fabricks 3.0.11__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (176) hide show
  1. fabricks/__init__.py +0 -0
  2. fabricks/api/__init__.py +11 -0
  3. fabricks/api/cdc/__init__.py +6 -0
  4. fabricks/api/cdc/nocdc.py +3 -0
  5. fabricks/api/cdc/scd1.py +3 -0
  6. fabricks/api/cdc/scd2.py +3 -0
  7. fabricks/api/context.py +27 -0
  8. fabricks/api/core.py +4 -0
  9. fabricks/api/deploy.py +3 -0
  10. fabricks/api/exceptions.py +19 -0
  11. fabricks/api/extenders.py +3 -0
  12. fabricks/api/job_schema.py +3 -0
  13. fabricks/api/log.py +3 -0
  14. fabricks/api/masks.py +3 -0
  15. fabricks/api/metastore/__init__.py +10 -0
  16. fabricks/api/metastore/database.py +3 -0
  17. fabricks/api/metastore/table.py +3 -0
  18. fabricks/api/metastore/view.py +6 -0
  19. fabricks/api/notebooks/__init__.py +0 -0
  20. fabricks/api/notebooks/cluster.py +6 -0
  21. fabricks/api/notebooks/initialize.py +42 -0
  22. fabricks/api/notebooks/process.py +54 -0
  23. fabricks/api/notebooks/run.py +59 -0
  24. fabricks/api/notebooks/schedule.py +75 -0
  25. fabricks/api/notebooks/terminate.py +31 -0
  26. fabricks/api/parsers.py +3 -0
  27. fabricks/api/schedules.py +3 -0
  28. fabricks/api/udfs.py +3 -0
  29. fabricks/api/utils.py +9 -0
  30. fabricks/api/version.py +3 -0
  31. fabricks/api/views.py +6 -0
  32. fabricks/cdc/__init__.py +14 -0
  33. fabricks/cdc/base/__init__.py +4 -0
  34. fabricks/cdc/base/_types.py +10 -0
  35. fabricks/cdc/base/cdc.py +5 -0
  36. fabricks/cdc/base/configurator.py +223 -0
  37. fabricks/cdc/base/generator.py +177 -0
  38. fabricks/cdc/base/merger.py +110 -0
  39. fabricks/cdc/base/processor.py +471 -0
  40. fabricks/cdc/cdc.py +5 -0
  41. fabricks/cdc/nocdc.py +20 -0
  42. fabricks/cdc/scd.py +22 -0
  43. fabricks/cdc/scd1.py +15 -0
  44. fabricks/cdc/scd2.py +15 -0
  45. fabricks/cdc/templates/__init__.py +0 -0
  46. fabricks/cdc/templates/ctes/base.sql.jinja +35 -0
  47. fabricks/cdc/templates/ctes/current.sql.jinja +28 -0
  48. fabricks/cdc/templates/ctes/deduplicate_hash.sql.jinja +32 -0
  49. fabricks/cdc/templates/ctes/deduplicate_key.sql.jinja +31 -0
  50. fabricks/cdc/templates/ctes/rectify.sql.jinja +113 -0
  51. fabricks/cdc/templates/ctes/slice.sql.jinja +1 -0
  52. fabricks/cdc/templates/filter.sql.jinja +4 -0
  53. fabricks/cdc/templates/filters/final.sql.jinja +4 -0
  54. fabricks/cdc/templates/filters/latest.sql.jinja +17 -0
  55. fabricks/cdc/templates/filters/update.sql.jinja +30 -0
  56. fabricks/cdc/templates/macros/bactick.sql.jinja +1 -0
  57. fabricks/cdc/templates/macros/hash.sql.jinja +18 -0
  58. fabricks/cdc/templates/merge.sql.jinja +3 -0
  59. fabricks/cdc/templates/merges/nocdc.sql.jinja +41 -0
  60. fabricks/cdc/templates/merges/scd1.sql.jinja +73 -0
  61. fabricks/cdc/templates/merges/scd2.sql.jinja +54 -0
  62. fabricks/cdc/templates/queries/__init__.py +0 -0
  63. fabricks/cdc/templates/queries/context.sql.jinja +186 -0
  64. fabricks/cdc/templates/queries/final.sql.jinja +1 -0
  65. fabricks/cdc/templates/queries/nocdc/complete.sql.jinja +10 -0
  66. fabricks/cdc/templates/queries/nocdc/update.sql.jinja +34 -0
  67. fabricks/cdc/templates/queries/scd1.sql.jinja +85 -0
  68. fabricks/cdc/templates/queries/scd2.sql.jinja +98 -0
  69. fabricks/cdc/templates/query.sql.jinja +15 -0
  70. fabricks/context/__init__.py +72 -0
  71. fabricks/context/_types.py +133 -0
  72. fabricks/context/config/__init__.py +92 -0
  73. fabricks/context/config/utils.py +53 -0
  74. fabricks/context/log.py +77 -0
