xparse-client 0.2.20__py3-none-any.whl → 0.3.0b1__py3-none-any.whl

tests/unit/connectors/test_smb.py

@@ -0,0 +1,442 @@
+"""SmbSource 测试
+
+测试 SMB/CIFS 数据源连接器。使用 mock 模式测试，无需真实 SMB 服务器。
+
+运行方式：
+    pytest tests/unit/connectors/test_smb.py -v
+"""
+
+from datetime import datetime
+from unittest.mock import MagicMock, Mock, patch
+
+import pytest
+
+from xparse_client.connectors import SmbSource
+from xparse_client.exceptions import SourceError
+
+# ============================================================================
+# Mock SMB 测试
+# ============================================================================
+
+def _setup_smb_mock(mock_get_smb):
+    """设置 SMB mock 的辅助函数
+
+    返回:
+        (mock_smb_instance, mock_smb_class)
+    """
+    mock_smb_class = MagicMock()
+    mock_smb_instance = MagicMock()
+    mock_smb_class.return_value = mock_smb_instance
+    mock_get_smb.return_value = mock_smb_class
+    return mock_smb_instance, mock_smb_class
+
+
+@patch('xparse_client.connectors.sources.smb._get_smb_connection')
+def test_smb_source_initialization(mock_get_smb):
+    """测试 SmbSource 初始化"""
+    mock_smb, _ = _setup_smb_mock(mock_get_smb)
+
+    source = SmbSource(
+        host="192.168.1.100",
+        share_name="documents",
+        username="testuser",
+        password="testpass",
+        domain="WORKGROUP",
+        port=445,
+        path="/reports"
+    )
+
+    assert source.host == "192.168.1.100"
+    assert source.share_name == "documents"
+    assert source.username == "testuser"
+    assert source.domain == "WORKGROUP"
+    assert source.port == 445
+    assert source.path == "reports"
+    assert source.pattern is None
+    assert source.recursive is False
+
+    # 验证连接
+    mock_smb.connect.assert_called_once_with("192.168.1.100", 445)
+
+
+@patch('xparse_client.connectors.sources.smb._get_smb_connection')
+def test_smb_source_list_files_basic(mock_get_smb):
+    """测试列出文件（非递归）"""
+    mock_smb, _ = _setup_smb_mock(mock_get_smb)
+
+    # Mock 文件列表
+    mock_file1 = Mock()
+    mock_file1.filename = 'file1.pdf'
+    mock_file1.isDirectory = False
+
+    mock_file2 = Mock()
+    mock_file2.filename = 'file2.docx'
+    mock_file2.isDirectory = False
+
+    mock_dir = Mock()
+    mock_dir.filename = 'subdir'
+    mock_dir.isDirectory = True
+
+    mock_dot = Mock()
+    mock_dot.filename = '.'
+    mock_dot.isDirectory = True
+
+    mock_dotdot = Mock()
+    mock_dotdot.filename = '..'
+    mock_dotdot.isDirectory = True
+
+    mock_smb.listPath.return_value = [
+        mock_file1, mock_file2, mock_dir, mock_dot, mock_dotdot
+    ]
+
+    source = SmbSource(
+        host="192.168.1.100",
+        share_name="documents",
+        username="test",
+        password="test",
+        recursive=False
+    )
+
+    files = source.list_files()
+
+    # 非递归，只列出文件，不包含目录和 . ..
