test_data_manager.py 34 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723
  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_can_create(self):
  40. dn_config = Config.configure_in_memory_data_node("dn", 10, scope=Scope.SCENARIO)
  41. global_dn_config = Config.configure_data_node(
  42. id="global_dn", storage_type="in_memory", scope=Scope.GLOBAL, data=10
  43. )
  44. reasons = _DataManager._can_create(global_dn_config)
  45. assert bool(reasons) is True
  46. assert reasons._reasons == {}
  47. reasons = _DataManager._can_create(dn_config)
  48. assert bool(reasons) is False
  49. assert reasons._reasons == {dn_config.id: {'Data node config "dn" does not have GLOBAL scope'}}
  50. reasons = _DataManager._can_create(1)
  51. assert bool(reasons) is False
  52. assert reasons._reasons == {"1": {'Object "1" is not a valid config to be created'}}
  53. def test_create_data_node_with_name_provided(self):
  54. dn_config = Config.configure_data_node(id="dn", foo="bar", name="acb")
  55. dn = _DataManager._create_and_set(dn_config, None, None)
  56. assert dn.name == "acb"
  57. def test_create_and_get_csv_data_node(self):
  58. # Test we can instantiate a CsvDataNode from DataNodeConfig with :
  59. # - a csv type
  60. # - a default scenario scope
  61. # - No owner_id
  62. csv_dn_config = Config.configure_data_node(id="foo", storage_type="csv", path="bar", has_header=True)
  63. csv_dn = _DataManager._create_and_set(csv_dn_config, None, None)
  64. assert isinstance(csv_dn, CSVDataNode)
  65. assert isinstance(_DataManager._get(csv_dn.id), CSVDataNode)
  66. assert _DataManager._exists(csv_dn.id)
  67. assert _DataManager._get(csv_dn.id) is not None
  68. assert _DataManager._get(csv_dn.id).id == csv_dn.id
  69. assert _DataManager._get(csv_dn.id).config_id == "foo"
  70. assert _DataManager._get(csv_dn.id).config_id == csv_dn.config_id
  71. assert _DataManager._get(csv_dn.id).scope == Scope.SCENARIO
  72. assert _DataManager._get(csv_dn.id).scope == csv_dn.scope
  73. assert _DataManager._get(csv_dn.id).owner_id is None
  74. assert _DataManager._get(csv_dn.id).owner_id == csv_dn.owner_id
  75. assert _DataManager._get(csv_dn.id).parent_ids == set()
  76. assert _DataManager._get(csv_dn.id).parent_ids == csv_dn.parent_ids
  77. assert _DataManager._get(csv_dn.id).last_edit_date is None
  78. assert _DataManager._get(csv_dn.id).last_edit_date == csv_dn.last_edit_date
  79. assert _DataManager._get(csv_dn.id).job_ids == []
  80. assert _DataManager._get(csv_dn.id).job_ids == csv_dn.job_ids
  81. assert not _DataManager._get(csv_dn.id).is_ready_for_reading
  82. assert _DataManager._get(csv_dn.id).is_ready_for_reading == csv_dn.is_ready_for_reading
  83. assert (
  84. len(_DataManager._get(csv_dn.id).properties) == 5
  85. ) # path, encoding, has_header, exposed_type, is_generated
  86. assert _DataManager._get(csv_dn.id).properties.get("path") == "bar"
  87. assert _DataManager._get(csv_dn.id).properties.get("encoding") == "utf-8"
  88. assert _DataManager._get(csv_dn.id).properties.get("has_header") is True
  89. assert _DataManager._get(csv_dn.id).properties.get("exposed_type") == "pandas"
  90. assert _DataManager._get(csv_dn.id).properties.get("is_generated") is False
  91. assert _DataManager._get(csv_dn.id).properties == csv_dn.properties
  92. assert _DataManager._get(csv_dn.id).edit_in_progress is False
  93. assert _DataManager._get(csv_dn.id)._editor_id is None
  94. assert _DataManager._get(csv_dn.id)._editor_expiration_date is None
  95. assert _DataManager._get(csv_dn) is not None
  96. assert _DataManager._get(csv_dn).id == csv_dn.id
  97. assert _DataManager._get(csv_dn).config_id == "foo"
  98. assert _DataManager._get(csv_dn).config_id == csv_dn.config_id
  99. assert _DataManager._get(csv_dn).scope == Scope.SCENARIO
  100. assert _DataManager._get(csv_dn).scope == csv_dn.scope
  101. assert _DataManager._get(csv_dn).owner_id is None
