test_data_manager.py 33 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705
  1. # Copyright 2021-2024 Avaiga Private Limited
  2. #
  3. # Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
  4. # the License. You may obtain a copy of the License at
  5. #
  6. # http://www.apache.org/licenses/LICENSE-2.0
  7. #
  8. # Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
  9. # an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
  10. # specific language governing permissions and limitations under the License.
  11. import os
  12. import pathlib
  13. import pytest
  14. from taipy.config.common.scope import Scope
  15. from taipy.config.config import Config
  16. from taipy.core._version._version_manager import _VersionManager
  17. from taipy.core.config.data_node_config import DataNodeConfig
  18. from taipy.core.data._data_manager import _DataManager
  19. from taipy.core.data.csv import CSVDataNode
  20. from taipy.core.data.data_node_id import DataNodeId
  21. from taipy.core.data.in_memory import InMemoryDataNode
  22. from taipy.core.data.pickle import PickleDataNode
  23. from taipy.core.exceptions.exceptions import InvalidDataNodeType, ModelNotFound
  24. from tests.core.utils.named_temporary_file import NamedTemporaryFile
  25. def file_exists(file_path: str) -> bool:
  26. return os.path.exists(file_path)
  27. class TestDataManager:
  28. def test_create_data_node_and_modify_properties_does_not_modify_config(self):
  29. dn_config = Config.configure_data_node(id="name", foo="bar")
  30. dn = _DataManager._create_and_set(dn_config, None, None)
  31. assert dn_config.properties.get("foo") == "bar"
  32. assert dn_config.properties.get("baz") is None
  33. dn.properties["baz"] = "qux"
  34. _DataManager._set(dn)
  35. assert dn_config.properties.get("foo") == "bar"
  36. assert dn_config.properties.get("baz") is None
  37. assert dn.properties.get("foo") == "bar"
  38. assert dn.properties.get("baz") == "qux"
  39. def test_create_data_node_with_name_provided(self):
  40. dn_config = Config.configure_data_node(id="dn", foo="bar", name="acb")
  41. dn = _DataManager._create_and_set(dn_config, None, None)
  42. assert dn.name == "acb"
  43. def test_create_and_get_csv_data_node(self):
  44. # Test we can instantiate a CsvDataNode from DataNodeConfig with :
  45. # - a csv type
  46. # - a default scenario scope
  47. # - No owner_id
  48. csv_dn_config = Config.configure_data_node(id="foo", storage_type="csv", path="bar", has_header=True)
  49. csv_dn = _DataManager._create_and_set(csv_dn_config, None, None)
  50. assert isinstance(csv_dn, CSVDataNode)
  51. assert isinstance(_DataManager._get(csv_dn.id), CSVDataNode)
  52. assert _DataManager._exists(csv_dn.id)
  53. assert _DataManager._get(csv_dn.id) is not None
  54. assert _DataManager._get(csv_dn.id).id == csv_dn.id
  55. assert _DataManager._get(csv_dn.id).config_id == "foo"
  56. assert _DataManager._get(csv_dn.id).config_id == csv_dn.config_id
  57. assert _DataManager._get(csv_dn.id).scope == Scope.SCENARIO
  58. assert _DataManager._get(csv_dn.id).scope == csv_dn.scope
  59. assert _DataManager._get(csv_dn.id).owner_id is None
  60. assert _DataManager._get(csv_dn.id).owner_id == csv_dn.owner_id
  61. assert _DataManager._get(csv_dn.id).parent_ids == set()
  62. assert _DataManager._get(csv_dn.id).parent_ids == csv_dn.parent_ids
  63. assert _DataManager._get(csv_dn.id).last_edit_date is None
  64. assert _DataManager._get(csv_dn.id).last_edit_date == csv_dn.last_edit_date
  65. assert _DataManager._get(csv_dn.id).job_ids == []
  66. assert _DataManager._get(csv_dn.id).job_ids == csv_dn.job_ids
  67. assert not _DataManager._get(csv_dn.id).is_ready_for_reading
  68. assert _DataManager._get(csv_dn.id).is_ready_for_reading == csv_dn.is_ready_for_reading
  69. assert (
  70. len(_DataManager._get(csv_dn.id).properties) == 5
  71. ) # path, encoding, has_header, exposed_type, is_generated
  72. assert _DataManager._get(csv_dn.id).properties.get("path") == "bar"
  73. assert _DataManager._get(csv_dn.id).properties.get("encoding") == "utf-8"