  75. fabricks/context/runtime.py +117 -0
  76. fabricks/context/secret.py +103 -0
  77. fabricks/context/spark_session.py +82 -0
  78. fabricks/context/utils.py +80 -0
  79. fabricks/core/__init__.py +4 -0
  80. fabricks/core/dags/__init__.py +9 -0
  81. fabricks/core/dags/base.py +99 -0
  82. fabricks/core/dags/generator.py +157 -0
  83. fabricks/core/dags/log.py +12 -0
  84. fabricks/core/dags/processor.py +228 -0
  85. fabricks/core/dags/run.py +39 -0
  86. fabricks/core/dags/terminator.py +25 -0
  87. fabricks/core/dags/utils.py +54 -0
  88. fabricks/core/extenders.py +33 -0
  89. fabricks/core/job_schema.py +32 -0
  90. fabricks/core/jobs/__init__.py +21 -0
  91. fabricks/core/jobs/base/__init__.py +10 -0
  92. fabricks/core/jobs/base/_types.py +284 -0
  93. fabricks/core/jobs/base/checker.py +139 -0
  94. fabricks/core/jobs/base/configurator.py +306 -0
  95. fabricks/core/jobs/base/exception.py +85 -0
  96. fabricks/core/jobs/base/generator.py +447 -0
  97. fabricks/core/jobs/base/invoker.py +206 -0
  98. fabricks/core/jobs/base/job.py +5 -0
  99. fabricks/core/jobs/base/processor.py +249 -0
  100. fabricks/core/jobs/bronze.py +395 -0
  101. fabricks/core/jobs/get_job.py +127 -0
  102. fabricks/core/jobs/get_job_conf.py +152 -0
  103. fabricks/core/jobs/get_job_id.py +31 -0
  104. fabricks/core/jobs/get_jobs.py +107 -0
  105. fabricks/core/jobs/get_schedule.py +10 -0
  106. fabricks/core/jobs/get_schedules.py +32 -0
  107. fabricks/core/jobs/gold.py +415 -0
  108. fabricks/core/jobs/silver.py +373 -0
  109. fabricks/core/masks.py +52 -0
  110. fabricks/core/parsers/__init__.py +12 -0
  111. fabricks/core/parsers/_types.py +6 -0
  112. fabricks/core/parsers/base.py +95 -0
  113. fabricks/core/parsers/decorator.py +11 -0
  114. fabricks/core/parsers/get_parser.py +26 -0
  115. fabricks/core/parsers/utils.py +69 -0
  116. fabricks/core/schedules/__init__.py +14 -0
  117. fabricks/core/schedules/diagrams.py +21 -0
  118. fabricks/core/schedules/generate.py +20 -0
  119. fabricks/core/schedules/get_schedule.py +5 -0
  120. fabricks/core/schedules/get_schedules.py +9 -0
  121. fabricks/core/schedules/process.py +9 -0
  122. fabricks/core/schedules/run.py +3 -0
  123. fabricks/core/schedules/terminate.py +6 -0
  124. fabricks/core/schedules/views.py +61 -0
  125. fabricks/core/steps/__init__.py +4 -0
  126. fabricks/core/steps/_types.py +7 -0
  127. fabricks/core/steps/base.py +423 -0
  128. fabricks/core/steps/get_step.py +10 -0
  129. fabricks/core/steps/get_step_conf.py +26 -0
  130. fabricks/core/udfs.py +106 -0
  131. fabricks/core/views.py +41 -0
  132. fabricks/deploy/__init__.py +92 -0
  133. fabricks/deploy/masks.py +8 -0
  134. fabricks/deploy/notebooks.py +71 -0
  135. fabricks/deploy/schedules.py +10 -0
  136. fabricks/deploy/tables.py +82 -0
  137. fabricks/deploy/udfs.py +19 -0
  138. fabricks/deploy/utils.py +36 -0
  139. fabricks/deploy/views.py +509 -0
  140. fabricks/metastore/README.md +3 -0
  141. fabricks/metastore/__init__.py +5 -0
  142. fabricks/metastore/_types.py +65 -0
  143. fabricks/metastore/database.py +65 -0
  144. fabricks/metastore/dbobject.py +66 -0
  145. fabricks/metastore/pyproject.toml +20 -0
  146. fabricks/metastore/table.py +768 -0
  147. fabricks/metastore/utils.py +51 -0
  148. fabricks/metastore/view.py +53 -0
  149. fabricks/utils/__init__.py +0 -0
  150. fabricks/utils/_types.py +6 -0
  151. fabricks/utils/azure_queue.py +93 -0
  152. fabricks/utils/azure_table.py +154 -0
  153. fabricks/utils/console.py +51 -0