  102. assert _DataManager._get(csv_dn).owner_id == csv_dn.owner_id
  103. assert _DataManager._get(csv_dn).parent_ids == set()
  104. assert _DataManager._get(csv_dn).parent_ids == csv_dn.parent_ids
  105. assert _DataManager._get(csv_dn).last_edit_date is None
  106. assert _DataManager._get(csv_dn).last_edit_date == csv_dn.last_edit_date
  107. assert _DataManager._get(csv_dn).job_ids == []
  108. assert _DataManager._get(csv_dn).job_ids == csv_dn.job_ids
  109. assert not _DataManager._get(csv_dn).is_ready_for_reading
  110. assert _DataManager._get(csv_dn).is_ready_for_reading == csv_dn.is_ready_for_reading
  111. assert len(_DataManager._get(csv_dn).properties) == 5 # path, encoding, has_header, exposed_type, is_generated
  112. assert _DataManager._get(csv_dn).properties.get("path") == "bar"
  113. assert _DataManager._get(csv_dn).properties.get("encoding") == "utf-8"
  114. assert _DataManager._get(csv_dn).properties.get("has_header") is True
  115. assert _DataManager._get(csv_dn.id).properties.get("exposed_type") == "pandas"
  116. assert _DataManager._get(csv_dn.id).properties.get("is_generated") is False
  117. assert _DataManager._get(csv_dn).properties == csv_dn.properties
  118. assert _DataManager._get(csv_dn.id).edit_in_progress is False
  119. assert _DataManager._get(csv_dn.id)._editor_id is None
  120. assert _DataManager._get(csv_dn.id)._editor_expiration_date is None
  121. def test_edit_and_get_data_node(self):
  122. config = Config.configure_pickle_data_node(id="foo")
  123. dn = _DataManager._create_and_set(config, None, None)
  124. assert _DataManager._get(dn.id).last_edit_date is None
  125. assert len(_DataManager._get(dn.id).properties) == 2 # is_generated and path
  126. assert isinstance(_DataManager._get(dn.id).properties.get("path"), str)
  127. assert _DataManager._get(dn.id).properties.get("is_generated") is True
  128. assert not _DataManager._get(dn.id).edit_in_progress
  129. assert _DataManager._get(dn.id)._editor_id is None
  130. assert _DataManager._get(dn.id)._editor_expiration_date is None
  131. dn.lock_edit("foo")
  132. assert _DataManager._get(dn.id).last_edit_date is None
  133. assert len(_DataManager._get(dn.id).properties) == 2 # is_generated and path
  134. assert isinstance(_DataManager._get(dn.id).properties.get("path"), str)
  135. assert _DataManager._get(dn.id).properties.get("is_generated") is True
  136. assert _DataManager._get(dn.id).edit_in_progress
  137. assert _DataManager._get(dn.id).editor_id == "foo"
  138. assert _DataManager._get(dn.id).editor_expiration_date is not None
  139. dn.unlock_edit("foo")
  140. assert _DataManager._get(dn.id).last_edit_date is None
  141. assert len(_DataManager._get(dn.id).properties) == 2 # is_generated and path
  142. assert isinstance(_DataManager._get(dn.id).properties.get("path"), str)
  143. assert _DataManager._get(dn.id).properties.get("is_generated") is True
  144. assert not _DataManager._get(dn.id).edit_in_progress
  145. assert _DataManager._get(dn.id).editor_id is None
  146. assert _DataManager._get(dn.id).editor_expiration_date is None
  147. def test_create_and_get_in_memory_data_node(self):
  148. # Test we can instantiate an InMemoryDataNode from DataNodeConfig with :
  149. # - an in_memory type
  150. # - a scenario scope
  151. # - an owner id
  152. # - some default data
  153. in_memory_dn_config = Config.configure_data_node(
  154. id="baz", storage_type="in_memory", scope=Scope.SCENARIO, default_data="qux", other_data="foo"
  155. )
  156. in_mem_dn = _DataManager._create_and_set(in_memory_dn_config, "Scenario_id", {"task_id"})
  157. assert isinstance(in_mem_dn, InMemoryDataNode)
  158. assert isinstance(_DataManager._get(in_mem_dn.id), InMemoryDataNode)
  159. assert _DataManager._exists(in_mem_dn.id)
  160. assert _DataManager._get(in_mem_dn.id) is not None
  161. assert _DataManager._get(in_mem_dn.id).id == in_mem_dn.id
