tests/integration/test_authentication_flow.py

"""Integration tests for authentication flow."""
import pytest
from fastapi.testclient import TestClient
from unittest.mock import Mock, patch, AsyncMock
import json
import base64
from typing import Dict, Any

from main import app
from infrastructure.services.jwt_validation_service import JWTValidationService


class TestAuthenticationFlow:
    """Test complete authentication flow from request to response."""
    
    def setup_method(self):
        """Set up test fixtures."""
        self.client = TestClient(app)
        
        # Create valid JWT token structure for testing
        self.valid_token_header = {"alg": "HS256", "typ": "JWT"}
        self.valid_token_payload = {"sub": "test-user", "iat": 1234567890}
        
        # Create properly formatted JWT token (header.payload.signature)
        header_b64 = base64.urlsafe_b64encode(
            json.dumps(self.valid_token_header).encode()
        ).decode().rstrip('=')
        payload_b64 = base64.urlsafe_b64encode(
            json.dumps(self.valid_token_payload).encode()
        ).decode().rstrip('=')
        signature = "test-signature"
        
        self.valid_jwt_token = f"{header_b64}.{payload_b64}.{signature}"
        self.valid_auth_header = f"Bearer {self.valid_jwt_token}"
    
    def test_missing_authorization_header_returns_401(self):
        """Test that missing Authorization header returns 401."""
        response = self.client.get("/api/v1/jobs/test-job-id")
        
        assert response.status_code == 401
        assert response.json()["error"] == "Missing Authorization header"
        assert "WWW-Authenticate" in response.headers
        assert response.headers["WWW-Authenticate"] == "Bearer"
    
    def test_invalid_authorization_header_format_returns_401(self):
        """Test that invalid Authorization header format returns 401."""
        invalid_headers = [
            {"Authorization": "Invalid token"},  # Missing Bearer prefix
            {"Authorization": "Basic dGVzdA=="},  # Wrong auth type
            {"Authorization": "Bearer"},  # Missing token
            {"Authorization": "Bearer "},  # Empty token
        ]
        
        for headers in invalid_headers:
            response = self.client.get("/api/v1/jobs/test-job-id", headers=headers)
            
            assert response.status_code == 401
            assert "Invalid Authorization header" in response.json()["error"] or \
                   "Empty bearer token" in response.json()["error"]
            assert "WWW-Authenticate" in response.headers
    
    def test_invalid_jwt_structure_returns_401(self):
        """Test that invalid JWT structure returns 401."""
        invalid_tokens = [
            "Bearer invalid-token",  # Not a JWT structure
            "Bearer one.two",  # Only two parts
            "Bearer one.two.three.four",  # Too many parts
            "Bearer .payload.signature",  # Empty header
            "Bearer header..signature",  # Empty payload
            "Bearer header.payload.",  # Empty signature
        ]
        
        for auth_header in invalid_tokens:
            response = self.client.get(
                "/api/v1/jobs/test-job-id", 
                headers={"Authorization": auth_header}
            )
            
            assert response.status_code == 401
            assert "Invalid JWT token structure" in response.json()["error"]
    
    @patch('infrastructure.services.jwt_validation_service.JWTValidationService.validate_structure')
    def test_valid_jwt_token_allows_access(self, mock_validate):
        """Test that valid JWT token allows access to protected endpoints."""
        mock_validate.return_value = True
        
        # Mock the job repository to avoid database dependency
        with patch('application.use_cases.check_job_status.CheckJobStatusUseCase.execute') as mock_execute:
            mock_execute.side_effect = Exception("Job not found: test-job-id")  # Expected for non-existent job
            
            response = self.client.get(
                "/api/v1/jobs/test-job-id",
                headers={"Authorization": self.valid_auth_header}
            )
            
            # Should get past authentication (even if job doesn't exist)
            assert response.status_code != 401
            mock_validate.assert_called_once_with(self.valid_jwt_token)
    
    def test_extraction_endpoint_requires_authentication(self):
        """Test that extraction endpoint requires authentication."""
        # Create a minimal video file for testing
        test_file_content = b"fake video content"
        
        response = self.client.post(
            "/api/v1/extract",
            files={"video": ("test.mp4", test_file_content, "video/mp4")},
            data={"output_format": "mp3", "quality": "medium"}
        )
        
        assert response.status_code == 401
        assert "Missing Authorization header" in response.json()["error"]
    
    @patch('infrastructure.services.jwt_validation_service.JWTValidationService.validate_structure')
    def test_extraction_endpoint_with_valid_token(self, mock_validate):
        """Test extraction endpoint with valid authentication."""
        mock_validate.return_value = True
        
        test_file_content = b"fake video content"
        
        # Mock all the dependencies to avoid actual processing
        with patch('interfaces.api.dependencies.get_use_cases') as mock_use_cases, \
             patch('interfaces.api.dependencies.get_services') as mock_services:
            
            # Mock services
            mock_file_repo = Mock()
            mock_file_repo.save_stream = AsyncMock(return_value="/tmp/test.mp4")
            mock_file_repo.delete_file = AsyncMock()
            mock_services.return_value.file_repository = mock_file_repo
            
