265 | insert_entities[-1][default_float_vec_field_name] = vectors |
||
| 266 | with pytest.raises(Exception) as e: |
||
| 267 | ids = connect.insert(collection, insert_entities) |
||
| 268 | |||
| 269 | |||
| 270 | class TestInsertBinary: |
||
| 271 | @pytest.fixture( |
||
| 272 | scope="function", |
||
| 273 | params=gen_binary_index() |
||
| 274 | ) |
||
| 275 | def get_binary_index(self, request): |
||
| 276 | request.param["metric_type"] = "JACCARD" |
||
| 277 | return request.param |
||
| 278 | |||
| 279 | def test_insert_binary_entities(self, connect, binary_collection): |
||
| 280 | ''' |
||
| 281 | target: test insert entities in binary collection |
||
| 282 | method: create collection and insert binary entities in it |
||
| 283 | expected: the collection row count equals to nb |
||
| 284 | ''' |
||
| 285 | ids = connect.insert(binary_collection, default_binary_entities_rows) |
||
| 286 | assert len(ids) == default_nb |
||
| 287 | connect.flush() |
||
| 288 | assert connect.count_entities(binary_collection) == default_nb |
||
| 289 | |||
| 290 | def test_insert_binary_tag(self, connect, binary_collection): |
||
| 291 | ''' |
||
| 292 | target: test insert entities and create partition tag |
||
| 293 | method: create collection and insert binary entities in it, with the partition_tag param |
||
| 294 | expected: the collection row count equals to nb |
||
| 295 | ''' |
||
| 296 | connect.create_partition(binary_collection, default_tag) |
||
| 297 | ids = connect.insert(binary_collection, default_binary_entities_rows, partition_tag=default_tag) |
||
| 298 | assert len(ids) == default_nb |
||
| 299 | assert connect.has_partition(binary_collection, default_tag) |
||
| 300 | |||
| 301 | # TODO |
||
| 302 | @pytest.mark.level(2) |
||
| 303 | def test_insert_binary_multi_times(self, connect, binary_collection): |
||
| 304 | ''' |
||
| 305 | target: test insert entities multi times and final flush |
||
| 306 | method: create collection and insert binary entity multi and final flush |
||
| 307 | expected: the collection row count equals to nb |
||
| 308 | ''' |
||
| 309 | for i in range(default_nb): |
||
| 310 | ids = connect.insert(binary_collection, default_binary_entity_row) |
||
| 311 | assert len(ids) == 1 |
||
| 312 | connect.flush([binary_collection]) |
||
| 313 | assert connect.count_entities(binary_collection) == default_nb |
||
| 314 | |||
| 315 | View Code Duplication | def test_insert_binary_after_create_index(self, connect, binary_collection, get_binary_index): |
|
| 316 | ''' |
||
| 317 | target: test insert binary entities after build index |
||
| 318 | method: build index and insert entities |
||
| 319 | expected: no error raised |
||
| 320 | ''' |
||
| 321 | connect.create_index(binary_collection, binary_field_name, get_binary_index) |
||
| 322 | ids = connect.insert(binary_collection, default_binary_entities_rows) |
||
| 323 | assert len(ids) == default_nb |
||
| 324 | connect.flush([binary_collection]) |
||
| 325 | info = connect.get_collection_info(binary_collection) |
||
| 326 | fields = info["fields"] |
||
| 327 | for field in fields: |
||
| 328 | if field["name"] == binary_field_name: |
||
| 329 | assert field["indexes"][0] == get_binary_index |
||
| 330 | |||
| 331 | View Code Duplication | @pytest.mark.timeout(ADD_TIMEOUT) |
|
| 332 | def test_insert_binary_create_index(self, connect, binary_collection, get_binary_index): |
||
| 333 | ''' |
||
| 334 | target: test build index insert after vector |
||
| 335 | method: insert vector and build index |
||
| 336 | expected: no error raised |
||
| 337 | ''' |
||
| 338 | ids = connect.insert(binary_collection, default_binary_entities_rows) |
||
| 339 | assert len(ids) == default_nb |
||
| 340 | connect.flush([binary_collection]) |
||
| 341 | connect.create_index(binary_collection, binary_field_name, get_binary_index) |
||
| 342 | info = connect.get_collection_info(binary_collection) |
||
| 343 | fields = info["fields"] |
||
| 344 | for field in fields: |
||
| 345 | if field["name"] == binary_field_name: |
||
| 346 | assert field["indexes"][0] == get_binary_index |
||
| 347 | |||
| 348 | |||
| 349 | class TestInsertInvalid(object): |
||
| 350 | """ |
||
| 351 | Test inserting vectors with invalid collection names |
||
| 352 | """ |
||