  162. assert _DataManager._get(in_mem_dn.id).config_id == "baz"
  163. assert _DataManager._get(in_mem_dn.id).config_id == in_mem_dn.config_id
  164. assert _DataManager._get(in_mem_dn.id).scope == Scope.SCENARIO
  165. assert _DataManager._get(in_mem_dn.id).scope == in_mem_dn.scope
  166. assert _DataManager._get(in_mem_dn.id).owner_id == "Scenario_id"
  167. assert _DataManager._get(in_mem_dn.id).owner_id == in_mem_dn.owner_id
  168. assert _DataManager._get(in_mem_dn.id).parent_ids == {"task_id"}
  169. assert _DataManager._get(in_mem_dn.id).parent_ids == in_mem_dn.parent_ids
  170. assert _DataManager._get(in_mem_dn.id).last_edit_date is not None
  171. assert _DataManager._get(in_mem_dn.id).last_edit_date == in_mem_dn.last_edit_date
  172. assert _DataManager._get(in_mem_dn.id).job_ids == []
  173. assert _DataManager._get(in_mem_dn.id).job_ids == in_mem_dn.job_ids
  174. assert _DataManager._get(in_mem_dn.id).is_ready_for_reading
  175. assert _DataManager._get(in_mem_dn.id).is_ready_for_reading == in_mem_dn.is_ready_for_reading
  176. assert len(_DataManager._get(in_mem_dn.id).properties) == 1
  177. assert _DataManager._get(in_mem_dn.id).properties.get("other_data") == "foo"
  178. assert _DataManager._get(in_mem_dn.id).properties == in_mem_dn.properties
  179. assert _DataManager._get(in_mem_dn) is not None
  180. assert _DataManager._get(in_mem_dn).id == in_mem_dn.id
  181. assert _DataManager._get(in_mem_dn).config_id == "baz"
  182. assert _DataManager._get(in_mem_dn).config_id == in_mem_dn.config_id
  183. assert _DataManager._get(in_mem_dn).scope == Scope.SCENARIO
  184. assert _DataManager._get(in_mem_dn).scope == in_mem_dn.scope
  185. assert _DataManager._get(in_mem_dn).owner_id == "Scenario_id"
  186. assert _DataManager._get(in_mem_dn).owner_id == in_mem_dn.owner_id
  187. assert _DataManager._get(in_mem_dn).parent_ids == {"task_id"}
  188. assert _DataManager._get(in_mem_dn).parent_ids == in_mem_dn.parent_ids
  189. assert _DataManager._get(in_mem_dn).last_edit_date is not None
  190. assert _DataManager._get(in_mem_dn).last_edit_date == in_mem_dn.last_edit_date
  191. assert _DataManager._get(in_mem_dn).job_ids == []
  192. assert _DataManager._get(in_mem_dn).job_ids == in_mem_dn.job_ids
  193. assert _DataManager._get(in_mem_dn).is_ready_for_reading
  194. assert _DataManager._get(in_mem_dn).is_ready_for_reading == in_mem_dn.is_ready_for_reading
  195. assert len(_DataManager._get(in_mem_dn).properties) == 1
  196. assert _DataManager._get(in_mem_dn).properties.get("other_data") == "foo"
  197. assert _DataManager._get(in_mem_dn).properties == in_mem_dn.properties
  198. def test_create_and_get_pickle_data_node(self):
  199. # Test we can instantiate a PickleDataNode from DataNodeConfig with :
  200. # - an in_memory type
  201. # - a business cycle scope
  202. # - No owner id
  203. # - no default data
  204. dn_config = Config.configure_data_node(id="plop", storage_type="pickle", scope=Scope.CYCLE)
  205. pickle_dn = _DataManager._create_and_set(dn_config, None, {"task_id_1", "task_id_2"})
  206. assert isinstance(pickle_dn, PickleDataNode)
  207. assert isinstance(_DataManager._get(pickle_dn.id), PickleDataNode)
  208. assert _DataManager._exists(pickle_dn.id)
  209. assert _DataManager._get(pickle_dn.id) is not None
  210. assert _DataManager._get(pickle_dn.id).id == pickle_dn.id
  211. assert _DataManager._get(pickle_dn.id).config_id == "plop"
  212. assert _DataManager._get(pickle_dn.id).config_id == pickle_dn.config_id
  213. assert _DataManager._get(pickle_dn.id).scope == Scope.CYCLE
  214. assert _DataManager._get(pickle_dn.id).scope == pickle_dn.scope
  215. assert _DataManager._get(pickle_dn.id).owner_id is None
  216. assert _DataManager._get(pickle_dn.id).owner_id == pickle_dn.owner_id
  217. assert _DataManager._get(pickle_dn.id).parent_ids == {"task_id_1", "task_id_2"}
  218. assert _DataManager._get(pickle_dn.id).parent_ids == pickle_dn.parent_ids
