test_data_manager.py 32 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671
  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. from unittest.mock import patch
  14. import pytest
  15. from taipy.config.common.scope import Scope
  16. from taipy.config.config import Config
  17. from taipy.core._version._version_manager import _VersionManager
  18. from taipy.core.config.data_node_config import DataNodeConfig
  19. from taipy.core.data._data_manager import _DataManager
  20. from taipy.core.data.csv import CSVDataNode
  21. from taipy.core.data.data_node_id import DataNodeId
  22. from taipy.core.data.in_memory import InMemoryDataNode
  23. from taipy.core.data.pickle import PickleDataNode
  24. from taipy.core.exceptions.exceptions import InvalidDataNodeType, ModelNotFound
  25. from tests.core.utils.named_temporary_file import NamedTemporaryFile
  26. def file_exists(file_path: str) -> bool:
  27. return os.path.exists(file_path)
  28. class TestDataManager:
  29. def test_create_data_node_and_modify_properties_does_not_modify_config(self):
  30. dn_config = Config.configure_data_node(id="name", foo="bar")
  31. dn = _DataManager._create_and_set(dn_config, None, None)
  32. assert dn_config.properties.get("foo") == "bar"
  33. assert dn_config.properties.get("baz") is None
  34. dn.properties["baz"] = "qux"
  35. _DataManager._set(dn)
  36. assert dn_config.properties.get("foo") == "bar"
  37. assert dn_config.properties.get("baz") is None
  38. assert dn.properties.get("foo") == "bar"
  39. assert dn.properties.get("baz") == "qux"
  40. def test_create_data_node_with_name_provided(self):
  41. dn_config = Config.configure_data_node(id="dn", foo="bar", name="acb")
  42. dn = _DataManager._create_and_set(dn_config, None, None)
  43. assert dn.name == "acb"
  44. def test_create_and_get_csv_data_node(self):
  45. # Test we can instantiate a CsvDataNode from DataNodeConfig with :
  46. # - a csv type
  47. # - a default scenario scope
  48. # - No owner_id
  49. csv_dn_config = Config.configure_data_node(id="foo", storage_type="csv", path="bar", has_header=True)
  50. csv_dn = _DataManager._create_and_set(csv_dn_config, None, None)
  51. assert isinstance(csv_dn, CSVDataNode)
  52. assert isinstance(_DataManager._get(csv_dn.id), CSVDataNode)
  53. assert _DataManager._exists(csv_dn.id)
  54. assert _DataManager._get(csv_dn.id) is not None
  55. assert _DataManager._get(csv_dn.id).id == csv_dn.id
  56. assert _DataManager._get(csv_dn.id).config_id == "foo"
  57. assert _DataManager._get(csv_dn.id).config_id == csv_dn.config_id
  58. assert _DataManager._get(csv_dn.id).scope == Scope.SCENARIO
  59. assert _DataManager._get(csv_dn.id).scope == csv_dn.scope
  60. assert _DataManager._get(csv_dn.id).owner_id is None
  61. assert _DataManager._get(csv_dn.id).owner_id == csv_dn.owner_id
  62. assert _DataManager._get(csv_dn.id).parent_ids == set()
  63. assert _DataManager._get(csv_dn.id).parent_ids == csv_dn.parent_ids
  64. assert _DataManager._get(csv_dn.id).last_edit_date is None
  65. assert _DataManager._get(csv_dn.id).last_edit_date == csv_dn.last_edit_date
  66. assert _DataManager._get(csv_dn.id).job_ids == []
  67. assert _DataManager._get(csv_dn.id).job_ids == csv_dn.job_ids
  68. assert not _DataManager._get(csv_dn.id).is_ready_for_reading
  69. assert _DataManager._get(csv_dn.id).is_ready_for_reading == csv_dn.is_ready_for_reading
  70. assert len(_DataManager._get(csv_dn.id).properties) == 4
  71. assert _DataManager._get(csv_dn.id).properties.get("path") == "bar"
  72. assert _DataManager._get(csv_dn.id).properties.get("encoding") == "utf-8"
  73. assert _DataManager._get(csv_dn.id).properties.get("has_header") is True
  74. assert _DataManager._get(csv_dn.id).properties.get("exposed_type") == "pandas"
  75. assert _DataManager._get(csv_dn.id).properties == csv_dn.properties
