| 1 | # SPDX-License-Identifier: Apache-2.0 |
| 2 | # SPDX-FileCopyrightText: Copyright contributors to the vLLM project |
| 3 | # ruff: noqa: E501 |
| 4 | |
| 5 | import json |
| 6 | import random |
| 7 | from unittest.mock import MagicMock |
| 8 | |
| 9 | import pytest |
| 10 | |
| 11 | from tests.tool_parsers.utils import run_tool_extraction_streaming |
| 12 | from vllm.entrypoints.generate.base.protocol import FunctionCall, ToolCall |
| 13 | from vllm.entrypoints.openai.chat_completion.protocol import ( |
| 14 | ChatCompletionRequest, |
| 15 | ChatCompletionToolsParam, |
| 16 | ) |
| 17 | from vllm.tool_parsers import ToolParser, ToolParserManager |
| 18 | from vllm.tool_parsers.minicpm5xml_tool_parser import MiniCPM5XMLToolParser |
| 19 | |
| 20 | |
| 21 | def _tool(name: str, parameters: dict) -> ChatCompletionToolsParam: |
| 22 | return ChatCompletionToolsParam( |
| 23 | type="function", |
| 24 | function={ |
| 25 | "name": name, |
| 26 | "parameters": parameters, |
| 27 | }, |
| 28 | ) |
| 29 | |
| 30 | |
| 31 | def make_tools_weather() -> list[ChatCompletionToolsParam]: |
| 32 | return [ |
| 33 | _tool( |
| 34 | "get_weather", |
| 35 | { |
| 36 | "type": "object", |
| 37 | "properties": { |
| 38 | "city": {"type": "string"}, |
| 39 | "date": {"type": "string"}, |
| 40 | }, |
| 41 | "required": ["city"], |
| 42 | }, |
| 43 | ) |
| 44 | ] |
| 45 | |
| 46 | |
| 47 | def make_tools_sum() -> list[ChatCompletionToolsParam]: |
| 48 | return [ |
| 49 | _tool( |
| 50 | "sum_values", |
| 51 | { |
| 52 | "type": "object", |
| 53 | "properties": { |
| 54 | "nums": {"type": "array"}, |
| 55 | "exact": {"type": "boolean"}, |
| 56 | }, |
| 57 | "required": ["nums"], |
| 58 | }, |
| 59 | ) |
| 60 | ] |
| 61 | |
| 62 | |
| 63 | def make_tools_no_required() -> list[ChatCompletionToolsParam]: |
| 64 | return [ |
| 65 | _tool( |
| 66 | "noop", |
| 67 | { |
| 68 | "type": "object", |
| 69 | "properties": {"note": {"type": "string"}}, |
| 70 | "required": [], |
| 71 | }, |
| 72 | ) |
| 73 | ] |
| 74 | |
| 75 | |
| 76 | def make_request( |
| 77 | tools: list[ChatCompletionToolsParam], |
| 78 | tool_choice: str = "auto", |
| 79 | ) -> ChatCompletionRequest: |
| 80 | return ChatCompletionRequest( |
| 81 | model="test-model", |
| 82 | messages=[{"role": "user", "content": "hi"}], |
| 83 | tools=tools, |
| 84 | tool_choice=tool_choice, |
| 85 | ) |
| 86 | |
| 87 | |
| 88 | def make_tool_call(name: str, arguments: dict) -> ToolCall: |
| 89 | return ToolCall( |
| 90 | type="function", |
| 91 | function=FunctionCall( |
| 92 | name=name, |
| 93 | arguments=json.dumps(arguments, ensure_ascii=False), |
| 94 | ), |
| 95 | ) |
| 96 | |
| 97 | |
| 98 | def assert_tool_calls( |
| 99 | actual: list[ToolCall], |
| 100 | expected: list[ToolCall], |
| 101 | ) -> None: |
| 102 | assert len(actual) == len(expected) |
| 103 | for act, exp in zip(actual, expected): |
| 104 | assert act.type == "function" |
| 105 | assert act.function.name == exp.function.name |
| 106 | assert json.loads(act.function.arguments) == json.loads(exp.function.arguments) |
| 107 | |
| 108 | |
| 109 | @pytest.fixture |
| 110 | def parser() -> ToolParser: |
| 111 | mock_tokenizer = MagicMock() |
| 112 | return MiniCPM5XMLToolParser(mock_tokenizer) |
| 113 | |
| 114 | |
| 115 | def test_registered_in_tool_parser_manager() -> None: |
| 116 | cls = ToolParserManager.get_tool_parser("minicpm5") |