  219. assert _DataManager._get(pickle_dn.id).last_edit_date is None
  220. assert _DataManager._get(pickle_dn.id).last_edit_date == pickle_dn.last_edit_date
  221. assert _DataManager._get(pickle_dn.id).job_ids == []
  222. assert _DataManager._get(pickle_dn.id).job_ids == pickle_dn.job_ids
  223. assert not _DataManager._get(pickle_dn.id).is_ready_for_reading
  224. assert _DataManager._get(pickle_dn.id).is_ready_for_reading == pickle_dn.is_ready_for_reading
  225. assert len(_DataManager._get(pickle_dn.id).properties) == 2 # is_generated and path
  226. assert _DataManager._get(pickle_dn.id).properties == pickle_dn.properties
  227. assert _DataManager._get(pickle_dn) is not None
  228. assert _DataManager._get(pickle_dn).id == pickle_dn.id
  229. assert _DataManager._get(pickle_dn).config_id == "plop"
  230. assert _DataManager._get(pickle_dn).config_id == pickle_dn.config_id
  231. assert _DataManager._get(pickle_dn).scope == Scope.CYCLE
  232. assert _DataManager._get(pickle_dn).scope == pickle_dn.scope
  233. assert _DataManager._get(pickle_dn).owner_id is None
  234. assert _DataManager._get(pickle_dn).owner_id == pickle_dn.owner_id
  235. assert _DataManager._get(pickle_dn).parent_ids == {"task_id_1", "task_id_2"}
  236. assert _DataManager._get(pickle_dn).parent_ids == pickle_dn.parent_ids
  237. assert _DataManager._get(pickle_dn).last_edit_date is None
  238. assert _DataManager._get(pickle_dn).last_edit_date == pickle_dn.last_edit_date
  239. assert _DataManager._get(pickle_dn).job_ids == []
  240. assert _DataManager._get(pickle_dn).job_ids == pickle_dn.job_ids
  241. assert not _DataManager._get(pickle_dn).is_ready_for_reading
  242. assert _DataManager._get(pickle_dn).is_ready_for_reading == pickle_dn.is_ready_for_reading
  243. assert len(_DataManager._get(pickle_dn).properties) == 2 # is_generated and path
  244. assert _DataManager._get(pickle_dn).properties == pickle_dn.properties
  245. def test_create_raises_exception_with_wrong_type(self):
  246. wrong_type_dn_config = DataNodeConfig(id="foo", storage_type="bar", scope=DataNodeConfig._DEFAULT_SCOPE)
  247. with pytest.raises(InvalidDataNodeType):
  248. _DataManager._create_and_set(wrong_type_dn_config, None, None)
  249. def test_create_from_same_config_generates_new_data_node_and_new_id(self):
  250. dn_config = Config.configure_data_node(id="foo", storage_type="in_memory")
  251. dn = _DataManager._create_and_set(dn_config, None, None)
  252. dn_2 = _DataManager._create_and_set(dn_config, None, None)
  253. assert dn_2.id != dn.id
  254. def test_create_uses_overridden_attributes_in_config_file(self):
  255. Config.override(os.path.join(pathlib.Path(__file__).parent.resolve(), "data_sample/config.toml"))
  256. csv_dn_cfg = Config.configure_data_node(id="foo", storage_type="csv", path="bar", has_header=True)
  257. csv_dn = _DataManager._create_and_set(csv_dn_cfg, None, None)
  258. assert csv_dn.config_id == "foo"
  259. assert isinstance(csv_dn, CSVDataNode)
  260. assert csv_dn._path == "path_from_config_file"
  261. assert csv_dn.has_header
  262. csv_dn_cfg = Config.configure_data_node(id="baz", storage_type="csv", path="bar", has_header=True)
  263. csv_dn = _DataManager._create_and_set(csv_dn_cfg, None, None)
  264. assert csv_dn.config_id == "baz"
  265. assert isinstance(csv_dn, CSVDataNode)
  266. assert csv_dn._path == "bar"
  267. assert csv_dn.has_header
  268. def test_get_if_not_exists(self):
  269. with pytest.raises(ModelNotFound):
  270. _DataManager._repository._load("test_data_node_2")
  271. def test_get_all(self):
  272. assert len(_DataManager._get_all()) == 0
  273. dn_config_1 = Config.configure_data_node(id="foo", storage_type="in_memory")
  274. _DataManager._create_and_set(dn_config_1, None, None)
  275. assert len(_DataManager._get_all()) == 1
  276. dn_config_2 = Config.configure_data_node(id="baz", storage_type="in_memory")
  277. _DataManager._create_and_set(dn_config_2, None, None)