  76. assert _DataManager._get(csv_dn.id).edit_in_progress is False
  77. assert _DataManager._get(csv_dn.id)._editor_id is None
  78. assert _DataManager._get(csv_dn.id)._editor_expiration_date is None
  79. assert _DataManager._get(csv_dn) is not None
  80. assert _DataManager._get(csv_dn).id == csv_dn.id
  81. assert _DataManager._get(csv_dn).config_id == "foo"
  82. assert _DataManager._get(csv_dn).config_id == csv_dn.config_id
  83. assert _DataManager._get(csv_dn).scope == Scope.SCENARIO
  84. assert _DataManager._get(csv_dn).scope == csv_dn.scope
  85. assert _DataManager._get(csv_dn).owner_id is None
  86. assert _DataManager._get(csv_dn).owner_id == csv_dn.owner_id
  87. assert _DataManager._get(csv_dn).parent_ids == set()
  88. assert _DataManager._get(csv_dn).parent_ids == csv_dn.parent_ids
  89. assert _DataManager._get(csv_dn).last_edit_date is None
  90. assert _DataManager._get(csv_dn).last_edit_date == csv_dn.last_edit_date
  91. assert _DataManager._get(csv_dn).job_ids == []
  92. assert _DataManager._get(csv_dn).job_ids == csv_dn.job_ids
  93. assert not _DataManager._get(csv_dn).is_ready_for_reading
  94. assert _DataManager._get(csv_dn).is_ready_for_reading == csv_dn.is_ready_for_reading
  95. assert len(_DataManager._get(csv_dn).properties) == 4
  96. assert _DataManager._get(csv_dn).properties.get("path") == "bar"
  97. assert _DataManager._get(csv_dn).properties.get("encoding") == "utf-8"
  98. assert _DataManager._get(csv_dn).properties.get("has_header") is True
  99. assert _DataManager._get(csv_dn.id).properties.get("exposed_type") == "pandas"
  100. assert _DataManager._get(csv_dn).properties == csv_dn.properties
  101. assert _DataManager._get(csv_dn.id).edit_in_progress is False
  102. assert _DataManager._get(csv_dn.id)._editor_id is None
  103. assert _DataManager._get(csv_dn.id)._editor_expiration_date is None
  104. def test_edit_and_get_data_node(self):
  105. config = Config.configure_pickle_data_node(id="foo")
  106. dn = _DataManager._create_and_set(config, None, None)
  107. assert _DataManager._get(dn.id).last_edit_date is None
  108. assert len(_DataManager._get(dn.id).properties) == 1
  109. assert _DataManager._get(dn.id).properties.get("is_generated")
  110. assert not _DataManager._get(dn.id).edit_in_progress
  111. assert _DataManager._get(dn.id)._editor_id is None
  112. assert _DataManager._get(dn.id)._editor_expiration_date is None
  113. dn.lock_edit("foo")
  114. assert _DataManager._get(dn.id).last_edit_date is None
  115. assert len(_DataManager._get(dn.id).properties) == 1
  116. assert _DataManager._get(dn.id).properties.get("is_generated")
  117. assert _DataManager._get(dn.id).edit_in_progress
  118. assert _DataManager._get(dn.id).editor_id == "foo"
  119. assert _DataManager._get(dn.id).editor_expiration_date is not None
  120. dn.unlock_edit("foo")
  121. assert _DataManager._get(dn.id).last_edit_date is None
  122. assert len(_DataManager._get(dn.id).properties) == 1
  123. assert _DataManager._get(dn.id).properties.get("is_generated")
  124. assert not _DataManager._get(dn.id).edit_in_progress
  125. assert _DataManager._get(dn.id).editor_id is None
  126. assert _DataManager._get(dn.id).editor_expiration_date is None
  127. def test_create_and_get_in_memory_data_node(self):
  128. # Test we can instantiate an InMemoryDataNode from DataNodeConfig with :
  129. # - an in_memory type
  130. # - a scenario scope
  131. # - an owner id
  132. # - some default data
  133. in_memory_dn_config = Config.configure_data_node(
  134. id="baz", storage_type="in_memory", scope=Scope.SCENARIO, default_data="qux", other_data="foo"
  135. )
  136. in_mem_dn = _DataManager._create_and_set(in_memory_dn_config, "Scenario_id", {"task_id"})
  137. assert isinstance(in_mem_dn, InMemoryDataNode)
  138. assert isinstance(_DataManager._get(in_mem_dn.id), InMemoryDataNode)
