Using an AI Large Model to Automate Report Comparison Testing

Background

When a new version of a reporting system is released, regression testing must verify that unchanged metrics are not inadvertently affected. Manually exporting multiple reports and comparing each field is time‑consuming and error‑prone.

Prompt 1: Define the Comparison Task

The author asks the Hunyuan model: “You have two reports to compare and need to mark differences in red. What steps are required?” The model replies with a high‑level approach, suggesting that Python code be generated to automate the comparison.

Prompt 2: Generate Python Code

Using the follow‑up prompt “Can you implement this comparison with ExcelA.xls and ExcelB.xls?”, the model produces a Python script that reads the two Excel files, detects differing cells, and prints the results. The author saves the script and runs it in Visual Studio Code, confirming that the output correctly lists the differences.

Iterative Optimization

To make the script more flexible, the author asks the model to replace hard‑coded filenames with user‑provided paths. The revised code now accepts command‑line arguments for the file locations. Further prompts request that differing cells be displayed in bold‑italic font and that the solution support more than two files. The model updates the script accordingly, adding input handling for an arbitrary number of Excel files and formatting the output.

Execution in VS Code

The final script is copied into Visual Studio Code and executed. Console output shows correct detection of single‑cell differences and, after modifying one file to contain multiple differing values, the script still reports all discrepancies. When the formatting prompt is applied, the generated Excel files highlight differing cells in bold‑italic style as expected.

Conclusion

The demonstration confirms that Tencent’s Hunyuan large model can quickly produce functional Python code for report comparison, adapt to user‑defined filenames, format results, and scale to multiple files. Leveraging a large‑model assistant thus accelerates automated testing, improves coverage, and reduces the manual effort traditionally required for large‑scale report validation.