  278. _DataManager._create_and_set(dn_config_2, None, None)
  279. assert len(_DataManager._get_all()) == 3
  280. assert len([dn for dn in _DataManager._get_all() if dn.config_id == "foo"]) == 1
  281. assert len([dn for dn in _DataManager._get_all() if dn.config_id == "baz"]) == 2
  282. def test_get_all_on_multiple_versions_environment(self):
  283. # Create 5 data nodes with 2 versions each
  284. # Only version 1.0 has the data node with config_id = "config_id_1"
  285. # Only version 2.0 has the data node with config_id = "config_id_6"
  286. for version in range(1, 3):
  287. for i in range(5):
  288. _DataManager._set(
  289. InMemoryDataNode(
  290. f"config_id_{i + version}",
  291. Scope.SCENARIO,
  292. id=DataNodeId(f"id{i}_v{version}"),
  293. version=f"{version}.0",
  294. )
  295. )
  296. _VersionManager._set_experiment_version("1.0")
  297. assert len(_DataManager._get_all()) == 5
  298. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_1"}])) == 1
  299. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_6"}])) == 0
  300. _VersionManager._set_development_version("1.0")
  301. assert len(_DataManager._get_all()) == 5
  302. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_1"}])) == 1
  303. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_6"}])) == 0
  304. _VersionManager._set_experiment_version("2.0")
  305. assert len(_DataManager._get_all()) == 5
  306. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_1"}])) == 0
  307. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_6"}])) == 1
  308. _VersionManager._set_development_version("2.0")
  309. assert len(_DataManager._get_all()) == 5
  310. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_1"}])) == 0
  311. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_6"}])) == 1
  312. def test_set(self):
  313. dn = InMemoryDataNode(
  314. "config_id",
  315. Scope.SCENARIO,
  316. id=DataNodeId("id"),
  317. owner_id=None,
  318. parent_ids={"task_id_1"},
  319. last_edit_date=None,
  320. edits=[],
  321. edit_in_progress=False,
  322. properties={"foo": "bar"},
  323. )
  324. assert len(_DataManager._get_all()) == 0
  325. assert not _DataManager._exists(dn.id)
  326. _DataManager._set(dn)
  327. assert len(_DataManager._get_all()) == 1
  328. assert _DataManager._exists(dn.id)
  329. # changing data node attribute
  330. dn._config_id = "foo"
  331. assert dn.config_id == "foo"
  332. _DataManager._set(dn)
  333. assert len(_DataManager._get_all()) == 1
  334. assert dn.config_id == "foo"
  335. assert _DataManager._get(dn.id).config_id == "foo"
  336. def test_delete(self):
  337. dn_1 = InMemoryDataNode("config_id", Scope.SCENARIO, id="id_1")
  338. dn_2 = InMemoryDataNode("config_id", Scope.SCENARIO, id="id_2")
  339. dn_3 = InMemoryDataNode("config_id", Scope.SCENARIO, id="id_3")
  340. assert len(_DataManager._get_all()) == 0
  341. _DataManager._set(dn_1)
  342. _DataManager._set(dn_2)
  343. _DataManager._set(dn_3)
  344. assert len(_DataManager._get_all()) == 3
  345. assert all(_DataManager._exists(dn.id) for dn in [dn_1, dn_2, dn_3])
  346. _DataManager._delete(dn_1.id)
  347. assert len(_DataManager._get_all()) == 2
  348. assert _DataManager._get(dn_2.id).id == dn_2.id
  349. assert _DataManager._get(dn_3.id).id == dn_3.id
  350. assert _DataManager._get(dn_1.id) is None
  351. assert all(_DataManager._exists(dn.id) for dn in [dn_2, dn_3])
  352. assert not _DataManager._exists(dn_1.id)
  353. _DataManager._delete_all()
  354. assert len(_DataManager._get_all()) == 0
  355. assert not any(_DataManager._exists(dn.id) for dn in [dn_2, dn_3])
  356. def test_get_or_create(self):
  357. def _get_or_create_dn(config, *args):
  358. return _DataManager._bulk_get_or_create([config], *args)[config]
  359. _DataManager._delete_all()
  360. global_dn_config = Config.configure_data_node(
  361. id="test_data_node", storage_type="in_memory", scope=Scope.GLOBAL, data="In memory Data Node"
