Spaces:
Paused
Paused
File size: 25,520 Bytes
4b03eed | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 | # -*- coding: utf-8 -*-
"""Unit tests for AgenticMemoryMiddleware with real Agent execution."""
import os
import shutil
import tempfile
from typing import Any, Type
from unittest.async_case import IsolatedAsyncioTestCase
from pydantic import BaseModel
from utils import AnyString, AnyValue, MockModel
from agentscope.agent import Agent
from agentscope.message import (
HintBlock,
Msg,
TextBlock,
ToolCallBlock,
UserMsg,
)
from agentscope.middleware import AgenticMemoryMiddleware
from agentscope.model import ChatResponse, StructuredResponse
from agentscope.permission import (
PermissionBehavior,
PermissionContext,
PermissionDecision,
)
from agentscope.tool import ToolBase, ToolChunk, Toolkit
class _RecordingMockModel(MockModel):
"""A ``MockModel`` that records chat and structured-output calls."""
def __init__(self, **kwargs: Any) -> None:
"""Initialize the recording mock model.
Args:
**kwargs (`Any`):
Keyword arguments forwarded to :class:`MockModel`.
"""
kwargs.setdefault("context_size", 100_000)
super().__init__(**kwargs)
self.chat_messages: list[list[Msg]] = []
self.structured_messages: list[list[Msg]] = []
async def _call_api(
self,
*args: Any,
**kwargs: Any,
) -> ChatResponse:
"""Record the chat messages and delegate to ``MockModel``.
Args:
*args (`Any`):
Positional arguments forwarded to ``MockModel``.
**kwargs (`Any`):
Keyword arguments forwarded to ``MockModel``.
Returns:
`ChatResponse`:
The configured mock chat response.
"""
self.chat_messages.append(kwargs["messages"])
return await super()._call_api(*args, **kwargs)
async def _call_api_with_structured_output(
self,
model_name: str,
messages: list[Msg],
structured_model: Type[BaseModel] | dict,
**kwargs: Any,
) -> StructuredResponse:
"""Record structured-output messages and delegate to ``MockModel``.
Args:
model_name (`str`):
The model name.
messages (`list[Msg]`):
The structured-output prompt messages.
structured_model (`Type[BaseModel] | dict`):
The expected structured-output schema.
**kwargs (`Any`):
Extra keyword arguments forwarded to ``MockModel``.
Returns:
`StructuredResponse`:
The configured mock structured response.
"""
self.structured_messages.append(messages)
return await super()._call_api_with_structured_output(
model_name,
messages,
structured_model,
**kwargs,
)
class _DummyTool(ToolBase):
"""A minimal tool that forces a second Agent reasoning iteration."""
name: str = "dummy"
description: str = "A dummy tool for middleware tests."
input_schema: dict[str, Any] = {"type": "object", "properties": {}}
is_concurrency_safe: bool = True
is_read_only: bool = True
is_external_tool: bool = False
is_mcp: bool = False
async def check_permissions(
self,
tool_input: dict[str, Any],
context: PermissionContext,
) -> PermissionDecision:
"""Allow every dummy tool call.
Args:
tool_input (`dict[str, Any]`):
The tool input.
context (`PermissionContext`):
The permission context.
Returns:
`PermissionDecision`:
The allow decision.
"""
return PermissionDecision(
behavior=PermissionBehavior.ALLOW,
decision_reason="Dummy tool always allows.",
message="Dummy tool always allows.",
)
async def __call__(self, **kwargs: Any) -> ToolChunk:
"""Return a fixed tool result.
Args:
**kwargs (`Any`):
Ignored tool arguments.
Returns:
`ToolChunk`:
The fixed tool output.
"""
return ToolChunk(content=[TextBlock(text="tool result")])
def _text_response(text: str) -> ChatResponse:
"""Build a text-only chat response.
Args:
text (`str`):
The response text.
Returns:
`ChatResponse`:
A complete chat response with one text block.
"""
return ChatResponse(content=[TextBlock(text=text)], is_last=True)
def _tool_response() -> ChatResponse:
"""Build a chat response that calls the dummy tool.
Returns:
`ChatResponse`:
A complete chat response with one tool call block.
"""
return ChatResponse(
content=[
ToolCallBlock(
id="call_dummy",
name="dummy",
input="{}",
),
],
is_last=True,
)
def _structured_response(selected_files: list[str]) -> StructuredResponse:
"""Build a structured memory-selection response.
Args:
selected_files (`list[str]`):
The selected memory filenames.
Returns:
`StructuredResponse`:
The structured response consumed by the middleware.
