tests/test_github.py

"""
Unit tests for the github module.
"""

import unittest
from unittest.mock import patch, MagicMock
import requests
from functions import github

# pylint: disable=protected-access

class TestExtractGitHubUsername(unittest.TestCase):
    """Test cases for the _extract_github_username function."""

    def test_valid_github_urls(self):
        """Test extraction from valid GitHub URLs."""
        test_cases = [
            ("https://github.com/octocat", "octocat"),
            ("https://github.com/octocat/", "octocat"),
            ("http://github.com/test-user", "test-user"),
            ("github.com/user_name", "user_name"),
            ("https://github.com/user123", "user123"),
            ("https://github.com/octocat/Hello-World", "octocat"),
        ]

        for url, expected in test_cases:
            with self.subTest(url=url):
                result = github._extract_github_username(url)
                self.assertEqual(result, expected)

    def test_invalid_github_urls(self):
        """Test handling of invalid GitHub URLs."""
        invalid_urls = [
            "",
            "https://gitlab.com/user",
            "https://github.com/",
            "[email protected]",
            "https://github.com/user-with-very-long-name-that-exceeds-github-limit",
            None
        ]

        for url in invalid_urls:
            with self.subTest(url=url):
                result = github._extract_github_username(url)
                self.assertIsNone(result)

    def test_username_validation(self):
        """Test username format validation."""
        # Valid usernames
        valid_usernames = ["user", "user-name", "user_name", "user123", "a" * 39]
        for username in valid_usernames:
            with self.subTest(username=username):
                result = github._extract_github_username(f"github.com/{username}")
                self.assertEqual(result, username)

        # Invalid usernames
        invalid_usernames = ["", "a" * 40, "user@name", "user.name"]
        for username in invalid_usernames:
            with self.subTest(username=username):
                result = github._extract_github_username(f"github.com/{username}")
                self.assertIsNone(result)


class TestGetGitHubUserInfo(unittest.TestCase):
    """Test cases for the _get_github_user_info function."""

    @patch('requests.get')
    def test_successful_user_info(self, mock_get):
        """Test successful user info retrieval."""
        mock_response = MagicMock()
        mock_response.status_code = 200
        mock_response.json.return_value = {
            "login": "octocat",
            "public_repos": 10,
            "name": "The Octocat"
        }
        mock_get.return_value = mock_response

        result = github._get_github_user_info("octocat")

        self.assertEqual(result["status"], "success")
        self.assertIn("data", result)
        self.assertEqual(result["data"]["login"], "octocat")

    @patch('requests.get')
    def test_user_not_found(self, mock_get):
        """Test handling of non-existent user."""
        mock_response = MagicMock()
        mock_response.status_code = 404
        mock_get.return_value = mock_response

        result = github._get_github_user_info("nonexistentuser")

        self.assertEqual(result["status"], "error")
        self.assertIn("not found", result["message"])

    @patch('requests.get')
    def test_rate_limit_exceeded(self, mock_get):
        """Test handling of rate limit exceeded."""
        mock_response = MagicMock()
        mock_response.status_code = 403
        mock_get.return_value = mock_response

        result = github._get_github_user_info("octocat")

        self.assertEqual(result["status"], "error")
        self.assertIn("rate limit", result["message"])

    @patch('requests.get')
    def test_network_error(self, mock_get):
        """Test handling of network errors."""
        mock_get.side_effect = requests.RequestException("Connection error")

        result = github._get_github_user_info("octocat")

        self.assertEqual(result["status"], "error")
        self.assertIn("Network error", result["message"])


class TestGetUserRepositories(unittest.TestCase):
    """Test cases for the _get_user_repositories function."""

