cit-pydata 0.0.1.dev2__tar.gz → 0.0.1.dev4__tar.gz

{cit-pydata-0.0.1.dev2 → cit-pydata-0.0.1.dev4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: cit-pydata
-Version: 0.0.1.dev2
+Version: 0.0.1.dev4
 Summary: Python clients for commonly used api services including Salesforce, SQL, Marketo, Box, SFSync*, and JobLog*
 Author-email: Dionis Wang <dwang@eab.com>
 Project-URL: Homepage, https://github.com/EAB-IT-DEV/cit-pydata

{cit-pydata-0.0.1.dev2 → cit-pydata-0.0.1.dev4}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "cit-pydata"
-version = "0.0.1.dev2"
+version = "0.0.1.dev4"
 authors = [
   { name="Dionis Wang", email="dwang@eab.com" },
 ]
@@ -21,7 +21,7 @@ Homepage = "https://github.com/EAB-IT-DEV/cit-pydata"
 Issues = "https://github.com/EAB-IT-DEV/cit-pydata/issues"
 
 [project.optional-dependencies]
-aws = ["aws-lambda-powertools"]
+aws = ["aws-lambda-powertools", "boto3"]
 box = ["boxsdk"]
 joblog = ["marketorestpython"]
 salesforce = ["simple_salesforce","pyforce"]
@@ -32,5 +32,5 @@ sql-pymssql = ["SQLAlchemy<2.0", "pymssql", "numpy", "pandas"]
 sql-psycopg2 = ["SQLAlchemy<2.0", "psycopg2-binary", "numpy", "pandas"]
 sql-mssql = ["SQLAlchemy<2.0", "pymssql", "numpy", "pandas"]
 sql-postgres = ["SQLAlchemy<2.0", "psycopg2-binary", "numpy", "pandas"]
-local = ["python-dotenv", "boto3"]
+local = ["python-dotenv"]
 all = ["aws-lambda-powertools","pandas","boxsdk","marketorestpython","simple_salesforce","pyforce","paramiko","SQLAlchemy<2.0", "pyodbc","pymssql", "numpy", "pandas","psycopg2-binary","python-dotenv", "boto3"]

{cit-pydata-0.0.1.dev2 → cit-pydata-0.0.1.dev4}/src/cit_pydata/aws/api.py

@@ -89,6 +89,7 @@ class S3Client:
         kwargs for method client.list_objects_v2
         https://boto3.amazonaws.com/v1/documentation/api/latest/reference/services/s3.html#S3.Client.list_objects_v2
         """
+        import pandas
 
         # self.logger.info(f'Getting S3 objects: {**kwargs}')
         object_list = []
@@ -110,8 +111,6 @@ class S3Client:
                 object_list.append(object_item)
             response.pop("Contents")
 
-        if not pandas in sys.modules:
-            import pandas
         df_s3_objects = pandas.DataFrame(object_list)
         # self.logger.debug(response)
         # self.logger.debug(df_s3_objects.info())
@@ -119,8 +118,7 @@ class S3Client:
         return response, df_s3_objects
 
     def metadata_to_sql(self, sql_client, table_name: str, dataframe):
-        if "sqlalchemy" not in sys.modules:
-            import sqlalchemy
+        import sqlalchemy
 
         try:
             sql_client.insert_df(

{cit-pydata-0.0.1.dev2 → cit-pydata-0.0.1.dev4}/src/cit_pydata/box/api.py

@@ -14,8 +14,8 @@ class BoxClient:
         self.client = None
 
     def _get_client(self):
-        if "boxsdk" not in sys.modules:
-            from boxsdk import Client, JWTAuth, OAuth2
+        from boxsdk import Client, JWTAuth, OAuth2
+
         # JWT Auth
         auth = JWTAuth.from_settings_file(self.box_jwt_auth_file_path)
 

{cit-pydata-0.0.1.dev2 → cit-pydata-0.0.1.dev4}/src/cit_pydata/joblog/api.py

@@ -13,8 +13,8 @@ class JobLogClient:
         self.log_guid = None
 
     def start(self):
-        if "uuid" not in sys.modules:
-            import uuid
+        import uuid
+
         if not self.log_guid:
             self.log_guid = str(uuid.uuid4())
             param_list = [self.job_name, "initializing", self.log_guid]

{cit-pydata-0.0.1.dev2 → cit-pydata-0.0.1.dev4}/src/cit_pydata/sql/api.py

@@ -8,6 +8,7 @@ from cit_pydata.aws import api as aws_api
 import time
 import numpy as np
 import pandas as pd
+import polars as pl
 
 from sqlalchemy import dialects, create_engine, text, MetaData, Table
 from sqlalchemy.orm import Session
@@ -449,6 +450,23 @@ class SQLClient:
 
         return dataframe
 
+    def execute_sql_select_pl(self, sql_statement, execute_options: dict = None) -> pl.DataFrame:
+        """ similar to execute_sql_select2,
+            but the function process the sql with polars and return a polars DataFrame"""
+            
+        dataframe = None
+        if not execute_options:
+            execute_options = {}
+
+        with self.sql_engine.connect() as connection:
+            dataframe = pl.read_database(
+                query=text(sql_statement),
+                connection=connection,
+                execute_options=execute_options,
+            )
+
+        return dataframe
+
     def get_dataframe_table(self, table_name, chunk_count=None, **kwargs):
         from itertools import islice
 

{cit-pydata-0.0.1.dev2 → cit-pydata-0.0.1.dev4}/src/cit_pydata.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: cit-pydata
-Version: 0.0.1.dev2
+Version: 0.0.1.dev4
 Summary: Python clients for commonly used api services including Salesforce, SQL, Marketo, Box, SFSync*, and JobLog*
 Author-email: Dionis Wang <dwang@eab.com>
 Project-URL: Homepage, https://github.com/EAB-IT-DEV/cit-pydata

{cit-pydata-0.0.1.dev2 → cit-pydata-0.0.1.dev4}/src/cit_pydata.egg-info/requires.txt

@@ -17,6 +17,7 @@ boto3
 
 [aws]
 aws-lambda-powertools
+boto3
 
 [box]
 boxsdk
@@ -26,7 +27,6 @@ marketorestpython
 
 [local]
 python-dotenv
-boto3
 
 [salesforce]
 simple_salesforce