Refactor devide logic by files and more tests

tests/test_cli.py

@@ -0,0 +1,107 @@
+import os
+import pytest
+import tempfile
+import shutil
+from typing import Generator
+from unittest.mock import patch, MagicMock
+from repo_to_text.cli.cli import (
+    create_default_settings_file,
+    parse_args,
+    main
+)
+
+@pytest.fixture
+def temp_dir() -> Generator[str, None, None]:
+    """Create a temporary directory for testing."""
+    temp_path = tempfile.mkdtemp()
+    yield temp_path
+    shutil.rmtree(temp_path)
+
+def test_parse_args_defaults() -> None:
+    """Test parsing command line arguments with default values."""
+    with patch('sys.argv', ['repo-to-text']):
+        args = parse_args()
+        assert args.input_dir == '.'
+        assert not args.debug
+        assert args.output_dir is None
+        assert not args.create_settings
+        assert not args.stdout
+        assert args.ignore_patterns is None
+
+def test_parse_args_with_values() -> None:
+    """Test parsing command line arguments with provided values."""
+    test_args = [
+        'repo-to-text',
+        'input/path',
+        '--debug',
+        '--output-dir', 'output/path',
+        '--ignore-patterns', '*.log', 'temp/'
+    ]
+    with patch('sys.argv', test_args):
+        args = parse_args()
+        assert args.input_dir == 'input/path'
+        assert args.debug
+        assert args.output_dir == 'output/path'
+        assert args.ignore_patterns == ['*.log', 'temp/']
+
+def test_create_default_settings_file(temp_dir: str) -> None:
+    """Test creation of default settings file."""
+    os.chdir(temp_dir)
+    create_default_settings_file()
+    
+    settings_file = '.repo-to-text-settings.yaml'
+    assert os.path.exists(settings_file)
+    
+    with open(settings_file, 'r') as f:
+        content = f.read()
+        assert 'gitignore-import-and-ignore: True' in content
+        assert 'ignore-tree-and-content:' in content
+        assert 'ignore-content:' in content
+
+def test_create_default_settings_file_already_exists(temp_dir: str) -> None:
+    """Test handling of existing settings file."""
+    os.chdir(temp_dir)
+    # Create the file first
+    create_default_settings_file()
+    
+    # Try to create it again
+    with pytest.raises(FileExistsError) as exc_info:
+        create_default_settings_file()
+    assert "already exists" in str(exc_info.value)
+
+@patch('repo_to_text.cli.cli.save_repo_to_text')
+def test_main_normal_execution(mock_save_repo: MagicMock) -> None:
+    """Test main function with normal execution."""
+    with patch('sys.argv', ['repo-to-text', '--stdout']):
+        with pytest.raises(SystemExit) as exc_info:
+            main()
+        assert exc_info.value.code == 0
+        mock_save_repo.assert_called_once_with(
+            path='.',
+            output_dir=None,
+            to_stdout=True,
+            cli_ignore_patterns=None
+        )
+
+@patch('repo_to_text.cli.cli.create_default_settings_file')
+def test_main_create_settings(mock_create_settings: MagicMock) -> None:
+    """Test main function with create settings option."""
+    with patch('sys.argv', ['repo-to-text', '--create-settings']):
+        with pytest.raises(SystemExit) as exc_info:
+            main()
+        assert exc_info.value.code == 0
+        mock_create_settings.assert_called_once()
+
+@patch('repo_to_text.cli.cli.setup_logging')
+@patch('repo_to_text.cli.cli.create_default_settings_file')
+def test_main_with_debug_logging(mock_create_settings: MagicMock, mock_setup_logging: MagicMock) -> None:
+    """Test main function with debug logging enabled."""