    @patch('requests.get')
    def test_successful_repository_retrieval(self, mock_get):
        """Test successful repository retrieval."""
        mock_response = MagicMock()
        mock_response.status_code = 200
        mock_response.json.return_value = [
            {
                "name": "Hello-World",
                "description": "My first repository",
                "language": "Python",
                "stargazers_count": 5,
                "forks_count": 2
            }
        ]
        mock_get.return_value = mock_response

        result = github._get_user_repositories("octocat")

        self.assertEqual(result["status"], "success")
        self.assertIn("data", result)
        self.assertEqual(len(result["data"]), 1)
        self.assertEqual(result["data"][0]["name"], "Hello-World")

    @patch('requests.get')
    def test_empty_repository_list(self, mock_get):
        """Test handling of empty repository list."""
        mock_response = MagicMock()
        mock_response.status_code = 200
        mock_response.json.return_value = []
        mock_get.return_value = mock_response

        result = github._get_user_repositories("octocat")

        self.assertEqual(result["status"], "success")
        self.assertEqual(len(result["data"]), 0)

    @patch('requests.get')
    def test_api_error(self, mock_get):
        """Test handling of API errors."""
        mock_response = MagicMock()
        mock_response.status_code = 500
        mock_get.return_value = mock_response

        result = github._get_user_repositories("octocat")

        self.assertEqual(result["status"], "error")
        self.assertIn("GitHub API error", result["message"])


class TestProcessRepositoryData(unittest.TestCase):
    """Test cases for the _process_repository_data function."""

    def test_basic_processing(self):
        """Test basic repository data processing."""
        raw_repos = [
            {
                "name": "test-repo",
                "description": "Test repository",
                "language": "Python",
                "stargazers_count": 10,
                "forks_count": 5,
                "updated_at": "2024-01-01T00:00:00Z",
                "html_url": "https://github.com/user/test-repo",
                "topics": ["python", "test"],
                "fork": False
            }
        ]

        result = github._process_repository_data(raw_repos)

        self.assertEqual(len(result), 1)
        processed_repo = result[0]
        self.assertEqual(processed_repo["name"], "test-repo")
        self.assertEqual(processed_repo["description"], "Test repository")
        self.assertEqual(processed_repo["language"], "Python")
        self.assertEqual(processed_repo["stars"], 10)
        self.assertEqual(processed_repo["forks"], 5)
        self.assertFalse(processed_repo["is_fork"])

    def test_fork_filtering(self):
        """Test filtering of unmodified forks."""
        raw_repos = [
            {
                "name": "original-repo",
                "fork": False,
                "stargazers_count": 5
            },
            {
                "name": "unmodified-fork",
                "fork": True,
                "stargazers_count": 0
            },
            {
                "name": "modified-fork",
                "fork": True,
                "stargazers_count": 3
            }
        ]

        result = github._process_repository_data(raw_repos)

        # Should include original repo and modified fork, exclude unmodified fork
        self.assertEqual(len(result), 2)
        repo_names = [repo["name"] for repo in result]
        self.assertIn("original-repo", repo_names)
        self.assertIn("modified-fork", repo_names)
        self.assertNotIn("unmodified-fork", repo_names)

    def test_missing_fields(self):
        """Test handling of missing fields in repository data."""
        raw_repos = [
            {
                "name": "minimal-repo"
                # Missing most optional fields
            }
        ]

        result = github._process_repository_data(raw_repos)

        self.assertEqual(len(result), 1)
        processed_repo = result[0]
        self.assertEqual(processed_repo["name"], "minimal-repo")
        self.assertEqual(processed_repo["description"], "")
        self.assertEqual(processed_repo["language"], "")
        self.assertEqual(processed_repo["stars"], 0)
        self.assertEqual(processed_repo["forks"], 0)


class TestGetGitHubRepositories(unittest.TestCase):
    """Test cases for the get_github_repositories function."""

    def test_empty_url(self):
        """Test handling of empty or None URL."""
        test_cases = [None, "", "   "]

        for url in test_cases:
            with self.subTest(url=url):
                result = github.get_github_repositories(url)
                self.assertEqual(result["status"], "error")
                self.assertIn("No GitHub URL provided", result["message"])