  74. assert _DataManager._get(csv_dn.id).properties.get("has_header") is True
  75. assert _DataManager._get(csv_dn.id).properties.get("exposed_type") == "pandas"
  76. assert _DataManager._get(csv_dn.id).properties.get("is_generated") is False
  77. assert _DataManager._get(csv_dn.id).properties == csv_dn.properties
  78. assert _DataManager._get(csv_dn.id).edit_in_progress is False
  79. assert _DataManager._get(csv_dn.id)._editor_id is None
  80. assert _DataManager._get(csv_dn.id)._editor_expiration_date is None
  81. assert _DataManager._get(csv_dn) is not None
  82. assert _DataManager._get(csv_dn).id == csv_dn.id
  83. assert _DataManager._get(csv_dn).config_id == "foo"
  84. assert _DataManager._get(csv_dn).config_id == csv_dn.config_id
  85. assert _DataManager._get(csv_dn).scope == Scope.SCENARIO
  86. assert _DataManager._get(csv_dn).scope == csv_dn.scope
  87. assert _DataManager._get(csv_dn).owner_id is None
  88. assert _DataManager._get(csv_dn).owner_id == csv_dn.owner_id
  89. assert _DataManager._get(csv_dn).parent_ids == set()
  90. assert _DataManager._get(csv_dn).parent_ids == csv_dn.parent_ids
  91. assert _DataManager._get(csv_dn).last_edit_date is None
  92. assert _DataManager._get(csv_dn).last_edit_date == csv_dn.last_edit_date
  93. assert _DataManager._get(csv_dn).job_ids == []
  94. assert _DataManager._get(csv_dn).job_ids == csv_dn.job_ids
  95. assert not _DataManager._get(csv_dn).is_ready_for_reading
  96. assert _DataManager._get(csv_dn).is_ready_for_reading == csv_dn.is_ready_for_reading
  97. assert len(_DataManager._get(csv_dn).properties) == 5 # path, encoding, has_header, exposed_type, is_generated
  98. assert _DataManager._get(csv_dn).properties.get("path") == "bar"
  99. assert _DataManager._get(csv_dn).properties.get("encoding") == "utf-8"
  100. assert _DataManager._get(csv_dn).properties.get("has_header") is True
  101. assert _DataManager._get(csv_dn.id).properties.get("exposed_type") == "pandas"
  102. assert _DataManager._get(csv_dn.id).properties.get("is_generated") is False
  103. assert _DataManager._get(csv_dn).properties == csv_dn.properties
  104. assert _DataManager._get(csv_dn.id).edit_in_progress is False
  105. assert _DataManager._get(csv_dn.id)._editor_id is None
  106. assert _DataManager._get(csv_dn.id)._editor_expiration_date is None
  107. def test_edit_and_get_data_node(self):
  108. config = Config.configure_pickle_data_node(id="foo")
  109. dn = _DataManager._create_and_set(config, None, None)
  110. assert _DataManager._get(dn.id).last_edit_date is None
  111. assert len(_DataManager._get(dn.id).properties) == 2 # is_generated and path
  112. assert isinstance(_DataManager._get(dn.id).properties.get("path"), str)
  113. assert _DataManager._get(dn.id).properties.get("is_generated") is True
  114. assert not _DataManager._get(dn.id).edit_in_progress
  115. assert _DataManager._get(dn.id)._editor_id is None
  116. assert _DataManager._get(dn.id)._editor_expiration_date is None
  117. dn.lock_edit("foo")
  118. assert _DataManager._get(dn.id).last_edit_date is None
  119. assert len(_DataManager._get(dn.id).properties) == 2 # is_generated and path
  120. assert isinstance(_DataManager._get(dn.id).properties.get("path"), str)
  121. assert _DataManager._get(dn.id).properties.get("is_generated") is True
  122. assert _DataManager._get(dn.id).edit_in_progress
  123. assert _DataManager._get(dn.id).editor_id == "foo"
  124. assert _DataManager._get(dn.id).editor_expiration_date is not None
  125. dn.unlock_edit("foo")
  126. assert _DataManager._get(dn.id).last_edit_date is None
  127. assert len(_DataManager._get(dn.id).properties) == 2 # is_generated and path
  128. assert isinstance(_DataManager._get(dn.id).properties.get("path"), str)
  129. assert _DataManager._get(dn.id).properties.get("is_generated") is True
  130. assert not _DataManager._get(dn.id).edit_in_progress
  131. assert _DataManager._get(dn.id).editor_id is None
  132. assert _DataManager._get(dn.id).editor_expiration_date is None
  133. def test_create_and_get_in_memory_data_node(self):