"""
return StructuredResponse(content={"selected_files": selected_files})
def _block_to_dict(block: Any) -> dict:
"""Convert a message block into a stable assertion dictionary.
Args:
block (`Any`):
The message block to convert.
Returns:
`dict`:
The stable block representation.
"""
if isinstance(block, TextBlock):
return {
"type": "text",
"text": block.text,
"id": AnyString(),
}
if isinstance(block, HintBlock):
return {
"type": "hint",
"hint": block.hint,
"id": AnyString(),
"source": block.source,
}
if isinstance(block, ToolCallBlock):
return {
"type": "tool_call",
"id": AnyString(),
"name": block.name,
"input": block.input,
"state": block.state,
"suggested_rules": block.suggested_rules,
}
return block.model_dump()
def _message_to_dict(msg: Msg) -> dict:
"""Convert a message into a stable assertion dictionary.
Args:
msg (`Msg`):
The message to convert.
Returns:
`dict`:
The stable message representation.
"""
return {
"id": AnyString(),
"name": msg.name,
"role": msg.role,
"content": [_block_to_dict(block) for block in msg.content],
"metadata": msg.metadata,
}
def _hint_texts(agent: Agent) -> list[str]:
"""Collect hint texts from an agent context.
Args:
agent (`Agent`):
The agent whose context is inspected.
Returns:
`list[str]`:
The hint texts in context order.
"""
return [
block.hint
for msg in agent.state.context
for block in msg.content
if isinstance(block, HintBlock)
]
def _write_memory_file(
memory_dir: str,
filename: str,
description: str,
memory_type: str,
body: str,
) -> None:
"""Write one Markdown memory file with frontmatter.
Args:
memory_dir (`str`):
The memory directory.
filename (`str`):
The memory filename relative to ``memory_dir``.
description (`str`):
The frontmatter description.
memory_type (`str`):
The frontmatter memory type.
body (`str`):
The Markdown body.
"""
path = os.path.join(memory_dir, filename)
os.makedirs(os.path.dirname(path), exist_ok=True)
with open(path, "w", encoding="utf-8") as f:
f.write(
"---\n"
f"name: {filename}\n"
f"description: {description}\n"
f"type: {memory_type}\n"
"---\n\n"
f"{body}\n",
)
class AgenticMemoryMiddlewareTest(IsolatedAsyncioTestCase):
"""Agent-level tests for :class:`AgenticMemoryMiddleware`."""
async def asyncSetUp(self) -> None:
"""Create a temporary workspace for each test."""
self.temp_dir = tempfile.mkdtemp()
async def asyncTearDown(self) -> None:
"""Remove the temporary workspace after each test."""
shutil.rmtree(self.temp_dir, ignore_errors=True)
def _make_agent(
self,
model: _RecordingMockModel,
middleware: AgenticMemoryMiddleware,
toolkit: Toolkit | None = None,
) -> Agent:
"""Build an Agent with the filesystem memory middleware attached.
Args:
model (`_RecordingMockModel`):
The mock model used by the agent.
middleware (`AgenticMemoryMiddleware`):
The middleware under test.
toolkit (`Toolkit | None`, optional):
The toolkit for the agent. Defaults to an empty toolkit.
Returns:
`Agent`:
The configured agent.
"""
return Agent(
name="assistant",
system_prompt="You are helpful.",
model=model,
toolkit=toolkit or Toolkit(),
middlewares=[middleware],
)
async def test_agent_reply_creates_layout_and_injects_memory_prompt(
self,
) -> None:
"""Agent reply should create layout and inject memory instructions."""
model = _RecordingMockModel()
model.set_responses([_text_response("done")])
middleware = AgenticMemoryMiddleware(workdir=self.temp_dir)
agent = self._make_agent(model, middleware)
reply = await agent.reply(UserMsg("user", "hello"))
memory_dir = os.path.join(self.temp_dir, "Memory")
system_prompt = model.chat_messages[0][0].get_text_content()
self.assertDictEqual(
{
"reply": _message_to_dict(reply),
"memory_dir_exists": os.path.isdir(memory_dir),
"memory_md_exists": os.path.isfile(
os.path.join(memory_dir, "MEMORY.md"),
),
"system_prompt": {
"has_memory_dir": memory_dir in system_prompt,
"has_placeholder": "{memory_dir}" in system_prompt,
"has_memory_header": "## MEMORY.md" in system_prompt,
"has_empty_memory_text": (
"Your MEMORY.md is currently empty" in system_prompt
),
},
},
{
"reply": {
"id": AnyString(),
"name": "assistant",
"role": "assistant",
"content": [
{
"type": "text",
"text": "done",
"id": AnyString(),
},
],
"metadata": {},
},
"memory_dir_exists": True,
"memory_md_exists": True,
"system_prompt": {
"has_memory_dir": True,
"has_placeholder": False,
"has_memory_header": True,
"has_empty_memory_text": True,
},
},
)
async def test_agent_reasoning_injects_selected_memory_hint(
self,
) -> None:
"""Agent reasoning should inject content selected by structured
output."""