  139. assert _DataManager._exists(in_mem_dn.id)
  140. assert _DataManager._get(in_mem_dn.id) is not None
  141. assert _DataManager._get(in_mem_dn.id).id == in_mem_dn.id
  142. assert _DataManager._get(in_mem_dn.id).config_id == "baz"
  143. assert _DataManager._get(in_mem_dn.id).config_id == in_mem_dn.config_id
  144. assert _DataManager._get(in_mem_dn.id).scope == Scope.SCENARIO
  145. assert _DataManager._get(in_mem_dn.id).scope == in_mem_dn.scope
  146. assert _DataManager._get(in_mem_dn.id).owner_id == "Scenario_id"
  147. assert _DataManager._get(in_mem_dn.id).owner_id == in_mem_dn.owner_id
  148. assert _DataManager._get(in_mem_dn.id).parent_ids == {"task_id"}
  149. assert _DataManager._get(in_mem_dn.id).parent_ids == in_mem_dn.parent_ids
  150. assert _DataManager._get(in_mem_dn.id).last_edit_date is not None
  151. assert _DataManager._get(in_mem_dn.id).last_edit_date == in_mem_dn.last_edit_date
  152. assert _DataManager._get(in_mem_dn.id).job_ids == []
  153. assert _DataManager._get(in_mem_dn.id).job_ids == in_mem_dn.job_ids
  154. assert _DataManager._get(in_mem_dn.id).is_ready_for_reading
  155. assert _DataManager._get(in_mem_dn.id).is_ready_for_reading == in_mem_dn.is_ready_for_reading
  156. assert len(_DataManager._get(in_mem_dn.id).properties) == 1
  157. assert _DataManager._get(in_mem_dn.id).properties.get("other_data") == "foo"
  158. assert _DataManager._get(in_mem_dn.id).properties == in_mem_dn.properties
  159. assert _DataManager._get(in_mem_dn) is not None
  160. assert _DataManager._get(in_mem_dn).id == in_mem_dn.id
  161. assert _DataManager._get(in_mem_dn).config_id == "baz"
  162. assert _DataManager._get(in_mem_dn).config_id == in_mem_dn.config_id
  163. assert _DataManager._get(in_mem_dn).scope == Scope.SCENARIO
  164. assert _DataManager._get(in_mem_dn).scope == in_mem_dn.scope
  165. assert _DataManager._get(in_mem_dn).owner_id == "Scenario_id"
  166. assert _DataManager._get(in_mem_dn).owner_id == in_mem_dn.owner_id
  167. assert _DataManager._get(in_mem_dn).parent_ids == {"task_id"}
  168. assert _DataManager._get(in_mem_dn).parent_ids == in_mem_dn.parent_ids
  169. assert _DataManager._get(in_mem_dn).last_edit_date is not None
  170. assert _DataManager._get(in_mem_dn).last_edit_date == in_mem_dn.last_edit_date
  171. assert _DataManager._get(in_mem_dn).job_ids == []
  172. assert _DataManager._get(in_mem_dn).job_ids == in_mem_dn.job_ids
  173. assert _DataManager._get(in_mem_dn).is_ready_for_reading
  174. assert _DataManager._get(in_mem_dn).is_ready_for_reading == in_mem_dn.is_ready_for_reading
  175. assert len(_DataManager._get(in_mem_dn).properties) == 1
  176. assert _DataManager._get(in_mem_dn).properties.get("other_data") == "foo"
  177. assert _DataManager._get(in_mem_dn).properties == in_mem_dn.properties
  178. def test_create_and_get_pickle_data_node(self):
  179. # Test we can instantiate a PickleDataNode from DataNodeConfig with :
  180. # - an in_memory type
  181. # - a business cycle scope
  182. # - No owner id
  183. # - no default data
  184. dn_config = Config.configure_data_node(id="plop", storage_type="pickle", scope=Scope.CYCLE)
  185. pickle_dn = _DataManager._create_and_set(dn_config, None, {"task_id_1", "task_id_2"})
  186. assert isinstance(pickle_dn, PickleDataNode)
  187. assert isinstance(_DataManager._get(pickle_dn.id), PickleDataNode)
  188. assert _DataManager._exists(pickle_dn.id)
  189. assert _DataManager._get(pickle_dn.id) is not None
  190. assert _DataManager._get(pickle_dn.id).id == pickle_dn.id
  191. assert _DataManager._get(pickle_dn.id).config_id == "plop"
  192. assert _DataManager._get(pickle_dn.id).config_id == pickle_dn.config_id
  193. assert _DataManager._get(pickle_dn.id).scope == Scope.CYCLE
  194. assert _DataManager._get(pickle_dn.id).scope == pickle_dn.scope
  195. assert _DataManager._get(pickle_dn.id).owner_id is None