  134. # Test we can instantiate an InMemoryDataNode from DataNodeConfig with :
  135. # - an in_memory type
  136. # - a scenario scope
  137. # - an owner id
  138. # - some default data
  139. in_memory_dn_config = Config.configure_data_node(
  140. id="baz", storage_type="in_memory", scope=Scope.SCENARIO, default_data="qux", other_data="foo"
  141. )
  142. in_mem_dn = _DataManager._create_and_set(in_memory_dn_config, "Scenario_id", {"task_id"})
  143. assert isinstance(in_mem_dn, InMemoryDataNode)
  144. assert isinstance(_DataManager._get(in_mem_dn.id), InMemoryDataNode)
  145. assert _DataManager._exists(in_mem_dn.id)
  146. assert _DataManager._get(in_mem_dn.id) is not None
  147. assert _DataManager._get(in_mem_dn.id).id == in_mem_dn.id
  148. assert _DataManager._get(in_mem_dn.id).config_id == "baz"
  149. assert _DataManager._get(in_mem_dn.id).config_id == in_mem_dn.config_id
  150. assert _DataManager._get(in_mem_dn.id).scope == Scope.SCENARIO
  151. assert _DataManager._get(in_mem_dn.id).scope == in_mem_dn.scope
  152. assert _DataManager._get(in_mem_dn.id).owner_id == "Scenario_id"
  153. assert _DataManager._get(in_mem_dn.id).owner_id == in_mem_dn.owner_id
  154. assert _DataManager._get(in_mem_dn.id).parent_ids == {"task_id"}
  155. assert _DataManager._get(in_mem_dn.id).parent_ids == in_mem_dn.parent_ids
  156. assert _DataManager._get(in_mem_dn.id).last_edit_date is not None
  157. assert _DataManager._get(in_mem_dn.id).last_edit_date == in_mem_dn.last_edit_date
  158. assert _DataManager._get(in_mem_dn.id).job_ids == []
  159. assert _DataManager._get(in_mem_dn.id).job_ids == in_mem_dn.job_ids
  160. assert _DataManager._get(in_mem_dn.id).is_ready_for_reading
  161. assert _DataManager._get(in_mem_dn.id).is_ready_for_reading == in_mem_dn.is_ready_for_reading
  162. assert len(_DataManager._get(in_mem_dn.id).properties) == 1
  163. assert _DataManager._get(in_mem_dn.id).properties.get("other_data") == "foo"
  164. assert _DataManager._get(in_mem_dn.id).properties == in_mem_dn.properties
  165. assert _DataManager._get(in_mem_dn) is not None
  166. assert _DataManager._get(in_mem_dn).id == in_mem_dn.id
  167. assert _DataManager._get(in_mem_dn).config_id == "baz"
  168. assert _DataManager._get(in_mem_dn).config_id == in_mem_dn.config_id
  169. assert _DataManager._get(in_mem_dn).scope == Scope.SCENARIO
  170. assert _DataManager._get(in_mem_dn).scope == in_mem_dn.scope
  171. assert _DataManager._get(in_mem_dn).owner_id == "Scenario_id"
  172. assert _DataManager._get(in_mem_dn).owner_id == in_mem_dn.owner_id
  173. assert _DataManager._get(in_mem_dn).parent_ids == {"task_id"}
  174. assert _DataManager._get(in_mem_dn).parent_ids == in_mem_dn.parent_ids
  175. assert _DataManager._get(in_mem_dn).last_edit_date is not None
  176. assert _DataManager._get(in_mem_dn).last_edit_date == in_mem_dn.last_edit_date
  177. assert _DataManager._get(in_mem_dn).job_ids == []
  178. assert _DataManager._get(in_mem_dn).job_ids == in_mem_dn.job_ids
  179. assert _DataManager._get(in_mem_dn).is_ready_for_reading
  180. assert _DataManager._get(in_mem_dn).is_ready_for_reading == in_mem_dn.is_ready_for_reading
  181. assert len(_DataManager._get(in_mem_dn).properties) == 1
  182. assert _DataManager._get(in_mem_dn).properties.get("other_data") == "foo"
  183. assert _DataManager._get(in_mem_dn).properties == in_mem_dn.properties
  184. def test_create_and_get_pickle_data_node(self):
  185. # Test we can instantiate a PickleDataNode from DataNodeConfig with :
  186. # - an in_memory type
  187. # - a business cycle scope
  188. # - No owner id
  189. # - no default data
  190. dn_config = Config.configure_data_node(id="plop", storage_type="pickle", scope=Scope.CYCLE)
  191. pickle_dn = _DataManager._create_and_set(dn_config, None, {"task_id_1", "task_id_2"})
  192. assert isinstance(pickle_dn, PickleDataNode)
  193. assert isinstance(_DataManager._get(pickle_dn.id), PickleDataNode)
  194. assert _DataManager._exists(pickle_dn.id)
  195. assert _DataManager._get(pickle_dn.id) is not None