            # Mock use cases
            mock_validation_service = Mock()
            mock_validation_service.validate_external_job_id = Mock()
            
            mock_extract_use_case = Mock()
            mock_extract_use_case.execute_with_job = AsyncMock(return_value=Mock(
                job_id="test-job-123",
                external_job_id=None,
                status="processing",
                message="Job created",
                check_url="/api/v1/jobs/test-job-123",
                file_size_mb=0.001
            ))
            
            mock_use_cases.return_value.validation_service = mock_validation_service
            mock_use_cases.return_value.extract_audio_async = mock_extract_use_case
            
            response = self.client.post(
                "/api/v1/extract",
                headers={"Authorization": self.valid_auth_header},
                files={"video": ("test.mp4", test_file_content, "video/mp4")},
                data={"output_format": "mp3", "quality": "medium"}
            )
            
            # Should get past authentication
            assert response.status_code != 401
            mock_validate.assert_called_with(self.valid_jwt_token)
    
    def test_download_endpoint_requires_authentication(self):
        """Test that download endpoint requires authentication."""
        response = self.client.get("/api/v1/jobs/test-job-id/download")
        
        assert response.status_code == 401
        assert "Missing Authorization header" in response.json()["error"]
    
    @patch('infrastructure.services.jwt_validation_service.JWTValidationService.validate_structure')
    def test_download_endpoint_with_valid_token(self, mock_validate):
        """Test download endpoint with valid authentication."""
        mock_validate.return_value = True
        
        # Mock the download use case to avoid database dependency
        with patch('application.use_cases.download_audio_result.DownloadAudioResultUseCase.execute') as mock_execute:
            mock_execute.side_effect = Exception("Job not found: test-job-id")  # Expected for non-existent job
            
            response = self.client.get(
                "/api/v1/jobs/test-job-id/download",
                headers={"Authorization": self.valid_auth_header}
            )
            
            # Should get past authentication (even if job doesn't exist)
            assert response.status_code != 401
            mock_validate.assert_called_once_with(self.valid_jwt_token)
    
    def test_info_endpoint_does_not_require_authentication(self):
        """Test that info endpoint is public and doesn't require authentication."""
        response = self.client.get("/api/v1/info")
        
        assert response.status_code == 200
        assert "version" in response.json()
        assert "supported_video_formats" in response.json()
    
    def test_authentication_error_response_format(self):
        """Test that authentication errors return proper response format."""
        response = self.client.get("/api/v1/jobs/test-job-id")
        
        assert response.status_code == 401
        response_data = response.json()
        
        # Verify error response structure
        assert "error" in response_data
        assert isinstance(response_data["error"], str)
        
        # Verify WWW-Authenticate header
        assert "WWW-Authenticate" in response.headers
        assert response.headers["WWW-Authenticate"] == "Bearer"
    
    @patch('infrastructure.services.jwt_validation_service.JWTValidationService.validate_structure')
    def test_bearer_token_passed_to_job_creation(self, mock_validate):
        """Test that bearer token is properly passed to job creation."""
        mock_validate.return_value = True
        
        test_file_content = b"fake video content"
        
        # Mock dependencies
        with patch('interfaces.api.dependencies.get_use_cases') as mock_use_cases, \
             patch('interfaces.api.dependencies.get_services') as mock_services, \
             patch('domain.entities.job.Job.create_new') as mock_create_job:
            
            # Setup mocks
            mock_services.return_value.file_repository.save_stream = AsyncMock(return_value="/tmp/test.mp4")
            mock_use_cases.return_value.validation_service.validate_external_job_id = Mock()
            
            mock_job = Mock()
            mock_job.id = "test-job-123"
            mock_create_job.return_value = mock_job
            
            mock_use_cases.return_value.extract_audio_async.execute_with_job = AsyncMock(
                return_value=Mock(
                    job_id="test-job-123",
                    external_job_id=None,
                    status="processing",
                    message="Job created",
                    check_url="/api/v1/jobs/test-job-123",
                    file_size_mb=0.001
                )
            )
            
            response = self.client.post(
                "/api/v1/extract",
                headers={"Authorization": self.valid_auth_header},
                files={"video": ("test.mp4", test_file_content, "video/mp4")},
                data={"output_format": "mp3", "quality": "medium"}
            )
            
            # Verify job creation was called with bearer token
            mock_create_job.assert_called_once()
            call_kwargs = mock_create_job.call_args.kwargs
            assert call_kwargs["bearer_token"] == self.valid_jwt_token
    
    def test_concurrent_authentication_requests(self):
        """Test that authentication works correctly under concurrent load."""
        import concurrent.futures
        import threading
        
        def make_authenticated_request(token_suffix: str):
            """Make a request with a unique token."""
            # Create unique token for each thread
            unique_token = f"{self.valid_jwt_token}-{token_suffix}"
            
            with patch('infrastructure.services.jwt_validation_service.JWTValidationService.validate_structure') as mock_validate:
                mock_validate.return_value = True
                
                response = self.client.get(
                    f"/api/v1/jobs/test-job-{token_suffix}",
                    headers={"Authorization": f"Bearer {unique_token}"}
                )
                
                return response.status_code, unique_token
        
        # Make multiple concurrent requests
        with concurrent.futures.ThreadPoolExecutor(max_workers=5) as executor:
            futures = [
                executor.submit(make_authenticated_request, str(i))
                for i in range(10)
            ]
            
            results = [future.result() for future in futures]
        