memory_dir = os.path.join(self.temp_dir, "Memory")
os.makedirs(memory_dir)
with open(
os.path.join(memory_dir, "MEMORY.md"),
"w",
encoding="utf-8",
) as f:
f.write("- [User profile](user_profile.md) — User profile.\n")
_write_memory_file(
memory_dir,
"user_profile.md",
"User profile details",
"user",
"The user prefers concise Chinese answers.",
)
model = _RecordingMockModel()
model.set_structured_response(
_structured_response(["user_profile.md"]),
)
model.set_responses([_tool_response(), _text_response("final answer")])
middleware = AgenticMemoryMiddleware(workdir=self.temp_dir)
agent = self._make_agent(
model,
middleware,
toolkit=Toolkit(tools=[_DummyTool()]),
)
reply = await agent.reply(UserMsg("user", "what do you remember?"))
hint_texts = _hint_texts(agent)
self.assertDictEqual(
{
"reply": _message_to_dict(reply),
"hints": [
{
"has_selected_content": (
"The user prefers concise Chinese answers." in hint
),
"has_selected_path": "user_profile.md" in hint,
}
for hint in hint_texts
],
"context": [
_message_to_dict(msg) for msg in agent.state.context
],
"structured_call_count": len(model.structured_messages),
},
{
"reply": {
"id": AnyString(),
"name": "assistant",
"role": "assistant",
"content": [
{
"type": "text",
"text": "final answer",
"id": AnyString(),
},
],
"metadata": {},
},
"hints": [
{
"has_selected_content": True,
"has_selected_path": True,
},
],
"context": [
{
"id": AnyString(),
"name": "user",
"role": "user",
"content": [
{
"type": "text",
"text": "what do you remember?",
"id": AnyString(),
},
],
"metadata": {},
},
{
"id": AnyString(),
"name": "assistant",
"role": "assistant",
"content": [
{
"type": "tool_call",
"id": AnyString(),
"name": "dummy",
"input": "{}",
"state": "finished",
"suggested_rules": [],
},
AnyValue(),
{
"type": "hint",
"hint": AnyString(),
"id": AnyString(),
"source": None,
},
{
"type": "text",
"text": "final answer",
"id": AnyString(),
},
],
"metadata": {},
},
],
"structured_call_count": 1,
},
)
async def test_agent_filters_hallucinated_memory_filenames(self) -> None:
"""Agent retrieval should ignore filenames not present in memory."""
memory_dir = os.path.join(self.temp_dir, "Memory")
os.makedirs(memory_dir)
with open(
os.path.join(memory_dir, "MEMORY.md"),
"w",
encoding="utf-8",
) as f:
f.write("- [User profile](user_profile.md) — User profile.\n")
_write_memory_file(
memory_dir,
"user_profile.md",
"User profile details",
"user",
"Only this real memory should be injected.",
)
model = _RecordingMockModel()
model.set_structured_response(
_structured_response(["user_profile.md", "missing.md"]),
)
model.set_responses([_tool_response(), _text_response("filtered")])
middleware = AgenticMemoryMiddleware(workdir=self.temp_dir)
agent = self._make_agent(
model,
middleware,
toolkit=Toolkit(tools=[_DummyTool()]),
)
await agent.reply(UserMsg("user", "recall memory"))
hint_texts = _hint_texts(agent)
self.assertListEqual(
[
{
"has_real_memory": (
"Only this real memory should be injected." in hint
),
"has_missing_memory": "missing.md" in hint,
}
for hint in hint_texts
],
[
{
"has_real_memory": True,
"has_missing_memory": False,
},
],
)
async def test_agent_does_not_inject_hint_when_no_file_selected(
self,
) -> None:
"""Agent retrieval should inject no hint when selection is empty."""