  196. assert _DataManager._get(pickle_dn.id).id == pickle_dn.id
  197. assert _DataManager._get(pickle_dn.id).config_id == "plop"
  198. assert _DataManager._get(pickle_dn.id).config_id == pickle_dn.config_id
  199. assert _DataManager._get(pickle_dn.id).scope == Scope.CYCLE
  200. assert _DataManager._get(pickle_dn.id).scope == pickle_dn.scope
  201. assert _DataManager._get(pickle_dn.id).owner_id is None
  202. assert _DataManager._get(pickle_dn.id).owner_id == pickle_dn.owner_id
  203. assert _DataManager._get(pickle_dn.id).parent_ids == {"task_id_1", "task_id_2"}
  204. assert _DataManager._get(pickle_dn.id).parent_ids == pickle_dn.parent_ids
  205. assert _DataManager._get(pickle_dn.id).last_edit_date is None
  206. assert _DataManager._get(pickle_dn.id).last_edit_date == pickle_dn.last_edit_date
  207. assert _DataManager._get(pickle_dn.id).job_ids == []
  208. assert _DataManager._get(pickle_dn.id).job_ids == pickle_dn.job_ids
  209. assert not _DataManager._get(pickle_dn.id).is_ready_for_reading
  210. assert _DataManager._get(pickle_dn.id).is_ready_for_reading == pickle_dn.is_ready_for_reading
  211. assert len(_DataManager._get(pickle_dn.id).properties) == 2 # is_generated and path
  212. assert _DataManager._get(pickle_dn.id).properties == pickle_dn.properties
  213. assert _DataManager._get(pickle_dn) is not None
  214. assert _DataManager._get(pickle_dn).id == pickle_dn.id
  215. assert _DataManager._get(pickle_dn).config_id == "plop"
  216. assert _DataManager._get(pickle_dn).config_id == pickle_dn.config_id
  217. assert _DataManager._get(pickle_dn).scope == Scope.CYCLE
  218. assert _DataManager._get(pickle_dn).scope == pickle_dn.scope
  219. assert _DataManager._get(pickle_dn).owner_id is None
  220. assert _DataManager._get(pickle_dn).owner_id == pickle_dn.owner_id
  221. assert _DataManager._get(pickle_dn).parent_ids == {"task_id_1", "task_id_2"}
  222. assert _DataManager._get(pickle_dn).parent_ids == pickle_dn.parent_ids
  223. assert _DataManager._get(pickle_dn).last_edit_date is None
  224. assert _DataManager._get(pickle_dn).last_edit_date == pickle_dn.last_edit_date
  225. assert _DataManager._get(pickle_dn).job_ids == []
  226. assert _DataManager._get(pickle_dn).job_ids == pickle_dn.job_ids
  227. assert not _DataManager._get(pickle_dn).is_ready_for_reading
  228. assert _DataManager._get(pickle_dn).is_ready_for_reading == pickle_dn.is_ready_for_reading
  229. assert len(_DataManager._get(pickle_dn).properties) == 2 # is_generated and path
  230. assert _DataManager._get(pickle_dn).properties == pickle_dn.properties
  231. def test_create_raises_exception_with_wrong_type(self):
  232. wrong_type_dn_config = DataNodeConfig(id="foo", storage_type="bar", scope=DataNodeConfig._DEFAULT_SCOPE)
  233. with pytest.raises(InvalidDataNodeType):
  234. _DataManager._create_and_set(wrong_type_dn_config, None, None)
  235. def test_create_from_same_config_generates_new_data_node_and_new_id(self):
  236. dn_config = Config.configure_data_node(id="foo", storage_type="in_memory")
  237. dn = _DataManager._create_and_set(dn_config, None, None)
  238. dn_2 = _DataManager._create_and_set(dn_config, None, None)
  239. assert dn_2.id != dn.id
  240. def test_create_uses_overridden_attributes_in_config_file(self):
  241. Config.override(os.path.join(pathlib.Path(__file__).parent.resolve(), "data_sample/config.toml"))
  242. csv_dn_cfg = Config.configure_data_node(id="foo", storage_type="csv", path="bar", has_header=True)
  243. csv_dn = _DataManager._create_and_set(csv_dn_cfg, None, None)
  244. assert csv_dn.config_id == "foo"
  245. assert isinstance(csv_dn, CSVDataNode)
  246. assert csv_dn._path == "path_from_config_file"
  247. assert csv_dn.has_header
  248. csv_dn_cfg = Config.configure_data_node(id="baz", storage_type="csv", path="bar", has_header=True)
  249. csv_dn = _DataManager._create_and_set(csv_dn_cfg, None, None)
  250. assert csv_dn.config_id == "baz"
  251. assert isinstance(csv_dn, CSVDataNode)
  252. assert csv_dn._path == "bar"
  253. assert csv_dn.has_header
