|
| 1 | +"""Tests for extraction_tool.py metadata and functionality.""" |
| 2 | + |
| 3 | +from unittest.mock import AsyncMock, MagicMock, patch |
| 4 | +from uuid import UUID |
| 5 | + |
| 6 | +import pytest |
| 7 | +from uipath.agent.models.agent import ( |
| 8 | + AgentIxpExtractionResourceConfig, |
| 9 | + AgentIxpExtractionToolProperties, |
| 10 | +) |
| 11 | +from uipath.platform.attachments import Attachment |
| 12 | +from uipath.platform.documents import ExtractionResponseIXP |
| 13 | + |
| 14 | +from uipath_langchain.agent.tools.extraction_tool import create_ixp_extraction_tool |
| 15 | + |
| 16 | + |
| 17 | +class TestExtractionToolMetadata: |
| 18 | + """Test that extraction tool has correct metadata for observability.""" |
| 19 | + |
| 20 | + @pytest.fixture |
| 21 | + def extraction_resource(self): |
| 22 | + """Create a minimal extraction tool resource config.""" |
| 23 | + return AgentIxpExtractionResourceConfig( |
| 24 | + name="test_extraction", |
| 25 | + description="Extract data from files", |
| 26 | + input_schema={ |
| 27 | + "type": "object", |
| 28 | + "properties": { |
| 29 | + "attachment": { |
| 30 | + "description": "the file uploaded as attachement", |
| 31 | + "$ref": "#/definitions/job-attachment", |
| 32 | + } |
| 33 | + }, |
| 34 | + "required": ["attachment"], |
| 35 | + "definitions": { |
| 36 | + "job-attachment": { |
| 37 | + "type": "object", |
| 38 | + "required": ["ID"], |
| 39 | + "x-uipath-resource-kind": "JobAttachment", |
| 40 | + "properties": { |
| 41 | + "ID": { |
| 42 | + "type": "string", |
| 43 | + "description": "Orchestrator attachment key", |
| 44 | + }, |
| 45 | + "FullName": {"type": "string", "description": "File name"}, |
| 46 | + "MimeType": { |
| 47 | + "type": "string", |
| 48 | + "description": 'The MIME type of the content, such as "application/json" or "image/png"', |
| 49 | + }, |
| 50 | + "Metadata": { |
| 51 | + "type": "object", |
| 52 | + "description": "Dictionary<string, string> of metadata", |
| 53 | + "additionalProperties": {"type": "string"}, |
| 54 | + }, |
| 55 | + }, |
| 56 | + } |
| 57 | + }, |
| 58 | + }, |
| 59 | + output_schema={"type": "object", "properties": {}}, |
| 60 | + properties=AgentIxpExtractionToolProperties( |
| 61 | + project_name="TestProject", |
| 62 | + version_tag="v1.0", |
| 63 | + ), |
| 64 | + ) |
| 65 | + |
| 66 | + def test_extraction_tool_has_correct_name(self, extraction_resource): |
| 67 | + """Test that extraction tool has sanitized name.""" |
| 68 | + tool = create_ixp_extraction_tool(extraction_resource) |
| 69 | + |
| 70 | + assert tool.name == "test_extraction" |
| 71 | + |
| 72 | + def test_extraction_tool_has_correct_description(self, extraction_resource): |
| 73 | + """Test that extraction tool has correct description.""" |
| 74 | + tool = create_ixp_extraction_tool(extraction_resource) |
| 75 | + |
| 76 | + assert tool.description == "Extract data from files" |
| 77 | + |
| 78 | + def test_extraction_tool_has_attachment_input_schema(self, extraction_resource): |
| 79 | + """Test that extraction tool uses Attachment as input schema.""" |
| 80 | + tool = create_ixp_extraction_tool(extraction_resource) |
| 81 | + |
| 82 | + assert tool.args_schema == Attachment |
| 83 | + |
| 84 | + def test_extraction_tool_has_extraction_response_output_type( |
| 85 | + self, extraction_resource |
| 86 | + ): |
| 87 | + """Test that extraction tool has ExtractionResponseIXP as output type.""" |
| 88 | + tool = create_ixp_extraction_tool(extraction_resource) |
| 89 | + |
| 90 | + assert hasattr(tool, "output_type") |
| 91 | + assert tool.output_type == ExtractionResponseIXP |
| 92 | + |
| 93 | + |
| 94 | +class TestExtractionToolFunctionality: |
| 95 | + """Test the extraction tool function behavior.""" |
| 96 | + |
| 97 | + @pytest.fixture |
| 98 | + def extraction_resource(self): |
| 99 | + """Create a minimal extraction tool resource config.""" |
| 100 | + return AgentIxpExtractionResourceConfig( |
| 101 | + name="test_extraction", |
| 102 | + description="Extract data from files", |
