test_data_manager.py 34 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730
  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 taipy.core.reason import NotGlobalScope, WrongConfigType
  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_can_create(self):
  41. dn_config = Config.configure_data_node("dn", 10, scope=Scope.SCENARIO)
  42. global_dn_config = Config.configure_data_node(
  43. id="global_dn", storage_type="in_memory", scope=Scope.GLOBAL, data=10
  44. )
  45. reasons = _DataManager._can_create()
  46. assert bool(reasons) is True
  47. assert reasons._reasons == {}
  48. reasons = _DataManager._can_create(global_dn_config)
  49. assert bool(reasons) is True
  50. assert reasons._reasons == {}
  51. reasons = _DataManager._can_create(dn_config)
  52. assert bool(reasons) is False
  53. assert isinstance(list(reasons._reasons[dn_config.id])[0], NotGlobalScope)
  54. assert list(reasons._reasons[dn_config.id])[0].reason == 'Data node config "dn" does not have GLOBAL scope'
  55. reasons = _DataManager._can_create(1)
  56. assert bool(reasons) is False
  57. assert isinstance(list(reasons._reasons["1"])[0], WrongConfigType)
  58. assert list(reasons._reasons["1"])[0].reason == 'Object "1" must be a valid DataNodeConfig'
  59. def test_create_data_node_with_name_provided(self):
  60. dn_config = Config.configure_data_node(id="dn", foo="bar", name="acb")
  61. dn = _DataManager._create_and_set(dn_config, None, None)
  62. assert dn.name == "acb"
  63. def test_create_and_get_csv_data_node(self):
  64. # Test we can instantiate a CsvDataNode from DataNodeConfig with :
  65. # - a csv type
  66. # - a default scenario scope
  67. # - No owner_id
  68. csv_dn_config = Config.configure_data_node(id="foo", storage_type="csv", path="bar", has_header=True)
  69. csv_dn = _DataManager._create_and_set(csv_dn_config, None, None)
  70. assert isinstance(csv_dn, CSVDataNode)
  71. assert isinstance(_DataManager._get(csv_dn.id), CSVDataNode)
  72. assert _DataManager._exists(csv_dn.id)
  73. assert _DataManager._get(csv_dn.id) is not None
  74. assert _DataManager._get(csv_dn.id).id == csv_dn.id
  75. assert _DataManager._get(csv_dn.id).config_id == "foo"
  76. assert _DataManager._get(csv_dn.id).config_id == csv_dn.config_id
  77. assert _DataManager._get(csv_dn.id).scope == Scope.SCENARIO
  78. assert _DataManager._get(csv_dn.id).scope == csv_dn.scope
  79. assert _DataManager._get(csv_dn.id).owner_id is None
  80. assert _DataManager._get(csv_dn.id).owner_id == csv_dn.owner_id
  81. assert _DataManager._get(csv_dn.id).parent_ids == set()
  82. assert _DataManager._get(csv_dn.id).parent_ids == csv_dn.parent_ids
  83. assert _DataManager._get(csv_dn.id).last_edit_date is None
  84. assert _DataManager._get(csv_dn.id).last_edit_date == csv_dn.last_edit_date
  85. assert _DataManager._get(csv_dn.id).job_ids == []
  86. assert _DataManager._get(csv_dn.id).job_ids == csv_dn.job_ids
  87. assert not _DataManager._get(csv_dn.id).is_ready_for_reading
  88. assert _DataManager._get(csv_dn.id).is_ready_for_reading == csv_dn.is_ready_for_reading
  89. assert (
  90. len(_DataManager._get(csv_dn.id).properties) == 5
  91. ) # path, encoding, has_header, exposed_type, is_generated
  92. assert _DataManager._get(csv_dn.id).properties.get("path") == "bar"
  93. assert _DataManager._get(csv_dn.id).properties.get("encoding") == "utf-8"
  94. assert _DataManager._get(csv_dn.id).properties.get("has_header") is True
  95. assert _DataManager._get(csv_dn.id).properties.get("exposed_type") == "pandas"
  96. assert _DataManager._get(csv_dn.id).properties.get("is_generated") is False
  97. assert _DataManager._get(csv_dn.id).properties == csv_dn.properties
  98. assert _DataManager._get(csv_dn.id).edit_in_progress is False
  99. assert _DataManager._get(csv_dn.id)._editor_id is None
  100. assert _DataManager._get(csv_dn.id)._editor_expiration_date is None
  101. assert _DataManager._get(csv_dn) is not None