  254. def test_get_if_not_exists(self):
  255. with pytest.raises(ModelNotFound):
  256. _DataManager._repository._load("test_data_node_2")
  257. def test_get_all(self):
  258. assert len(_DataManager._get_all()) == 0
  259. dn_config_1 = Config.configure_data_node(id="foo", storage_type="in_memory")
  260. _DataManager._create_and_set(dn_config_1, None, None)
  261. assert len(_DataManager._get_all()) == 1
  262. dn_config_2 = Config.configure_data_node(id="baz", storage_type="in_memory")
  263. _DataManager._create_and_set(dn_config_2, None, None)
  264. _DataManager._create_and_set(dn_config_2, None, None)
  265. assert len(_DataManager._get_all()) == 3
  266. assert len([dn for dn in _DataManager._get_all() if dn.config_id == "foo"]) == 1
  267. assert len([dn for dn in _DataManager._get_all() if dn.config_id == "baz"]) == 2
  268. def test_get_all_on_multiple_versions_environment(self):
  269. # Create 5 data nodes with 2 versions each
  270. # Only version 1.0 has the data node with config_id = "config_id_1"
  271. # Only version 2.0 has the data node with config_id = "config_id_6"
  272. for version in range(1, 3):
  273. for i in range(5):
  274. _DataManager._set(
  275. InMemoryDataNode(
  276. f"config_id_{i + version}",
  277. Scope.SCENARIO,
  278. id=DataNodeId(f"id{i}_v{version}"),
  279. version=f"{version}.0",
  280. )
  281. )
  282. _VersionManager._set_experiment_version("1.0")
  283. assert len(_DataManager._get_all()) == 5
  284. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_1"}])) == 1
  285. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_6"}])) == 0
  286. _VersionManager._set_development_version("1.0")
  287. assert len(_DataManager._get_all()) == 5
  288. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_1"}])) == 1
  289. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_6"}])) == 0
  290. _VersionManager._set_experiment_version("2.0")
  291. assert len(_DataManager._get_all()) == 5
  292. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_1"}])) == 0
  293. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_6"}])) == 1
  294. _VersionManager._set_development_version("2.0")
  295. assert len(_DataManager._get_all()) == 5
  296. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_1"}])) == 0
  297. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_6"}])) == 1
  298. def test_set(self):
  299. dn = InMemoryDataNode(
  300. "config_id",
  301. Scope.SCENARIO,
  302. id=DataNodeId("id"),
  303. owner_id=None,
  304. parent_ids={"task_id_1"},
  305. last_edit_date=None,
  306. edits=[],
  307. edit_in_progress=False,
  308. properties={"foo": "bar"},
  309. )
  310. assert len(_DataManager._get_all()) == 0
  311. assert not _DataManager._exists(dn.id)
  312. _DataManager._set(dn)
  313. assert len(_DataManager._get_all()) == 1
  314. assert _DataManager._exists(dn.id)
  315. # changing data node attribute
  316. dn._config_id = "foo"
  317. assert dn.config_id == "foo"
  318. _DataManager._set(dn)
  319. assert len(_DataManager._get_all()) == 1
  320. assert dn.config_id == "foo"
  321. assert _DataManager._get(dn.id).config_id == "foo"
  322. def test_delete(self):
  323. dn_1 = InMemoryDataNode("config_id", Scope.SCENARIO, id="id_1")
  324. dn_2 = InMemoryDataNode("config_id", Scope.SCENARIO, id="id_2")
  325. dn_3 = InMemoryDataNode("config_id", Scope.SCENARIO, id="id_3")
  326. assert len(_DataManager._get_all()) == 0
  327. _DataManager._set(dn_1)
  328. _DataManager._set(dn_2)
  329. _DataManager._set(dn_3)
  330. assert len(_DataManager._get_all()) == 3
  331. assert all(_DataManager._exists(dn.id) for dn in [dn_1, dn_2, dn_3])
  332. _DataManager._delete(dn_1.id)
  333. assert len(_DataManager._get_all()) == 2
  334. assert _DataManager._get(dn_2.id).id == dn_2.id
  335. assert _DataManager._get(dn_3.id).id == dn_3.id
  336. assert _DataManager._get(dn_1.id) is None
  337. assert all(_DataManager._exists(dn.id) for dn in [dn_2, dn_3])
  338. assert not _DataManager._exists(dn_1.id)
  339. _DataManager._delete_all()
  340. assert len(_DataManager._get_all()) == 0