| 103 | + input_schema={ |
| 104 | + "type": "object", |
| 105 | + "properties": { |
| 106 | + "attachment": { |
| 107 | + "description": "the file uploaded as attachment", |
| 108 | + "$ref": "#/definitions/job-attachment", |
| 109 | + } |
| 110 | + }, |
| 111 | + "required": ["attachment"], |
| 112 | + "definitions": { |
| 113 | + "job-attachment": { |
| 114 | + "type": "object", |
| 115 | + "required": ["ID"], |
| 116 | + "x-uipath-resource-kind": "JobAttachment", |
| 117 | + "properties": { |
| 118 | + "ID": { |
| 119 | + "type": "string", |
| 120 | + "description": "Orchestrator attachment key", |
| 121 | + }, |
| 122 | + "FullName": {"type": "string", "description": "File name"}, |
| 123 | + "MimeType": { |
| 124 | + "type": "string", |
| 125 | + "description": "The MIME type of the content", |
| 126 | + }, |
| 127 | + }, |
| 128 | + } |
| 129 | + }, |
| 130 | + }, |
| 131 | + output_schema={"type": "object", "properties": {}}, |
| 132 | + properties=AgentIxpExtractionToolProperties( |
| 133 | + project_name="TestProject", |
| 134 | + version_tag="v1.0", |
| 135 | + ), |
| 136 | + ) |
| 137 | + |
| 138 | + @pytest.mark.asyncio |
| 139 | + @patch("uipath.platform.UiPath") |
| 140 | + @patch("uipath_langchain.agent.tools.extraction_tool.interrupt") |
| 141 | + async def test_extraction_tool_downloads_attachment_and_calls_interrupt( |
| 142 | + self, mock_interrupt, mock_uipath_class, extraction_resource |
| 143 | + ): |
| 144 | + """Test that extraction tool downloads attachment and calls interrupt with correct params.""" |
| 145 | + mock_client = MagicMock() |
| 146 | + mock_uipath_class.return_value = mock_client |
| 147 | + mock_client.attachments.download_async = AsyncMock( |
| 148 | + return_value="/path/to/document.pdf" |
| 149 | + ) |
| 150 | + mock_interrupt.return_value = {"extracted_data": {"field1": "value1"}} |
| 151 | + |
| 152 | + tool = create_ixp_extraction_tool(extraction_resource) |
| 153 | + |
| 154 | + result = await tool.ainvoke( |
| 155 | + { |
| 156 | + "id": "fa93f4ca-bd3f-473a-93e5-e6e5b5a8f27f", |
| 157 | + "full_name": "document.pdf", |
| 158 | + "mime_type": "application/pdf", |
| 159 | + } |
| 160 | + ) |
| 161 | + |
| 162 | + mock_client.attachments.download_async.assert_called_once_with( |
| 163 | + key=UUID("fa93f4ca-bd3f-473a-93e5-e6e5b5a8f27f"), |
| 164 | + destination_path="document.pdf", |
| 165 | + ) |
| 166 | + |
| 167 | + assert mock_interrupt.called |
| 168 | + interrupt_arg = mock_interrupt.call_args[0][0] |
| 169 | + assert interrupt_arg.project_name == "TestProject" |
| 170 | + assert interrupt_arg.tag == "v1.0" |
| 171 | + assert interrupt_arg.file_path == "/path/to/document.pdf" |
| 172 | + |
| 173 | + assert result == {"extracted_data": {"field1": "value1"}} |
| 174 | + |
| 175 | + @pytest.mark.asyncio |
| 176 | + @patch("uipath.platform.UiPath") |
| 177 | + @patch("uipath_langchain.agent.tools.extraction_tool.interrupt") |
| 178 | + async def test_extraction_tool_handles_missing_attachment_id( |
| 179 | + self, mock_interrupt, mock_uipath_class, extraction_resource |
| 180 | + ): |
| 181 | + """Test that extraction tool handles None attachment_id.""" |
| 182 | + mock_client = MagicMock() |
| 183 | + mock_uipath_class.return_value = mock_client |
| 184 | + mock_client.attachments.download_async = AsyncMock( |
| 185 | + return_value="/path/to/file.pdf" |
| 186 | + ) |
| 187 | + mock_interrupt.return_value = {"extracted_data": {}} |
| 188 | + |
| 189 | + tool = create_ixp_extraction_tool(extraction_resource) |
| 190 | + |
| 191 | + await tool.ainvoke({"full_name": "file.pdf", "mime_type": "application/pdf"}) |
| 192 | + |
| 193 | + mock_client.attachments.download_async.assert_called_once_with( |
| 194 | + key=None, destination_path="file.pdf" |
| 195 | + ) |
| 196 | + |
| 197 | + @pytest.mark.asyncio |
| 198 | + @patch("uipath.platform.UiPath") |
| 199 | + @patch("uipath_langchain.agent.tools.extraction_tool.interrupt") |
| 200 | + async def test_extraction_tool_with_different_version_tag( |