  102. assert _DataManager._get(csv_dn).id == csv_dn.id
  103. assert _DataManager._get(csv_dn).config_id == "foo"
  104. assert _DataManager._get(csv_dn).config_id == csv_dn.config_id
  105. assert _DataManager._get(csv_dn).scope == Scope.SCENARIO
  106. assert _DataManager._get(csv_dn).scope == csv_dn.scope
  107. assert _DataManager._get(csv_dn).owner_id is None
  108. assert _DataManager._get(csv_dn).owner_id == csv_dn.owner_id
  109. assert _DataManager._get(csv_dn).parent_ids == set()
  110. assert _DataManager._get(csv_dn).parent_ids == csv_dn.parent_ids
  111. assert _DataManager._get(csv_dn).last_edit_date is None
  112. assert _DataManager._get(csv_dn).last_edit_date == csv_dn.last_edit_date
  113. assert _DataManager._get(csv_dn).job_ids == []
  114. assert _DataManager._get(csv_dn).job_ids == csv_dn.job_ids
  115. assert not _DataManager._get(csv_dn).is_ready_for_reading
  116. assert _DataManager._get(csv_dn).is_ready_for_reading == csv_dn.is_ready_for_reading
  117. assert len(_DataManager._get(csv_dn).properties) == 5 # path, encoding, has_header, exposed_type, is_generated
  118. assert _DataManager._get(csv_dn).properties.get("path") == "bar"
  119. assert _DataManager._get(csv_dn).properties.get("encoding") == "utf-8"
  120. assert _DataManager._get(csv_dn).properties.get("has_header") is True
  121. assert _DataManager._get(csv_dn.id).properties.get("exposed_type") == "pandas"
  122. assert _DataManager._get(csv_dn.id).properties.get("is_generated") is False
  123. assert _DataManager._get(csv_dn).properties == csv_dn.properties
  124. assert _DataManager._get(csv_dn.id).edit_in_progress is False
  125. assert _DataManager._get(csv_dn.id)._editor_id is None
  126. assert _DataManager._get(csv_dn.id)._editor_expiration_date is None
  127. def test_edit_and_get_data_node(self):
  128. config = Config.configure_pickle_data_node(id="foo")
  129. dn = _DataManager._create_and_set(config, None, None)
  130. assert _DataManager._get(dn.id).last_edit_date is None
  131. assert len(_DataManager._get(dn.id).properties) == 2 # is_generated and path
  132. assert isinstance(_DataManager._get(dn.id).properties.get("path"), str)
  133. assert _DataManager._get(dn.id).properties.get("is_generated") is True
  134. assert not _DataManager._get(dn.id).edit_in_progress
  135. assert _DataManager._get(dn.id)._editor_id is None
  136. assert _DataManager._get(dn.id)._editor_expiration_date is None
  137. dn.lock_edit("foo")
  138. assert _DataManager._get(dn.id).last_edit_date is None
  139. assert len(_DataManager._get(dn.id).properties) == 2 # is_generated and path
  140. assert isinstance(_DataManager._get(dn.id).properties.get("path"), str)
  141. assert _DataManager._get(dn.id).properties.get("is_generated") is True
  142. assert _DataManager._get(dn.id).edit_in_progress
  143. assert _DataManager._get(dn.id).editor_id == "foo"
  144. assert _DataManager._get(dn.id).editor_expiration_date is not None
  145. dn.unlock_edit("foo")
  146. assert _DataManager._get(dn.id).last_edit_date is None
  147. assert len(_DataManager._get(dn.id).properties) == 2 # is_generated and path
  148. assert isinstance(_DataManager._get(dn.id).properties.get("path"), str)
  149. assert _DataManager._get(dn.id).properties.get("is_generated") is True
  150. assert not _DataManager._get(dn.id).edit_in_progress
  151. assert _DataManager._get(dn.id).editor_id is None
  152. assert _DataManager._get(dn.id).editor_expiration_date is None
  153. def test_create_and_get_in_memory_data_node(self):
  154. # Test we can instantiate an InMemoryDataNode from DataNodeConfig with :
  155. # - an in_memory type
  156. # - a scenario scope
  157. # - an owner id
  158. # - some default data
  159. in_memory_dn_config = Config.configure_data_node(
  160. id="baz", storage_type="in_memory", scope=Scope.SCENARIO, default_data="qux", other_data="foo"
  161. )
  162. in_mem_dn = _DataManager._create_and_set(in_memory_dn_config, "Scenario_id", {"task_id"})
