Author: Marcus Bravidor ([email protected]) Version: PyDSout 0.1 Date: 2018-04-04 License: MIT License

Purpose: Automatically request firm-level financial data (e.g., WorldScope) from Thomson Reuters DataStream and save the results to a long format CSV file (separator: semicolon or ";"). The resulting file will look this:

Requirements: Valid DataStream license and credentials (usually supplied by your company or institution).

Acknowledgements: This script is based on PyDataStream by Vladimir Filimonov.

1. Setup
2. Customize the request
3. Resulting file
4. Troubleshooting

Install the necessary packages via the pip installer:

pip install pandas
pip install suds
pip install json
pip install pydatastream

Copy PyDSout.ipynb, firms.txt, and items.txt in the same folder.

Start Jupyter Notebook and you're ready to go.

Use the following steps to start a customized request:

• Time span: Change the key variables to your desired values:

  • enddate: Date for last request in ISO format (YYYY-MM-DD)
  • startdate: Date for first request in ISO format (YYYY-MM-DD)
  • freq: Frequency of the request. Possible values: Yearly (Y), Monthly (M), Weekly (W), Daily (D)

• Firms to request: Add a file named firms.txt to the root directory. Format:

DSID1
DSID2
DSID3

756944
866922
278419

• Items to be requested per firm: Add a file named items.txt to the root directory. Format:

code1,code2,code3,
NOSHFF,WC05301,WC07101,

The resulting file is a semicolon-separated (";") long-format CSV file called ds_output.csv.

Unexpected errors or connections breaks while requesting data

1. Continue with the conversion to a structured list.
2. Since the last entry in dobject is incomplete, the export to CSV would result in an error. Therefore, change the line for i in range(1,len(dobject)): to for i in range(1,len(dobject)-1):.
3. Remove the IDs of firms with completed requests from firms.txt.

Important note: PyDSout automatically replaces existing export files. To save your export, you should either rename the name of automatically created CSV file or name of the next export file by editing line df.to_csv('ds_output.csv', sep=';', encoding='utf-8'). The numerous resulting files can be merged later one with your preferred statistical software or using the pandas package in Python.