| 117 | assert cls is MiniCPM5XMLToolParser |
| 118 | |
| 119 | |
| 120 | def test_adjust_request_skip_special_tokens(parser: ToolParser) -> None: |
| 121 | request = make_request(make_tools_weather()) |
| 122 | assert request.skip_special_tokens is True |
| 123 | adjusted = parser.adjust_request(request) |
| 124 | assert adjusted.skip_special_tokens is False |
| 125 | |
| 126 | |
| 127 | def test_adjust_request_tool_choice_none(parser: ToolParser) -> None: |
| 128 | request = make_request(make_tools_weather(), tool_choice="none") |
| 129 | adjusted = parser.adjust_request(request) |
| 130 | assert adjusted.skip_special_tokens is True |
| 131 | |
| 132 | |
| 133 | def test_no_tool_call(parser: ToolParser) -> None: |
| 134 | request = make_request(make_tools_weather()) |
| 135 | out = parser.extract_tool_calls("How can I help you?", request) |
| 136 | assert not out.tools_called |
| 137 | assert out.tool_calls == [] |
| 138 | assert out.content == "How can I help you?" |
| 139 | |
| 140 | |
| 141 | def test_single_call_with_surrounding_text(parser: ToolParser) -> None: |
| 142 | request = make_request(make_tools_weather()) |
| 143 | text = ( |
| 144 | "Intro before.\n" |
| 145 | '<function name="get_weather">' |
| 146 | '<param name="city">上海</param>' |
| 147 | '<param name="date">2024-06-27</param>' |
| 148 | "</function>\n" |
| 149 | "Outro after.\n" |
| 150 | ) |
| 151 | out = parser.extract_tool_calls(text, request) |
| 152 | assert out.tools_called |
| 153 | assert_tool_calls( |
| 154 | out.tool_calls, |
| 155 | [ |
| 156 | make_tool_call( |
| 157 | "get_weather", |
| 158 | { |
| 159 | "city": "上海", |
| 160 | "date": "2024-06-27", |
| 161 | }, |
| 162 | ) |
| 163 | ], |
| 164 | ) |
| 165 | assert out.content is None |
| 166 | |
| 167 | |
| 168 | def test_cdata_multiline(parser: ToolParser) -> None: |
| 169 | request = make_request(make_tools_weather()) |
| 170 | text = ( |
| 171 | '<function name="get_weather">' |
| 172 | '<param name="city"><![CDATA[北\n京]]></param>' |
| 173 | '<param name="date">2024-06-27</param>' |
| 174 | "</function>\n" |
| 175 | ) |
| 176 | out = parser.extract_tool_calls(text, request) |
| 177 | assert len(out.tool_calls) == 1 |
| 178 | args = json.loads(out.tool_calls[0].function.arguments) |
| 179 | assert args["city"] == "北\n京" |
| 180 | assert args["date"] == "2024-06-27" |
| 181 | |
| 182 | |
| 183 | def test_tokenizer_space_marker(parser: ToolParser) -> None: |
| 184 | request = make_request(make_tools_weather()) |
| 185 | text = ( |
| 186 | '<function\u0120name="get_weather">' |
| 187 | '<param\u0120name="city">上海</param>' |
| 188 | '<param\u0120name="date">2024-06-27</param>' |
| 189 | "</function>\n" |
| 190 | ) |
| 191 | out = parser.extract_tool_calls(text, request) |
| 192 | assert out.tools_called |
| 193 | assert_tool_calls( |
| 194 | out.tool_calls, |
| 195 | [ |
| 196 | make_tool_call( |
| 197 | "get_weather", |
| 198 | { |
| 199 | "city": "上海", |
| 200 | "date": "2024-06-27", |
| 201 | }, |
| 202 | ) |
| 203 | ], |
| 204 | ) |
| 205 | |
| 206 | |
| 207 | def test_collapsed_function_and_param_tags(parser: ToolParser) -> None: |
| 208 | request = make_request(make_tools_weather()) |
| 209 | text = ( |
| 210 | '<functionname="get_weather">' |
| 211 | '<paramname="city">上海</param>' |
| 212 | '<paramname="date">2024-06-27</param>' |
| 213 | "</function>\n" |
| 214 | ) |
| 215 | out = parser.extract_tool_calls(text, request) |