  163. assert isinstance(in_mem_dn, InMemoryDataNode)
  164. assert isinstance(_DataManager._get(in_mem_dn.id), InMemoryDataNode)
  165. assert _DataManager._exists(in_mem_dn.id)
  166. assert _DataManager._get(in_mem_dn.id) is not None
  167. assert _DataManager._get(in_mem_dn.id).id == in_mem_dn.id
  168. assert _DataManager._get(in_mem_dn.id).config_id == "baz"
  169. assert _DataManager._get(in_mem_dn.id).config_id == in_mem_dn.config_id
  170. assert _DataManager._get(in_mem_dn.id).scope == Scope.SCENARIO
  171. assert _DataManager._get(in_mem_dn.id).scope == in_mem_dn.scope
  172. assert _DataManager._get(in_mem_dn.id).owner_id == "Scenario_id"
  173. assert _DataManager._get(in_mem_dn.id).owner_id == in_mem_dn.owner_id
  174. assert _DataManager._get(in_mem_dn.id).parent_ids == {"task_id"}
  175. assert _DataManager._get(in_mem_dn.id).parent_ids == in_mem_dn.parent_ids
  176. assert _DataManager._get(in_mem_dn.id).last_edit_date is not None
  177. assert _DataManager._get(in_mem_dn.id).last_edit_date == in_mem_dn.last_edit_date
  178. assert _DataManager._get(in_mem_dn.id).job_ids == []
  179. assert _DataManager._get(in_mem_dn.id).job_ids == in_mem_dn.job_ids
  180. assert _DataManager._get(in_mem_dn.id).is_ready_for_reading
  181. assert _DataManager._get(in_mem_dn.id).is_ready_for_reading == in_mem_dn.is_ready_for_reading
  182. assert len(_DataManager._get(in_mem_dn.id).properties) == 1
  183. assert _DataManager._get(in_mem_dn.id).properties.get("other_data") == "foo"
  184. assert _DataManager._get(in_mem_dn.id).properties == in_mem_dn.properties
  185. assert _DataManager._get(in_mem_dn) is not None
  186. assert _DataManager._get(in_mem_dn).id == in_mem_dn.id
  187. assert _DataManager._get(in_mem_dn).config_id == "baz"
  188. assert _DataManager._get(in_mem_dn).config_id == in_mem_dn.config_id
  189. assert _DataManager._get(in_mem_dn).scope == Scope.SCENARIO
  190. assert _DataManager._get(in_mem_dn).scope == in_mem_dn.scope
  191. assert _DataManager._get(in_mem_dn).owner_id == "Scenario_id"
  192. assert _DataManager._get(in_mem_dn).owner_id == in_mem_dn.owner_id
  193. assert _DataManager._get(in_mem_dn).parent_ids == {"task_id"}
  194. assert _DataManager._get(in_mem_dn).parent_ids == in_mem_dn.parent_ids
  195. assert _DataManager._get(in_mem_dn).last_edit_date is not None
  196. assert _DataManager._get(in_mem_dn).last_edit_date == in_mem_dn.last_edit_date
  197. assert _DataManager._get(in_mem_dn).job_ids == []
  198. assert _DataManager._get(in_mem_dn).job_ids == in_mem_dn.job_ids
  199. assert _DataManager._get(in_mem_dn).is_ready_for_reading
  200. assert _DataManager._get(in_mem_dn).is_ready_for_reading == in_mem_dn.is_ready_for_reading
  201. assert len(_DataManager._get(in_mem_dn).properties) == 1
  202. assert _DataManager._get(in_mem_dn).properties.get("other_data") == "foo"
  203. assert _DataManager._get(in_mem_dn).properties == in_mem_dn.properties
  204. def test_create_and_get_pickle_data_node(self):
  205. # Test we can instantiate a PickleDataNode from DataNodeConfig with :
  206. # - an in_memory type
  207. # - a business cycle scope
  208. # - No owner id
  209. # - no default data
  210. dn_config = Config.configure_data_node(id="plop", storage_type="pickle", scope=Scope.CYCLE)
  211. pickle_dn = _DataManager._create_and_set(dn_config, None, {"task_id_1", "task_id_2"})
  212. assert isinstance(pickle_dn, PickleDataNode)
  213. assert isinstance(_DataManager._get(pickle_dn.id), PickleDataNode)
  214. assert _DataManager._exists(pickle_dn.id)
  215. assert _DataManager._get(pickle_dn.id) is not None
  216. assert _DataManager._get(pickle_dn.id).id == pickle_dn.id
  217. assert _DataManager._get(pickle_dn.id).config_id == "plop"
  218. assert _DataManager._get(pickle_dn.id).config_id == pickle_dn.config_id
  219. assert _DataManager._get(pickle_dn.id).scope == Scope.CYCLE
  220. assert _DataManager._get(pickle_dn.id).scope == pickle_dn.scope
