Web Testing Adoption Dynamics Analyzer

This repository contains the source code and methodology developed for the thesis: "A Software Framework to Investigate End-to-End Web Testing Adoption Dynamics".

The project implements a systematic workflow to investigate the motivations behind adoption and migration of web GUI testing frameworks (Selenium, Cypress, Puppeteer, Playwright) in open-source projects.

📖 Project Overview

This software was designed to enrich E2EGit dataset with qualitative insights on why projects choose to adopt or migrate between web testing frameworks. Starting from a set of manually identified adoption and migration events (provided as input Excel files), the framework automates the retrieval of historical context to reconstruct the decision-making process.

The methodology, described in the thesis, follows a mixed-methods approach:

1. Automated Context Extraction: Retrieving specific commit sequences and issue discussions surrounding the known transition dates.
2. Noise Reduction: Applying a specialized keyword-based taxonomy to filter relevant content.
3. Manual Classification: Classifying the motivations through a custom "Human-in-the-loop" GUI.
4. Taxonomic Analysis: Categorizing the rationale (e.g., Developer Experience & Usability, Performance & Efficiency) to characterize the transition landscape.

📂 Project Structure

The codebase is organized into modular components reflecting the research phases:

Module	Description
01_data_mining	Retrieval & Filtering: Processes the input Excel files to fetch and filter commits/issues from GitHub/Git.
02_manual_labeling	Validation Interface: A Flask Web GUI for the manual review of the filtered candidates.
03_db_integration	Data Enrichment: Integrates the classified qualitative data back into the E2EGit dataset schema.
04_statistical_analysis	Characterization: Generates the empirical statistics and visualizations presented in the results.
core	Shared Logic: Database models, configuration, and utility functions.
resources	Data Storage: Contains the Input Excel Files (events), intermediate JSONs, and the SQLite database.

⚠️ Execution Rule

IMPORTANT: This project is structured as a Python package. All commands must be executed from the ROOT directory using the -m flag.

• ✅ Correct: python -m module_name.cli ...
• ❌ Incorrect: cd module_name && python cli.py

For more details, refer to each module's help command:

python -m module_name.cli --help

🚀 Setup

1. (optional) Create and activate a virtual environment:

python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

2. Install the required dependencies:

pip install -r requirements.txt

3. Configure Environment:

Create a .env file in resources/ (use resources/.env.example as a template) to configure the DB path and GitHub Token.

⚙️ Workflow

To replicate the study, ensure the input files (creation-adoption-gui.xlsx, migration_analysis.xlsx) are present in the resources/ folder.

1. Data Retrieval

The system reads the input Excel files containing the transition events.

1.1. Retrieve transition-related commits and 10 preceding commits:

python -m 01_data_mining.cli --task retrieve --target commit --type adoption
python -m 01_data_mining.cli --task retrieve --target commit --type migration

1.2. Retrieve all issues of the repositories involved in transition events:

python -m 01_data_mining.cli --task retrieve --target issue

If you need to download issues only for a specific repository, add the --repo owner/repo argument

python -m 01_data_mining.cli --task retrieve --target issue --repo owner/repo

2. Data Filtering

2.1. Filter commits by keywords:

python -m 01_data_mining.cli --task filter --target commit --type adoption
python -m 01_data_mining.cli --task filter --target commit --type migration

2.2. Filter issues by time window (10 days before transition) and keywords:

python -m 01_data_mining.cli --task filter --target issue --type adoption
python -m 01_data_mining.cli --task filter --target issue --type migration

3. Manual Review

3.1. Convert filtered JSONs to Excel for manual labeling:

python -m 02_manual_labeling.cli --task convert --target commit --type adoption
python -m 02_manual_labeling.cli --task convert --target commit --type migration
python -m 02_manual_labeling.cli --task convert --target issue --type adoption
python -m 02_manual_labeling.cli --task convert --target issue --type migration

3.2. Launch the Flask Web Application:

Load the manual labeling interface to classify the filtered candidates.

python -m 02_manual_labeling.cli --task server --target commit --type adoption
python -m 02_manual_labeling.cli --task server --target commit --type migration
python -m 02_manual_labeling.cli --task server --target issue --type adoption
python -m 02_manual_labeling.cli --task server --target issue --type migration

4. Data Refinement

After manual review define the taxonomic labels in the Excel files and create a new Excel file containing transition events linked to their motivations and associated issue number/commit hash. You can find the files we used for the thesis in the resources folder.

5. Database Integration

Integrate the classified data back into the E2EGit dataset SQLite database.

python -m 03_db_integration.cli

6. Statistical Analysis

Generate the statistics and visualizations for the final analysis.

python -m 04_statistical_analysis.cli

🎓 Reference

If you use this tool, please cite:

@mastersthesis{dimartino2025software,
  author  = {Giuseppe Di Martino},
  title   = {A Software Framework to Investigate End-to-End Web Testing Adoption Dynamics},
  school  = {University of Naples Federico II},
  year    = {2025},
  type    = {Bachelor's Thesis},
  note    = {B.Sc. Thesis in Computer Science}
}