A Module to enable Hepsiburada Data Science Team to utilize different tools.
Project description
Hepsiburada Data Science Utilities
This module includes utilities for Hepsiburada Data Science Team.
- Library is available via PyPi.
- Library can be downloaded using pip as follows:
pip install heps-ds-utils
- Existing library can be upgraded using pip as follows:
pip install heps-ds-utils --upgrade
Available Modules
- Hive Operations
from heps_ds_utils import HiveOperations
# A connection is needed to be generated in a specific runtime.
# There are 3 ways to set credentials for connection.
# 1) Instance try to set default credentials from Environment Variables.
hive_ds = HiveOperations()
hive_ds.connect_to_hive()
# 2) One can pass credentials to instance initiation to override default.
hive_ds = HiveOperations(HIVE_HOST="XXX", HIVE_PORT="YYY", HIVE_USER="ZZZ", HIVE_PASS="WWW", HADOOP_EDGE_HOST="QQQ")
hive_ds = HiveOperations(HIVE_USER="ZZZ", HIVE_PASS="WWW")
hive_ds.connect_to_hive()
# 3) One can change any of the credentials after initiation using appropriate attribute.
hive_ds = HiveOperations()
hive_ds.hive_username = 'XXX'
hive_ds.hive_password = 'YYY'
hive_ds.connect_to_hive()
# Execute an SQL query to retrieve data.
# Currently Implemented Types: DataFrame, Numpy Array, Dictionary, List.
SQL_QUERY = "SELECT * FROM {db}.{table}"
data, columns = hive_ds.execute_query(SQL_QUERY, return_type="dataframe", return_columns=False)
# Execute an SQL query to create and insert data into table.
SQL_QUERY = "INSERT INTO .."
hive_ds.create_insert_table(SQL_QUERY)
# Send Files to Hive and Create a Table with the Data.
# Currently DataFrame or Numpy Array can be sent to Hive.
# While sending Numpy Array columns have to be provided.
SQL_QUERY = "INSERT INTO .."
hive_ds.send_files_to_hive("{db}.{table}", data, columns=None)
# Close the connection at the end of the runtime.
hive_ds.disconnect_from_hive()
- BigQuery Operations
from heps_ds_utils import BigQueryOperations, execute_from_bq_file
# A connection is needed to be generated in a specific runtime.
# There are 3 ways to set credentials for connection.
# 1) Instance try to set default credentials from Environment Variables.
bq_ds = BigQueryOperations()
# 2) One can pass credentials to instance initiation to override default.
bq_ds = BigQueryOperations(gcp_key_path="/tmp/keys/ds_qos.json")
# Unlike HiveOperations, initiation creates a direct connection. Absence of
# credentials will throw an error.
# Execute an SQL query to retrieve data.
# Currently Implemented Types: DataFrame.
QUERY_STRING = """SELECT * FROM `[project_name].[dataset_name].[table_name]` LIMIT 20"""
data = bq_ds.execute_query(QUERY_STRING, return_type='dataframe')
# Create a Dataset in BigQuery.
bq_ds.create_new_dataset("example_dataset")
# Create a Table under a Dataset in BigQuery.
schema = [
{"field_name": "id", "field_type": "INTEGER", "field_mode": "REQUIRED"},
{"field_name": "first_name", "field_type": "STRING", "field_mode": "REQUIRED"},
{"field_name": "last_name", "field_type": "STRING", "field_mode": "REQUIRED"},
{"field_name": "email", "field_type": "STRING", "field_mode": "REQUIRED"},
{"field_name": "gender", "field_type": "STRING", "field_mode": "REQUIRED"},
{"field_name": "ip_address", "field_type": "STRING", "field_mode": "REQUIRED"}]
bq_ds.create_new_table(dataset='example_dataset', table_name='mock_data', schema=schema)
# Insert into an existing Table from Dataframe.
# Don't create and insert in the same runtime.
# Google throws an error when creation and insertion time is close.
bq_ds.insert_rows_into_existing_table(dataset='example_dataset', table='mock_data', data=df)
# Delete a Table.
bq_ds.delete_existing_table('example_dataset', 'mock_data')
# Delete a Dataset.
# Trying to delete a dataset consisting of tables will throw an error.
bq_ds.delete_existing_dataset('example_dataset')
# Load Dataframe As a Table. BigQuery will infer the data types.
bq_ds.load_data_to_table('example_dataset', 'mock_data', df, overwrite=False)
# To execute BQ commands sequentially from a BigQuery Script without a return statement !
execute_from_bq_file(bq_client=bq_ds, bq_file_path="tests/test_data/test_case_2.bq", verbose=True)
- Logging Operations
from heps_ds_utils import LoggingOperations
# A connection is needed to be generated in a specific runtime.
# There are 3 ways to set credentials for connection.
# 1) Instance try to set default credentials from Environment Variables.
logger_ds = LoggingOperations()
# 2) One can pass credentials to instance initiation to override default.
logger_ds = LoggingOperations(gcp_key_path="/tmp/keys/ds_qos.json")
# Unlike HiveOperations, initiation creates a direct connection. Absence of
# credentials will throw an error.
Release Notes:
0.4.4:
- BigQueryOperations:
- insert_rows_into_existing_table: insertion exception handling added.
- insert_rows_into_existing_table: retry added.
- Put time between table creation and insertion.
- execute_query: total_bytes_processed info added.
- execute_query: max allowed total_bytes_processed set to 100GB.
- execute_query: return_type=None for Queries w/o any return.
- load_data_to_table: kwargs['overwrite'] is added.
- load_data_to_table(..., overwrite=True) to overwrite to table.
- load_data_to_table(..., overwrite=False) to append to table.
- not passing overwrite kwarg will print a DeprecationWarning.
- execute_from_bq_file: sequential execution of BigQuery commands from
a file. It has its own parser.
- execute_from_bq_file(..., verbose=True) to print BigQuery commands to console.
- execute_from_bq_file(..., verbose=False) not to print BigQuery commands to console.
0.4.5:
- LoggingOperations
- Bug Fix in Authentication to GCP Logging !
- BigQueryOperations
- Executing BQ files for different environments !
0.4.6:
- BigQueryOperations
- BQ Parser bug fix !
- BQ File Execution dependent queries
- Some of the queries depends on the previous command executions. For these cases: dependent_queries is needed to be set to True ! execute_from_bq_file( bq_ds, "tests/test_data/test_case_4.bq", verbose=True, config=configs, dependent_queries=True)
- LoggingOperations
- protobuf dependency issue resolved!
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
heps_ds_utils-0.4.6a1.tar.gz
(14.3 kB
view hashes)
Built Distribution
Close
Hashes for heps_ds_utils-0.4.6a1-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 4dfb96ac62bd4b479382fcc1fc88196a6f937868781f2e1186afad2cfaa361b5 |
|
MD5 | 82447847e9c13130fa98f9f1653ad073 |
|
BLAKE2b-256 | 9e97cf6cf07d92874c007e20940345e459e22d5ca702e011e5638c0a519611a3 |