test_data_manager.py 34 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727
  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_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()
  45. assert bool(reasons) is True
  46. assert reasons._reasons == {}
  47. reasons = _DataManager._can_create(global_dn_config)
  48. assert bool(reasons) is True
  49. assert reasons._reasons == {}
  50. reasons = _DataManager._can_create(dn_config)
  51. assert bool(reasons) is False
  52. assert reasons._reasons == {dn_config.id: {'Data node config "dn" does not have GLOBAL scope'}}
  53. reasons = _DataManager._can_create(1)
  54. assert bool(reasons) is False
  55. assert reasons._reasons == {"1": {'Object "1" must be a valid DataNodeConfig'}}
  56. def test_create_data_node_with_name_provided(self):
  57. dn_config = Config.configure_data_node(id="dn", foo="bar", name="acb")
  58. dn = _DataManager._create_and_set(dn_config, None, None)
  59. assert dn.name == "acb"
  60. def test_create_and_get_csv_data_node(self):
  61. # Test we can instantiate a CsvDataNode from DataNodeConfig with :
  62. # - a csv type
  63. # - a default scenario scope
  64. # - No owner_id
  65. csv_dn_config = Config.configure_data_node(id="foo", storage_type="csv", path="bar", has_header=True)
  66. csv_dn = _DataManager._create_and_set(csv_dn_config, None, None)
  67. assert isinstance(csv_dn, CSVDataNode)
  68. assert isinstance(_DataManager._get(csv_dn.id), CSVDataNode)
  69. assert _DataManager._exists(csv_dn.id)
  70. assert _DataManager._get(csv_dn.id) is not None
  71. assert _DataManager._get(csv_dn.id).id == csv_dn.id
  72. assert _DataManager._get(csv_dn.id).config_id == "foo"
  73. assert _DataManager._get(csv_dn.id).config_id == csv_dn.config_id
  74. assert _DataManager._get(csv_dn.id).scope == Scope.SCENARIO
  75. assert _DataManager._get(csv_dn.id).scope == csv_dn.scope
  76. assert _DataManager._get(csv_dn.id).owner_id is None
  77. assert _DataManager._get(csv_dn.id).owner_id == csv_dn.owner_id
  78. assert _DataManager._get(csv_dn.id).parent_ids == set()
  79. assert _DataManager._get(csv_dn.id).parent_ids == csv_dn.parent_ids
  80. assert _DataManager._get(csv_dn.id).last_edit_date is None
  81. assert _DataManager._get(csv_dn.id).last_edit_date == csv_dn.last_edit_date
  82. assert _DataManager._get(csv_dn.id).job_ids == []
  83. assert _DataManager._get(csv_dn.id).job_ids == csv_dn.job_ids
  84. assert not _DataManager._get(csv_dn.id).is_ready_for_reading
  85. assert _DataManager._get(csv_dn.id).is_ready_for_reading == csv_dn.is_ready_for_reading
  86. assert (
  87. len(_DataManager._get(csv_dn.id).properties) == 5
  88. ) # path, encoding, has_header, exposed_type, is_generated
  89. assert _DataManager._get(csv_dn.id).properties.get("path") == "bar"
  90. assert _DataManager._get(csv_dn.id).properties.get("encoding") == "utf-8"
  91. assert _DataManager._get(csv_dn.id).properties.get("has_header") is True
  92. assert _DataManager._get(csv_dn.id).properties.get("exposed_type") == "pandas"
  93. assert _DataManager._get(csv_dn.id).properties.get("is_generated") is False
  94. assert _DataManager._get(csv_dn.id).properties == csv_dn.properties
  95. assert _DataManager._get(csv_dn.id).edit_in_progress is False
  96. assert _DataManager._get(csv_dn.id)._editor_id is None
  97. assert _DataManager._get(csv_dn.id)._editor_expiration_date is None
  98. assert _DataManager._get(csv_dn) is not None
  99. assert _DataManager._get(csv_dn).id == csv_dn.id
  100. assert _DataManager._get(csv_dn).config_id == "foo"
  101. assert _DataManager._get(csv_dn).config_id == csv_dn.config_id
  102. assert _DataManager._get(csv_dn).scope == Scope.SCENARIO
  103. assert _DataManager._get(csv_dn).scope == csv_dn.scope
  104. assert _DataManager._get(csv_dn).owner_id is None
  105. assert _DataManager._get(csv_dn).owner_id == csv_dn.owner_id