  341. assert not any(_DataManager._exists(dn.id) for dn in [dn_2, dn_3])
  342. def test_get_or_create(self):
  343. def _get_or_create_dn(config, *args):
  344. return _DataManager._bulk_get_or_create([config], *args)[config]
  345. _DataManager._delete_all()
  346. global_dn_config = Config.configure_data_node(
  347. id="test_data_node", storage_type="in_memory", scope=Scope.GLOBAL, data="In memory Data Node"
  348. )
  349. cycle_dn_config = Config.configure_data_node(
  350. id="test_data_node1", storage_type="in_memory", scope=Scope.CYCLE, data="In memory Data Node"
  351. )
  352. scenario_dn_config = Config.configure_data_node(
  353. id="test_data_node2", storage_type="in_memory", scope=Scope.SCENARIO, data="In memory scenario"
  354. )
  355. assert len(_DataManager._get_all()) == 0
  356. global_dn = _get_or_create_dn(global_dn_config, None, None)
  357. assert len(_DataManager._get_all()) == 1
  358. global_dn_bis = _get_or_create_dn(global_dn_config, None)
  359. assert len(_DataManager._get_all()) == 1
  360. assert global_dn.id == global_dn_bis.id
  361. scenario_dn = _get_or_create_dn(scenario_dn_config, None, "scenario_id")
  362. assert len(_DataManager._get_all()) == 2
  363. scenario_dn_bis = _get_or_create_dn(scenario_dn_config, None, "scenario_id")
  364. assert len(_DataManager._get_all()) == 2
  365. assert scenario_dn.id == scenario_dn_bis.id
  366. scenario_dn_ter = _get_or_create_dn(scenario_dn_config, None, "scenario_id")
  367. assert len(_DataManager._get_all()) == 2
  368. assert scenario_dn.id == scenario_dn_bis.id
  369. assert scenario_dn_bis.id == scenario_dn_ter.id
  370. scenario_dn_quater = _get_or_create_dn(scenario_dn_config, None, "scenario_id_2")
  371. assert len(_DataManager._get_all()) == 3
  372. assert scenario_dn.id == scenario_dn_bis.id
  373. assert scenario_dn_bis.id == scenario_dn_ter.id
  374. assert scenario_dn_ter.id != scenario_dn_quater.id
  375. assert len(_DataManager._get_all()) == 3
  376. cycle_dn = _get_or_create_dn(cycle_dn_config, "cycle_id", None)
  377. assert len(_DataManager._get_all()) == 4
  378. cycle_dn_1 = _get_or_create_dn(cycle_dn_config, "cycle_id", None)
  379. assert len(_DataManager._get_all()) == 4
  380. assert cycle_dn.id == cycle_dn_1.id
  381. cycle_dn_2 = _get_or_create_dn(cycle_dn_config, "cycle_id", "scenario_id")
  382. assert len(_DataManager._get_all()) == 4
  383. assert cycle_dn.id == cycle_dn_2.id
  384. cycle_dn_3 = _get_or_create_dn(cycle_dn_config, "cycle_id", None)
  385. assert len(_DataManager._get_all()) == 4
  386. assert cycle_dn.id == cycle_dn_3.id
  387. cycle_dn_4 = _get_or_create_dn(cycle_dn_config, "cycle_id", "scenario_id")
  388. assert len(_DataManager._get_all()) == 4
  389. assert cycle_dn.id == cycle_dn_4.id
  390. cycle_dn_5 = _get_or_create_dn(cycle_dn_config, "cycle_id", "scenario_id_2")
  391. assert len(_DataManager._get_all()) == 4
  392. assert cycle_dn.id == cycle_dn_5.id
  393. assert cycle_dn_1.id == cycle_dn_2.id
  394. assert cycle_dn_2.id == cycle_dn_3.id
  395. assert cycle_dn_3.id == cycle_dn_4.id
  396. assert cycle_dn_4.id == cycle_dn_5.id
  397. def test_ensure_persistence_of_data_node(self):
  398. dm = _DataManager()
  399. dm._delete_all()
  400. dn_config_1 = Config.configure_data_node(
  401. id="data_node_1", storage_type="in_memory", data="In memory sequence 2"
  402. )
  403. dn_config_2 = Config.configure_data_node(
  404. id="data_node_2", storage_type="in_memory", data="In memory sequence 2"
  405. )
  406. dm._bulk_get_or_create([dn_config_1, dn_config_2])
  407. assert len(dm._get_all()) == 2
  408. # Delete the DataManager to ensure it's get from the storage system
  409. del dm
  410. dm = _DataManager()
  411. dm._bulk_get_or_create([dn_config_1])
  412. assert len(dm._get_all()) == 2
  413. dm._delete_all()
  414. @pytest.mark.parametrize(
  415. "storage_type,path",
  416. [
  417. ("pickle", "pickle_file_path"),
  418. ("csv", "csv_file"),
  419. ("excel", "excel_file"),
  420. ("json", "json_file"),
  421. ("parquet", "parquet_file_path"),
  422. ],