| 353 | |||
| 354 | @pytest.fixture( |
||
| 355 | scope="function", |
||
| 356 | params=gen_invalid_strs() |
||
| 357 | ) |
||
| 358 | def get_collection_name(self, request): |
||
| 359 | yield request.param |
||
| 360 | |||
| 361 | @pytest.fixture( |
||
| 362 | scope="function", |
||
| 363 | params=gen_invalid_strs() |
||
| 364 | ) |
||
| 365 | def get_tag_name(self, request): |
||
| 366 | yield request.param |
||
| 367 | |||
| 368 | @pytest.fixture( |
||
| 369 | scope="function", |
||
| 370 | params=gen_invalid_strs() |
||
| 371 | ) |
||
| 372 | def get_field_name(self, request): |
||
| 373 | yield request.param |
||
| 374 | |||
| 375 | @pytest.fixture( |
||
| 376 | scope="function", |
||
| 377 | params=gen_invalid_strs() |
||
| 378 | ) |
||
| 379 | def get_field_type(self, request): |
||
| 380 | yield request.param |
||
| 381 | |||
| 382 | @pytest.fixture( |
||
| 383 | scope="function", |
||
| 384 | params=gen_invalid_strs() |
||
| 385 | ) |
||
| 386 | def get_field_int_value(self, request): |
||
| 387 | yield request.param |
||
| 388 | |||
| 389 | @pytest.fixture( |
||
| 390 | scope="function", |
||
| 391 | params=gen_invalid_ints() |
||
| 392 | ) |
||
| 393 | def get_entity_id(self, request): |
||
| 394 | yield request.param |
||
| 395 | |||
| 396 | @pytest.fixture( |
||
| 397 | scope="function", |
||
| 398 | params=gen_invalid_vectors() |
||
| 399 | ) |
||
| 400 | def get_field_vectors_value(self, request): |
||
| 401 | yield request.param |
||
| 402 | |||
| 403 | def test_insert_field_name_not_match(self, connect, collection): |
||
| 404 | ''' |
||
| 405 | target: test insert, with field name not matched |
||
| 406 | method: create collection and insert entities in it |
||
| 407 | expected: raise an exception |
||
| 408 | ''' |
||
| 409 | tmp_entity = copy.deepcopy(default_entity_row) |
||
| 410 | tmp_entity[0]["string"] = "string" |
||
| 411 | with pytest.raises(Exception): |
||
| 412 | connect.insert(collection, tmp_entity) |
||
| 413 | |||
| 414 | def test_insert_with_invalid_collection_name(self, connect, get_collection_name): |
||
| 415 | collection_name = get_collection_name |
||
| 416 | with pytest.raises(Exception): |
||
| 417 | connect.insert(collection_name, default_entity_row) |
||
| 418 | |||
| 419 | def test_insert_with_invalid_tag_name(self, connect, collection, get_tag_name): |
||
| 420 | tag_name = get_tag_name |
||
| 421 | connect.create_partition(collection, default_tag) |
||
| 422 | if tag_name is not None: |
||
| 423 | with pytest.raises(Exception): |
||
| 424 | connect.insert(collection, default_entity_row, partition_tag=tag_name) |
||
| 425 | else: |
||
| 426 | connect.insert(collection, default_entity_row, partition_tag=tag_name) |
||
| 427 | |||
| 428 | def test_insert_with_less_field(self, connect, collection): |
||
| 429 | tmp_entity = copy.deepcopy(default_entity_row) |
||
| 430 | tmp_entity[0].pop(default_float_vec_field_name) |
||
| 431 | with pytest.raises(Exception): |
||
| 432 | connect.insert(collection, tmp_entity) |
||
| 433 | |||
| 434 | def test_insert_with_less_field_id(self, connect, id_collection): |
||
| 435 | tmp_entity = copy.deepcopy(gen_entities_rows(default_nb, _id=False)) |
||
| 436 | tmp_entity[0].pop("_id") |
||
| 437 | with pytest.raises(Exception): |
||
| 438 | connect.insert(id_collection, tmp_entity) |
||
| 439 | |||
| 440 | def test_insert_with_more_field(self, connect, collection): |
||
| 441 | tmp_entity = copy.deepcopy(default_entity_row) |
||
| 442 | tmp_entity[0]["new_field"] = 1 |
||
| 443 | with pytest.raises(Exception): |
||
| 444 | connect.insert(collection, tmp_entity) |
||
| 445 | |||
| 446 | def test_insert_with_more_field_id(self, connect, collection): |
||
| 447 | tmp_entity = copy.deepcopy(default_entity_row) |
||
| 448 | tmp_entity[0]["_id"] = 1 |
||
| 449 | with pytest.raises(Exception): |
||
| 450 | connect.insert(collection, tmp_entity) |
||
| 451 | |||
| 452 | def test_insert_with_invalid_field_vector_value(self, connect, collection, get_field_vectors_value): |
||
| 453 | tmp_entity = copy.deepcopy(default_entity_row) |
||
| 454 | tmp_entity[0][default_float_vec_field_name][1] = get_field_vectors_value |
||
| 455 | with pytest.raises(Exception): |
||
| 456 | connect.insert(collection, tmp_entity) |
||
| 457 |
Loading history...
