Automating Pricing Estimator Testing with a Flask‑Based Python Service Integrated into Java

In the previous article the pricing estimator system was introduced and the challenges of testing its mapping logic were outlined; this follow‑up explains how the identified problems were solved and further optimizations were made.

Background

The estimator workflow involves many mapping steps (SPU, external inspection items, etc.) configured manually in Excel files, making validation time‑consuming and error‑prone when done manually.

Overall Process

The first version used a Python script to filter Excel data. The new version service‑ifies the script, adds a front‑end UI, and connects Java, Python, and the inspection service via HTTP.

Users trigger a request from the front end, the Java service calls the inspection service, parses the report, and forwards the request to the Python service, which performs the mapping and writes results to the server; the Java service then reads the results and returns them for display.

Solution 3.1 – Python Script Maintenance

Problem

The script was embedded in the Java service, requiring a full redeployment for any change.

Solution

Encapsulate the script as an HTTP endpoint using Flask, allowing independent updates without affecting other services.

Implementation

Flask was chosen for its lightweight nature. A minimal Flask app is shown below:

from flask import Flask, request
app = Flask(__name__)

@app.route('/out')
def hello_world():
    return 'Hello World!'

if __name__ == '__main__':
    app.run(host='127.0.0.1', port=5000, debug=True)

Running this script and visiting http://127.0.0.1:5000/out displays "Hello World!".

The service defines an endpoint /req that accepts GET and POST, extracts parameters via request.values, and performs the mapping logic.

Solution 3.2 – Excel File Handling

Problem

Excel mapping templates were packaged inside the Java service, causing frequent redeployments when they changed and path issues on the server.

Solution

Store Excel files on the server separately; updating the mapping only requires uploading a new Excel file.

Overall Effect

The first‑phase tool wrote mapping results to Excel; the new front‑end now displays results directly. In a test scenario, entering category, scenario, and inspection code and clicking execute completes the entire pricing flow in about 3 seconds, showing the final price and each mapping condition clearly.

Future Improvements

Decouple Java and Python services to avoid requiring them on the same server for Excel access.

Enable diffing of expected vs. actual results and display differences on the front end.