  106. assert _DataManager._get(csv_dn).parent_ids == set()
  107. assert _DataManager._get(csv_dn).parent_ids == csv_dn.parent_ids
  108. assert _DataManager._get(csv_dn).last_edit_date is None
  109. assert _DataManager._get(csv_dn).last_edit_date == csv_dn.last_edit_date
  110. assert _DataManager._get(csv_dn).job_ids == []
  111. assert _DataManager._get(csv_dn).job_ids == csv_dn.job_ids
  112. assert not _DataManager._get(csv_dn).is_ready_for_reading
  113. assert _DataManager._get(csv_dn).is_ready_for_reading == csv_dn.is_ready_for_reading
  114. assert len(_DataManager._get(csv_dn).properties) == 5 # path, encoding, has_header, exposed_type, is_generated
  115. assert _DataManager._get(csv_dn).properties.get("path") == "bar"
  116. assert _DataManager._get(csv_dn).properties.get("encoding") == "utf-8"
  117. assert _DataManager._get(csv_dn).properties.get("has_header") is True
  118. assert _DataManager._get(csv_dn.id).properties.get("exposed_type") == "pandas"
  119. assert _DataManager._get(csv_dn.id).properties.get("is_generated") is False
  120. assert _DataManager._get(csv_dn).properties == csv_dn.properties
  121. assert _DataManager._get(csv_dn.id).edit_in_progress is False
  122. assert _DataManager._get(csv_dn.id)._editor_id is None
  123. assert _DataManager._get(csv_dn.id)._editor_expiration_date is None
  124. def test_edit_and_get_data_node(self):
  125. config = Config.configure_pickle_data_node(id="foo")
  126. dn = _DataManager._create_and_set(config, None, None)
  127. assert _DataManager._get(dn.id).last_edit_date is None
  128. assert len(_DataManager._get(dn.id).properties) == 2 # is_generated and path
  129. assert isinstance(_DataManager._get(dn.id).properties.get("path"), str)
  130. assert _DataManager._get(dn.id).properties.get("is_generated") is True
  131. assert not _DataManager._get(dn.id).edit_in_progress
  132. assert _DataManager._get(dn.id)._editor_id is None
  133. assert _DataManager._get(dn.id)._editor_expiration_date is None
  134. dn.lock_edit("foo")
  135. assert _DataManager._get(dn.id).last_edit_date is None
  136. assert len(_DataManager._get(dn.id).properties) == 2 # is_generated and path
  137. assert isinstance(_DataManager._get(dn.id).properties.get("path"), str)
  138. assert _DataManager._get(dn.id).properties.get("is_generated") is True
  139. assert _DataManager._get(dn.id).edit_in_progress
  140. assert _DataManager._get(dn.id).editor_id == "foo"
  141. assert _DataManager._get(dn.id).editor_expiration_date is not None
  142. dn.unlock_edit("foo")
  143. assert _DataManager._get(dn.id).last_edit_date is None
  144. assert len(_DataManager._get(dn.id).properties) == 2 # is_generated and path
  145. assert isinstance(_DataManager._get(dn.id).properties.get("path"), str)
  146. assert _DataManager._get(dn.id).properties.get("is_generated") is True
  147. assert not _DataManager._get(dn.id).edit_in_progress
  148. assert _DataManager._get(dn.id).editor_id is None
  149. assert _DataManager._get(dn.id).editor_expiration_date is None
  150. def test_create_and_get_in_memory_data_node(self):
  151. # Test we can instantiate an InMemoryDataNode from DataNodeConfig with :
  152. # - an in_memory type
  153. # - a scenario scope
  154. # - an owner id
  155. # - some default data
  156. in_memory_dn_config = Config.configure_data_node(
  157. id="baz", storage_type="in_memory", scope=Scope.SCENARIO, default_data="qux", other_data="foo"
  158. )
  159. in_mem_dn = _DataManager._create_and_set(in_memory_dn_config, "Scenario_id", {"task_id"})
  160. assert isinstance(in_mem_dn, InMemoryDataNode)
  161. assert isinstance(_DataManager._get(in_mem_dn.id), InMemoryDataNode)
  162. assert _DataManager._exists(in_mem_dn.id)
  163. assert _DataManager._get(in_mem_dn.id) is not None
  164. assert _DataManager._get(in_mem_dn.id).id == in_mem_dn.id
  165. assert _DataManager._get(in_mem_dn.id).config_id == "baz"