| 201 | + self, mock_interrupt, mock_uipath_class |
| 202 | + ): |
| 203 | + """Test extraction tool with different version tag.""" |
| 204 | + extraction_resource = AgentIxpExtractionResourceConfig( |
| 205 | + name="test_extraction_v2", |
| 206 | + description="Extract data from files v2", |
| 207 | + input_schema={"type": "object", "properties": {}}, |
| 208 | + output_schema={"type": "object", "properties": {}}, |
| 209 | + properties=AgentIxpExtractionToolProperties( |
| 210 | + project_name="TestProjectV2", |
| 211 | + version_tag="staging", |
| 212 | + ), |
| 213 | + ) |
| 214 | + |
| 215 | + mock_client = MagicMock() |
| 216 | + mock_uipath_class.return_value = mock_client |
| 217 | + mock_client.attachments.download_async = AsyncMock( |
| 218 | + return_value="/path/to/document.pdf" |
| 219 | + ) |
| 220 | + mock_interrupt.return_value = {"extracted_data": {}} |
| 221 | + |
| 222 | + tool = create_ixp_extraction_tool(extraction_resource) |
| 223 | + |
| 224 | + await tool.ainvoke( |
| 225 | + { |
| 226 | + "id": "fa93f4ca-bd3f-473a-93e5-e6e5b5a8f27f", |
| 227 | + "full_name": "document.pdf", |
| 228 | + "mime_type": "application/pdf", |
| 229 | + } |
| 230 | + ) |
| 231 | + |
| 232 | + interrupt_arg = mock_interrupt.call_args[0][0] |
| 233 | + assert interrupt_arg.tag == "staging" |
| 234 | + |
| 235 | + @pytest.mark.asyncio |
| 236 | + @patch("uipath.platform.UiPath") |
| 237 | + async def test_extraction_tool_propagates_download_exception( |
| 238 | + self, mock_uipath_class, extraction_resource |
| 239 | + ): |
| 240 | + """Test that exceptions from attachment download are propagated.""" |
| 241 | + mock_client = MagicMock() |
| 242 | + mock_uipath_class.return_value = mock_client |
| 243 | + mock_client.attachments.download_async = AsyncMock( |
| 244 | + side_effect=Exception("Download failed") |
| 245 | + ) |
| 246 | + |
| 247 | + tool = create_ixp_extraction_tool(extraction_resource) |
| 248 | + |
| 249 | + with pytest.raises(Exception) as exc_info: |
| 250 | + await tool.ainvoke( |
| 251 | + { |
| 252 | + "id": "fa93f4ca-bd3f-473a-93e5-e6e5b5a8f27f", |
| 253 | + "full_name": "file.pdf", |
| 254 | + "mime_type": "application/pdf", |
| 255 | + } |
| 256 | + ) |
| 257 | + |
| 258 | + assert "Download failed" in str(exc_info.value) |
| 259 | + |
| 260 | + |
| 261 | +class TestExtractionToolNameSanitization: |
| 262 | + """Test that extraction tool names are properly sanitized.""" |
| 263 | + |
| 264 | + @pytest.mark.asyncio |
| 265 | + async def test_extraction_tool_name_with_spaces(self): |
| 266 | + """Test that tool names with spaces are sanitized.""" |
| 267 | + resource = AgentIxpExtractionResourceConfig( |
| 268 | + name="Invoice Extraction Tool", |
| 269 | + description="Extract invoices", |
| 270 | + input_schema={"type": "object", "properties": {}}, |
| 271 | + output_schema={"type": "object", "properties": {}}, |
| 272 | + properties=AgentIxpExtractionToolProperties( |
| 273 | + project_name="InvoiceExtraction", |
| 274 | + version_tag="v1.0", |
| 275 | + ), |
| 276 | + ) |
| 277 | + |
| 278 | + tool = create_ixp_extraction_tool(resource) |
| 279 | + |
| 280 | + assert " " not in tool.name |
| 281 | + |
| 282 | + @pytest.mark.asyncio |
| 283 | + async def test_extraction_tool_name_with_special_chars(self): |
| 284 | + """Test that tool names with special characters are sanitized.""" |
| 285 | + resource = AgentIxpExtractionResourceConfig( |
| 286 | + name="invoice-extraction@v1", |
| 287 | + description="Extract invoices", |
| 288 | + input_schema={"type": "object", "properties": {}}, |
| 289 | + output_schema={"type": "object", "properties": {}}, |
| 290 | + properties=AgentIxpExtractionToolProperties( |
| 291 | + project_name="InvoiceExtraction", |
| 292 | + version_tag="v1.0", |
| 293 | + ), |
| 294 | + ) |
| 295 | + |
| 296 | + tool = create_ixp_extraction_tool(resource) |
| 297 | + |
| 298 | + # Tool name should be sanitized |
| 299 | + assert tool.name is not None |
| 300 | + assert len(tool.name) > 0 |
0 commit comments