  221. assert _DataManager._get(pickle_dn.id).owner_id is None
  222. assert _DataManager._get(pickle_dn.id).owner_id == pickle_dn.owner_id
  223. assert _DataManager._get(pickle_dn.id).parent_ids == {"task_id_1", "task_id_2"}
  224. assert _DataManager._get(pickle_dn.id).parent_ids == pickle_dn.parent_ids
  225. assert _DataManager._get(pickle_dn.id).last_edit_date is None
  226. assert _DataManager._get(pickle_dn.id).last_edit_date == pickle_dn.last_edit_date
  227. assert _DataManager._get(pickle_dn.id).job_ids == []
  228. assert _DataManager._get(pickle_dn.id).job_ids == pickle_dn.job_ids
  229. assert not _DataManager._get(pickle_dn.id).is_ready_for_reading
  230. assert _DataManager._get(pickle_dn.id).is_ready_for_reading == pickle_dn.is_ready_for_reading
  231. assert len(_DataManager._get(pickle_dn.id).properties) == 2 # is_generated and path
  232. assert _DataManager._get(pickle_dn.id).properties == pickle_dn.properties
  233. assert _DataManager._get(pickle_dn) is not None
  234. assert _DataManager._get(pickle_dn).id == pickle_dn.id
  235. assert _DataManager._get(pickle_dn).config_id == "plop"
  236. assert _DataManager._get(pickle_dn).config_id == pickle_dn.config_id
  237. assert _DataManager._get(pickle_dn).scope == Scope.CYCLE
  238. assert _DataManager._get(pickle_dn).scope == pickle_dn.scope
  239. assert _DataManager._get(pickle_dn).owner_id is None
  240. assert _DataManager._get(pickle_dn).owner_id == pickle_dn.owner_id
  241. assert _DataManager._get(pickle_dn).parent_ids == {"task_id_1", "task_id_2"}
  242. assert _DataManager._get(pickle_dn).parent_ids == pickle_dn.parent_ids
  243. assert _DataManager._get(pickle_dn).last_edit_date is None
  244. assert _DataManager._get(pickle_dn).last_edit_date == pickle_dn.last_edit_date
  245. assert _DataManager._get(pickle_dn).job_ids == []
  246. assert _DataManager._get(pickle_dn).job_ids == pickle_dn.job_ids
  247. assert not _DataManager._get(pickle_dn).is_ready_for_reading
  248. assert _DataManager._get(pickle_dn).is_ready_for_reading == pickle_dn.is_ready_for_reading
  249. assert len(_DataManager._get(pickle_dn).properties) == 2 # is_generated and path
  250. assert _DataManager._get(pickle_dn).properties == pickle_dn.properties
  251. def test_create_raises_exception_with_wrong_type(self):
  252. wrong_type_dn_config = DataNodeConfig(id="foo", storage_type="bar", scope=DataNodeConfig._DEFAULT_SCOPE)
  253. with pytest.raises(InvalidDataNodeType):
  254. _DataManager._create_and_set(wrong_type_dn_config, None, None)
  255. def test_create_from_same_config_generates_new_data_node_and_new_id(self):
  256. dn_config = Config.configure_data_node(id="foo", storage_type="in_memory")
  257. dn = _DataManager._create_and_set(dn_config, None, None)
  258. dn_2 = _DataManager._create_and_set(dn_config, None, None)
  259. assert dn_2.id != dn.id
  260. def test_create_uses_overridden_attributes_in_config_file(self):
  261. Config.override(os.path.join(pathlib.Path(__file__).parent.resolve(), "data_sample/config.toml"))
  262. csv_dn_cfg = Config.configure_data_node(id="foo", 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 == "foo"
  265. assert isinstance(csv_dn, CSVDataNode)
  266. assert csv_dn._path == "path_from_config_file"
  267. assert csv_dn.has_header
  268. csv_dn_cfg = Config.configure_data_node(id="baz", storage_type="csv", path="bar", has_header=True)
  269. csv_dn = _DataManager._create_and_set(csv_dn_cfg, None, None)
  270. assert csv_dn.config_id == "baz"
  271. assert isinstance(csv_dn, CSVDataNode)
  272. assert csv_dn._path == "bar"
  273. assert csv_dn.has_header
  274. def test_get_if_not_exists(self):
  275. with pytest.raises(ModelNotFound):
  276. _DataManager._repository._load("test_data_node_2")
  277. def test_get_all(self):
  278. assert len(_DataManager._get_all()) == 0
  279. dn_config_1 = Config.configure_data_node(id="foo", storage_type="in_memory")
  280. _DataManager._create_and_set(dn_config_1, None, None)