+    with patch('sys.argv', ['repo-to-text', '--debug', '--create-settings']):
+        with pytest.raises(SystemExit) as exc_info:
+            main()
+        assert exc_info.value.code == 0
+        mock_setup_logging.assert_called_once_with(debug=True)
+        mock_create_settings.assert_called_once()
+
+if __name__ == "__main__":
+    pytest.main([__file__]) 

tests/test_core.py

@@ -0,0 +1,285 @@
+import os
+import tempfile
+import shutil
+import pytest
+from typing import Generator
+from repo_to_text.core.core import (
+    get_tree_structure,
+    load_ignore_specs,
+    should_ignore_file,
+    is_ignored_path,
+    remove_empty_dirs,
+    save_repo_to_text
+)
+
+@pytest.fixture
+def temp_dir() -> Generator[str, None, None]:
+    """Create a temporary directory for testing."""
+    temp_path = tempfile.mkdtemp()
+    yield temp_path
+    shutil.rmtree(temp_path)
+
+@pytest.fixture
+def sample_repo(temp_dir: str) -> str:
+    """Create a sample repository structure for testing."""
+    # Create directories
+    os.makedirs(os.path.join(temp_dir, "src"))
+    os.makedirs(os.path.join(temp_dir, "tests"))
+    
+    # Create sample files
+    files = {
+        "README.md": "# Test Project",
+        ".gitignore": """
+*.pyc
+__pycache__/
+.git/
+""",
+        "src/main.py": "print('Hello World')",
+        "tests/test_main.py": "def test_sample(): pass",
+        ".repo-to-text-settings.yaml": """
+gitignore-import-and-ignore: True
+ignore-tree-and-content:
+  - ".git/"
+  - ".repo-to-text-settings.yaml"
+ignore-content:
+  - "README.md"
+"""
+    }
+    
+    for file_path, content in files.items():
+        full_path = os.path.join(temp_dir, file_path)
+        os.makedirs(os.path.dirname(full_path), exist_ok=True)
+        with open(full_path, "w") as f:
+            f.write(content)
+    
+    return temp_dir
+
+def test_is_ignored_path() -> None:
+    """Test the is_ignored_path function."""
+    assert is_ignored_path(".git/config") is True
+    assert is_ignored_path("repo-to-text_output.txt") is True
+    assert is_ignored_path("src/main.py") is False
+    assert is_ignored_path("normal_file.txt") is False
+
+def test_load_ignore_specs(sample_repo: str) -> None:
+    """Test loading ignore specifications from files."""
+    gitignore_spec, content_ignore_spec, tree_and_content_ignore_spec = load_ignore_specs(sample_repo)
+    
+    assert gitignore_spec is not None
+    assert content_ignore_spec is not None
+    assert tree_and_content_ignore_spec is not None
+    
+    # Test gitignore patterns
+    assert gitignore_spec.match_file("test.pyc") is True
+    assert gitignore_spec.match_file("__pycache__/cache.py") is True
+    assert gitignore_spec.match_file(".git/config") is True
+    
+    # Test content ignore patterns
+    assert content_ignore_spec.match_file("README.md") is True
+    
+    # Test tree and content ignore patterns
+    assert tree_and_content_ignore_spec.match_file(".git/config") is True
+
+def test_should_ignore_file(sample_repo: str) -> None:
+    """Test file ignoring logic."""
+    gitignore_spec, content_ignore_spec, tree_and_content_ignore_spec = load_ignore_specs(sample_repo)
+    
+    # Test various file paths
+    assert should_ignore_file(
+        ".git/config",
+        ".git/config",
+        gitignore_spec,
+        content_ignore_spec,
+        tree_and_content_ignore_spec
+    ) is True
+    
+    assert should_ignore_file(
+        "src/main.py",
+        "src/main.py",
+        gitignore_spec,
+        content_ignore_spec,
+        tree_and_content_ignore_spec
+    ) is False
+
+def test_get_tree_structure(sample_repo: str) -> None:
+    """Test tree structure generation."""
+    gitignore_spec, _, tree_and_content_ignore_spec = load_ignore_specs(sample_repo)
+    tree_output = get_tree_structure(sample_repo, gitignore_spec, tree_and_content_ignore_spec)
+    
+    # Basic structure checks
+    assert "src" in tree_output
+    assert "tests" in tree_output
+    assert "main.py" in tree_output
+    assert "test_main.py" in tree_output
+    assert ".git" not in tree_output
+
+def test_remove_empty_dirs(temp_dir: str) -> None:
+    """Test removal of empty directories from tree output."""