  362. )
  363. cycle_dn_config = Config.configure_data_node(
  364. id="test_data_node1", storage_type="in_memory", scope=Scope.CYCLE, data="In memory Data Node"
  365. )
  366. scenario_dn_config = Config.configure_data_node(
  367. id="test_data_node2", storage_type="in_memory", scope=Scope.SCENARIO, data="In memory scenario"
  368. )
  369. assert len(_DataManager._get_all()) == 0
  370. global_dn = _get_or_create_dn(global_dn_config, None, None)
  371. assert len(_DataManager._get_all()) == 1
  372. global_dn_bis = _get_or_create_dn(global_dn_config, None)
  373. assert len(_DataManager._get_all()) == 1
  374. assert global_dn.id == global_dn_bis.id
  375. scenario_dn = _get_or_create_dn(scenario_dn_config, None, "scenario_id")
  376. assert len(_DataManager._get_all()) == 2
  377. scenario_dn_bis = _get_or_create_dn(scenario_dn_config, None, "scenario_id")
  378. assert len(_DataManager._get_all()) == 2
  379. assert scenario_dn.id == scenario_dn_bis.id
  380. scenario_dn_ter = _get_or_create_dn(scenario_dn_config, None, "scenario_id")
  381. assert len(_DataManager._get_all()) == 2
  382. assert scenario_dn.id == scenario_dn_bis.id
  383. assert scenario_dn_bis.id == scenario_dn_ter.id
  384. scenario_dn_quater = _get_or_create_dn(scenario_dn_config, None, "scenario_id_2")
  385. assert len(_DataManager._get_all()) == 3
  386. assert scenario_dn.id == scenario_dn_bis.id
  387. assert scenario_dn_bis.id == scenario_dn_ter.id
  388. assert scenario_dn_ter.id != scenario_dn_quater.id
  389. assert len(_DataManager._get_all()) == 3
  390. cycle_dn = _get_or_create_dn(cycle_dn_config, "cycle_id", None)
  391. assert len(_DataManager._get_all()) == 4
  392. cycle_dn_1 = _get_or_create_dn(cycle_dn_config, "cycle_id", None)
  393. assert len(_DataManager._get_all()) == 4
  394. assert cycle_dn.id == cycle_dn_1.id
  395. cycle_dn_2 = _get_or_create_dn(cycle_dn_config, "cycle_id", "scenario_id")
  396. assert len(_DataManager._get_all()) == 4
  397. assert cycle_dn.id == cycle_dn_2.id
  398. cycle_dn_3 = _get_or_create_dn(cycle_dn_config, "cycle_id", None)
  399. assert len(_DataManager._get_all()) == 4
  400. assert cycle_dn.id == cycle_dn_3.id
  401. cycle_dn_4 = _get_or_create_dn(cycle_dn_config, "cycle_id", "scenario_id")
  402. assert len(_DataManager._get_all()) == 4
  403. assert cycle_dn.id == cycle_dn_4.id
  404. cycle_dn_5 = _get_or_create_dn(cycle_dn_config, "cycle_id", "scenario_id_2")
  405. assert len(_DataManager._get_all()) == 4
  406. assert cycle_dn.id == cycle_dn_5.id
  407. assert cycle_dn_1.id == cycle_dn_2.id
  408. assert cycle_dn_2.id == cycle_dn_3.id
  409. assert cycle_dn_3.id == cycle_dn_4.id
  410. assert cycle_dn_4.id == cycle_dn_5.id
  411. def test_ensure_persistence_of_data_node(self):
  412. dm = _DataManager()
  413. dm._delete_all()
  414. dn_config_1 = Config.configure_data_node(
  415. id="data_node_1", storage_type="in_memory", data="In memory sequence 2"
  416. )
  417. dn_config_2 = Config.configure_data_node(
  418. id="data_node_2", storage_type="in_memory", data="In memory sequence 2"
  419. )
  420. dm._bulk_get_or_create([dn_config_1, dn_config_2])
  421. assert len(dm._get_all()) == 2
  422. # Delete the DataManager to ensure it's get from the storage system
  423. del dm
  424. dm = _DataManager()
  425. dm._bulk_get_or_create([dn_config_1])
  426. assert len(dm._get_all()) == 2
  427. dm._delete_all()
  428. @pytest.mark.parametrize(
  429. "storage_type,path",
  430. [
  431. ("pickle", "pickle_file_path"),
  432. ("csv", "csv_file"),
  433. ("excel", "excel_file"),
  434. ("json", "json_file"),
  435. ("parquet", "parquet_file_path"),
  436. ],
  437. )
  438. def test_clean_generated_files(self, storage_type, path, request):
  439. path = request.getfixturevalue(path)
  440. user_dn_config = Config.configure_data_node(
  441. id="d1", storage_type=storage_type, path=path, default_data={"a": [1], "b": [2]}
