Why This Matters

Test generation one-by-one is slow. A pipeline that processes a directory, generates tests for each file, runs them, and reports results scales to large codebases.

Concept Explained

A script that walks a source directory. For each file, calls Copilot to generate tests. Writes tests to a test directory. Runs the test suite. Reports pass/fail per file.

How It Works

Write a Python or Node script. Walk src/. For each .ts/.py file, call Copilot CLI to generate tests. Write to tests/. Run test suite. Report.

Step-by-Step Tutorial

1. Walk source directory

2. Generate tests

3. Write tests

4. Run test suite

5. Report

Real-World Example

A team ran the pipeline on a 50-file module with 0% test coverage. Generated tests for all 50. 35 passed first try. 15 needed iteration (manual fix or re-generate). Final: 45/50 files with passing tests. Coverage: 0% → 65% in 2 hours.

Example Prompts / Commands / Code

#!/usr/bin/env python3
# generate-tests.py
import subprocess
import os
from pathlib import Path

SRC_DIR = 'src'
TEST_DIR = 'tests'

for src_file in Path(SRC_DIR).rglob('*.ts'):
    if src_file.name.endswith('.test.ts'):
        continue

    test_file = Path(TEST_DIR) / (src_file.stem + '.test.ts')
    if test_file.exists():
        print(f'Skip {src_file} (test exists)')
        continue

    print(f'Generating tests for {src_file}')
    result = subprocess.run([
        'copilot', '--no-interactive',
        f'Generate vitest tests for {src_file}. Output as a single code block.'
    ], capture_output=True, text=True)

    # Extract code block from response
    code = extract_code_block(result.stdout)
    test_file.write_text(code)

print('Running tests...')
subprocess.run(['npm', 'test'])

# Report pass/fail per file

Common Mistakes

• Overwriting existing tests — destroys manual work.
• Not running tests after generation.
• Trusting generated tests without review.
• Not iterating on failures.

Best Practices

• Skip files with existing tests.
• Run tests after generation.
• Review generated tests before committing.
• Iterate on failures (manual fix or re-generate).
• Report pass/fail per file.

Troubleshooting

Practical Exercise

Professional Challenge

Frequently Asked Questions

Further Reading

• Writing tests with Copilot

Official References

• Writing tests with Copilot

Related lessons: BE-25, IN-33