+    # Create test directory structure
+    os.makedirs(os.path.join(temp_dir, "src"))
+    os.makedirs(os.path.join(temp_dir, "empty_dir"))
+    os.makedirs(os.path.join(temp_dir, "tests"))
+    
+    # Create some files
+    with open(os.path.join(temp_dir, "src/main.py"), "w") as f:
+        f.write("print('test')")
+    with open(os.path.join(temp_dir, "tests/test_main.py"), "w") as f:
+        f.write("def test(): pass")
+    
+    # Create a mock tree output that matches the actual tree command format
+    tree_output = (
+        f"{temp_dir}\n"
+        f"├── {os.path.join(temp_dir, 'src')}\n"
+        f"│   └── {os.path.join(temp_dir, 'src/main.py')}\n"
+        f"├── {os.path.join(temp_dir, 'empty_dir')}\n"
+        f"└── {os.path.join(temp_dir, 'tests')}\n"
+        f"    └── {os.path.join(temp_dir, 'tests/test_main.py')}\n"
+    )
+    
+    filtered_output = remove_empty_dirs(tree_output, temp_dir)
+    
+    # Check that empty_dir is removed but other directories remain
+    assert "empty_dir" not in filtered_output
+    assert os.path.join(temp_dir, "src") in filtered_output
+    assert os.path.join(temp_dir, "tests") in filtered_output
+    assert os.path.join(temp_dir, "src/main.py") in filtered_output
+    assert os.path.join(temp_dir, "tests/test_main.py") in filtered_output
+
+def test_save_repo_to_text(sample_repo: str) -> None:
+    """Test the main save_repo_to_text function."""
+    # Create output directory
+    output_dir = os.path.join(sample_repo, "output")
+    os.makedirs(output_dir, exist_ok=True)
+    
+    # Create .git directory to ensure it's properly ignored
+    os.makedirs(os.path.join(sample_repo, ".git"))
+    with open(os.path.join(sample_repo, ".git/config"), "w") as f:
+        f.write("[core]\n\trepositoryformatversion = 0\n")
+    
+    # Test file output
+    output_file = save_repo_to_text(sample_repo, output_dir=output_dir)
+    assert os.path.exists(output_file)
+    assert os.path.dirname(output_file) == output_dir
+    
+    # Check file contents
+    with open(output_file, 'r') as f:
+        content = f.read()
+        
+        # Basic content checks
+        assert "Directory Structure:" in content
+        
+        # Check for expected files
+        assert "src/main.py" in content
+        assert "tests/test_main.py" in content
+        
+        # Check for file contents
+        assert "print('Hello World')" in content
+        assert "def test_sample(): pass" in content
+        
+        # Ensure ignored patterns are not in output
+        assert ".git/config" not in content  # Check specific file
+        assert "repo-to-text_" not in content
+        assert ".repo-to-text-settings.yaml" not in content
+        
+        # Check that .gitignore content is not included
+        assert "*.pyc" not in content
+        assert "__pycache__" not in content
+
+def test_save_repo_to_text_stdout(sample_repo: str) -> None:
+    """Test save_repo_to_text with stdout output."""
+    output = save_repo_to_text(sample_repo, to_stdout=True)
+    assert isinstance(output, str)
+    assert "Directory Structure:" in output
+    assert "src/main.py" in output
+    assert "tests/test_main.py" in output
+
+def test_load_ignore_specs_with_cli_patterns(sample_repo: str) -> None:
+    """Test loading ignore specs with CLI patterns."""
+    cli_patterns = ["*.log", "temp/"]
+    gitignore_spec, content_ignore_spec, tree_and_content_ignore_spec = load_ignore_specs(sample_repo, cli_patterns)
+    
+    assert tree_and_content_ignore_spec.match_file("test.log") is True
+    assert tree_and_content_ignore_spec.match_file("temp/file.txt") is True
+    assert tree_and_content_ignore_spec.match_file("normal.txt") is False
+
+def test_load_ignore_specs_without_gitignore(temp_dir: str) -> None:
+    """Test loading ignore specs when .gitignore is missing."""