  281. assert len(_DataManager._get_all()) == 1
  282. dn_config_2 = Config.configure_data_node(id="baz", storage_type="in_memory")
  283. _DataManager._create_and_set(dn_config_2, None, None)
  284. _DataManager._create_and_set(dn_config_2, None, None)
  285. assert len(_DataManager._get_all()) == 3
  286. assert len([dn for dn in _DataManager._get_all() if dn.config_id == "foo"]) == 1
  287. assert len([dn for dn in _DataManager._get_all() if dn.config_id == "baz"]) == 2
  288. def test_get_all_on_multiple_versions_environment(self):
  289. # Create 5 data nodes with 2 versions each
  290. # Only version 1.0 has the data node with config_id = "config_id_1"
  291. # Only version 2.0 has the data node with config_id = "config_id_6"
  292. for version in range(1, 3):
  293. for i in range(5):
  294. _DataManager._set(
  295. InMemoryDataNode(
  296. f"config_id_{i + version}",
  297. Scope.SCENARIO,
  298. id=DataNodeId(f"id{i}_v{version}"),
  299. version=f"{version}.0",
  300. )
  301. )
  302. _VersionManager._set_experiment_version("1.0")
  303. assert len(_DataManager._get_all()) == 5
  304. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_1"}])) == 1
  305. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_6"}])) == 0
  306. _VersionManager._set_development_version("1.0")
  307. assert len(_DataManager._get_all()) == 5
  308. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_1"}])) == 1
  309. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_6"}])) == 0
  310. _VersionManager._set_experiment_version("2.0")
  311. assert len(_DataManager._get_all()) == 5
  312. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_1"}])) == 0
  313. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_6"}])) == 1
  314. _VersionManager._set_development_version("2.0")
  315. assert len(_DataManager._get_all()) == 5
  316. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_1"}])) == 0
  317. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_6"}])) == 1
  318. def test_set(self):
  319. dn = InMemoryDataNode(
  320. "config_id",
  321. Scope.SCENARIO,
  322. id=DataNodeId("id"),
  323. owner_id=None,
  324. parent_ids={"task_id_1"},
  325. last_edit_date=None,
  326. edits=[],
  327. edit_in_progress=False,
  328. properties={"foo": "bar"},
  329. )
  330. assert len(_DataManager._get_all()) == 0
  331. assert not _DataManager._exists(dn.id)
  332. _DataManager._set(dn)
  333. assert len(_DataManager._get_all()) == 1
  334. assert _DataManager._exists(dn.id)
  335. # changing data node attribute
  336. dn._config_id = "foo"
  337. assert dn.config_id == "foo"
  338. _DataManager._set(dn)
  339. assert len(_DataManager._get_all()) == 1
  340. assert dn.config_id == "foo"
  341. assert _DataManager._get(dn.id).config_id == "foo"
  342. def test_delete(self):
  343. dn_1 = InMemoryDataNode("config_id", Scope.SCENARIO, id="id_1")
  344. dn_2 = InMemoryDataNode("config_id", Scope.SCENARIO, id="id_2")
  345. dn_3 = InMemoryDataNode("config_id", Scope.SCENARIO, id="id_3")
  346. assert len(_DataManager._get_all()) == 0
  347. _DataManager._set(dn_1)
  348. _DataManager._set(dn_2)
  349. _DataManager._set(dn_3)
  350. assert len(_DataManager._get_all()) == 3
  351. assert all(_DataManager._exists(dn.id) for dn in [dn_1, dn_2, dn_3])
  352. _DataManager._delete(dn_1.id)
  353. assert len(_DataManager._get_all()) == 2
  354. assert _DataManager._get(dn_2.id).id == dn_2.id
  355. assert _DataManager._get(dn_3.id).id == dn_3.id
  356. assert _DataManager._get(dn_1.id) is None
  357. assert all(_DataManager._exists(dn.id) for dn in [dn_2, dn_3])
  358. assert not _DataManager._exists(dn_1.id)
  359. _DataManager._delete_all()
  360. assert len(_DataManager._get_all()) == 0
  361. assert not any(_DataManager._exists(dn.id) for dn in [dn_2, dn_3])
  362. def test_get_or_create(self):
  363. def _get_or_create_dn(config, *args):
  364. return _DataManager._bulk_get_or_create([config], *args)[config]