  166. assert _DataManager._get(in_mem_dn.id).config_id == in_mem_dn.config_id
  167. assert _DataManager._get(in_mem_dn.id).scope == Scope.SCENARIO
  168. assert _DataManager._get(in_mem_dn.id).scope == in_mem_dn.scope
  169. assert _DataManager._get(in_mem_dn.id).owner_id == "Scenario_id"
  170. assert _DataManager._get(in_mem_dn.id).owner_id == in_mem_dn.owner_id
  171. assert _DataManager._get(in_mem_dn.id).parent_ids == {"task_id"}
  172. assert _DataManager._get(in_mem_dn.id).parent_ids == in_mem_dn.parent_ids
  173. assert _DataManager._get(in_mem_dn.id).last_edit_date is not None
  174. assert _DataManager._get(in_mem_dn.id).last_edit_date == in_mem_dn.last_edit_date
  175. assert _DataManager._get(in_mem_dn.id).job_ids == []
  176. assert _DataManager._get(in_mem_dn.id).job_ids == in_mem_dn.job_ids
  177. assert _DataManager._get(in_mem_dn.id).is_ready_for_reading
  178. assert _DataManager._get(in_mem_dn.id).is_ready_for_reading == in_mem_dn.is_ready_for_reading
  179. assert len(_DataManager._get(in_mem_dn.id).properties) == 1
  180. assert _DataManager._get(in_mem_dn.id).properties.get("other_data") == "foo"
  181. assert _DataManager._get(in_mem_dn.id).properties == in_mem_dn.properties
  182. assert _DataManager._get(in_mem_dn) is not None
  183. assert _DataManager._get(in_mem_dn).id == in_mem_dn.id
  184. assert _DataManager._get(in_mem_dn).config_id == "baz"
  185. assert _DataManager._get(in_mem_dn).config_id == in_mem_dn.config_id
  186. assert _DataManager._get(in_mem_dn).scope == Scope.SCENARIO
  187. assert _DataManager._get(in_mem_dn).scope == in_mem_dn.scope
  188. assert _DataManager._get(in_mem_dn).owner_id == "Scenario_id"
  189. assert _DataManager._get(in_mem_dn).owner_id == in_mem_dn.owner_id
  190. assert _DataManager._get(in_mem_dn).parent_ids == {"task_id"}
  191. assert _DataManager._get(in_mem_dn).parent_ids == in_mem_dn.parent_ids
  192. assert _DataManager._get(in_mem_dn).last_edit_date is not None
  193. assert _DataManager._get(in_mem_dn).last_edit_date == in_mem_dn.last_edit_date
  194. assert _DataManager._get(in_mem_dn).job_ids == []
  195. assert _DataManager._get(in_mem_dn).job_ids == in_mem_dn.job_ids
  196. assert _DataManager._get(in_mem_dn).is_ready_for_reading
  197. assert _DataManager._get(in_mem_dn).is_ready_for_reading == in_mem_dn.is_ready_for_reading
  198. assert len(_DataManager._get(in_mem_dn).properties) == 1
  199. assert _DataManager._get(in_mem_dn).properties.get("other_data") == "foo"
  200. assert _DataManager._get(in_mem_dn).properties == in_mem_dn.properties
  201. def test_create_and_get_pickle_data_node(self):
  202. # Test we can instantiate a PickleDataNode from DataNodeConfig with :
  203. # - an in_memory type
  204. # - a business cycle scope
  205. # - No owner id
  206. # - no default data
  207. dn_config = Config.configure_data_node(id="plop", storage_type="pickle", scope=Scope.CYCLE)
  208. pickle_dn = _DataManager._create_and_set(dn_config, None, {"task_id_1", "task_id_2"})
  209. assert isinstance(pickle_dn, PickleDataNode)
  210. assert isinstance(_DataManager._get(pickle_dn.id), PickleDataNode)
  211. assert _DataManager._exists(pickle_dn.id)
  212. assert _DataManager._get(pickle_dn.id) is not None
  213. assert _DataManager._get(pickle_dn.id).id == pickle_dn.id
  214. assert _DataManager._get(pickle_dn.id).config_id == "plop"
  215. assert _DataManager._get(pickle_dn.id).config_id == pickle_dn.config_id
  216. assert _DataManager._get(pickle_dn.id).scope == Scope.CYCLE
  217. assert _DataManager._get(pickle_dn.id).scope == pickle_dn.scope
  218. assert _DataManager._get(pickle_dn.id).owner_id is None
  219. assert _DataManager._get(pickle_dn.id).owner_id == pickle_dn.owner_id
  220. assert _DataManager._get(pickle_dn.id).parent_ids == {"task_id_1", "task_id_2"}
  221. assert _DataManager._get(pickle_dn.id).parent_ids == pickle_dn.parent_ids