    @patch('functions.github._get_user_repositories')
    @patch('functions.github._get_github_user_info')
    @patch('functions.github._extract_github_username')
    def test_successful_retrieval(self, mock_extract, mock_user_info, mock_repos):
        """Test successful repository retrieval."""
        mock_extract.return_value = "octocat"
        mock_user_info.return_value = {
            "status": "success",
            "data": {"public_repos": 10}
        }
        mock_repos.return_value = {
            "status": "success",
            "data": [{"name": "Hello-World", "fork": False, "stargazers_count": 5}]
        }

        result = github.get_github_repositories("https://github.com/octocat")

        self.assertEqual(result["status"], "success")
        self.assertIn("repositories", result)
        self.assertIn("metadata", result)
        self.assertEqual(result["metadata"]["username"], "octocat")

    def test_invalid_url_format(self):
        """Test handling of invalid URL format."""
        result = github.get_github_repositories("https://gitlab.com/user")

        self.assertEqual(result["status"], "error")
        self.assertIn("Invalid GitHub URL format", result["message"])


class TestFormatRepositoriesForLLM(unittest.TestCase):
    """Test cases for the format_repositories_for_llm function."""

    def test_successful_formatting(self):
        """Test successful formatting of repository data."""
        github_result = {
            "status": "success",
            "repositories": [
                {
                    "name": "test-repo",
                    "description": "A test repository",
                    "language": "Python",
                    "stars": 10,
                    "forks": 5,
                    "updated_at": "2024-01-01T00:00:00Z",
                    "html_url": "https://github.com/user/test-repo",
                    "topics": ["python", "test"]
                }
            ],
            "metadata": {
                "username": "testuser",
                "profile_url": "https://github.com/testuser"
            }
        }

        result = github.format_repositories_for_llm(github_result)

        self.assertIn("=== GITHUB REPOSITORIES ===", result)
        self.assertIn("testuser", result)
        self.assertIn("test-repo", result)
        self.assertIn("A test repository", result)
        self.assertIn("Python", result)
        self.assertIn("=== END GITHUB REPOSITORIES ===", result)

    def test_error_status(self):
        """Test formatting when there's an error status."""
        github_result = {
            "status": "error",
            "message": "User not found"
        }

        result = github.format_repositories_for_llm(github_result)

        self.assertIn("could not be retrieved", result)
        self.assertIn("User not found", result)

    def test_no_repositories(self):
        """Test formatting when no repositories are found."""
        github_result = {
            "status": "success",
            "repositories": [],
            "metadata": {"username": "emptyuser"}
        }

        result = github.format_repositories_for_llm(github_result)

        self.assertIn("No public repositories found", result)
        self.assertIn("emptyuser", result)

    def test_many_repositories_limit(self):
        """Test that formatting limits to 20 repositories."""
        repositories = []
        for i in range(25):
            repositories.append({
                "name": f"repo-{i}",
                "description": f"Repository {i}",
                "language": "Python",
                "stars": i,
                "forks": 0,
                "updated_at": "2024-01-01T00:00:00Z",
                "html_url": f"https://github.com/user/repo-{i}",
                "topics": []
            })

        github_result = {
            "status": "success",
            "repositories": repositories,
            "metadata": {"username": "manyrepos"}
        }

        result = github.format_repositories_for_llm(github_result)

        # Should mention "and 5 more repositories"
        self.assertIn("and 5 more repositories", result)
        # Should contain the first 20 repos
        self.assertIn("repo-0", result)
        self.assertIn("repo-19", result)
        # Should not contain repos beyond 20
        self.assertNotIn("repo-20", result)


class TestGetRepositoryDetails(unittest.TestCase):
    """Test cases for the get_repository_details function."""

    def test_invalid_repository_url(self):
        """Test handling of invalid repository URLs."""
        invalid_urls = [
            "",
            None,
            "https://gitlab.com/user/repo",
            "https://github.com/",
            "https://github.com/user",
            "not-a-url",
        ]

        for url in invalid_urls:
            with self.subTest(url=url):
                result = github.get_repository_details(url)
                self.assertEqual(result["status"], "error")
                self.assertIn("message", result)

    @patch('functions.github._get_repository_info')
    @patch('functions.github._extract_repo_info')
    def test_repository_not_found(self, mock_extract, mock_get_info):
        """Test handling of non-existent repository."""
        mock_extract.return_value = ("user", "nonexistent-repo")
        mock_get_info.return_value = {
            "status": "error",
            "message": "Repository 'user/nonexistent-repo' not found"
        }

        result = github.get_repository_details("https://github.com/user/nonexistent-repo")

        self.assertEqual(result["status"], "error")
        self.assertIn("not found", result["message"])