  365. _DataManager._delete_all()
  366. global_dn_config = Config.configure_data_node(
  367. id="test_data_node", storage_type="in_memory", scope=Scope.GLOBAL, data="In memory Data Node"
  368. )
  369. cycle_dn_config = Config.configure_data_node(
  370. id="test_data_node1", storage_type="in_memory", scope=Scope.CYCLE, data="In memory Data Node"
  371. )
  372. scenario_dn_config = Config.configure_data_node(
  373. id="test_data_node2", storage_type="in_memory", scope=Scope.SCENARIO, data="In memory scenario"
  374. )
  375. assert len(_DataManager._get_all()) == 0
  376. global_dn = _get_or_create_dn(global_dn_config, None, None)
  377. assert len(_DataManager._get_all()) == 1
  378. global_dn_bis = _get_or_create_dn(global_dn_config, None)
  379. assert len(_DataManager._get_all()) == 1
  380. assert global_dn.id == global_dn_bis.id
  381. scenario_dn = _get_or_create_dn(scenario_dn_config, None, "scenario_id")
  382. assert len(_DataManager._get_all()) == 2
  383. scenario_dn_bis = _get_or_create_dn(scenario_dn_config, None, "scenario_id")
  384. assert len(_DataManager._get_all()) == 2
  385. assert scenario_dn.id == scenario_dn_bis.id
  386. scenario_dn_ter = _get_or_create_dn(scenario_dn_config, None, "scenario_id")
  387. assert len(_DataManager._get_all()) == 2
  388. assert scenario_dn.id == scenario_dn_bis.id
  389. assert scenario_dn_bis.id == scenario_dn_ter.id
  390. scenario_dn_quater = _get_or_create_dn(scenario_dn_config, None, "scenario_id_2")
  391. assert len(_DataManager._get_all()) == 3
  392. assert scenario_dn.id == scenario_dn_bis.id
  393. assert scenario_dn_bis.id == scenario_dn_ter.id
  394. assert scenario_dn_ter.id != scenario_dn_quater.id
  395. assert len(_DataManager._get_all()) == 3
  396. cycle_dn = _get_or_create_dn(cycle_dn_config, "cycle_id", None)
  397. assert len(_DataManager._get_all()) == 4
  398. cycle_dn_1 = _get_or_create_dn(cycle_dn_config, "cycle_id", None)
  399. assert len(_DataManager._get_all()) == 4
  400. assert cycle_dn.id == cycle_dn_1.id
  401. cycle_dn_2 = _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_2.id
  404. cycle_dn_3 = _get_or_create_dn(cycle_dn_config, "cycle_id", None)
  405. assert len(_DataManager._get_all()) == 4
  406. assert cycle_dn.id == cycle_dn_3.id
  407. cycle_dn_4 = _get_or_create_dn(cycle_dn_config, "cycle_id", "scenario_id")
  408. assert len(_DataManager._get_all()) == 4
  409. assert cycle_dn.id == cycle_dn_4.id
  410. cycle_dn_5 = _get_or_create_dn(cycle_dn_config, "cycle_id", "scenario_id_2")
  411. assert len(_DataManager._get_all()) == 4
  412. assert cycle_dn.id == cycle_dn_5.id
  413. assert cycle_dn_1.id == cycle_dn_2.id
  414. assert cycle_dn_2.id == cycle_dn_3.id
  415. assert cycle_dn_3.id == cycle_dn_4.id
  416. assert cycle_dn_4.id == cycle_dn_5.id
  417. def test_ensure_persistence_of_data_node(self):
  418. dm = _DataManager()
  419. dm._delete_all()
  420. dn_config_1 = Config.configure_data_node(
  421. id="data_node_1", storage_type="in_memory", data="In memory sequence 2"
  422. )
  423. dn_config_2 = Config.configure_data_node(
  424. id="data_node_2", storage_type="in_memory", data="In memory sequence 2"
  425. )
  426. dm._bulk_get_or_create([dn_config_1, dn_config_2])
  427. assert len(dm._get_all()) == 2
  428. # Delete the DataManager to ensure it's get from the storage system
  429. del dm
  430. dm = _DataManager()
  431. dm._bulk_get_or_create([dn_config_1])
  432. assert len(dm._get_all()) == 2
  433. dm._delete_all()
  434. @pytest.mark.parametrize(
  435. "storage_type,path",
  436. [
  437. ("pickle", "pickle_file_path"),
  438. ("csv", "csv_file"),
  439. ("excel", "excel_file"),
  440. ("json", "json_file"),
  441. ("parquet", "parquet_file_path"),
  442. ],
  443. )
  444. def test_clean_generated_files(self, storage_type, path, request):
  445. path = request.getfixturevalue(path)
  446. user_dn_config = Config.configure_data_node(