  222. assert _DataManager._get(pickle_dn.id).last_edit_date is None
  223. assert _DataManager._get(pickle_dn.id).last_edit_date == pickle_dn.last_edit_date
  224. assert _DataManager._get(pickle_dn.id).job_ids == []
  225. assert _DataManager._get(pickle_dn.id).job_ids == pickle_dn.job_ids
  226. assert not _DataManager._get(pickle_dn.id).is_ready_for_reading
  227. assert _DataManager._get(pickle_dn.id).is_ready_for_reading == pickle_dn.is_ready_for_reading
  228. assert len(_DataManager._get(pickle_dn.id).properties) == 2 # is_generated and path
  229. assert _DataManager._get(pickle_dn.id).properties == pickle_dn.properties
  230. assert _DataManager._get(pickle_dn) is not None
  231. assert _DataManager._get(pickle_dn).id == pickle_dn.id
  232. assert _DataManager._get(pickle_dn).config_id == "plop"
  233. assert _DataManager._get(pickle_dn).config_id == pickle_dn.config_id
  234. assert _DataManager._get(pickle_dn).scope == Scope.CYCLE
  235. assert _DataManager._get(pickle_dn).scope == pickle_dn.scope
  236. assert _DataManager._get(pickle_dn).owner_id is None
  237. assert _DataManager._get(pickle_dn).owner_id == pickle_dn.owner_id
  238. assert _DataManager._get(pickle_dn).parent_ids == {"task_id_1", "task_id_2"}
  239. assert _DataManager._get(pickle_dn).parent_ids == pickle_dn.parent_ids
  240. assert _DataManager._get(pickle_dn).last_edit_date is None
  241. assert _DataManager._get(pickle_dn).last_edit_date == pickle_dn.last_edit_date
  242. assert _DataManager._get(pickle_dn).job_ids == []
  243. assert _DataManager._get(pickle_dn).job_ids == pickle_dn.job_ids
  244. assert not _DataManager._get(pickle_dn).is_ready_for_reading
  245. assert _DataManager._get(pickle_dn).is_ready_for_reading == pickle_dn.is_ready_for_reading
  246. assert len(_DataManager._get(pickle_dn).properties) == 2 # is_generated and path
  247. assert _DataManager._get(pickle_dn).properties == pickle_dn.properties
  248. def test_create_raises_exception_with_wrong_type(self):
  249. wrong_type_dn_config = DataNodeConfig(id="foo", storage_type="bar", scope=DataNodeConfig._DEFAULT_SCOPE)
  250. with pytest.raises(InvalidDataNodeType):
  251. _DataManager._create_and_set(wrong_type_dn_config, None, None)
  252. def test_create_from_same_config_generates_new_data_node_and_new_id(self):
  253. dn_config = Config.configure_data_node(id="foo", storage_type="in_memory")
  254. dn = _DataManager._create_and_set(dn_config, None, None)
  255. dn_2 = _DataManager._create_and_set(dn_config, None, None)
  256. assert dn_2.id != dn.id
  257. def test_create_uses_overridden_attributes_in_config_file(self):
  258. Config.override(os.path.join(pathlib.Path(__file__).parent.resolve(), "data_sample/config.toml"))
  259. csv_dn_cfg = Config.configure_data_node(id="foo", storage_type="csv", path="bar", has_header=True)
  260. csv_dn = _DataManager._create_and_set(csv_dn_cfg, None, None)
  261. assert csv_dn.config_id == "foo"
  262. assert isinstance(csv_dn, CSVDataNode)
  263. assert csv_dn._path == "path_from_config_file"
  264. assert csv_dn.has_header
  265. csv_dn_cfg = Config.configure_data_node(id="baz", storage_type="csv", path="bar", has_header=True)
  266. csv_dn = _DataManager._create_and_set(csv_dn_cfg, None, None)
  267. assert csv_dn.config_id == "baz"
  268. assert isinstance(csv_dn, CSVDataNode)
  269. assert csv_dn._path == "bar"
  270. assert csv_dn.has_header
  271. def test_get_if_not_exists(self):
  272. with pytest.raises(ModelNotFound):
  273. _DataManager._repository._load("test_data_node_2")
  274. def test_get_all(self):
  275. assert len(_DataManager._get_all()) == 0
  276. dn_config_1 = Config.configure_data_node(id="foo", storage_type="in_memory")
  277. _DataManager._create_and_set(dn_config_1, None, None)
  278. assert len(_DataManager._get_all()) == 1
  279. dn_config_2 = Config.configure_data_node(id="baz", storage_type="in_memory")
  280. _DataManager._create_and_set(dn_config_2, None, None)