  196. assert _DataManager._get(pickle_dn.id).owner_id == pickle_dn.owner_id
  197. assert _DataManager._get(pickle_dn.id).parent_ids == {"task_id_1", "task_id_2"}
  198. assert _DataManager._get(pickle_dn.id).parent_ids == pickle_dn.parent_ids
  199. assert _DataManager._get(pickle_dn.id).last_edit_date is None
  200. assert _DataManager._get(pickle_dn.id).last_edit_date == pickle_dn.last_edit_date
  201. assert _DataManager._get(pickle_dn.id).job_ids == []
  202. assert _DataManager._get(pickle_dn.id).job_ids == pickle_dn.job_ids
  203. assert not _DataManager._get(pickle_dn.id).is_ready_for_reading
  204. assert _DataManager._get(pickle_dn.id).is_ready_for_reading == pickle_dn.is_ready_for_reading
  205. assert len(_DataManager._get(pickle_dn.id).properties) == 1
  206. assert _DataManager._get(pickle_dn.id).properties == pickle_dn.properties
  207. assert _DataManager._get(pickle_dn) is not None
  208. assert _DataManager._get(pickle_dn).id == pickle_dn.id
  209. assert _DataManager._get(pickle_dn).config_id == "plop"
  210. assert _DataManager._get(pickle_dn).config_id == pickle_dn.config_id
  211. assert _DataManager._get(pickle_dn).scope == Scope.CYCLE
  212. assert _DataManager._get(pickle_dn).scope == pickle_dn.scope
  213. assert _DataManager._get(pickle_dn).owner_id is None
  214. assert _DataManager._get(pickle_dn).owner_id == pickle_dn.owner_id
  215. assert _DataManager._get(pickle_dn).parent_ids == {"task_id_1", "task_id_2"}
  216. assert _DataManager._get(pickle_dn).parent_ids == pickle_dn.parent_ids
  217. assert _DataManager._get(pickle_dn).last_edit_date is None
  218. assert _DataManager._get(pickle_dn).last_edit_date == pickle_dn.last_edit_date
  219. assert _DataManager._get(pickle_dn).job_ids == []
  220. assert _DataManager._get(pickle_dn).job_ids == pickle_dn.job_ids
  221. assert not _DataManager._get(pickle_dn).is_ready_for_reading
  222. assert _DataManager._get(pickle_dn).is_ready_for_reading == pickle_dn.is_ready_for_reading
  223. assert len(_DataManager._get(pickle_dn).properties) == 1
  224. assert _DataManager._get(pickle_dn).properties == pickle_dn.properties
  225. def test_create_raises_exception_with_wrong_type(self):
  226. wrong_type_dn_config = DataNodeConfig(id="foo", storage_type="bar", scope=DataNodeConfig._DEFAULT_SCOPE)
  227. with pytest.raises(InvalidDataNodeType):
  228. _DataManager._create_and_set(wrong_type_dn_config, None, None)
  229. def test_create_from_same_config_generates_new_data_node_and_new_id(self):
  230. dn_config = Config.configure_data_node(id="foo", storage_type="in_memory")
  231. dn = _DataManager._create_and_set(dn_config, None, None)
  232. dn_2 = _DataManager._create_and_set(dn_config, None, None)
  233. assert dn_2.id != dn.id
  234. def test_create_uses_overridden_attributes_in_config_file(self):
  235. Config.override(os.path.join(pathlib.Path(__file__).parent.resolve(), "data_sample/config.toml"))
  236. csv_dn_cfg = Config.configure_data_node(id="foo", storage_type="csv", path="bar", has_header=True)
  237. csv_dn = _DataManager._create_and_set(csv_dn_cfg, None, None)
  238. assert csv_dn.config_id == "foo"
  239. assert isinstance(csv_dn, CSVDataNode)
  240. assert csv_dn._path == "path_from_config_file"
  241. assert csv_dn.has_header
  242. csv_dn_cfg = Config.configure_data_node(id="baz", 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 == "baz"
  245. assert isinstance(csv_dn, CSVDataNode)
  246. assert csv_dn._path == "bar"
  247. assert csv_dn.has_header
  248. def test_get_if_not_exists(self):
  249. with pytest.raises(ModelNotFound):
  250. _DataManager._repository._load("test_data_node_2")
  251. def test_get_all(self):
  252. assert len(_DataManager._get_all()) == 0
  253. dn_config_1 = Config.configure_data_node(id="foo", storage_type="in_memory")
  254. _DataManager._create_and_set(dn_config_1, None, None)
