feat(test): add test command to e2e test the review command

[fabienzucchet]

Fixes #19

Behaviour of this feature:

• new command test running test
• tests are defined in natural language (test case) and a snippet is generated (i.e. a code snippet corresponding to the code description). An example of the expected answer is also provided (snapshot).
• The models runs a review of the tests code snippets, then compare the result to the snapshots
• if the two reviews are close enough (threshold defined in constants.ts), the test PASS
• if they are relatively close, the tests is WARN and the two reviews are displayed for manual review
• else, the test fails and the two reviews are also displayed for debugging purpose
• ci integration

[github-actions]

LOGAF Level 3 - /home/runner/work/code-review-gpt/code-review-gpt/src/review/prompt/constructPrompt.ts

Consider using a more efficient algorithm for splitting files into batches. The current implementation has a time complexity of O(n^2), which could be improved to O(n) by keeping track of the current batch size.

---

LOGAF Level 3 - /home/runner/work/code-review-gpt/code-review-gpt/src/test/run/generateTestReport.ts

1. The generateTestReport function has a lot of repeated code in the if-else statements. This could be simplified by creating a helper function that takes the test result and message as parameters and returns the formatted string. For example:

function formatTestReport(result: testResult, message: string) {
  switch (result) {
    case testResult.PASS:
      return chalk.green(message);
    case testResult.WARN:
      return chalk.yellow(message);
    case testResult.FAIL:
      return chalk.red(message);
  }
}

2. The generateTestResultsSummary function could be simplified by using the reduce function to calculate the counts of each test result. For example:

const counts = Object.values(testResults).reduce((counts, result) => {
  counts[result]++;
  return counts;
}, { PASS: 0, WARN: 0, FAIL: 0 });

---

LOGAF Level 3 - /home/runner/work/code-review-gpt/code-review-gpt/src/test/load/loadTestCodeSnippets.ts

Consider handling errors when generating code snippets. Currently, if there is an error when generating a code snippet, the entire process will fail. It would be better to catch the error, log it, and continue with the remaining test cases.

---

🔄🧹🔍

---

Powered by Code Review GPT

[github-actions]

Test results summary:

[PASS] - Test case: Bad variable name
[PASS] - Test case: Exposed secret
[PASS] - Test case: Too many nested loops
[PASS] - Test case: Unawaited Promise

SUMMARY: ✅ PASS: 4, ⚠️ WARN: 0, ❌ FAIL: 0

---

Tests Powered by Code Review GPT

[StanGirard]

WTF ?! that is a fantastic feature ! This is a killer feature !!!!