  281. _DataManager._create_and_set(dn_config_2, None, None)
  282. assert len(_DataManager._get_all()) == 3
  283. assert len([dn for dn in _DataManager._get_all() if dn.config_id == "foo"]) == 1
  284. assert len([dn for dn in _DataManager._get_all() if dn.config_id == "baz"]) == 2
  285. def test_get_all_on_multiple_versions_environment(self):
  286. # Create 5 data nodes with 2 versions each
  287. # Only version 1.0 has the data node with config_id = "config_id_1"
  288. # Only version 2.0 has the data node with config_id = "config_id_6"
  289. for version in range(1, 3):
  290. for i in range(5):
  291. _DataManager._set(
  292. InMemoryDataNode(
  293. f"config_id_{i + version}",
  294. Scope.SCENARIO,
  295. id=DataNodeId(f"id{i}_v{version}"),
  296. version=f"{version}.0",
  297. )
  298. )
  299. _VersionManager._set_experiment_version("1.0")
  300. assert len(_DataManager._get_all()) == 5
  301. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_1"}])) == 1
  302. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_6"}])) == 0
  303. _VersionManager._set_development_version("1.0")
  304. assert len(_DataManager._get_all()) == 5
  305. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_1"}])) == 1
  306. assert len(_DataManager._get_all_by(filters=[{"version": "1.0", "config_id": "config_id_6"}])) == 0
  307. _VersionManager._set_experiment_version("2.0")
  308. assert len(_DataManager._get_all()) == 5
  309. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_1"}])) == 0
  310. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_6"}])) == 1
  311. _VersionManager._set_development_version("2.0")
  312. assert len(_DataManager._get_all()) == 5
  313. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_1"}])) == 0
  314. assert len(_DataManager._get_all_by(filters=[{"version": "2.0", "config_id": "config_id_6"}])) == 1
  315. def test_set(self):
  316. dn = InMemoryDataNode(
  317. "config_id",
  318. Scope.SCENARIO,
  319. id=DataNodeId("id"),
  320. owner_id=None,
  321. parent_ids={"task_id_1"},
  322. last_edit_date=None,
  323. edits=[],
  324. edit_in_progress=False,
  325. properties={"foo": "bar"},
  326. )
  327. assert len(_DataManager._get_all()) == 0
  328. assert not _DataManager._exists(dn.id)
  329. _DataManager._set(dn)
  330. assert len(_DataManager._get_all()) == 1
  331. assert _DataManager._exists(dn.id)
  332. # changing data node attribute
  333. dn._config_id = "foo"
  334. assert dn.config_id == "foo"
  335. _DataManager._set(dn)
  336. assert len(_DataManager._get_all()) == 1
  337. assert dn.config_id == "foo"
  338. assert _DataManager._get(dn.id).config_id == "foo"
  339. def test_delete(self):
  340. dn_1 = InMemoryDataNode("config_id", Scope.SCENARIO, id="id_1")
  341. dn_2 = InMemoryDataNode("config_id", Scope.SCENARIO, id="id_2")
  342. dn_3 = InMemoryDataNode("config_id", Scope.SCENARIO, id="id_3")
  343. assert len(_DataManager._get_all()) == 0
  344. _DataManager._set(dn_1)
  345. _DataManager._set(dn_2)
  346. _DataManager._set(dn_3)
  347. assert len(_DataManager._get_all()) == 3
  348. assert all(_DataManager._exists(dn.id) for dn in [dn_1, dn_2, dn_3])
  349. _DataManager._delete(dn_1.id)
  350. assert len(_DataManager._get_all()) == 2
  351. assert _DataManager._get(dn_2.id).id == dn_2.id
  352. assert _DataManager._get(dn_3.id).id == dn_3.id
  353. assert _DataManager._get(dn_1.id) is None
  354. assert all(_DataManager._exists(dn.id) for dn in [dn_2, dn_3])
  355. assert not _DataManager._exists(dn_1.id)
  356. _DataManager._delete_all()
  357. assert len(_DataManager._get_all()) == 0
  358. assert not any(_DataManager._exists(dn.id) for dn in [dn_2, dn_3])
  359. def test_get_or_create(self):
  360. def _get_or_create_dn(config, *args):
  361. return _DataManager._bulk_get_or_create([config], *args)[config]
  362. _DataManager._delete_all()