  255. assert len(_DataManager._get_all()) == 1
  256. dn_config_2 = Config.configure_data_node(id="baz", storage_type="in_memory")
  257. _DataManager._create_and_set(dn_config_2, None, None)
  258. _DataManager._create_and_set(dn_config_2, None, None)
  259. assert len(_DataManager._get_all()) == 3
  260. assert len([dn for dn in _DataManager._get_all() if dn.config_id == "foo"]) == 1
  261. assert len([dn for dn in _DataManager._get_all() if dn.config_id == "baz"]) == 2
  262. def test_get_all_on_multiple_versions_environment(self):
  263. # Create 5 data nodes with 2 versions each
  264. # Only version 1.0 has the data node with config_id = "config_id_1"
  265. # Only version 2.0 has the data node with config_id = "config_id_6"
  266. for version in range(1, 3):
  267. for i in range(5):
  268. _DataManager._set(
  269. InMemoryDataNode(
  270. f"config_id_{i + version}",
  271. Scope.SCENARIO,
  272. id=DataNodeId(f"id{i}_v{version}"),
  273. version=f"{version}.0",
  274. )
  275. )
  276. _VersionManager._set_experiment_version("1.0")
  277. assert len(_DataManager._get_all()) == 5
  278. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_1"}])) == 1
  279. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_6"}])) == 0
  280. _VersionManager._set_development_version("1.0")
  281. assert len(_DataManager._get_all()) == 5
  282. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_1"}])) == 1
  283. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_6"}])) == 0
  284. _VersionManager._set_experiment_version("2.0")
  285. assert len(_DataManager._get_all()) == 5
  286. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_1"}])) == 0
  287. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_6"}])) == 1
  288. _VersionManager._set_development_version("2.0")
  289. assert len(_DataManager._get_all()) == 5
  290. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_1"}])) == 0
  291. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_6"}])) == 1
  292. def test_set(self):
  293. dn = InMemoryDataNode(
  294. "config_id",
  295. Scope.SCENARIO,
  296. id=DataNodeId("id"),
  297. owner_id=None,
  298. parent_ids={"task_id_1"},
  299. last_edit_date=None,
  300. edits=[],
  301. edit_in_progress=False,
  302. properties={"foo": "bar"},
  303. )
  304. assert len(_DataManager._get_all()) == 0
  305. assert not _DataManager._exists(dn.id)
  306. _DataManager._set(dn)
  307. assert len(_DataManager._get_all()) == 1
  308. assert _DataManager._exists(dn.id)
  309. # changing data node attribute
  310. dn._config_id = "foo"
  311. assert dn.config_id == "foo"
  312. _DataManager._set(dn)
  313. assert len(_DataManager._get_all()) == 1
  314. assert dn.config_id == "foo"
  315. assert _DataManager._get(dn.id).config_id == "foo"
  316. def test_delete(self):
  317. dn_1 = InMemoryDataNode("config_id", Scope.SCENARIO, id="id_1")
  318. dn_2 = InMemoryDataNode("config_id", Scope.SCENARIO, id="id_2")
  319. dn_3 = InMemoryDataNode("config_id", Scope.SCENARIO, id="id_3")
  320. assert len(_DataManager._get_all()) == 0
  321. _DataManager._set(dn_1)
  322. _DataManager._set(dn_2)
  323. _DataManager._set(dn_3)
  324. assert len(_DataManager._get_all()) == 3
  325. assert all(_DataManager._exists(dn.id) for dn in [dn_1, dn_2, dn_3])
  326. _DataManager._delete(dn_1.id)
  327. assert len(_DataManager._get_all()) == 2
  328. assert _DataManager._get(dn_2.id).id == dn_2.id
  329. assert _DataManager._get(dn_3.id).id == dn_3.id
  330. assert _DataManager._get(dn_1.id) is None
  331. assert all(_DataManager._exists(dn.id) for dn in [dn_2, dn_3])
  332. assert not _DataManager._exists(dn_1.id)
  333. _DataManager._delete_all()
  334. assert len(_DataManager._get_all()) == 0
  335. assert not any(_DataManager._exists(dn.id) for dn in [dn_2, dn_3])
  336. def test_get_or_create(self):
  337. def _get_or_create_dn(config, *args):