| 216 | assert out.tools_called |
| 217 | assert_tool_calls( |
| 218 | out.tool_calls, |
| 219 | [ |
| 220 | make_tool_call( |
| 221 | "get_weather", |
| 222 | { |
| 223 | "city": "上海", |
| 224 | "date": "2024-06-27", |
| 225 | }, |
| 226 | ) |
| 227 | ], |
| 228 | ) |
| 229 | |
| 230 | |
| 231 | def test_collapsed_param_tags_with_tokenizer_space_function( |
| 232 | parser: ToolParser, |
| 233 | ) -> None: |
| 234 | request = make_request(make_tools_weather()) |
| 235 | text = ( |
| 236 | '<function\u0120name="get_weather">' |
| 237 | '<paramname="city">上海</param>' |
| 238 | '<paramname="date">2024-06-27</param>' |
| 239 | "</function>\n" |
| 240 | ) |
| 241 | out = parser.extract_tool_calls(text, request) |
| 242 | assert out.tools_called |
| 243 | assert_tool_calls( |
| 244 | out.tool_calls, |
| 245 | [ |
| 246 | make_tool_call( |
| 247 | "get_weather", |
| 248 | { |
| 249 | "city": "上海", |
| 250 | "date": "2024-06-27", |
| 251 | }, |
| 252 | ) |
| 253 | ], |
| 254 | ) |
| 255 | |
| 256 | |
| 257 | def test_extract_tool_calls_streaming_partial_chunks(parser: ToolParser) -> None: |
| 258 | request = make_request(make_tools_weather()) |
| 259 | chunks = [ |
| 260 | '<function name="get_weather">', |
| 261 | '<param name="city">', |
| 262 | '上海</param><param name="date">2024-06-27</param></function>\n', |
| 263 | ] |
| 264 | reconstructor = run_tool_extraction_streaming( |
| 265 | parser, |
| 266 | chunks, |
| 267 | request, |
| 268 | assert_one_tool_per_delta=False, |
| 269 | ) |
| 270 | assert len(reconstructor.tool_calls) == 1 |
| 271 | assert reconstructor.tool_calls[0].function.name == "get_weather" |
| 272 | assert json.loads(reconstructor.tool_calls[0].function.arguments) == { |
| 273 | "city": "上海", |
| 274 | "date": "2024-06-27", |
| 275 | } |
| 276 | |
| 277 | |
| 278 | def test_extract_tool_calls_streaming_tokenizer_space_marker( |
| 279 | parser: ToolParser, |
| 280 | ) -> None: |
| 281 | request = make_request(make_tools_weather()) |
| 282 | chunks = [ |
| 283 | '<function\u0120name="get_weather">', |
| 284 | '<param\u0120name="city">', |
| 285 | '上海</param><param\u0120name="date">2024-06-27</param></function>\n', |
| 286 | ] |
| 287 | reconstructor = run_tool_extraction_streaming( |
| 288 | parser, |
| 289 | chunks, |
| 290 | request, |
| 291 | assert_one_tool_per_delta=False, |
| 292 | ) |
| 293 | assert len(reconstructor.tool_calls) == 1 |
| 294 | assert json.loads(reconstructor.tool_calls[0].function.arguments) == { |
| 295 | "city": "上海", |
| 296 | "date": "2024-06-27", |
| 297 | } |
| 298 | |
| 299 | |
| 300 | def test_extract_tool_calls_streaming_collapsed_tags_weather( |
| 301 | parser: ToolParser, |
| 302 | ) -> None: |
| 303 | request = make_request(make_tools_weather()) |
| 304 | text = ( |
| 305 | '<functionname="get_weather">' |
| 306 | '<paramname="city">上海</param>' |
| 307 | '<paramname="date">2024-06-27</param>' |
| 308 | "</function>\n" |
| 309 | ) |
| 310 | random.seed(2) |
| 311 | reconstructor = run_tool_extraction_streaming( |
| 312 | parser, |
| 313 | _random_chunks(text, 1, 4), |
| 314 | request, |
| 315 | ) |
| 316 | assert len(reconstructor.tool_calls) == 1 |
| 317 | assert reconstructor.tool_calls[0].function.name == "get_weather" |
| 318 | assert json.loads(reconstructor.tool_calls[0].function.arguments) == { |
| 319 | "city": "上海", |
| 320 | "date": "2024-06-27", |
| 321 | } |
| 322 | |