    @patch('functions.github._get_repository_contributors')
    @patch('functions.github._get_repository_releases')
    @patch('functions.github._get_repository_contents')
    @patch('functions.github._get_repository_readme')
    @patch('functions.github._get_repository_languages')
    @patch('functions.github._get_repository_info')
    @patch('functions.github._extract_repo_info')
    def test_successful_repository_details(self, mock_extract, mock_get_info,
                                         mock_languages, mock_readme, mock_contents,
                                         mock_releases, mock_contributors):
        """Test successful repository details retrieval."""
        # Mock URL extraction
        mock_extract.return_value = ("octocat", "Hello-World")

        # Mock basic repository info
        mock_get_info.return_value = {
            "status": "success",
            "data": {
                "name": "Hello-World",
                "full_name": "octocat/Hello-World",
                "description": "This your first repo!",
                "language": "C",
                "stargazers_count": 80,
                "forks_count": 9,
                "watchers_count": 80,
                "size": 108,
                "created_at": "2011-01-26T19:01:12Z",
                "updated_at": "2011-01-26T19:14:43Z",
                "pushed_at": "2011-01-26T19:06:43Z",
                "html_url": "https://github.com/octocat/Hello-World",
                "clone_url": "https://github.com/octocat/Hello-World.git",
                "ssh_url": "[email protected]:octocat/Hello-World.git",
                "topics": ["example", "tutorial"],
                "license": {"name": "MIT License", "spdx_id": "MIT"},
                "fork": False,
                "archived": False,
                "private": False,
                "default_branch": "master",
                "open_issues_count": 0,
                "has_issues": True,
                "has_wiki": True,
                "has_pages": False,
                "has_projects": True,
                "visibility": "public"
            }
        }

        # Mock additional data
        mock_languages.return_value = {
            "status": "success",
            "data": {"C": 78.1, "Makefile": 21.9}
        }

        mock_readme.return_value = {
            "status": "success",
            "data": "# Hello World\n\nThis is a test repository."
        }

        mock_contents.return_value = {
            "status": "success",
            "data": ["README.md", "hello.c", "Makefile"]
        }

        mock_releases.return_value = {
            "status": "success",
            "data": [
                {
                    "tag_name": "v1.0.0",
                    "name": "First Release",
                    "published_at": "2011-01-26T19:14:43Z",
                    "prerelease": False,
                    "draft": False
                }
            ]
        }

        mock_contributors.return_value = {
            "status": "success",
            "data": [
                {
                    "login": "octocat",
                    "contributions": 32,
                    "html_url": "https://github.com/octocat",
                    "type": "User"
                }
            ]
        }

        result = github.get_repository_details("https://github.com/octocat/Hello-World")

        # Verify success
        self.assertEqual(result["status"], "success")
        self.assertIn("repository", result)

        repo = result["repository"]

        # Verify basic info
        self.assertEqual(repo["name"], "Hello-World")
        self.assertEqual(repo["full_name"], "octocat/Hello-World")
        self.assertEqual(repo["description"], "This your first repo!")
        self.assertEqual(repo["language"], "C")
        self.assertEqual(repo["stars"], 80)
        self.assertEqual(repo["forks"], 9)

        # Verify additional data
        self.assertEqual(repo["languages"], {"C": 78.1, "Makefile": 21.9})
        self.assertIn("Hello World", repo["readme"])
        self.assertEqual(repo["file_structure"], ["README.md", "hello.c", "Makefile"])
        self.assertEqual(len(repo["releases"]), 1)
        self.assertEqual(repo["releases"][0]["tag_name"], "v1.0.0")
        self.assertEqual(len(repo["contributors"]), 1)
        self.assertEqual(repo["contributors"][0]["login"], "octocat")