+    gitignore_spec, content_ignore_spec, tree_and_content_ignore_spec = load_ignore_specs(temp_dir)
+    assert gitignore_spec is None
+    assert content_ignore_spec is None
+    assert tree_and_content_ignore_spec is not None
+
+def test_get_tree_structure_with_special_chars(temp_dir: str) -> None:
+    """Test tree structure generation with special characters in paths."""
+    # Create files with special characters
+    special_dir = os.path.join(temp_dir, "special chars")
+    os.makedirs(special_dir)
+    with open(os.path.join(special_dir, "file with spaces.txt"), "w") as f:
+        f.write("test")
+    
+    tree_output = get_tree_structure(temp_dir)
+    assert "special chars" in tree_output
+    assert "file with spaces.txt" in tree_output
+
+def test_should_ignore_file_edge_cases(sample_repo: str) -> None:
+    """Test edge cases for should_ignore_file function."""
+    gitignore_spec, content_ignore_spec, tree_and_content_ignore_spec = load_ignore_specs(sample_repo)
+    
+    # Test with dot-prefixed paths
+    assert should_ignore_file(
+        "./src/main.py",
+        "./src/main.py",
+        gitignore_spec,
+        content_ignore_spec,
+        tree_and_content_ignore_spec
+    ) is False
+    
+    # Test with absolute paths
+    abs_path = os.path.join(sample_repo, "src/main.py")
+    rel_path = "src/main.py"
+    assert should_ignore_file(
+        abs_path,
+        rel_path,
+        gitignore_spec,
+        content_ignore_spec,
+        tree_and_content_ignore_spec
+    ) is False
+
+def test_save_repo_to_text_with_binary_files(temp_dir: str) -> None:
+    """Test handling of binary files in save_repo_to_text."""
+    # Create a binary file
+    binary_path = os.path.join(temp_dir, "binary.bin")
+    binary_content = b'\x00\x01\x02\x03'
+    with open(binary_path, "wb") as f:
+        f.write(binary_content)
+    
+    output = save_repo_to_text(temp_dir, to_stdout=True)
+    
+    # Check that the binary file is listed in the structure
+    assert "binary.bin" in output
+    # Check that the file content section exists with raw binary content
+    expected_content = f"Contents of binary.bin:\n```\n{binary_content.decode('latin1')}\n```"
+    assert expected_content in output
+
+def test_save_repo_to_text_custom_output_dir(temp_dir: str) -> None:
+    """Test save_repo_to_text with custom output directory."""
+    # Create a simple file structure
+    with open(os.path.join(temp_dir, "test.txt"), "w") as f:
+        f.write("test content")
+    
+    # Create custom output directory
+    output_dir = os.path.join(temp_dir, "custom_output")
+    output_file = save_repo_to_text(temp_dir, output_dir=output_dir)
+    
+    assert os.path.exists(output_file)
+    assert os.path.dirname(output_file) == output_dir
+    assert output_file.startswith(output_dir)
+
+def test_get_tree_structure_empty_directory(temp_dir: str) -> None:
+    """Test tree structure generation for empty directory."""
+    tree_output = get_tree_structure(temp_dir)
+    # Should only contain the directory itself
+    assert tree_output.strip() == "" or tree_output.strip() == temp_dir
+
+if __name__ == "__main__":
+    pytest.main([__file__])

tests/test_utils.py

@@ -0,0 +1,142 @@
+import logging
+import pytest
+from typing import Generator
+from repo_to_text.utils.utils import setup_logging
+
+@pytest.fixture(autouse=True)
+def reset_logger() -> Generator[None, None, None]:
+    """Reset root logger before each test."""
+    root_logger = logging.getLogger()
+    for handler in root_logger.handlers[:]:
+        root_logger.removeHandler(handler)
+    root_logger.setLevel(logging.WARNING)  # Default level
+    yield
+    for handler in root_logger.handlers[:]:
+        root_logger.removeHandler(handler)
+    root_logger.setLevel(logging.WARNING)  # Reset after test
+
+def test_setup_logging_debug() -> None:
+    """Test setup_logging with debug mode."""