  338. return _DataManager._bulk_get_or_create([config], *args)[config]
  339. _DataManager._delete_all()
  340. global_dn_config = Config.configure_data_node(
  341. id="test_data_node", storage_type="in_memory", scope=Scope.GLOBAL, data="In memory Data Node"
  342. )
  343. cycle_dn_config = Config.configure_data_node(
  344. id="test_data_node1", storage_type="in_memory", scope=Scope.CYCLE, data="In memory Data Node"
  345. )
  346. scenario_dn_config = Config.configure_data_node(
  347. id="test_data_node2", storage_type="in_memory", scope=Scope.SCENARIO, data="In memory scenario"
  348. )
  349. assert len(_DataManager._get_all()) == 0
  350. global_dn = _get_or_create_dn(global_dn_config, None, None)
  351. assert len(_DataManager._get_all()) == 1
  352. global_dn_bis = _get_or_create_dn(global_dn_config, None)
  353. assert len(_DataManager._get_all()) == 1
  354. assert global_dn.id == global_dn_bis.id
  355. scenario_dn = _get_or_create_dn(scenario_dn_config, None, "scenario_id")
  356. assert len(_DataManager._get_all()) == 2
  357. scenario_dn_bis = _get_or_create_dn(scenario_dn_config, None, "scenario_id")
  358. assert len(_DataManager._get_all()) == 2
  359. assert scenario_dn.id == scenario_dn_bis.id
  360. scenario_dn_ter = _get_or_create_dn(scenario_dn_config, None, "scenario_id")
  361. assert len(_DataManager._get_all()) == 2
  362. assert scenario_dn.id == scenario_dn_bis.id
  363. assert scenario_dn_bis.id == scenario_dn_ter.id
  364. scenario_dn_quater = _get_or_create_dn(scenario_dn_config, None, "scenario_id_2")
  365. assert len(_DataManager._get_all()) == 3
  366. assert scenario_dn.id == scenario_dn_bis.id
  367. assert scenario_dn_bis.id == scenario_dn_ter.id
  368. assert scenario_dn_ter.id != scenario_dn_quater.id
  369. assert len(_DataManager._get_all()) == 3
  370. cycle_dn = _get_or_create_dn(cycle_dn_config, "cycle_id", None)
  371. assert len(_DataManager._get_all()) == 4
  372. cycle_dn_1 = _get_or_create_dn(cycle_dn_config, "cycle_id", None)
  373. assert len(_DataManager._get_all()) == 4
  374. assert cycle_dn.id == cycle_dn_1.id
  375. cycle_dn_2 = _get_or_create_dn(cycle_dn_config, "cycle_id", "scenario_id")
  376. assert len(_DataManager._get_all()) == 4
  377. assert cycle_dn.id == cycle_dn_2.id
  378. cycle_dn_3 = _get_or_create_dn(cycle_dn_config, "cycle_id", None)
  379. assert len(_DataManager._get_all()) == 4
  380. assert cycle_dn.id == cycle_dn_3.id
  381. cycle_dn_4 = _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_4.id
  384. cycle_dn_5 = _get_or_create_dn(cycle_dn_config, "cycle_id", "scenario_id_2")
  385. assert len(_DataManager._get_all()) == 4
  386. assert cycle_dn.id == cycle_dn_5.id
  387. assert cycle_dn_1.id == cycle_dn_2.id
  388. assert cycle_dn_2.id == cycle_dn_3.id
  389. assert cycle_dn_3.id == cycle_dn_4.id
  390. assert cycle_dn_4.id == cycle_dn_5.id
  391. def test_ensure_persistence_of_data_node(self):
  392. dm = _DataManager()
  393. dm._delete_all()
  394. dn_config_1 = Config.configure_data_node(
  395. id="data_node_1", storage_type="in_memory", data="In memory sequence 2"
  396. )
  397. dn_config_2 = Config.configure_data_node(
  398. id="data_node_2", storage_type="in_memory", data="In memory sequence 2"
  399. )
  400. dm._bulk_get_or_create([dn_config_1, dn_config_2])
  401. assert len(dm._get_all()) == 2
  402. # Delete the DataManager to ensure it's get from the storage system
  403. del dm
  404. dm = _DataManager()
  405. dm._bulk_get_or_create([dn_config_1])
  406. assert len(dm._get_all()) == 2
  407. dm._delete_all()
  408. def test_clean_generated_pickle_files(self, pickle_file_path):
  409. user_pickle_dn_config = Config.configure_data_node(
  410. id="d1", storage_type="pickle", path=pickle_file_path, default_data="d"
  411. )
  412. generated_pickle_dn_1_config = Config.configure_data_node(id="d2", storage_type="pickle", default_data="d")