        # Verify boolean flags
        self.assertFalse(repo["is_fork"])
        self.assertFalse(repo["is_archived"])
        self.assertFalse(repo["is_private"])
        self.assertTrue(repo["has_issues"])

    def test_extract_repo_info_valid_urls(self):
        """Test _extract_repo_info with valid repository URLs."""
        test_cases = [
            ("https://github.com/octocat/Hello-World", ("octocat", "Hello-World")),
            ("https://github.com/user/repo.git", ("user", "repo")),
            ("https://github.com/org/project/", ("org", "project")),
            ("github.com/test/example", ("test", "example")),
            ("https://github.com/user/repo/issues", ("user", "repo")),
        ]

        for url, expected in test_cases:
            with self.subTest(url=url):
                result = github._extract_repo_info(url)
                self.assertEqual(result, expected)

    def test_extract_repo_info_invalid_urls(self):
        """Test _extract_repo_info with invalid repository URLs."""
        invalid_urls = [
            "",
            "https://gitlab.com/user/repo",
            "https://github.com/user",
            "https://github.com/",
            "not-a-url",
        ]

        for url in invalid_urls:
            with self.subTest(url=url):
                result = github._extract_repo_info(url)
                self.assertEqual(result, (None, None))

    @patch('requests.get')
    def test_get_repository_info_success(self, mock_get):
        """Test _get_repository_info with successful response."""
        mock_response = MagicMock()
        mock_response.status_code = 200
        mock_response.json.return_value = {
            "name": "test-repo",
            "full_name": "user/test-repo"
        }
        mock_get.return_value = mock_response

        result = github._get_repository_info("user", "test-repo")

        self.assertEqual(result["status"], "success")
        self.assertIn("data", result)
        self.assertEqual(result["data"]["name"], "test-repo")

    @patch('requests.get')
    def test_get_repository_info_not_found(self, mock_get):
        """Test _get_repository_info with 404 response."""
        mock_response = MagicMock()
        mock_response.status_code = 404
        mock_get.return_value = mock_response

        result = github._get_repository_info("user", "nonexistent")

        self.assertEqual(result["status"], "error")
        self.assertIn("not found", result["message"])

    @patch('requests.get')
    def test_get_repository_languages_success(self, mock_get):
        """Test _get_repository_languages with successful response."""
        mock_response = MagicMock()
        mock_response.status_code = 200
        mock_response.json.return_value = {
            "Python": 50000,
            "JavaScript": 25000,
            "CSS": 25000
        }
        mock_get.return_value = mock_response

        result = github._get_repository_languages("user", "repo")

        self.assertEqual(result["status"], "success")
        expected_percentages = {"Python": 50.0, "JavaScript": 25.0, "CSS": 25.0}
        self.assertEqual(result["data"], expected_percentages)

    @patch('requests.get')
    def test_get_repository_readme_success(self, mock_get):
        """Test _get_repository_readme with successful response."""
        # Mock the README metadata response
        readme_response = MagicMock()
        readme_response.status_code = 200
        readme_response.json.return_value = {
            "download_url": "https://raw.githubusercontent.com/user/repo/main/README.md"
        }

        # Mock the README content response
        content_response = MagicMock()
        content_response.status_code = 200
        content_response.text = "# Test Repository\n\nThis is a test."

        mock_get.side_effect = [readme_response, content_response]

        result = github._get_repository_readme("user", "repo")

        self.assertEqual(result["status"], "success")
        self.assertIn("Test Repository", result["data"])

    @patch('requests.get')
    def test_get_repository_contents_success(self, mock_get):
        """Test _get_repository_contents with successful response."""
        mock_response = MagicMock()
        mock_response.status_code = 200
        mock_response.json.return_value = [
            {"name": "src", "type": "dir"},
            {"name": "README.md", "type": "file"},
            {"name": "setup.py", "type": "file"},
            {"name": "tests", "type": "dir"}
        ]
        mock_get.return_value = mock_response

        result = github._get_repository_contents("user", "repo")

        self.assertEqual(result["status"], "success")
        expected_structure = ["src/", "tests/", "README.md", "setup.py"]
        self.assertEqual(result["data"], expected_structure)


if __name__ == '__main__':
    unittest.main()