  447. id="d1", storage_type=storage_type, path=path, default_data={"a": [1], "b": [2]}
  448. )
  449. generated_dn_1_config = Config.configure_data_node(
  450. id="d2", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  451. )
  452. generated_dn_2_config = Config.configure_data_node(
  453. id="d3", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  454. )
  455. dns = _DataManager._bulk_get_or_create([user_dn_config, generated_dn_1_config, generated_dn_2_config])
  456. user_dn = dns[user_dn_config]
  457. generated_dn_1 = dns[generated_dn_1_config]
  458. generated_dn_2 = dns[generated_dn_2_config]
  459. _DataManager._clean_generated_file(user_dn.id)
  460. assert file_exists(user_dn.path)
  461. _DataManager._clean_generated_files([generated_dn_1, generated_dn_2])
  462. assert not file_exists(generated_dn_1.path)
  463. assert not file_exists(generated_dn_2.path)
  464. @pytest.mark.parametrize(
  465. "storage_type,path",
  466. [
  467. ("pickle", "pickle_file_path"),
  468. ("csv", "csv_file"),
  469. ("excel", "excel_file"),
  470. ("json", "json_file"),
  471. ("parquet", "parquet_file_path"),
  472. ],
  473. )
  474. def test_delete_does_clean_generated_pickle_files(self, storage_type, path, request):
  475. path = request.getfixturevalue(path)
  476. user_dn_config = Config.configure_data_node(
  477. id="d1", storage_type=storage_type, path=path, default_data={"a": [1], "b": [2]}
  478. )
  479. generated_dn_config_1 = Config.configure_data_node(
  480. id="d2", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  481. )
  482. generated_dn_config_2 = Config.configure_data_node(
  483. id="d3", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  484. )
  485. generated_dn_config_3 = Config.configure_data_node(
  486. id="d4", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  487. )
  488. dns = _DataManager._bulk_get_or_create(
  489. [
  490. user_dn_config,
  491. generated_dn_config_1,
  492. generated_dn_config_2,
  493. generated_dn_config_3,
  494. ]
  495. )
  496. user_dn = dns[user_dn_config]
  497. generated_dn_1 = dns[generated_dn_config_1]
  498. generated_dn_2 = dns[generated_dn_config_2]
  499. generated_dn_3 = dns[generated_dn_config_3]
  500. _DataManager._delete(user_dn.id)
  501. assert file_exists(user_dn.path)
  502. _DataManager._delete_many([generated_dn_1.id, generated_dn_2.id])
  503. assert not file_exists(generated_dn_1.path)
  504. assert not file_exists(generated_dn_2.path)
  505. _DataManager._delete_all()
  506. assert not file_exists(generated_dn_3.path)
  507. def test_create_dn_from_loaded_config_no_scope(self):
  508. file_config = NamedTemporaryFile(
  509. """
  510. [TAIPY]
  511. [DATA_NODE.a]
  512. default_data = "4:int"
  513. [DATA_NODE.b]
  514. [TASK.t]
  515. function = "math.sqrt:function"
  516. inputs = [ "a:SECTION",]
  517. outputs = [ "b:SECTION",]
  518. skippable = "False:bool"
  519. [SCENARIO.s]
  520. tasks = [ "t:SECTION",]
  521. sequences.s_sequence = [ "t:SECTION",]
  522. [SCENARIO.s.comparators]
  523. """
  524. )
  525. from taipy import core as tp
  526. Config.override(file_config.filename)
  527. tp.create_scenario(Config.scenarios["s"])
  528. tp.create_scenario(Config.scenarios["s"])
  529. assert len(tp.get_data_nodes()) == 4
  530. def test_create_dn_from_loaded_config_no_storage_type(self):
  531. file_config = NamedTemporaryFile(
  532. """
  533. [TAIPY]
  534. [DATA_NODE.input_dn]
  535. scope = "SCENARIO:SCOPE"
  536. default_data = "21:int"
  537. [DATA_NODE.output_dn]
  538. storage_type = "in_memory"
  539. scope = "SCENARIO:SCOPE"
  540. [TASK.double]
  541. inputs = [ "input_dn:SECTION",]
  542. function = "math.sqrt:function"
  543. outputs = [ "output_dn:SECTION",]
  544. skippable = "False:bool"
  545. [SCENARIO.my_scenario]
  546. tasks = [ "double:SECTION",]
  547. sequences.my_sequence = [ "double:SECTION",]
  548. [SCENARIO.my_scenario.comparators]
  549. """
  550. )
  551. from taipy import core as tp