  413. generated_pickle_dn_2_config = Config.configure_data_node(id="d3", storage_type="pickle", default_data="d")
  414. dns = _DataManager._bulk_get_or_create(
  415. [user_pickle_dn_config, generated_pickle_dn_1_config, generated_pickle_dn_2_config]
  416. )
  417. user_pickle_dn = dns[user_pickle_dn_config]
  418. generated_pickle_dn_1 = dns[generated_pickle_dn_1_config]
  419. generated_pickle_dn_2 = dns[generated_pickle_dn_2_config]
  420. _DataManager._clean_pickle_file(user_pickle_dn.id)
  421. assert file_exists(user_pickle_dn.path)
  422. _DataManager._clean_pickle_files([generated_pickle_dn_1, generated_pickle_dn_2])
  423. assert not file_exists(generated_pickle_dn_1.path)
  424. assert not file_exists(generated_pickle_dn_2.path)
  425. def test_delete_does_clean_generated_pickle_files(self, pickle_file_path):
  426. user_pickle_dn_config = Config.configure_data_node(
  427. id="d1", storage_type="pickle", path=pickle_file_path, default_data="d"
  428. )
  429. generated_pickle_dn_config_1 = Config.configure_data_node(id="d2", storage_type="pickle", default_data="d")
  430. generated_pickle_dn_config_2 = Config.configure_data_node(id="d3", storage_type="pickle", default_data="d")
  431. generated_pickle_dn_config_3 = Config.configure_data_node(id="d4", storage_type="pickle", default_data="d")
  432. dns = _DataManager._bulk_get_or_create(
  433. [
  434. user_pickle_dn_config,
  435. generated_pickle_dn_config_1,
  436. generated_pickle_dn_config_2,
  437. generated_pickle_dn_config_3,
  438. ]
  439. )
  440. user_pickle_dn = dns[user_pickle_dn_config]
  441. generated_pickle_dn_1 = dns[generated_pickle_dn_config_1]
  442. generated_pickle_dn_2 = dns[generated_pickle_dn_config_2]
  443. generated_pickle_dn_3 = dns[generated_pickle_dn_config_3]
  444. _DataManager._delete(user_pickle_dn.id)
  445. assert file_exists(user_pickle_dn.path)
  446. _DataManager._delete_many([generated_pickle_dn_1.id, generated_pickle_dn_2.id])
  447. assert not file_exists(generated_pickle_dn_1.path)
  448. assert not file_exists(generated_pickle_dn_2.path)
  449. _DataManager._delete_all()
  450. assert not file_exists(generated_pickle_dn_3.path)
  451. def test_create_dn_from_loaded_config_no_scope(self):
  452. with patch("sys.argv", ["prog"]):
  453. file_config = NamedTemporaryFile(
  454. """
  455. [TAIPY]
  456. [DATA_NODE.a]
  457. default_data = "4:int"
  458. [DATA_NODE.b]
  459. [TASK.t]
  460. function = "math.sqrt:function"
  461. inputs = [ "a:SECTION",]
  462. outputs = [ "b:SECTION",]
  463. skippable = "False:bool"
  464. [SCENARIO.s]
  465. tasks = [ "t:SECTION",]
  466. sequences.s_sequence = [ "t:SECTION",]
  467. [SCENARIO.s.comparators]
  468. """
  469. )
  470. from taipy import core as tp
  471. Config.override(file_config.filename)
  472. tp.create_scenario(Config.scenarios["s"])
  473. tp.create_scenario(Config.scenarios["s"])
  474. assert len(tp.get_data_nodes()) == 4
  475. def test_create_dn_from_loaded_config_no_storage_type(self):
  476. with patch("sys.argv", ["prog"]):
  477. file_config = NamedTemporaryFile(
  478. """
  479. [TAIPY]
  480. [DATA_NODE.input_dn]
  481. scope = "SCENARIO:SCOPE"
  482. default_data = "21:int"
  483. [DATA_NODE.output_dn]
  484. storage_type = "in_memory"
  485. scope = "SCENARIO:SCOPE"
  486. [TASK.double]
  487. inputs = [ "input_dn:SECTION",]
  488. function = "math.sqrt:function"
  489. outputs = [ "output_dn:SECTION",]
  490. skippable = "False:bool"
  491. [SCENARIO.my_scenario]
  492. tasks = [ "double:SECTION",]
  493. sequences.my_sequence = [ "double:SECTION",]
  494. [SCENARIO.my_scenario.comparators]
  495. """
  496. )
  497. from taipy import core as tp
  498. Config.override(file_config.filename)
  499. scenario = tp.create_scenario(Config.scenarios["my_scenario"])
  500. assert isinstance(scenario.input_dn, PickleDataNode)