| 323 | |
| 324 | def test_collapsed_tags_current_weather(parser: ToolParser) -> None: |
| 325 | request = make_request( |
| 326 | [ |
| 327 | _tool( |
| 328 | "get_current_weather", |
| 329 | { |
| 330 | "type": "object", |
| 331 | "properties": { |
| 332 | "city": {"type": "string"}, |
| 333 | "state": {"type": "string"}, |
| 334 | "unit": {"type": "string"}, |
| 335 | }, |
| 336 | "required": ["city", "state", "unit"], |
| 337 | }, |
| 338 | ) |
| 339 | ] |
| 340 | ) |
| 341 | text = ( |
| 342 | '<functionname="get_current_weather">' |
| 343 | '<paramname="city">Dallas</param>' |
| 344 | '<paramname="state">TX</param>' |
| 345 | '<paramname="unit">fahrenheit</param>' |
| 346 | "</function>" |
| 347 | ) |
| 348 | out = parser.extract_tool_calls(text, request) |
| 349 | assert out.tools_called |
| 350 | assert json.loads(out.tool_calls[0].function.arguments) == { |
| 351 | "city": "Dallas", |
| 352 | "state": "TX", |
| 353 | "unit": "fahrenheit", |
| 354 | } |
| 355 | |
| 356 | |
| 357 | def test_extract_tool_calls_streaming_incremental_arguments( |
| 358 | parser: ToolParser, |
| 359 | ) -> None: |
| 360 | request = make_request( |
| 361 | [ |
| 362 | _tool( |
| 363 | "get_current_weather", |
| 364 | { |
| 365 | "type": "object", |
| 366 | "properties": { |
| 367 | "city": {"type": "string"}, |
| 368 | "state": {"type": "string"}, |
| 369 | "unit": {"type": "string"}, |
| 370 | }, |
| 371 | "required": ["city", "state", "unit"], |
| 372 | }, |
| 373 | ) |
| 374 | ] |
| 375 | ) |
| 376 | text = ( |
| 377 | '<function name="get_current_weather">' |
| 378 | '<param name="city">Dallas</param>' |
| 379 | '<param name="state">TX</param>' |
| 380 | '<param name="unit">fahrenheit</param>' |
| 381 | "</function>" |
| 382 | ) |
| 383 | prev = "" |
| 384 | arguments = "" |
| 385 | for chunk in [text[:37], text[37:70], text[70:]]: |
| 386 | current = prev + chunk |
| 387 | delta = parser.extract_tool_calls_streaming( |
| 388 | prev, |
| 389 | current, |
| 390 | chunk, |
| 391 | [], |
| 392 | [], |
| 393 | [], |
| 394 | request, |
| 395 | ) |
| 396 | if delta and delta.tool_calls: |
| 397 | arg_delta = delta.tool_calls[0].function.arguments |
| 398 | if arg_delta: |
| 399 | arguments += arg_delta |
| 400 | prev = current |
| 401 | |
| 402 | assert json.loads(arguments) == { |
| 403 | "city": "Dallas", |
| 404 | "state": "TX", |
| 405 | "unit": "fahrenheit", |
| 406 | } |
| 407 | |
| 408 | |
| 409 | def test_unknown_tool_block_preserved(parser: ToolParser) -> None: |
| 410 | request = make_request(make_tools_weather()) |
| 411 | text = '<function name="unknown"><param name="x">1</param></function>\n' |
| 412 | out = parser.extract_tool_calls(text, request) |
| 413 | assert not out.tools_called |
| 414 | assert "unknown" in (out.content or "") |
| 415 | |
| 416 | |
| 417 | def test_non_string_types(parser: ToolParser) -> None: |
| 418 | request = make_request(make_tools_sum()) |
| 419 | text = ( |
| 420 | '<function name="sum_values">' |
| 421 | '<param name="nums">[1, 2, 3]</param>' |
| 422 | '<param name="exact">true</param>' |
| 423 | "</function>\n" |
| 424 | ) |
| 425 | out = parser.extract_tool_calls(text, request) |
| 426 | assert len(out.tool_calls) == 1 |
| 427 | args = json.loads(out.tool_calls[0].function.arguments) |
| 428 | assert args["nums"] == [1, 2, 3] |
| 429 | assert args["exact"] is True |
| 430 | |
| 431 | |
