pdf-accessibility/tests/test_checker.py

"""
Unit tests for enterprise_pdf_checker.py
"""

import pytest
from pathlib import Path
from unittest.mock import Mock, patch, MagicMock


class TestEnterprisePDFChecker:
    """Test suite for EnterprisePDFChecker class"""

    def test_checker_initialization_valid_pdf(self, sample_good_pdf):
        """Test that checker initializes with valid PDF"""
        from enterprise_pdf_checker import EnterprisePDFChecker

        config = {'anthropic_api_key': 'test-key', 'google_api_key': None}
        checker = EnterprisePDFChecker(str(sample_good_pdf), config)
        assert checker.pdf_path.exists()
        assert checker.pdf_path.suffix == '.pdf'

    def test_checker_initialization_missing_file(self):
        """Test that checker initializes but path does not exist"""
        from enterprise_pdf_checker import EnterprisePDFChecker

        checker = EnterprisePDFChecker("nonexistent.pdf")
        assert not checker.pdf_path.exists()

    def test_severity_levels(self):
        """Test that Severity enum has required levels"""
        from enterprise_pdf_checker import Severity

        assert hasattr(Severity, 'CRITICAL')
        assert hasattr(Severity, 'ERROR')
        assert hasattr(Severity, 'WARNING')
        assert hasattr(Severity, 'INFO')
        assert hasattr(Severity, 'SUCCESS')

    @patch('enterprise_pdf_checker.anthropic')
    def test_quick_check_without_api(self, mock_anthropic, sample_good_pdf):
        """Test quick check runs without actual API calls"""
        # Mock Anthropic to avoid real API calls
        mock_anthropic.Anthropic.return_value = MagicMock()

        from enterprise_pdf_checker import EnterprisePDFChecker

        config = {'anthropic_api_key': 'test-key', 'google_api_key': None}
        checker = EnterprisePDFChecker(str(sample_good_pdf), config)

        # Quick check should skip expensive API calls
        # Note: This will still try to analyze the PDF structure
        # but won't make external API calls if properly configured
        try:
            # Test that the method exists and is callable
            assert hasattr(checker, 'run_full_check')
            assert callable(checker.run_full_check)
        except Exception as e:
            pytest.skip(f"Skipping due to: {e}")

    def test_accessibility_issue_creation(self):
        """Test AccessibilityIssue dataclass"""
        from enterprise_pdf_checker import AccessibilityIssue, Severity

        issue = AccessibilityIssue(
            severity=Severity.ERROR,
            category="Test Category",
            description="Test description",
            wcag_criterion="1.1.1",
            recommendation="Test recommendation"
        )

        assert issue.severity == Severity.ERROR
        assert issue.category == "Test Category"
        assert issue.wcag_criterion == "1.1.1"

    def test_check_result_structure(self):
        """Test CheckResult dataclass"""
        from enterprise_pdf_checker import CheckResult

        result = CheckResult(
            check_name="Test Check",
            passed=True,
            issues=[],
            metadata={'test': 'data'}
        )

        assert result.check_name == "Test Check"
        assert result.passed is True
        assert isinstance(result.issues, list)
        assert isinstance(result.metadata, dict)


class TestCacheManager:
    """Test suite for CacheManager class"""

    def test_cache_key_generation(self):
        """Test that cache keys are generated correctly"""
        from enterprise_pdf_checker import CacheManager

        cache_manager = CacheManager()

        # Test with same content
        key1 = cache_manager.get_cache_key(b"test content")
        key2 = cache_manager.get_cache_key(b"test content")

        assert key1 == key2
        assert isinstance(key1, str)
        assert len(key1) > 0

    def test_cache_key_different_content(self):
        """Test that different content produces different keys"""
        from enterprise_pdf_checker import CacheManager

        cache_manager = CacheManager()

        key1 = cache_manager.get_cache_key(b"content 1")
        key2 = cache_manager.get_cache_key(b"content 2")

        assert key1 != key2


class TestRetryLogic:
    """Test retry logic integration"""

    def test_retry_decorator_exists(self):
        """Test that retry decorators are applied"""
        from enterprise_pdf_checker import EnterprisePDFChecker
        import inspect

        # Check that methods exist
        assert hasattr(EnterprisePDFChecker, '_analyze_image_with_claude')
        assert hasattr(EnterprisePDFChecker, '_analyze_image_with_google')

    def test_logger_initialized(self):
        """Test that logger is properly initialized"""
        import enterprise_pdf_checker

        assert hasattr(enterprise_pdf_checker, 'logger')
        assert enterprise_pdf_checker.logger is not None


# Integration test (requires actual PDF processing)
@pytest.mark.integration
class TestFullCheck:
    """Integration tests for full PDF checking"""

    def test_full_workflow_exists(self, sample_good_pdf):
        """Test that full workflow methods exist"""
        from enterprise_pdf_checker import EnterprisePDFChecker

        checker = EnterprisePDFChecker(str(sample_good_pdf))

        # Check that main methods exist
        assert hasattr(checker, 'run_full_check')
        assert hasattr(checker, 'to_dict')
        assert callable(checker.run_full_check)
        assert callable(checker.to_dict)


if __name__ == "__main__":
    pytest.main([__file__, "-v"])