        # All requests should get past authentication (they'll fail on job lookup, but that's expected)
        for status_code, token in results:
            assert status_code != 401, f"Authentication failed for token {token}"


class TestJWTValidationService:
    """Test JWT validation service directly."""
    
    def setup_method(self):
        """Set up test fixtures."""
        self.service = JWTValidationService()
    
    def test_validate_structure_with_valid_jwt(self):
        """Test JWT structure validation with valid tokens."""
        valid_tokens = [
            "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzdWIiOiIxMjM0NTY3ODkwIiwibmFtZSI6IkpvaG4gRG9lIiwiaWF0IjoxNTE2MjM5MDIyfQ.SflKxwRJSMeKKF2QT4fwpMeJf36POk6yJV_adQssw5c",
            "header.payload.signature",
            "a.b.c",
        ]
        
        for token in valid_tokens:
            assert self.service.validate_structure(token) is True
    
    def test_validate_structure_with_invalid_jwt(self):
        """Test JWT structure validation with invalid tokens."""
        invalid_tokens = [
            "",  # Empty
            "invalid",  # Single part
            "one.two",  # Two parts
            "one.two.three.four",  # Four parts
            ".payload.signature",  # Empty header
            "header..signature",  # Empty payload
            "header.payload.",  # Empty signature
            None,  # None value
        ]
        
        for token in invalid_tokens:
            assert self.service.validate_structure(token) is False
    
    def test_validate_structure_edge_cases(self):
        """Test JWT validation with edge cases."""
        edge_cases = [
            ("a" * 1000 + "." + "b" * 1000 + "." + "c" * 1000, True),  # Very long token
            ("1.2.3", True),  # Numeric parts
            ("a-b_c.d-e_f.g-h_i", True),  # Special characters in parts
        ]
        
        for token, expected in edge_cases:
            assert self.service.validate_structure(token) is expected


class TestAuthenticationMiddleware:
    """Test authentication middleware behavior."""
    
    def test_authentication_preserves_request_data(self):
        """Test that authentication doesn't modify request data."""
        client = TestClient(app)
        
        original_data = {"output_format": "mp3", "quality": "high"}
        test_file = ("test.mp4", b"fake content", "video/mp4")
        
        # Make request without auth (will fail auth, but request should be preserved)
        response = client.post(
            "/api/v1/extract",
            files={"video": test_file},
            data=original_data
        )
        
        # Should fail on auth, not on request parsing
        assert response.status_code == 401
        assert "Missing Authorization header" in response.json()["error"]
    
    @patch('infrastructure.services.jwt_validation_service.JWTValidationService.validate_structure')
    def test_authentication_allows_request_processing(self, mock_validate):
        """Test that valid authentication allows normal request processing."""
        mock_validate.return_value = True
        client = TestClient(app)
        
        with patch('interfaces.api.dependencies.get_use_cases'), \
             patch('interfaces.api.dependencies.get_services'):
            
            response = client.get(
                "/api/v1/jobs/test-job",
                headers={"Authorization": "Bearer valid.jwt.token"}
            )
            
            # Should get past authentication (will fail on job lookup, but that's post-auth)
            assert response.status_code != 401


class TestTokenSecurity:
    """Test token security and handling."""
    
    def test_token_not_logged_in_error_responses(self):
        """Test that tokens are not included in error response bodies."""
        client = TestClient(app)
        
        sensitive_token = "Bearer eyJhbGciOiJIUzI1NiJ9.sensitive-payload.secret-signature"
        
        response = client.get(
            "/api/v1/jobs/test-job",
            headers={"Authorization": sensitive_token}
        )
        
        response_text = response.text
        
        # Token should not appear in response
        assert "sensitive-payload" not in response_text
        assert "secret-signature" not in response_text
        assert "eyJhbGciOiJIUzI1NiJ9" not in response_text
    
    def test_malformed_token_handling(self):
        """Test that malformed tokens are handled securely."""
        client = TestClient(app)
        
        malformed_tokens = [
            "Bearer <script>alert('xss')</script>",
            "Bearer ' OR 1=1 --",
            "Bearer ../../../etc/passwd",
            "Bearer " + "A" * 10000,  # Very long token
        ]
        
        for token in malformed_tokens:
            response = client.get(
                "/api/v1/jobs/test-job",
                headers={"Authorization": token}
            )
            
            assert response.status_code == 401
            # Ensure no token content appears in response
            response_text = response.text.lower()
            assert "script" not in response_text
            assert "alert" not in response_text
            assert "etc/passwd" not in response_text