memory_dir = os.path.join(self.temp_dir, "Memory")
os.makedirs(memory_dir)
with open(
os.path.join(memory_dir, "MEMORY.md"),
"w",
encoding="utf-8",
) as f:
f.write("- [User profile](user_profile.md) — User profile.\n")
_write_memory_file(
memory_dir,
"user_profile.md",
"User profile details",
"user",
"This memory is available but not selected.",
)
model = _RecordingMockModel()
model.set_structured_response(_structured_response([]))
model.set_responses([_tool_response(), _text_response("no hint")])
middleware = AgenticMemoryMiddleware(workdir=self.temp_dir)
agent = self._make_agent(
model,
middleware,
toolkit=Toolkit(tools=[_DummyTool()]),
)
reply = await agent.reply(UserMsg("user", "ignore memories"))
self.assertDictEqual(
{
"reply": _message_to_dict(reply),
"hints": _hint_texts(agent),
"structured_call_count": len(model.structured_messages),
},
{
"reply": {
"id": AnyString(),
"name": "assistant",
"role": "assistant",
"content": [
{
"type": "text",
"text": "no hint",
"id": AnyString(),
},
],
"metadata": {},
},
"hints": [],
"structured_call_count": 1,
},
)
async def test_agent_does_not_retrieve_when_only_memory_index_exists(
self,
) -> None:
"""Agent retrieval should skip structured output without topic
files."""
model = _RecordingMockModel()
model.set_structured_response(_structured_response(["missing.md"]))
model.set_responses([_tool_response(), _text_response("index only")])
middleware = AgenticMemoryMiddleware(workdir=self.temp_dir)
agent = self._make_agent(
model,
middleware,
toolkit=Toolkit(tools=[_DummyTool()]),
)
reply = await agent.reply(UserMsg("user", "hello"))
self.assertDictEqual(
{
"reply": _message_to_dict(reply),
"hints": _hint_texts(agent),
"structured_call_count": len(model.structured_messages),
"memory_md_exists": os.path.isfile(
os.path.join(self.temp_dir, "Memory", "MEMORY.md"),
),
},
{
"reply": {
"id": AnyString(),
"name": "assistant",
"role": "assistant",
"content": [
{
"type": "text",
"text": "index only",
"id": AnyString(),
},
],
"metadata": {},
},
"hints": [],
"structured_call_count": 0,
"memory_md_exists": True,
},
)
async def test_agent_system_prompt_contains_truncation_reminder(
self,
) -> None:
"""Agent system prompt should contain reminder for truncated index."""
memory_dir = os.path.join(self.temp_dir, "Memory")
os.makedirs(memory_dir)
with open(
os.path.join(memory_dir, "MEMORY.md"),
"w",
encoding="utf-8",
) as f:
f.write("0123456789" * 80)
model = _RecordingMockModel()
model.set_responses([_text_response("truncated")])
middleware = AgenticMemoryMiddleware(
workdir=self.temp_dir,
parameters=AgenticMemoryMiddleware.Parameters(
memory_max_tokens=10,
retrieval_async=False,
),
)
agent = self._make_agent(model, middleware)
await agent.reply(UserMsg("user", "hello"))
system_prompt = model.chat_messages[0][0].get_text_content()
self.assertDictEqual(
{
"has_truncated_marker": "<<<TRUNCATED>>>" in system_prompt,
"has_offset_reminder": "Use the `Read` tool with offset"
in system_prompt,
"has_memory_path": os.path.join(memory_dir, "MEMORY.md")
in system_prompt,
},
{
"has_truncated_marker": True,
"has_offset_reminder": True,
"has_memory_path": True,
},
)
async def test_agent_skips_retrieval_when_async_retrieval_disabled(
self,
) -> None:
"""Agent should not run retrieval when ``retrieval_async`` is false."""
memory_dir = os.path.join(self.temp_dir, "Memory")
os.makedirs(memory_dir)
with open(
os.path.join(memory_dir, "MEMORY.md"),
"w",
encoding="utf-8",
) as f:
f.write("- [User profile](user_profile.md) — User profile.\n")
_write_memory_file(
memory_dir,
"user_profile.md",
"User profile details",
"user",
"This memory should not be retrieved.",
)
model = _RecordingMockModel()
model.set_structured_response(
_structured_response(["user_profile.md"]),
)
model.set_responses([_tool_response(), _text_response("disabled")])
middleware = AgenticMemoryMiddleware(
workdir=self.temp_dir,
parameters=AgenticMemoryMiddleware.Parameters(
retrieval_async=False,
),
)
agent = self._make_agent(
model,
middleware,
toolkit=Toolkit(tools=[_DummyTool()]),
)
reply = await agent.reply(UserMsg("user", "remember?"))
self.assertDictEqual(
{
"reply": _message_to_dict(reply),
"hints": _hint_texts(agent),
"structured_call_count": len(model.structured_messages),
},
{
"reply": {
"id": AnyString(),
"name": "assistant",
"role": "assistant",
"content": [
{
"type": "text",
"text": "disabled",
"id": AnyString(),
},
],
"metadata": {},
},
"hints": [],
"structured_call_count": 0,
},
)
|