  363. global_dn_config = Config.configure_data_node(
  364. id="test_data_node", storage_type="in_memory", scope=Scope.GLOBAL, data="In memory Data Node"
  365. )
  366. cycle_dn_config = Config.configure_data_node(
  367. id="test_data_node1", storage_type="in_memory", scope=Scope.CYCLE, data="In memory Data Node"
  368. )
  369. scenario_dn_config = Config.configure_data_node(
  370. id="test_data_node2", storage_type="in_memory", scope=Scope.SCENARIO, data="In memory scenario"
  371. )
  372. assert len(_DataManager._get_all()) == 0
  373. global_dn = _get_or_create_dn(global_dn_config, None, None)
  374. assert len(_DataManager._get_all()) == 1
  375. global_dn_bis = _get_or_create_dn(global_dn_config, None)
  376. assert len(_DataManager._get_all()) == 1
  377. assert global_dn.id == global_dn_bis.id
  378. scenario_dn = _get_or_create_dn(scenario_dn_config, None, "scenario_id")
  379. assert len(_DataManager._get_all()) == 2
  380. scenario_dn_bis = _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. scenario_dn_ter = _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. assert scenario_dn_bis.id == scenario_dn_ter.id
  387. scenario_dn_quater = _get_or_create_dn(scenario_dn_config, None, "scenario_id_2")
  388. assert len(_DataManager._get_all()) == 3
  389. assert scenario_dn.id == scenario_dn_bis.id
  390. assert scenario_dn_bis.id == scenario_dn_ter.id
  391. assert scenario_dn_ter.id != scenario_dn_quater.id
  392. assert len(_DataManager._get_all()) == 3
  393. cycle_dn = _get_or_create_dn(cycle_dn_config, "cycle_id", None)
  394. assert len(_DataManager._get_all()) == 4
  395. cycle_dn_1 = _get_or_create_dn(cycle_dn_config, "cycle_id", None)
  396. assert len(_DataManager._get_all()) == 4
  397. assert cycle_dn.id == cycle_dn_1.id
  398. cycle_dn_2 = _get_or_create_dn(cycle_dn_config, "cycle_id", "scenario_id")
  399. assert len(_DataManager._get_all()) == 4
  400. assert cycle_dn.id == cycle_dn_2.id
  401. cycle_dn_3 = _get_or_create_dn(cycle_dn_config, "cycle_id", None)
  402. assert len(_DataManager._get_all()) == 4
  403. assert cycle_dn.id == cycle_dn_3.id
  404. cycle_dn_4 = _get_or_create_dn(cycle_dn_config, "cycle_id", "scenario_id")
  405. assert len(_DataManager._get_all()) == 4
  406. assert cycle_dn.id == cycle_dn_4.id
  407. cycle_dn_5 = _get_or_create_dn(cycle_dn_config, "cycle_id", "scenario_id_2")
  408. assert len(_DataManager._get_all()) == 4
  409. assert cycle_dn.id == cycle_dn_5.id
  410. assert cycle_dn_1.id == cycle_dn_2.id
  411. assert cycle_dn_2.id == cycle_dn_3.id
  412. assert cycle_dn_3.id == cycle_dn_4.id
  413. assert cycle_dn_4.id == cycle_dn_5.id
  414. def test_ensure_persistence_of_data_node(self):
  415. dm = _DataManager()
  416. dm._delete_all()
  417. dn_config_1 = Config.configure_data_node(
  418. id="data_node_1", storage_type="in_memory", data="In memory sequence 2"
  419. )
  420. dn_config_2 = Config.configure_data_node(
  421. id="data_node_2", storage_type="in_memory", data="In memory sequence 2"
  422. )
  423. dm._bulk_get_or_create([dn_config_1, dn_config_2])
  424. assert len(dm._get_all()) == 2
  425. # Delete the DataManager to ensure it's get from the storage system
  426. del dm
  427. dm = _DataManager()
  428. dm._bulk_get_or_create([dn_config_1])
  429. assert len(dm._get_all()) == 2
  430. dm._delete_all()
  431. @pytest.mark.parametrize(
  432. "storage_type,path",
  433. [
  434. ("pickle", "pickle_file_path"),
  435. ("csv", "csv_file"),
  436. ("excel", "excel_file"),
  437. ("json", "json_file"),
  438. ("parquet", "parquet_file_path"),
  439. ],
  440. )
  441. def test_clean_generated_files(self, storage_type, path, request):
  442. path = request.getfixturevalue(path)
  443. user_dn_config = Config.configure_data_node(
  444. id="d1", storage_type=storage_type, path=path, default_data={"a": [1], "b": [2]}
  445. )
  446. generated_dn_1_config = Config.configure_data_node(