| 432 | def test_multiple_calls_interleaved_text(parser: ToolParser) -> None: |
| 433 | tools = make_tools_weather() + make_tools_sum() |
| 434 | request = make_request(tools) |
| 435 | text = ( |
| 436 | "Head\n" |
| 437 | '<function name="get_weather"><param name="city">北京</param></function>\n' |
| 438 | "TXT\n" |
| 439 | '<function name="sum_values"><param name="nums">[7,8,9]</param>' |
| 440 | '<param name="exact">false</param></function>\n' |
| 441 | "Tail\n" |
| 442 | ) |
| 443 | out = parser.extract_tool_calls(text, request) |
| 444 | assert len(out.tool_calls) == 2 |
| 445 | args0 = json.loads(out.tool_calls[0].function.arguments) |
| 446 | assert args0["city"] == "北京" |
| 447 | args1 = json.loads(out.tool_calls[1].function.arguments) |
| 448 | assert args1["nums"] == [7, 8, 9] |
| 449 | assert args1["exact"] is False |
| 450 | assert out.content is None |
| 451 | |
| 452 | |
| 453 | def test_incomplete_missing_function_end(parser: ToolParser) -> None: |
| 454 | request = make_request(make_tools_weather()) |
| 455 | text = '<function name="get_weather"><param name="city">北京</param>' |
| 456 | out = parser.extract_tool_calls(text, request) |
| 457 | assert not out.tools_called |
| 458 | assert "get_weather" in (out.content or "") |
| 459 | |
| 460 | |
| 461 | def test_param_missing_name_invalid(parser: ToolParser) -> None: |
| 462 | request = make_request(make_tools_weather()) |
| 463 | text = ( |
| 464 | '<function name="get_weather">' |
| 465 | "<param>北京</param>" |
| 466 | '<param name="date">2024-06-27</param>' |
| 467 | "</function>\n" |
| 468 | ) |
| 469 | out = parser.extract_tool_calls(text, request) |
| 470 | assert not out.tools_called |
| 471 | assert "<param>北京</param>" in (out.content or "") |
| 472 | |
| 473 | |
| 474 | def test_duplicate_param_names_invalid(parser: ToolParser) -> None: |
| 475 | request = make_request(make_tools_weather()) |
| 476 | text = ( |
| 477 | '<function name="get_weather">' |
| 478 | '<param name="city">北京</param>' |
| 479 | '<param name="city">上海</param>' |
| 480 | "</function>\n" |
| 481 | ) |
| 482 | out = parser.extract_tool_calls(text, request) |
| 483 | assert not out.tools_called |
| 484 | |
| 485 | |
| 486 | def test_case_sensitive_param_name_invalid(parser: ToolParser) -> None: |
| 487 | request = make_request(make_tools_weather()) |
| 488 | text = '<function name="get_weather"><param name="City">北京</param></function>\n' |
| 489 | out = parser.extract_tool_calls(text, request) |
| 490 | assert not out.tools_called |
| 491 | |
| 492 | |
| 493 | def test_no_required_and_zero_param_valid(parser: ToolParser) -> None: |
| 494 | request = make_request(make_tools_no_required()) |
| 495 | text = '<function name="noop"></function>\n' |
| 496 | out = parser.extract_tool_calls(text, request) |
| 497 | assert len(out.tool_calls) == 1 |
| 498 | args = json.loads(out.tool_calls[0].function.arguments) |
| 499 | assert args == {} |
| 500 | |
| 501 | |
| 502 | def test_thinking_only_sentencepiece_normalized_in_content( |
| 503 | parser: ToolParser, |
| 504 | ) -> None: |
| 505 | request = make_request(make_tools_weather()) |
| 506 | text = "\u010aFirst,\u0120I\u0120need\u0120to\u0120check\u0120the\u0120weather." |
| 507 | out = parser.extract_tool_calls(text, request) |
| 508 | assert not out.tools_called |
| 509 | assert out.content is not None |
| 510 | assert "\u0120" not in out.content |
| 511 | assert "\u010a" not in out.content |