  552. Config.override(file_config.filename)
  553. scenario = tp.create_scenario(Config.scenarios["my_scenario"])
  554. assert isinstance(scenario.input_dn, PickleDataNode)
  555. assert isinstance(scenario.output_dn, InMemoryDataNode)
  556. def test_create_dn_from_loaded_config_modified_default_config(self):
  557. file_config = NamedTemporaryFile(
  558. """
  559. [TAIPY]
  560. [DATA_NODE.input_dn]
  561. scope = "SCENARIO:SCOPE"
  562. default_path="fake/path.csv"
  563. [DATA_NODE.output_dn]
  564. storage_type = "in_memory"
  565. scope = "SCENARIO:SCOPE"
  566. [TASK.double]
  567. inputs = [ "input_dn:SECTION",]
  568. function = "math.sqrt:function"
  569. outputs = [ "output_dn:SECTION",]
  570. skippable = "False:bool"
  571. [SCENARIO.my_scenario]
  572. tasks = [ "double:SECTION",]
  573. sequences.my_sequence = [ "double:SECTION",]
  574. [SCENARIO.my_scenario.comparators]
  575. """
  576. )
  577. from taipy import core as tp
  578. Config.set_default_data_node_configuration(storage_type="csv")
  579. Config.override(file_config.filename)
  580. scenario = tp.create_scenario(Config.scenarios["my_scenario"])
  581. assert isinstance(scenario.input_dn, CSVDataNode)
  582. assert isinstance(scenario.output_dn, InMemoryDataNode)
  583. def test_get_tasks_by_config_id(self):
  584. dn_config_1 = Config.configure_data_node("dn_1", scope=Scope.SCENARIO)
  585. dn_config_2 = Config.configure_data_node("dn_2", scope=Scope.SCENARIO)
  586. dn_config_3 = Config.configure_data_node("dn_3", scope=Scope.SCENARIO)
  587. dn_1_1 = _DataManager._create_and_set(dn_config_1, None, None)
  588. dn_1_2 = _DataManager._create_and_set(dn_config_1, None, None)
  589. dn_1_3 = _DataManager._create_and_set(dn_config_1, None, None)
  590. assert len(_DataManager._get_all()) == 3
  591. dn_2_1 = _DataManager._create_and_set(dn_config_2, None, None)
  592. dn_2_2 = _DataManager._create_and_set(dn_config_2, None, None)
  593. assert len(_DataManager._get_all()) == 5
  594. dn_3_1 = _DataManager._create_and_set(dn_config_3, None, None)
  595. assert len(_DataManager._get_all()) == 6
  596. dn_1_datanodes = _DataManager._get_by_config_id(dn_config_1.id)
  597. assert len(dn_1_datanodes) == 3
  598. assert sorted([dn_1_1.id, dn_1_2.id, dn_1_3.id]) == sorted([sequence.id for sequence in dn_1_datanodes])
  599. dn_2_datanodes = _DataManager._get_by_config_id(dn_config_2.id)
  600. assert len(dn_2_datanodes) == 2
  601. assert sorted([dn_2_1.id, dn_2_2.id]) == sorted([sequence.id for sequence in dn_2_datanodes])
  602. dn_3_datanodes = _DataManager._get_by_config_id(dn_config_3.id)
  603. assert len(dn_3_datanodes) == 1
  604. assert sorted([dn_3_1.id]) == sorted([sequence.id for sequence in dn_3_datanodes])
  605. def test_get_data_nodes_by_config_id_in_multiple_versions_environment(self):
  606. dn_config_1 = Config.configure_data_node("dn_1", scope=Scope.SCENARIO)
  607. dn_config_2 = Config.configure_data_node("dn_2", scope=Scope.SCENARIO)
  608. _VersionManager._set_experiment_version("1.0")
  609. _DataManager._create_and_set(dn_config_1, None, None)
  610. _DataManager._create_and_set(dn_config_1, None, None)
  611. _DataManager._create_and_set(dn_config_1, None, None)
  612. _DataManager._create_and_set(dn_config_2, None, None)
  613. _DataManager._create_and_set(dn_config_2, None, None)
  614. assert len(_DataManager._get_by_config_id(dn_config_1.id)) == 3
  615. assert len(_DataManager._get_by_config_id(dn_config_2.id)) == 2
  616. _VersionManager._set_experiment_version("2.0")
  617. _DataManager._create_and_set(dn_config_1, None, None)
  618. _DataManager._create_and_set(dn_config_1, None, None)
  619. _DataManager._create_and_set(dn_config_1, None, None)
  620. _DataManager._create_and_set(dn_config_2, None, None)
  621. _DataManager._create_and_set(dn_config_2, None, None)
  622. assert len(_DataManager._get_by_config_id(dn_config_1.id)) == 3
  623. assert len(_DataManager._get_by_config_id(dn_config_2.id)) == 2