  447. id="d2", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  448. )
  449. generated_dn_2_config = Config.configure_data_node(
  450. id="d3", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  451. )
  452. dns = _DataManager._bulk_get_or_create([user_dn_config, generated_dn_1_config, generated_dn_2_config])
  453. user_dn = dns[user_dn_config]
  454. generated_dn_1 = dns[generated_dn_1_config]
  455. generated_dn_2 = dns[generated_dn_2_config]
  456. _DataManager._clean_generated_file(user_dn.id)
  457. assert file_exists(user_dn.path)
  458. _DataManager._clean_generated_files([generated_dn_1, generated_dn_2])
  459. assert not file_exists(generated_dn_1.path)
  460. assert not file_exists(generated_dn_2.path)
  461. @pytest.mark.parametrize(
  462. "storage_type,path",
  463. [
  464. ("pickle", "pickle_file_path"),
  465. ("csv", "csv_file"),
  466. ("excel", "excel_file"),
  467. ("json", "json_file"),
  468. ("parquet", "parquet_file_path"),
  469. ],
  470. )
  471. def test_delete_does_clean_generated_pickle_files(self, storage_type, path, request):
  472. path = request.getfixturevalue(path)
  473. user_dn_config = Config.configure_data_node(
  474. id="d1", storage_type=storage_type, path=path, default_data={"a": [1], "b": [2]}
  475. )
  476. generated_dn_config_1 = Config.configure_data_node(
  477. id="d2", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  478. )
  479. generated_dn_config_2 = Config.configure_data_node(
  480. id="d3", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  481. )
  482. generated_dn_config_3 = Config.configure_data_node(
  483. id="d4", storage_type=storage_type, default_data={"a": [1], "b": [2]}
  484. )
  485. dns = _DataManager._bulk_get_or_create(
  486. [
  487. user_dn_config,
  488. generated_dn_config_1,
  489. generated_dn_config_2,
  490. generated_dn_config_3,
  491. ]
  492. )
  493. user_dn = dns[user_dn_config]
  494. generated_dn_1 = dns[generated_dn_config_1]
  495. generated_dn_2 = dns[generated_dn_config_2]
  496. generated_dn_3 = dns[generated_dn_config_3]
  497. _DataManager._delete(user_dn.id)
  498. assert file_exists(user_dn.path)
  499. _DataManager._delete_many([generated_dn_1.id, generated_dn_2.id])
  500. assert not file_exists(generated_dn_1.path)
  501. assert not file_exists(generated_dn_2.path)
  502. _DataManager._delete_all()
  503. assert not file_exists(generated_dn_3.path)
  504. def test_create_dn_from_loaded_config_no_scope(self):
  505. file_config = NamedTemporaryFile(
  506. """
  507. [TAIPY]
  508. [DATA_NODE.a]
  509. default_data = "4:int"
  510. [DATA_NODE.b]
  511. [TASK.t]
  512. function = "math.sqrt:function"
  513. inputs = [ "a:SECTION",]
  514. outputs = [ "b:SECTION",]
  515. skippable = "False:bool"
  516. [SCENARIO.s]
  517. tasks = [ "t:SECTION",]
  518. sequences.s_sequence = [ "t:SECTION",]
  519. [SCENARIO.s.comparators]
  520. """
  521. )
  522. from taipy import core as tp
  523. Config.override(file_config.filename)
  524. tp.create_scenario(Config.scenarios["s"])
  525. tp.create_scenario(Config.scenarios["s"])
  526. assert len(tp.get_data_nodes()) == 4
  527. def test_create_dn_from_loaded_config_no_storage_type(self):
  528. file_config = NamedTemporaryFile(
  529. """
  530. [TAIPY]
  531. [DATA_NODE.input_dn]
  532. scope = "SCENARIO:SCOPE"
  533. default_data = "21:int"
  534. [DATA_NODE.output_dn]
  535. storage_type = "in_memory"
  536. scope = "SCENARIO:SCOPE"
  537. [TASK.double]
  538. inputs = [ "input_dn:SECTION",]
  539. function = "math.sqrt:function"
  540. outputs = [ "output_dn:SECTION",]
  541. skippable = "False:bool"
  542. [SCENARIO.my_scenario]
  543. tasks = [ "double:SECTION",]
  544. sequences.my_sequence = [ "double:SECTION",]
  545. [SCENARIO.my_scenario.comparators]
  546. """
  547. )
  548. from taipy import core as tp
  549. Config.override(file_config.filename)