  423. )
  424. def test_clean_generated_files(self, storage_type, path, request):
  425. path = request.getfixturevalue(path)
  426. user_dn_config = Config.configure_data_node(
  427. id="d1", storage_type=storage_type, path=path, default_data={"a": [1], "b": [2]}
  428. )
  429. generated_dn_1_config = Config.configure_data_node(
  430. id="d2", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  431. )
  432. generated_dn_2_config = Config.configure_data_node(
  433. id="d3", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  434. )
  435. dns = _DataManager._bulk_get_or_create([user_dn_config, generated_dn_1_config, generated_dn_2_config])
  436. user_dn = dns[user_dn_config]
  437. generated_dn_1 = dns[generated_dn_1_config]
  438. generated_dn_2 = dns[generated_dn_2_config]
  439. _DataManager._clean_generated_file(user_dn.id)
  440. assert file_exists(user_dn.path)
  441. _DataManager._clean_generated_files([generated_dn_1, generated_dn_2])
  442. assert not file_exists(generated_dn_1.path)
  443. assert not file_exists(generated_dn_2.path)
  444. @pytest.mark.parametrize(
  445. "storage_type,path",
  446. [
  447. ("pickle", "pickle_file_path"),
  448. ("csv", "csv_file"),
  449. ("excel", "excel_file"),
  450. ("json", "json_file"),
  451. ("parquet", "parquet_file_path"),
  452. ],
  453. )
  454. def test_delete_does_clean_generated_pickle_files(self, storage_type, path, request):
  455. path = request.getfixturevalue(path)
  456. user_dn_config = Config.configure_data_node(
  457. id="d1", storage_type=storage_type, path=path, default_data={"a": [1], "b": [2]}
  458. )
  459. generated_dn_config_1 = Config.configure_data_node(
  460. id="d2", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  461. )
  462. generated_dn_config_2 = Config.configure_data_node(
  463. id="d3", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  464. )
  465. generated_dn_config_3 = Config.configure_data_node(
  466. id="d4", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  467. )
  468. dns = _DataManager._bulk_get_or_create(
  469. [
  470. user_dn_config,
  471. generated_dn_config_1,
  472. generated_dn_config_2,
  473. generated_dn_config_3,
  474. ]
  475. )
  476. user_dn = dns[user_dn_config]
  477. generated_dn_1 = dns[generated_dn_config_1]
  478. generated_dn_2 = dns[generated_dn_config_2]
  479. generated_dn_3 = dns[generated_dn_config_3]
  480. _DataManager._delete(user_dn.id)
  481. assert file_exists(user_dn.path)
  482. _DataManager._delete_many([generated_dn_1.id, generated_dn_2.id])
  483. assert not file_exists(generated_dn_1.path)
  484. assert not file_exists(generated_dn_2.path)
  485. _DataManager._delete_all()
  486. assert not file_exists(generated_dn_3.path)
  487. def test_create_dn_from_loaded_config_no_scope(self):
  488. file_config = NamedTemporaryFile(
  489. """
  490. [TAIPY]
  491. [DATA_NODE.a]
  492. default_data = "4:int"
  493. [DATA_NODE.b]
  494. [TASK.t]
  495. function = "math.sqrt:function"
  496. inputs = [ "a:SECTION",]
  497. outputs = [ "b:SECTION",]
  498. skippable = "False:bool"
  499. [SCENARIO.s]
  500. tasks = [ "t:SECTION",]
  501. sequences.s_sequence = [ "t:SECTION",]
  502. [SCENARIO.s.comparators]
  503. """
  504. )
  505. from taipy import core as tp
  506. Config.override(file_config.filename)
  507. tp.create_scenario(Config.scenarios["s"])
  508. tp.create_scenario(Config.scenarios["s"])
  509. assert len(tp.get_data_nodes()) == 4
  510. def test_create_dn_from_loaded_config_no_storage_type(self):
  511. file_config = NamedTemporaryFile(
  512. """
  513. [TAIPY]
  514. [DATA_NODE.input_dn]
  515. scope = "SCENARIO:SCOPE"
  516. default_data = "21:int"
  517. [DATA_NODE.output_dn]
  518. storage_type = "in_memory"
  519. scope = "SCENARIO:SCOPE"
  520. [TASK.double]
  521. inputs = [ "input_dn:SECTION",]
  522. function = "math.sqrt:function"
  523. outputs = [ "output_dn:SECTION",]
  524. skippable = "False:bool"
  525. [SCENARIO.my_scenario]
  526. tasks = [ "double:SECTION",]
  527. sequences.my_sequence = [ "double:SECTION",]