+    assert len(files) == 2
+    assert 'file1.pdf' in files
+    assert 'file2.docx' in files
+    assert 'subdir' not in files
+
+
+@patch('xparse_client.connectors.sources.smb._get_smb_connection')
+def test_smb_source_list_files_with_pattern(mock_get_smb):
+    """测试带 pattern 的文件列表"""
+    mock_smb, _ = _setup_smb_mock(mock_get_smb)
+
+    mock_file1 = Mock()
+    mock_file1.filename = 'file1.pdf'
+    mock_file1.isDirectory = False
+
+    mock_file2 = Mock()
+    mock_file2.filename = 'file2.docx'
+    mock_file2.isDirectory = False
+
+    mock_smb.listPath.return_value = [mock_file1, mock_file2]
+
+    source = SmbSource(
+        host="192.168.1.100",
+        share_name="documents",
+        username="test",
+        password="test",
+        pattern=["*.pdf"]
+    )
+
+    files = source.list_files()
+
+    assert len(files) == 1
+    assert 'file1.pdf' in files
+    assert 'file2.docx' not in files
+
+
+@patch('xparse_client.connectors.sources.smb._get_smb_connection')
+def test_smb_source_list_files_recursive(mock_get_smb):
+    """测试递归列出文件"""
+    mock_smb, _ = _setup_smb_mock(mock_get_smb)
+
+    # Mock 顶层文件
+    mock_file1 = Mock()
+    mock_file1.filename = 'file1.pdf'
+    mock_file1.isDirectory = False
+
+    mock_dir = Mock()
+    mock_dir.filename = 'subdir'
+    mock_dir.isDirectory = True
+
+    # Mock 子目录文件
+    mock_file2 = Mock()
+    mock_file2.filename = 'file2.pdf'
+    mock_file2.isDirectory = False
+
+    # 根据路径返回不同的文件列表
+    def mock_list_path(share, path):
+        if path == '/':
+            return [mock_file1, mock_dir]
+        elif 'subdir' in path:
+            return [mock_file2]
+        return []
+
+    mock_smb.listPath.side_effect = mock_list_path
+
+    source = SmbSource(
+        host="192.168.1.100",
+        share_name="documents",
+        username="test",
+        password="test",
+        recursive=True
+    )
+
+    files = source.list_files()
+
+    # 应该包含两级目录的文件
+    assert len(files) == 2
+    assert 'file1.pdf' in files
+    assert 'subdir/file2.pdf' in files
+
+
+@patch('xparse_client.connectors.sources.smb._get_smb_connection')
+def test_smb_source_list_files_ignore_hidden(mock_get_smb):
+    """测试忽略隐藏文件"""
+    mock_smb, _ = _setup_smb_mock(mock_get_smb)
+
+    mock_visible = Mock()
+    mock_visible.filename = 'visible.pdf'
+    mock_visible.isDirectory = False
+
+    mock_hidden = Mock()
+    mock_hidden.filename = '.hidden'
+    mock_hidden.isDirectory = False
+
+    mock_smb.listPath.return_value = [mock_visible, mock_hidden]
+
+    source = SmbSource(
+        host="192.168.1.100",
+        share_name="documents",
+        username="test",
+        password="test"
+    )
+
+    files = source.list_files()
+
+    # 应该过滤掉隐藏文件
+    assert len(files) == 1
+    assert 'visible.pdf' in files
+    assert '.hidden' not in files
+
+
+@patch('xparse_client.connectors.sources.smb._get_smb_connection')
+def test_smb_source_read_file(mock_get_smb):
+    """测试读取文件"""
+    mock_smb, _ = _setup_smb_mock(mock_get_smb)
+
+    # Mock retrieveFile
+    test_content = b'SMB file content'
+
+    def mock_retrieve_file(share, path, file_obj):
+        file_obj.write(test_content)
+
+    mock_smb.retrieveFile.side_effect = mock_retrieve_file
+
+    # Mock getAttributes
+    mock_attrs = Mock()
+    mock_attrs.create_time = datetime(2024, 1, 28, 12, 0, 0)
+    mock_attrs.last_write_time = datetime(2024, 1, 28, 14, 0, 0)
+    mock_smb.getAttributes.return_value = mock_attrs
+
+    source = SmbSource(
+        host="192.168.1.100",
+        share_name="documents",
+        username="test",
+        password="test"
+    )
+
+    file_bytes, data_source = source.read_file('test.pdf')
+
+    assert file_bytes == test_content
+    assert 'smb://192.168.1.100/documents/' in data_source['url']
+    assert data_source['record_locator']['protocol'] == 'smb'
+    assert data_source['record_locator']['server'] == '192.168.1.100'
+    assert data_source['record_locator']['share'] == 'documents'
+    assert 'version' in data_source
+    assert 'date_created' in data_source
+    assert 'date_modified' in data_source
+
+
+@patch('xparse_client.connectors.sources.smb._get_smb_connection')
+def test_smb_source_read_file_with_path(mock_get_smb):
+    """测试从指定路径读取文件"""
+    mock_smb, _ = _setup_smb_mock(mock_get_smb)
+
+    test_content = b'Content'
+
+    def mock_retrieve_file(share, path, file_obj):
+        # 验证路径包含 base path
+        assert '/reports/' in path or path.startswith('/reports/')
+        file_obj.write(test_content)
+
+    mock_smb.retrieveFile.side_effect = mock_retrieve_file
+
+    source = SmbSource(
+        host="192.168.1.100",
+        share_name="documents",
+        username="test",
+        password="test",
+        path="/reports"