| 512 | assert "First, I need to check the weather." in out.content |
| 513 | |
| 514 | |
| 515 | def test_properties_wrapped_arguments(parser: ToolParser) -> None: |
| 516 | tools = [ |
| 517 | _tool( |
| 518 | "get_customer_by_phone", |
| 519 | { |
| 520 | "type": "object", |
| 521 | "properties": {"phone_number": {"type": "string"}}, |
| 522 | "required": ["phone_number"], |
| 523 | }, |
| 524 | ) |
| 525 | ] |
| 526 | request = make_request(tools) |
| 527 | text = ( |
| 528 | '<function name="get_customer_by_phone">' |
| 529 | "<param name=\"properties\">{'phone_number': '555-123-2002'}</param>" |
| 530 | "</function>" |
| 531 | ) |
| 532 | out = parser.extract_tool_calls(text, request) |
| 533 | assert len(out.tool_calls) == 1 |
| 534 | assert out.tool_calls[0].function.name == "get_customer_by_phone" |
| 535 | args = json.loads(out.tool_calls[0].function.arguments) |
| 536 | assert args == {"phone_number": "555-123-2002"} |
| 537 | |
| 538 | |
| 539 | def test_arguments_wrapped_arguments(parser: ToolParser) -> None: |
| 540 | request = make_request(make_tools_weather()) |
| 541 | text = ( |
| 542 | '<function name="get_weather">' |
| 543 | '<param name="arguments">{"city": "上海", "date": "2024-06-27"}</param>' |
| 544 | "</function>" |
| 545 | ) |
| 546 | out = parser.extract_tool_calls(text, request) |
| 547 | assert len(out.tool_calls) == 1 |
| 548 | args = json.loads(out.tool_calls[0].function.arguments) |
| 549 | assert args == {"city": "上海", "date": "2024-06-27"} |
| 550 | |
| 551 | |
| 552 | def test_wrapped_arguments_still_validate_schema(parser: ToolParser) -> None: |
| 553 | request = make_request(make_tools_weather()) |
| 554 | text = ( |
| 555 | '<function name="get_weather">' |
| 556 | '<param name="properties">{"unknown": "x"}</param>' |
| 557 | "</function>" |
| 558 | ) |
| 559 | out = parser.extract_tool_calls(text, request) |
| 560 | assert not out.tools_called |
| 561 | |
| 562 | |
| 563 | def test_extra_arguments_ignored_when_required_present(parser: ToolParser) -> None: |
| 564 | request = make_request(make_tools_weather()) |
| 565 | text = ( |
| 566 | '<function name="get_weather">' |
| 567 | '<param name="city">上海</param>' |
| 568 | '<param name="unknown">ignored</param>' |
| 569 | "</function>" |
| 570 | ) |
| 571 | out = parser.extract_tool_calls(text, request) |
| 572 | assert len(out.tool_calls) == 1 |
| 573 | args = json.loads(out.tool_calls[0].function.arguments) |
| 574 | assert args == {"city": "上海"} |
| 575 | |
| 576 | |
| 577 | def test_extra_arguments_do_not_satisfy_required(parser: ToolParser) -> None: |
| 578 | request = make_request(make_tools_weather()) |
| 579 | text = ( |
| 580 | '<function name="get_weather"><param name="unknown">ignored</param></function>' |
| 581 | ) |
| 582 | out = parser.extract_tool_calls(text, request) |
| 583 | assert not out.tools_called |
| 584 | |
| 585 | |
| 586 | def test_zero_arg_tool_ignores_extra_arguments(parser: ToolParser) -> None: |
| 587 | request = make_request(make_tools_no_required()) |
| 588 | text = ( |
| 589 | '<function name="noop">' |
| 590 | '<param name="note">ignored</param>' |
| 591 | '<param name="extra">ignored</param>' |
| 592 | "</function>" |
| 593 | ) |
| 594 | out = parser.extract_tool_calls(text, request) |
| 595 | assert len(out.tool_calls) == 1 |
| 596 | args = json.loads(out.tool_calls[0].function.arguments) |