  528. [SCENARIO.my_scenario.comparators]
  529. """
  530. )
  531. from taipy import core as tp
  532. Config.override(file_config.filename)
  533. scenario = tp.create_scenario(Config.scenarios["my_scenario"])
  534. assert isinstance(scenario.input_dn, PickleDataNode)
  535. assert isinstance(scenario.output_dn, InMemoryDataNode)
  536. def test_create_dn_from_loaded_config_modified_default_config(self):
  537. file_config = NamedTemporaryFile(
  538. """
  539. [TAIPY]
  540. [DATA_NODE.input_dn]
  541. scope = "SCENARIO:SCOPE"
  542. default_path="fake/path.csv"
  543. [DATA_NODE.output_dn]
  544. storage_type = "in_memory"
  545. scope = "SCENARIO:SCOPE"
  546. [TASK.double]
  547. inputs = [ "input_dn:SECTION",]
  548. function = "math.sqrt:function"
  549. outputs = [ "output_dn:SECTION",]
  550. skippable = "False:bool"
  551. [SCENARIO.my_scenario]
  552. tasks = [ "double:SECTION",]
  553. sequences.my_sequence = [ "double:SECTION",]
  554. [SCENARIO.my_scenario.comparators]
  555. """
  556. )
  557. from taipy import core as tp
  558. Config.set_default_data_node_configuration(storage_type="csv")
  559. Config.override(file_config.filename)
  560. scenario = tp.create_scenario(Config.scenarios["my_scenario"])
  561. assert isinstance(scenario.input_dn, CSVDataNode)
  562. assert isinstance(scenario.output_dn, InMemoryDataNode)
  563. def test_get_tasks_by_config_id(self):
  564. dn_config_1 = Config.configure_data_node("dn_1", scope=Scope.SCENARIO)
  565. dn_config_2 = Config.configure_data_node("dn_2", scope=Scope.SCENARIO)
  566. dn_config_3 = Config.configure_data_node("dn_3", scope=Scope.SCENARIO)
  567. dn_1_1 = _DataManager._create_and_set(dn_config_1, None, None)
  568. dn_1_2 = _DataManager._create_and_set(dn_config_1, None, None)
  569. dn_1_3 = _DataManager._create_and_set(dn_config_1, None, None)
  570. assert len(_DataManager._get_all()) == 3
  571. dn_2_1 = _DataManager._create_and_set(dn_config_2, None, None)
  572. dn_2_2 = _DataManager._create_and_set(dn_config_2, None, None)
  573. assert len(_DataManager._get_all()) == 5
  574. dn_3_1 = _DataManager._create_and_set(dn_config_3, None, None)
  575. assert len(_DataManager._get_all()) == 6
  576. dn_1_datanodes = _DataManager._get_by_config_id(dn_config_1.id)
  577. assert len(dn_1_datanodes) == 3
  578. assert sorted([dn_1_1.id, dn_1_2.id, dn_1_3.id]) == sorted([sequence.id for sequence in dn_1_datanodes])
  579. dn_2_datanodes = _DataManager._get_by_config_id(dn_config_2.id)
  580. assert len(dn_2_datanodes) == 2
  581. assert sorted([dn_2_1.id, dn_2_2.id]) == sorted([sequence.id for sequence in dn_2_datanodes])
  582. dn_3_datanodes = _DataManager._get_by_config_id(dn_config_3.id)
  583. assert len(dn_3_datanodes) == 1
  584. assert sorted([dn_3_1.id]) == sorted([sequence.id for sequence in dn_3_datanodes])
  585. def test_get_data_nodes_by_config_id_in_multiple_versions_environment(self):
  586. dn_config_1 = Config.configure_data_node("dn_1", scope=Scope.SCENARIO)
  587. dn_config_2 = Config.configure_data_node("dn_2", scope=Scope.SCENARIO)
  588. _VersionManager._set_experiment_version("1.0")
  589. _DataManager._create_and_set(dn_config_1, None, None)
  590. _DataManager._create_and_set(dn_config_1, None, None)
  591. _DataManager._create_and_set(dn_config_1, None, None)
  592. _DataManager._create_and_set(dn_config_2, None, None)
  593. _DataManager._create_and_set(dn_config_2, None, None)
  594. assert len(_DataManager._get_by_config_id(dn_config_1.id)) == 3
  595. assert len(_DataManager._get_by_config_id(dn_config_2.id)) == 2
  596. _VersionManager._set_experiment_version("2.0")
  597. _DataManager._create_and_set(dn_config_1, None, None)
  598. _DataManager._create_and_set(dn_config_1, None, None)
  599. _DataManager._create_and_set(dn_config_1, None, None)
  600. _DataManager._create_and_set(dn_config_2, None, None)
  601. _DataManager._create_and_set(dn_config_2, None, None)
  602. assert len(_DataManager._get_by_config_id(dn_config_1.id)) == 3
  603. assert len(_DataManager._get_by_config_id(dn_config_2.id)) == 2