  550. scenario = tp.create_scenario(Config.scenarios["my_scenario"])
  551. assert isinstance(scenario.input_dn, PickleDataNode)
  552. assert isinstance(scenario.output_dn, InMemoryDataNode)
  553. def test_create_dn_from_loaded_config_modified_default_config(self):
  554. file_config = NamedTemporaryFile(
  555. """
  556. [TAIPY]
  557. [DATA_NODE.input_dn]
  558. scope = "SCENARIO:SCOPE"
  559. default_path="fake/path.csv"
  560. [DATA_NODE.output_dn]
  561. storage_type = "in_memory"
  562. scope = "SCENARIO:SCOPE"
  563. [TASK.double]
  564. inputs = [ "input_dn:SECTION",]
  565. function = "math.sqrt:function"
  566. outputs = [ "output_dn:SECTION",]
  567. skippable = "False:bool"
  568. [SCENARIO.my_scenario]
  569. tasks = [ "double:SECTION",]
  570. sequences.my_sequence = [ "double:SECTION",]
  571. [SCENARIO.my_scenario.comparators]
  572. """
  573. )
  574. from taipy import core as tp
  575. Config.set_default_data_node_configuration(storage_type="csv")
  576. Config.override(file_config.filename)
  577. scenario = tp.create_scenario(Config.scenarios["my_scenario"])
  578. assert isinstance(scenario.input_dn, CSVDataNode)
  579. assert isinstance(scenario.output_dn, InMemoryDataNode)
  580. def test_get_tasks_by_config_id(self):
  581. dn_config_1 = Config.configure_data_node("dn_1", scope=Scope.SCENARIO)
  582. dn_config_2 = Config.configure_data_node("dn_2", scope=Scope.SCENARIO)
  583. dn_config_3 = Config.configure_data_node("dn_3", scope=Scope.SCENARIO)
  584. dn_1_1 = _DataManager._create_and_set(dn_config_1, None, None)
  585. dn_1_2 = _DataManager._create_and_set(dn_config_1, None, None)
  586. dn_1_3 = _DataManager._create_and_set(dn_config_1, None, None)
  587. assert len(_DataManager._get_all()) == 3
  588. dn_2_1 = _DataManager._create_and_set(dn_config_2, None, None)
  589. dn_2_2 = _DataManager._create_and_set(dn_config_2, None, None)
  590. assert len(_DataManager._get_all()) == 5
  591. dn_3_1 = _DataManager._create_and_set(dn_config_3, None, None)
  592. assert len(_DataManager._get_all()) == 6
  593. dn_1_datanodes = _DataManager._get_by_config_id(dn_config_1.id)
  594. assert len(dn_1_datanodes) == 3
  595. assert sorted([dn_1_1.id, dn_1_2.id, dn_1_3.id]) == sorted([sequence.id for sequence in dn_1_datanodes])
  596. dn_2_datanodes = _DataManager._get_by_config_id(dn_config_2.id)
  597. assert len(dn_2_datanodes) == 2
  598. assert sorted([dn_2_1.id, dn_2_2.id]) == sorted([sequence.id for sequence in dn_2_datanodes])
  599. dn_3_datanodes = _DataManager._get_by_config_id(dn_config_3.id)
  600. assert len(dn_3_datanodes) == 1
  601. assert sorted([dn_3_1.id]) == sorted([sequence.id for sequence in dn_3_datanodes])
  602. def test_get_data_nodes_by_config_id_in_multiple_versions_environment(self):
  603. dn_config_1 = Config.configure_data_node("dn_1", scope=Scope.SCENARIO)
  604. dn_config_2 = Config.configure_data_node("dn_2", scope=Scope.SCENARIO)
  605. _VersionManager._set_experiment_version("1.0")
  606. _DataManager._create_and_set(dn_config_1, None, None)
  607. _DataManager._create_and_set(dn_config_1, None, None)
  608. _DataManager._create_and_set(dn_config_1, None, None)
  609. _DataManager._create_and_set(dn_config_2, None, None)
  610. _DataManager._create_and_set(dn_config_2, None, None)
  611. assert len(_DataManager._get_by_config_id(dn_config_1.id)) == 3
  612. assert len(_DataManager._get_by_config_id(dn_config_2.id)) == 2
  613. _VersionManager._set_experiment_version("2.0")
  614. _DataManager._create_and_set(dn_config_1, None, None)
  615. _DataManager._create_and_set(dn_config_1, None, None)
  616. _DataManager._create_and_set(dn_config_1, None, None)
  617. _DataManager._create_and_set(dn_config_2, None, None)
  618. _DataManager._create_and_set(dn_config_2, None, None)
  619. assert len(_DataManager._get_by_config_id(dn_config_1.id)) == 3
  620. assert len(_DataManager._get_by_config_id(dn_config_2.id)) == 2