Skip to content

Latest commit

 

History

History
99 lines (78 loc) · 3.14 KB

README.md

File metadata and controls

99 lines (78 loc) · 3.14 KB
Raw

bcpy

Latest Release latest release
License license
Build Status (master) travis build status

What is it?

This package is a wrapper for Microsoft's SQL Server bcp utility. Current database drivers available in Python are not fast enough for transferring millions of records (yes, I have tried pyodbc fast_execute_many). Despite the IO hits, the fastest option by far is saving the data to a CSV file in file system (preferably /dev/shm tmpfs) and using the bcp utility to transfer the CSV file to SQL Server.

How Can I Install It?

  1. Make sure your computeer has the requirements.
  2. You can download and install this package from PyPI repository by running the command below.
pip install bcpy

Examples

Following examples show you how to load (1) flat files and (2) DataFrame objects to SQL Server using this package.

Flat File

Following example assumes that you have a comma separated file with no qualifier in path 'tests/data1.csv'. The code below sends the the file to SQL Server.

import bcpy


sql_config = {
    'server': 'sql_server_hostname',
    'database': 'database_name',
    'username': 'test_user',
    'password': 'test_user_password1234'
}
sql_table_name = 'test_data1'
csv_file_path = 'tests/data1.csv'
flat_file = bcpy.FlatFile(qualifier='', path=csv_file_path)
sql_table = bcpy.SqlTable(sql_config, table=sql_table_name)
flat_file.to_sql(sql_table)

DataFrame

The following example creates a DataFrame with 100 rows and 4 columns populated with random data and then it sends it to SQL Server.

import bcpy
import numpy as np
import pandas as pd


sql_config = {
    'server': 'sql_server_hostname',
    'database': 'database_name',
    'username': 'test_user',
    'password': 'test_user_password1234'
}
table_name = 'test_dataframe'
df = pd.DataFrame(np.random.randint(-100, 100, size=(100, 4)),
                  columns=list('ABCD'))
bdf = bcpy.DataFrame(df)
sql_table = bcpy.SqlTable(sql_config, table=table_name)
bdf.to_sql(sql_table)

Requirements

You need a working version of Microsoft bcp installed in your system. Your PATH environment variable should contain the directory of the bcp utility. Following are the installation tutorials for different operating systems.