  154. fabricks/utils/fdict.py +240 -0
  155. fabricks/utils/helpers.py +228 -0
  156. fabricks/utils/log.py +236 -0
  157. fabricks/utils/mermaid.py +32 -0
  158. fabricks/utils/path.py +242 -0
  159. fabricks/utils/pip.py +61 -0
  160. fabricks/utils/pydantic.py +94 -0
  161. fabricks/utils/read/__init__.py +11 -0
  162. fabricks/utils/read/_types.py +3 -0
  163. fabricks/utils/read/read.py +305 -0
  164. fabricks/utils/read/read_excel.py +5 -0
  165. fabricks/utils/read/read_yaml.py +33 -0
  166. fabricks/utils/schema/__init__.py +7 -0
  167. fabricks/utils/schema/get_json_schema_for_type.py +161 -0
  168. fabricks/utils/schema/get_schema_for_type.py +99 -0
  169. fabricks/utils/spark.py +76 -0
  170. fabricks/utils/sqlglot.py +56 -0
  171. fabricks/utils/write/__init__.py +8 -0
  172. fabricks/utils/write/delta.py +46 -0
  173. fabricks/utils/write/stream.py +27 -0
  174. fabricks-3.0.11.dist-info/METADATA +23 -0
  175. fabricks-3.0.11.dist-info/RECORD +176 -0
  176. fabricks-3.0.11.dist-info/WHEEL +4 -0
@@ -0,0 +1,66 @@
1
+ from typing import Optional
2
+
3
+ from pyspark.errors.exceptions.base import AnalysisException
4
+ from pyspark.sql import SparkSession
5
+
6
+ from fabricks.context.log import DEFAULT_LOGGER
7
+ from fabricks.metastore.database import Database
8
+
9
+
10
+ class DbObject:
11
+ def __init__(self, database: str, *levels: str, spark: Optional[SparkSession] = None):
12
+ self.database = Database(database)
13
+ self.levels = levels
14
+
15
+ if spark is None:
16
+ from fabricks.utils.spark import spark
17
+
18
+ spark = spark
19
+
20
+ assert spark is not None
21
+ self.spark = spark
22
+
23
+ @property
24
+ def name(self) -> str:
25
+ return "_".join(self.levels)
26
+
27
+ @property
28
+ def qualified_name(self) -> str:
29
+ return f"{self.database.name}.{self.name}"
30
+
31
+ @property
32
+ def registered(self) -> bool:
33
+ try:
34
+ df = self.spark.sql(f"show tables in {self.database}").where(f"tableName == '{self.name}'")
35
+ return not df.isEmpty()
36
+ # not found
37
+ except AnalysisException:
38
+ return False
39
+
40
+ @property
41
+ def is_view(self) -> bool:
42
+ try:
43
+ df = self.spark.sql(f"show views in {self.database}").where(f"viewName == '{self.name}'")
44
+ return not df.isEmpty()
45
+ # not found
46
+ except AnalysisException:
47
+ return False
48
+
49
+ @property
50
+ def is_table(self) -> bool:
51
+ if self.is_view:
52
+ return False
53
+ else:
54
+ return self.registered
55
+
56
+ def drop(self):
57
+ if self.is_view:
58
+ DEFAULT_LOGGER.warning("drop view from metastore", extra={"label": self})
59
+ self.spark.sql(f"drop view if exists {self}")
60
+
61
+ elif self.is_table:
62
+ DEFAULT_LOGGER.warning("drop table from metastore", extra={"label": self})
63
+ self.spark.sql(f"drop table if exists {self}")
64
+
65
+ def __str__(self):
66
+ return self.qualified_name
@@ -0,0 +1,20 @@
1
+ [build-system]
2
+ requires = [ "poetry_core>=1.0.0",]
3
+ build-backend = "poetry.core.masonry.api"
4
+
5
+ [tool.poetry]
6
+ name = "fabricks-metastore"
7
+ version = "2024.7.1.5"
8
+ description = "Fabricks - Metastore"
9
+ license = "MIT"
10
+ authors = [ "BMS DWH Team <bi_support@bmsuisse.ch>",]
11
+ readme = "README.md"
12
+ packages = [{include="fabricks"}]
13
+
14
+ [tool.black]
15
+ line-length = 119
16
+
17
+ [tool.poetry.dependencies]
18
+ python = ">=3.9,<4"
19
+ "fabricks.utils" = { path = "../utils", develop = true }
20
+ "fabricks.context" = { path = "../context", develop = true }