+    root_logger = logging.getLogger()
+    root_logger.handlers.clear()  # Clear existing handlers
+    root_logger.setLevel(logging.WARNING)  # Reset to default
+    
+    setup_logging(debug=True)
+    assert len(root_logger.handlers) > 0
+    assert root_logger.level == logging.DEBUG
+
+def test_setup_logging_info() -> None:
+    """Test setup_logging with info mode."""
+    root_logger = logging.getLogger()
+    root_logger.handlers.clear()  # Clear existing handlers
+    root_logger.setLevel(logging.WARNING)  # Reset to default
+    
+    setup_logging(debug=False)
+    assert len(root_logger.handlers) > 0
+    assert root_logger.level == logging.INFO
+
+def test_setup_logging_formatter() -> None:
+    """Test logging formatter setup."""
+    setup_logging(debug=True)
+    logger = logging.getLogger()
+    handlers = logger.handlers
+    
+    # Check if there's at least one handler
+    assert len(handlers) > 0
+    
+    # Check formatter
+    formatter = handlers[0].formatter
+    assert formatter is not None
+    
+    # Test format string
+    test_record = logging.LogRecord(
+        name='test',
+        level=logging.DEBUG,
+        pathname='test.py',
+        lineno=1,
+        msg='Test message',
+        args=(),
+        exc_info=None
+    )
+    formatted = formatter.format(test_record)
+    assert 'Test message' in formatted
+    assert test_record.levelname in formatted
+
+def test_setup_logging_multiple_calls() -> None:
+    """Test that multiple calls to setup_logging don't create duplicate handlers."""
+    root_logger = logging.getLogger()
+    root_logger.handlers.clear()
+    
+    setup_logging(debug=True)
+    initial_handler_count = len(root_logger.handlers)
+    
+    # Call setup_logging again
+    setup_logging(debug=True)
+    assert len(root_logger.handlers) == initial_handler_count, "Should not create duplicate handlers"
+
+def test_setup_logging_level_change() -> None:
+    """Test changing log levels between setup_logging calls."""
+    root_logger = logging.getLogger()
+    root_logger.handlers.clear()
+    
+    # Start with debug
+    setup_logging(debug=True)
+    assert root_logger.level == logging.DEBUG
+    
+    # Clear handlers before next setup
+    root_logger.handlers.clear()
+    
+    # Switch to info
+    setup_logging(debug=False)
+    assert root_logger.level == logging.INFO
+
+def test_setup_logging_message_format() -> None:
+    """Test the actual format of logged messages."""
+    setup_logging(debug=True)
+    logger = logging.getLogger()
+    
+    # Create a temporary handler to capture output
+    import io
+    log_capture = io.StringIO()
+    handler = logging.StreamHandler(log_capture)
+    # Use formatter that includes pathname
+    handler.setFormatter(logging.Formatter('%(levelname)s %(name)s:%(pathname)s:%(lineno)d %(message)s'))
+    logger.addHandler(handler)
+    
+    # Ensure debug level is set
+    logger.setLevel(logging.DEBUG)
+    handler.setLevel(logging.DEBUG)
+    
+    # Log a test message
+    test_message = "Test log message"
+    logger.debug(test_message)
+    log_output = log_capture.getvalue()
+    
+    # Verify format components
+    assert test_message in log_output
+    assert "DEBUG" in log_output
+    assert "test_utils.py" in log_output
+
+def test_setup_logging_error_messages() -> None:
+    """Test logging of error messages."""
+    setup_logging(debug=False)
+    logger = logging.getLogger()
+    
+    # Create a temporary handler to capture output
+    import io
+    log_capture = io.StringIO()
+    handler = logging.StreamHandler(log_capture)
+    handler.setFormatter(logger.handlers[0].formatter)
+    logger.addHandler(handler)
+    
+    # Log an error message
+    error_message = "Test error message"
+    logger.error(error_message)
+    log_output = log_capture.getvalue()
+    
+    # Error messages should always be logged regardless of debug setting
+    assert error_message in log_output
+    assert "ERROR" in log_output
+
+if __name__ == "__main__":
+    pytest.main([__file__])