| 597 | assert args == {"note": "ignored"} |
| 598 | |
| 599 | |
| 600 | def test_alias_get_details_by_phone(parser: ToolParser) -> None: |
| 601 | tools = [ |
| 602 | _tool( |
| 603 | "get_customer_by_phone", |
| 604 | { |
| 605 | "type": "object", |
| 606 | "properties": {"phone_number": {"type": "string"}}, |
| 607 | "required": ["phone_number"], |
| 608 | }, |
| 609 | ) |
| 610 | ] |
| 611 | request = make_request(tools) |
| 612 | text = ( |
| 613 | '<function name="get_details_by_phone">' |
| 614 | '<param name="phone_number">555-123-2002</param>' |
| 615 | "</function>" |
| 616 | ) |
| 617 | out = parser.extract_tool_calls(text, request) |
| 618 | assert len(out.tool_calls) == 1 |
| 619 | assert out.tool_calls[0].function.name == "get_customer_by_phone" |
| 620 | args = json.loads(out.tool_calls[0].function.arguments) |
| 621 | assert args == {"phone_number": "555-123-2002"} |
| 622 | |
| 623 | |
| 624 | def test_alias_get_line_details(parser: ToolParser) -> None: |
| 625 | tools = [ |
| 626 | _tool( |
| 627 | "get_details_by_id", |
| 628 | { |
| 629 | "type": "object", |
| 630 | "properties": {"id": {"type": "string"}}, |
| 631 | "required": ["id"], |
| 632 | }, |
| 633 | ) |
| 634 | ] |
| 635 | request = make_request(tools) |
| 636 | text = ( |
| 637 | '<function name="get_line_details">' |
| 638 | '<param name="customer_id">C1001</param>' |
| 639 | '<param name="line_id">L1001</param>' |
| 640 | "</function>" |
| 641 | ) |
| 642 | out = parser.extract_tool_calls(text, request) |
| 643 | assert len(out.tool_calls) == 1 |
| 644 | assert out.tool_calls[0].function.name == "get_details_by_id" |
| 645 | args = json.loads(out.tool_calls[0].function.arguments) |
| 646 | assert args == {"id": "L1001"} |
| 647 | |
| 648 | |
| 649 | def test_alias_enable_roaming(parser: ToolParser) -> None: |
| 650 | tools = [ |
| 651 | _tool( |
| 652 | "toggle_roaming", |
| 653 | { |
| 654 | "type": "object", |
| 655 | "properties": { |
| 656 | "line_id": {"type": "string"}, |
| 657 | "enabled": {"type": "boolean"}, |
| 658 | }, |
| 659 | "required": ["line_id", "enabled"], |
| 660 | }, |
| 661 | ) |
| 662 | ] |
| 663 | request = make_request(tools) |
| 664 | text = ( |
| 665 | '<function name="enable_roaming"><param name="line_id">L1001</param></function>' |
| 666 | ) |
| 667 | out = parser.extract_tool_calls(text, request) |
| 668 | assert len(out.tool_calls) == 1 |
| 669 | assert out.tool_calls[0].function.name == "toggle_roaming" |
| 670 | args = json.loads(out.tool_calls[0].function.arguments) |
| 671 | assert args == {"line_id": "L1001", "enabled": True} |
| 672 | |
| 673 | |
| 674 | def _random_chunks(text: str, min_len: int, max_len: int) -> list[str]: |
| 675 | chunks: list[str] = [] |
| 676 | index = 0 |
| 677 | while index < len(text): |
| 678 | size = random.randint(min_len, max_len) |
| 679 | chunks.append(text[index : index + size]) |
| 680 | index += size |
| 681 | return chunks |
| 682 | |
| 683 | |
| 684 | def test_extract_tool_calls_streaming_single(parser: ToolParser) -> None: |
| 685 | request = make_request(make_tools_weather()) |
| 686 | text = ( |
| 687 | "Intro before.\n" |
| 688 | '<function name="get_weather">' |
| 689 | '<param name="city">上海</param>' |
| 690 | '<param name="date">2024-06-27</param>' |
| 691 | "</function>\n" |
| 692 | "Outro after.\n" |
| 693 | ) |
| 694 | random.seed(0) |
| 695 | reconstructor = run_tool_extraction_streaming( |