  442. )
  443. generated_dn_1_config = Config.configure_data_node(
  444. id="d2", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  445. )
  446. generated_dn_2_config = Config.configure_data_node(
  447. id="d3", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  448. )
  449. dns = _DataManager._bulk_get_or_create([user_dn_config, generated_dn_1_config, generated_dn_2_config])
  450. user_dn = dns[user_dn_config]
  451. generated_dn_1 = dns[generated_dn_1_config]
  452. generated_dn_2 = dns[generated_dn_2_config]
  453. _DataManager._clean_generated_file(user_dn.id)
  454. assert file_exists(user_dn.path)
  455. _DataManager._clean_generated_files([generated_dn_1, generated_dn_2])
  456. assert not file_exists(generated_dn_1.path)
  457. assert not file_exists(generated_dn_2.path)
  458. @pytest.mark.parametrize(
  459. "storage_type,path",
  460. [
  461. ("pickle", "pickle_file_path"),
  462. ("csv", "csv_file"),
  463. ("excel", "excel_file"),
  464. ("json", "json_file"),
  465. ("parquet", "parquet_file_path"),
  466. ],
  467. )
  468. def test_delete_does_clean_generated_pickle_files(self, storage_type, path, request):
  469. path = request.getfixturevalue(path)
  470. user_dn_config = Config.configure_data_node(
  471. id="d1", storage_type=storage_type, path=path, default_data={"a": [1], "b": [2]}
  472. )
  473. generated_dn_config_1 = Config.configure_data_node(
  474. id="d2", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  475. )
  476. generated_dn_config_2 = Config.configure_data_node(
  477. id="d3", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  478. )
  479. generated_dn_config_3 = Config.configure_data_node(
  480. id="d4", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  481. )
  482. dns = _DataManager._bulk_get_or_create(
  483. [
  484. user_dn_config,
  485. generated_dn_config_1,
  486. generated_dn_config_2,
  487. generated_dn_config_3,
  488. ]
  489. )
  490. user_dn = dns[user_dn_config]
  491. generated_dn_1 = dns[generated_dn_config_1]
  492. generated_dn_2 = dns[generated_dn_config_2]
  493. generated_dn_3 = dns[generated_dn_config_3]
  494. _DataManager._delete(user_dn.id)
  495. assert file_exists(user_dn.path)
  496. _DataManager._delete_many([generated_dn_1.id, generated_dn_2.id])
  497. assert not file_exists(generated_dn_1.path)
  498. assert not file_exists(generated_dn_2.path)
  499. _DataManager._delete_all()
  500. assert not file_exists(generated_dn_3.path)
  501. def test_create_dn_from_loaded_config_no_scope(self):
  502. file_config = NamedTemporaryFile(
  503. """
  504. [TAIPY]
  505. [DATA_NODE.a]
  506. default_data = "4:int"
  507. [DATA_NODE.b]
  508. [TASK.t]
  509. function = "math.sqrt:function"
  510. inputs = [ "a:SECTION",]
  511. outputs = [ "b:SECTION",]
  512. skippable = "False:bool"
  513. [SCENARIO.s]
  514. tasks = [ "t:SECTION",]
  515. sequences.s_sequence = [ "t:SECTION",]
  516. [SCENARIO.s.comparators]
  517. """
  518. )
  519. from taipy import core as tp
  520. Config.override(file_config.filename)
  521. tp.create_scenario(Config.scenarios["s"])
  522. tp.create_scenario(Config.scenarios["s"])
  523. assert len(tp.get_data_nodes()) == 4
  524. def test_create_dn_from_loaded_config_no_storage_type(self):
  525. file_config = NamedTemporaryFile(
  526. """
  527. [TAIPY]
  528. [DATA_NODE.input_dn]
  529. scope = "SCENARIO:SCOPE"
  530. default_data = "21:int"
  531. [DATA_NODE.output_dn]
  532. storage_type = "in_memory"
  533. scope = "SCENARIO:SCOPE"
  534. [TASK.double]
  535. inputs = [ "input_dn:SECTION",]
  536. function = "math.sqrt:function"
  537. outputs = [ "output_dn:SECTION",]
  538. skippable = "False:bool"
  539. [SCENARIO.my_scenario]
  540. tasks = [ "double:SECTION",]
  541. sequences.my_sequence = [ "double:SECTION",]
  542. [SCENARIO.my_scenario.comparators]
  543. """
  544. )
  545. from taipy import core as tp
  546. Config.override(file_config.filename)
