datapipelab 0.2.0__py3-none-any.whl → 0.2.2__py3-none-any.whl

datapipelab/app/node/sink/hive_node.py

@@ -1,16 +1,16 @@
 from datapipelab.app.node.tnode import TNode
-
+from datapipelab.logger import logger
 
 class HiveSinkNode(TNode):
     def __init__(self, spark, tnode_config, df):
         from pyspark.sql import DataFrame
         super().__init__(spark=spark)
-        self.mode = tnode_config['mode']
-        self.stream = tnode_config['stream']
+        self.mode = tnode_config.get('mode', None)
+        self.stream = tnode_config.get('stream', None)
         self.database_name = tnode_config['options']['database']
         self.table_name = tnode_config['options']['table']
-        self.partition_by = tnode_config['options'].get('partition_by')
-        self.partition_count = tnode_config['options'].get('partition_count')
+        self.partition_by = tnode_config['options'].get('partition_by', None)
+        self.partition_count = tnode_config['options'].get('partition_count', None)
         self.overwrite = tnode_config['options']['overwrite']
         self.df = df
 
@@ -20,6 +20,7 @@ class HiveSinkNode(TNode):
                 self.df = self.df.repartition(int(self.partition_count))
             else:
                 self.df = self.df.repartition(int(self.partition_count), self.partition_by)
+        logger.info("Start writing to Hive")
         self.df.write.insertInto(f'{self.database_name}.{self.table_name}', overwrite=self.overwrite)
 
     def _process(self):

datapipelab/pipeline_handler.py

@@ -1,5 +1,6 @@
 from datapipelab.app.node.processor.custom_node import CustomNode
 from datapipelab.app.node.processor.shell_node import ShellProcessorNode
+from datapipelab.app.node.source.hive_node import HiveSourceNode
 from datapipelab.app.node.source.spark_node import SparkSourceNode
 from datapipelab.app.node.source.delta_node import DeltaSourceNode
 from datapipelab.app.node.processor.spark_node import SparkProcessorNode
@@ -33,6 +34,9 @@ class PipelineHandler:
         if input_type == 'spark':
             if input_format == 'spark':
                 source_df = SparkSourceNode(self.spark, tnode_config).run()
+        if input_type == 'hive':
+            if input_format == 'hive':
+                source_df = HiveSourceNode(self.spark, tnode_config).run()
         if input_type == 'adls_path':
             if input_format == 'delta':
                 source_df = DeltaSourceNode(self.spark, tnode_config).run()
@@ -80,5 +84,10 @@ class PipelineHandler:
         if tnode_type == 'custom':
             from datapipelab.app.node import custom_node
             processor_df = custom_node.CustomNode(self.spark, tnode_config).run()
+        if tnode_type == 'spark':
+            if tnode_format == 'hive':
+                from datapipelab.app.node.sink import hive_node
+                processor_df = hive_node.HiveSinkNode(self.spark, tnode_config, t_df[tnode_name_df]).run()
+
 
 

{datapipelab-0.2.0.dist-info → datapipelab-0.2.2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datapipelab
-Version: 0.2.0
+Version: 0.2.2
 Summary: A data pipeline library with connectors, sources, processors, and sinks.
 Requires-Dist: json5
 Requires-Dist: loguru

{datapipelab-0.2.0.dist-info → datapipelab-0.2.2.dist-info}/RECORD

@@ -3,7 +3,7 @@ datapipelab/engine.py,sha256=3QRsedRYNov6xIDOZ1tukinFE-SKv39Fn3sNCnD3L6g,442
 datapipelab/logger.py,sha256=Ugv0A4TfD3JWCWXNWu0lURcnfAEyuVrK3IrvVVgcHBo,864
 datapipelab/pipeline.py,sha256=dw9D9KM_hztt9g_YzqoNgQBRyCYR92cRZwrU5duP_Pg,1464
 datapipelab/pipeline_config.py,sha256=2bFAJepViE7rT7CaRANZU07aeQpOYcZ954ISujm9pXA,3816
-datapipelab/pipeline_handler.py,sha256=oWhpnzMQJp_5Dhy2_EiGbq8W0twsIPLpnU3OOJkYxT8,4058
+datapipelab/pipeline_handler.py,sha256=dMFGpp8WgdTHlD4LuWkNzrmh3RqO2nQ1t_RyTCqIPIg,4515
 datapipelab/app/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datapipelab/app/connector_node/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datapipelab/app/node/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -18,14 +18,14 @@ datapipelab/app/node/processor/spark_node.py,sha256=jzqdffIHUCgOfMFcoqjXdl8wFag-
 datapipelab/app/node/sink/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datapipelab/app/node/sink/csv_node.py,sha256=d2hyufP5_Nmql0pfD0KeC4rFu1wXTnBxVsoGl7sWbhM,1681
 datapipelab/app/node/sink/delta_node.py,sha256=iKEdiTjJ7SHJZMrbm0jR5tms5JZ5iCFfQklZbI-Yr2o,2044
-datapipelab/app/node/sink/hive_node.py,sha256=BKSSYb1AexQD1Jl6zP5ak_ibvd0wYRNzC_zhTtNTDRg,1106
+datapipelab/app/node/sink/hive_node.py,sha256=ycknOPBBwZGH3oHram_6LjHy-ygFjhuFNvVoPaNGaCU,1220
 datapipelab/app/node/sink/pandas_csv_node.py,sha256=JsJFt2XRpwxGeJyt_PDUgqZafiQROf1Sk5TUhQPxh4c,870
 datapipelab/app/node/sink/teams_notification_node.py,sha256=6ZufdbhVvRXi3QTQafLo5uKl9kLyDnkYIE_VZFT0QNw,3581
 datapipelab/app/node/source/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 datapipelab/app/node/source/delta_node.py,sha256=gg7SfuKBAAfjk6OX2jNrot9XX61HoBe3us3D8O-dscE,529
 datapipelab/app/node/source/hive_node.py,sha256=h_AMCnnmon7uLRIGsaHAPWEReD3VaWZXnz9r0TpLGNM,478
 datapipelab/app/node/source/spark_node.py,sha256=S_x2atRFPDnXmhCUtcmaLc4BDFd2H4uQq6wnEJb7Uug,480
-datapipelab-0.2.0.dist-info/METADATA,sha256=_vL7fMDr9xkCojabJuY6-RUAQoyle8LxH1aRdeWVe1o,220
-datapipelab-0.2.0.dist-info/WHEEL,sha256=zaaOINJESkSfm_4HQVc5ssNzHCPXhJm0kEUakpsEHaU,91
-datapipelab-0.2.0.dist-info/top_level.txt,sha256=HgeBjHvXorKzvNqU5BNPutoI771HtiqVit9_-0Zyrb4,12
-datapipelab-0.2.0.dist-info/RECORD,,
+datapipelab-0.2.2.dist-info/METADATA,sha256=aRJaLx-19yUY47qYHcWyJ6pm8WCGr7YwPkeL6AH-FXg,220
+datapipelab-0.2.2.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+datapipelab-0.2.2.dist-info/top_level.txt,sha256=HgeBjHvXorKzvNqU5BNPutoI771HtiqVit9_-0Zyrb4,12
+datapipelab-0.2.2.dist-info/RECORD,,

{datapipelab-0.2.0.dist-info → datapipelab-0.2.2.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (80.8.0)
+Generator: setuptools (80.9.0)
 Root-Is-Purelib: true
 Tag: py3-none-any