| 696 | parser, |
| 697 | _random_chunks(text, 1, 4), |
| 698 | request, |
| 699 | ) |
| 700 | assert "Intro before." in reconstructor.other_content |
| 701 | assert "Outro after." in reconstructor.other_content |
| 702 | assert len(reconstructor.tool_calls) == 1 |
| 703 | assert reconstructor.tool_calls[0].function.name == "get_weather" |
| 704 | assert json.loads(reconstructor.tool_calls[0].function.arguments) == { |
| 705 | "city": "上海", |
| 706 | "date": "2024-06-27", |
| 707 | } |
| 708 | |
| 709 | |
| 710 | def test_extract_tool_calls_streaming_multiple(parser: ToolParser) -> None: |
| 711 | tools = make_tools_weather() + make_tools_sum() |
| 712 | request = make_request(tools) |
| 713 | text = ( |
| 714 | "Head\n" |
| 715 | '<function name="get_weather"><param name="city">北京</param></function>\n' |
| 716 | "TXT\n" |
| 717 | '<function name="sum_values"><param name="nums">[7,8,9]</param>' |
| 718 | '<param name="exact">false</param></function>\n' |
| 719 | "Tail\n" |
| 720 | ) |
| 721 | random.seed(1) |
| 722 | reconstructor = run_tool_extraction_streaming( |
| 723 | parser, |
| 724 | _random_chunks(text, 1, 5), |
| 725 | request, |
| 726 | ) |
| 727 | assert "Head" in reconstructor.other_content |
| 728 | assert "TXT" in reconstructor.other_content |
| 729 | assert "Tail" in reconstructor.other_content |
| 730 | assert len(reconstructor.tool_calls) == 2 |
| 731 | assert json.loads(reconstructor.tool_calls[0].function.arguments)["city"] == "北京" |
| 732 | assert json.loads(reconstructor.tool_calls[1].function.arguments) == { |
| 733 | "nums": [7, 8, 9], |
| 734 | "exact": False, |
| 735 | } |
| 736 | |
| 737 | |
| 738 | def make_tools_write() -> list[ChatCompletionToolsParam]: |
| 739 | return [ |
| 740 | _tool( |
| 741 | "write_file", |
| 742 | { |
| 743 | "type": "object", |
| 744 | "properties": {"content": {"type": "string"}}, |
| 745 | "required": ["content"], |
| 746 | }, |
| 747 | ) |
| 748 | ] |
| 749 | |
| 750 | |
| 751 | class TestParameterWhitespace: |
| 752 | """CDATA is verbatim, so its whitespace must survive.""" |
| 753 | |
| 754 | def test_cdata_whitespace_preserved(self, parser: ToolParser) -> None: |
| 755 | request = make_request(make_tools_write()) |
| 756 | text = ( |
| 757 | '<function name="write_file">' |
| 758 | '<param name="content"><![CDATA[ def foo():\n pass\n]]></param>' |
| 759 | "</function>\n" |
| 760 | ) |
| 761 | |
| 762 | out = parser.extract_tool_calls(text, request) |
| 763 | |
| 764 | assert len(out.tool_calls) == 1 |
| 765 | args = json.loads(out.tool_calls[0].function.arguments) |
| 766 | assert args["content"] == " def foo():\n pass\n" |
| 767 | |
| 768 | def test_cdata_whitespace_preserved_streaming(self, parser: ToolParser) -> None: |
| 769 | """Value split across chunks, i.e. the partial path.""" |
| 770 | request = make_request(make_tools_write()) |
| 771 | chunks = [ |
| 772 | '<function name="write_file">', |
| 773 | '<param name="content"><![CDATA[ def foo():\n', |
| 774 | " pass\n]]></param></function>\n", |
| 775 | ] |
| 776 | |
| 777 | reconstructor = run_tool_extraction_streaming( |
| 778 | parser, |
| 779 | chunks, |
| 780 | request, |
| 781 | assert_one_tool_per_delta=False, |
| 782 | ) |
| 783 | |
| 784 | assert len(reconstructor.tool_calls) == 1 |
| 785 | assert json.loads(reconstructor.tool_calls[0].function.arguments) == { |
| 786 | "content": " def foo():\n pass\n" |
| 787 | } |