+    )
+
+    file_bytes, data_source = source.read_file('test.pdf')
+
+    assert file_bytes == test_content
+
+
+@patch('xparse_client.connectors.sources.smb._get_smb_connection')
+def test_smb_source_read_file_not_found(mock_get_smb):
+    """测试读取不存在的文件"""
+    mock_smb, _ = _setup_smb_mock(mock_get_smb)
+
+    # Mock retrieveFile 抛出异常
+    mock_smb.retrieveFile.side_effect = Exception("File not found")
+
+    source = SmbSource(
+        host="192.168.1.100",
+        share_name="documents",
+        username="test",
+        password="test"
+    )
+
+    with pytest.raises(SourceError) as exc_info:
+        source.read_file('nonexistent.pdf')
+
+    assert "读取 SMB 文件失败" in str(exc_info.value)
+
+
+@patch('xparse_client.connectors.sources.smb._get_smb_connection')
+def test_smb_source_connection_error(mock_get_smb):
+    """测试连接错误"""
+    mock_smb, _ = _setup_smb_mock(mock_get_smb)
+
+    # Mock connect 抛出异常
+    mock_smb.connect.side_effect = Exception("Connection refused")
+
+    with pytest.raises(SourceError) as exc_info:
+        SmbSource(
+            host="invalid-host",
+            share_name="documents",
+            username="test",
+            password="test"
+        )
+
+    assert "SMB 连接失败" in str(exc_info.value)
+
+
+@patch('xparse_client.connectors.sources.smb._get_smb_connection')
+def test_smb_source_authentication_error(mock_get_smb):
+    """测试认证错误"""
+    mock_smb, _ = _setup_smb_mock(mock_get_smb)
+
+    # Mock connect 抛出认证异常
+    mock_smb.connect.side_effect = Exception("Authentication failed")
+
+    with pytest.raises(SourceError) as exc_info:
+        SmbSource(
+            host="192.168.1.100",
+            share_name="documents",
+            username="wrong",
+            password="wrong"
+        )
+
+    assert "SMB 连接失败" in str(exc_info.value)
+
+
+@patch('xparse_client.connectors.sources.smb._get_smb_connection')
+def test_smb_source_context_manager(mock_get_smb):
+    """测试上下文管理器"""
+    mock_smb, _ = _setup_smb_mock(mock_get_smb)
+
+    with SmbSource(
+        host="192.168.1.100",
+        share_name="documents",
+        username="test",
+        password="test"
+    ) as source:
+        assert source is not None
+
+    # 验证连接被关闭
+    mock_smb.close.assert_called_once()
+
+
+@patch('xparse_client.connectors.sources.smb._get_smb_connection')
+def test_smb_source_close(mock_get_smb):
+    """测试手动关闭连接"""
+    mock_smb, _ = _setup_smb_mock(mock_get_smb)
+
+    source = SmbSource(
+        host="192.168.1.100",
+        share_name="documents",
+        username="test",
+        password="test"
+    )
+
+    source.close()
+
+    # 验证 close 被调用
+    mock_smb.close.assert_called_once()
+
+
+@patch('xparse_client.connectors.sources.smb._get_smb_connection')
+def test_smb_source_custom_parameters(mock_get_smb):
+    """测试自定义参数"""
+    mock_smb, mock_smb_class = _setup_smb_mock(mock_get_smb)
+
+    source = SmbSource(
+        host="192.168.1.100",
+        share_name="documents",
+        username="test",
+        password="test",
+        domain="COMPANY",
+        port=1445,
+        path="/files",
+        pattern=["*.pdf", "*.docx"],
+        recursive=True
+    )
+
+    assert source.host == "192.168.1.100"
+    assert source.share_name == "documents"
+    assert source.domain == "COMPANY"
+    assert source.port == 1445
+    assert source.path == "files"
+    assert source.pattern == ["*.pdf", "*.docx"]
+    assert source.recursive is True
+
+    # 验证 SMBConnection 使用了正确的参数
+    mock_smb_class.assert_called_once()
+    call_args = mock_smb_class.call_args
+    assert call_args[1]['domain'] == "COMPANY"
+    assert call_args[1]['use_ntlm_v2'] is True
+
+
+@patch('xparse_client.connectors.sources.smb._get_smb_connection')
+def test_smb_source_repr(mock_get_smb):
+    """测试 __repr__"""
+    mock_smb, _ = _setup_smb_mock(mock_get_smb)
+
+    source = SmbSource(
+        host="192.168.1.100",
+        share_name="documents",
+        username="test",
+        password="test"
+    )
+
+    assert repr(source) == "<SmbSource host=192.168.1.100 share=documents>"
+
+
+@patch('xparse_client.connectors.sources.smb._get_smb_connection')
+def test_smb_source_to_timestamp(mock_get_smb):
+    """测试 _to_timestamp 方法"""
+    mock_smb, _ = _setup_smb_mock(mock_get_smb)
+
+    source = SmbSource(
+        host="192.168.1.100",
+        share_name="documents",
+        username="test",
+        password="test"
+    )
+
+    # None 值
+    assert source._to_timestamp(None) is None
+
+    # datetime 值
+    dt = datetime(2024, 1, 28, 12, 0, 0)
+    timestamp = source._to_timestamp(dt)
+    assert isinstance(timestamp, float)
+
+    # 数值
+    assert source._to_timestamp(1234567890) == 1234567890.0
+    assert source._to_timestamp(1234567890.123) == 1234567890.123