  501. assert isinstance(scenario.output_dn, InMemoryDataNode)
  502. def test_create_dn_from_loaded_config_modified_default_config(self):
  503. with patch("sys.argv", ["prog"]):
  504. file_config = NamedTemporaryFile(
  505. """
  506. [TAIPY]
  507. [DATA_NODE.input_dn]
  508. scope = "SCENARIO:SCOPE"
  509. default_path="fake/path.csv"
  510. [DATA_NODE.output_dn]
  511. storage_type = "in_memory"
  512. scope = "SCENARIO:SCOPE"
  513. [TASK.double]
  514. inputs = [ "input_dn:SECTION",]
  515. function = "math.sqrt:function"
  516. outputs = [ "output_dn:SECTION",]
  517. skippable = "False:bool"
  518. [SCENARIO.my_scenario]
  519. tasks = [ "double:SECTION",]
  520. sequences.my_sequence = [ "double:SECTION",]
  521. [SCENARIO.my_scenario.comparators]
  522. """
  523. )
  524. from taipy import core as tp
  525. Config.set_default_data_node_configuration(storage_type="csv")
  526. Config.override(file_config.filename)
  527. scenario = tp.create_scenario(Config.scenarios["my_scenario"])
  528. assert isinstance(scenario.input_dn, CSVDataNode)
  529. assert isinstance(scenario.output_dn, InMemoryDataNode)
  530. def test_get_tasks_by_config_id(self):
  531. dn_config_1 = Config.configure_data_node("dn_1", scope=Scope.SCENARIO)
  532. dn_config_2 = Config.configure_data_node("dn_2", scope=Scope.SCENARIO)
  533. dn_config_3 = Config.configure_data_node("dn_3", scope=Scope.SCENARIO)
  534. dn_1_1 = _DataManager._create_and_set(dn_config_1, None, None)
  535. dn_1_2 = _DataManager._create_and_set(dn_config_1, None, None)
  536. dn_1_3 = _DataManager._create_and_set(dn_config_1, None, None)
  537. assert len(_DataManager._get_all()) == 3
  538. dn_2_1 = _DataManager._create_and_set(dn_config_2, None, None)
  539. dn_2_2 = _DataManager._create_and_set(dn_config_2, None, None)
  540. assert len(_DataManager._get_all()) == 5
  541. dn_3_1 = _DataManager._create_and_set(dn_config_3, None, None)
  542. assert len(_DataManager._get_all()) == 6
  543. dn_1_datanodes = _DataManager._get_by_config_id(dn_config_1.id)
  544. assert len(dn_1_datanodes) == 3
  545. assert sorted([dn_1_1.id, dn_1_2.id, dn_1_3.id]) == sorted([sequence.id for sequence in dn_1_datanodes])
  546. dn_2_datanodes = _DataManager._get_by_config_id(dn_config_2.id)
  547. assert len(dn_2_datanodes) == 2
  548. assert sorted([dn_2_1.id, dn_2_2.id]) == sorted([sequence.id for sequence in dn_2_datanodes])
  549. dn_3_datanodes = _DataManager._get_by_config_id(dn_config_3.id)
  550. assert len(dn_3_datanodes) == 1
  551. assert sorted([dn_3_1.id]) == sorted([sequence.id for sequence in dn_3_datanodes])
  552. def test_get_data_nodes_by_config_id_in_multiple_versions_environment(self):
  553. dn_config_1 = Config.configure_data_node("dn_1", scope=Scope.SCENARIO)
  554. dn_config_2 = Config.configure_data_node("dn_2", scope=Scope.SCENARIO)
  555. _VersionManager._set_experiment_version("1.0")
  556. _DataManager._create_and_set(dn_config_1, None, None)
  557. _DataManager._create_and_set(dn_config_1, None, None)
  558. _DataManager._create_and_set(dn_config_1, None, None)
  559. _DataManager._create_and_set(dn_config_2, None, None)
  560. _DataManager._create_and_set(dn_config_2, None, None)
  561. assert len(_DataManager._get_by_config_id(dn_config_1.id)) == 3
  562. assert len(_DataManager._get_by_config_id(dn_config_2.id)) == 2
  563. _VersionManager._set_experiment_version("2.0")
  564. _DataManager._create_and_set(dn_config_1, None, None)
  565. _DataManager._create_and_set(dn_config_1, None, None)
  566. _DataManager._create_and_set(dn_config_1, None, None)
  567. _DataManager._create_and_set(dn_config_2, None, None)
  568. _DataManager._create_and_set(dn_config_2, None, None)
  569. assert len(_DataManager._get_by_config_id(dn_config_1.id)) == 3
  570. assert len(_DataManager._get_by_config_id(dn_config_2.id)) == 2