  547. scenario = tp.create_scenario(Config.scenarios["my_scenario"])
  548. assert isinstance(scenario.input_dn, PickleDataNode)
  549. assert isinstance(scenario.output_dn, InMemoryDataNode)
  550. def test_create_dn_from_loaded_config_modified_default_config(self):
  551. file_config = NamedTemporaryFile(
  552. """
  553. [TAIPY]
  554. [DATA_NODE.input_dn]
  555. scope = "SCENARIO:SCOPE"
  556. default_path="fake/path.csv"
  557. [DATA_NODE.output_dn]
  558. storage_type = "in_memory"
  559. scope = "SCENARIO:SCOPE"
  560. [TASK.double]
  561. inputs = [ "input_dn:SECTION",]
  562. function = "math.sqrt:function"
  563. outputs = [ "output_dn:SECTION",]
  564. skippable = "False:bool"
  565. [SCENARIO.my_scenario]
  566. tasks = [ "double:SECTION",]
  567. sequences.my_sequence = [ "double:SECTION",]
  568. [SCENARIO.my_scenario.comparators]
  569. """
  570. )
  571. from taipy import core as tp
  572. Config.set_default_data_node_configuration(storage_type="csv")
  573. Config.override(file_config.filename)
  574. scenario = tp.create_scenario(Config.scenarios["my_scenario"])
  575. assert isinstance(scenario.input_dn, CSVDataNode)
  576. assert isinstance(scenario.output_dn, InMemoryDataNode)
  577. def test_get_tasks_by_config_id(self):
  578. dn_config_1 = Config.configure_data_node("dn_1", scope=Scope.SCENARIO)
  579. dn_config_2 = Config.configure_data_node("dn_2", scope=Scope.SCENARIO)
  580. dn_config_3 = Config.configure_data_node("dn_3", scope=Scope.SCENARIO)
  581. dn_1_1 = _DataManager._create_and_set(dn_config_1, None, None)
  582. dn_1_2 = _DataManager._create_and_set(dn_config_1, None, None)
  583. dn_1_3 = _DataManager._create_and_set(dn_config_1, None, None)
  584. assert len(_DataManager._get_all()) == 3
  585. dn_2_1 = _DataManager._create_and_set(dn_config_2, None, None)
  586. dn_2_2 = _DataManager._create_and_set(dn_config_2, None, None)
  587. assert len(_DataManager._get_all()) == 5
  588. dn_3_1 = _DataManager._create_and_set(dn_config_3, None, None)
  589. assert len(_DataManager._get_all()) == 6
  590. dn_1_datanodes = _DataManager._get_by_config_id(dn_config_1.id)
  591. assert len(dn_1_datanodes) == 3
  592. assert sorted([dn_1_1.id, dn_1_2.id, dn_1_3.id]) == sorted([sequence.id for sequence in dn_1_datanodes])
  593. dn_2_datanodes = _DataManager._get_by_config_id(dn_config_2.id)
  594. assert len(dn_2_datanodes) == 2
  595. assert sorted([dn_2_1.id, dn_2_2.id]) == sorted([sequence.id for sequence in dn_2_datanodes])
  596. dn_3_datanodes = _DataManager._get_by_config_id(dn_config_3.id)
  597. assert len(dn_3_datanodes) == 1
  598. assert sorted([dn_3_1.id]) == sorted([sequence.id for sequence in dn_3_datanodes])
  599. def test_get_data_nodes_by_config_id_in_multiple_versions_environment(self):
  600. dn_config_1 = Config.configure_data_node("dn_1", scope=Scope.SCENARIO)
  601. dn_config_2 = Config.configure_data_node("dn_2", scope=Scope.SCENARIO)
  602. _VersionManager._set_experiment_version("1.0")
  603. _DataManager._create_and_set(dn_config_1, None, None)
  604. _DataManager._create_and_set(dn_config_1, None, None)
  605. _DataManager._create_and_set(dn_config_1, None, None)
  606. _DataManager._create_and_set(dn_config_2, None, None)
  607. _DataManager._create_and_set(dn_config_2, None, None)
  608. assert len(_DataManager._get_by_config_id(dn_config_1.id)) == 3
  609. assert len(_DataManager._get_by_config_id(dn_config_2.id)) == 2
  610. _VersionManager._set_experiment_version("2.0")
  611. _DataManager._create_and_set(dn_config_1, None, None)
  612. _DataManager._create_and_set(dn_config_1, None, None)
  613. _DataManager._create_and_set(dn_config_1, None, None)
  614. _DataManager._create_and_set(dn_config_2, None, None)
  615. _DataManager._create_and_set(dn_config_2, None, None)
  616. assert len(_DataManager._get_by_config_id(dn_config_1.id)) == 3
  617. assert len(_DataManager._get_by_config_id(dn_config_2.id)) == 2