File size: 18,965 Bytes
63256f9 |
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 |
"""
Integration tests for ai-me agent.
Tests the complete setup including vectorstore, agent configuration, and agent responses.
"""
import pytest
import pytest_asyncio
import re
import sys
import os
import logging
from datetime import datetime
from unittest.mock import AsyncMock, patch
# Something about these tests makes me feel yucky. Big, brittle, and slow. BBS?
# In the future we should run inference locally with docker-compose models.
# Set temperature and seed for deterministic test results
os.environ["TEMPERATURE"] = "0"
os.environ["SEED"] = "42"
# Point our RAG to the tests/data directory
project_root = os.path.abspath(os.path.join(os.path.dirname(__file__), "../.."))
test_data_dir = os.path.join(project_root, "tests", "data")
os.environ["DOC_ROOT"] = test_data_dir
os.environ["LOCAL_DOCS"] = "**/*.md"
from config import setup_logger, Config
from agent import AIMeAgent
from data import DataManager, DataManagerConfig
logger = setup_logger(__name__)
# ============================================================================
# SHARED CACHING - Initialize on first use, then reuse
# ============================================================================
_config = None
_vectorstore = None
_data_manager = None
def _get_shared_config():
"""Lazy initialization of shared config."""
global _config
if _config is None:
_config = Config() # type: ignore
logger.info(f"Initialized shared config: {_config.bot_full_name}")
return _config
def _get_shared_vectorstore():
"""Lazy initialization of shared vectorstore."""
global _vectorstore, _data_manager
if _vectorstore is None:
logger.info("Initializing shared vectorstore (first test)...")
test_data_dir = os.path.join(project_root, "tests", "data")
_data_config = DataManagerConfig(
doc_root=test_data_dir
)
_data_manager = DataManager(config=_data_config)
_vectorstore = _data_manager.setup_vectorstore()
logger.info(f"Shared vectorstore ready: {_vectorstore._collection.count()} documents")
return _vectorstore
@pytest_asyncio.fixture(scope="function")
async def ai_me_agent():
"""
Setup fixture for ai-me agent with vectorstore and MCP servers.
CRITICAL: Function-scoped fixture prevents hanging/blocking issues.
Each test gets its own agent instance with proper cleanup.
Reuses shared config and vectorstore (lazy-initialized on first use).
This fixture:
- Reuses shared config and vectorstore
- Creates agent WITH real subprocess MCP servers (GitHub, Time, Memory)
- Yields agent for test
- Cleans up MCP servers after test completes
"""
config = _get_shared_config()
vectorstore = _get_shared_vectorstore()
# Initialize agent config with shared vectorstore
aime_agent = AIMeAgent(
bot_full_name=config.bot_full_name,
model=config.model,
vectorstore=vectorstore,
github_token=config.github_token,
session_id="test-session"
)
# Create the agent WITH MCP servers enabled
logger.info("Creating ai-me agent with MCP servers...")
assert aime_agent.session_id is not None, "session_id should be set"
await aime_agent.create_ai_me_agent(
mcp_params=[
aime_agent.mcp_github_params,
aime_agent.mcp_time_params,
aime_agent.get_mcp_memory_params(aime_agent.session_id),
]
)
logger.info("Agent created successfully with MCP servers")
logger.info(f"Temperature set to {config.temperature}")
logger.info(f"Seed set to {config.seed}")
# Yield the agent for the test
yield aime_agent
# CRITICAL: Cleanup after test completes to prevent hanging
logger.info("Cleaning up MCP servers after test...")
await aime_agent.cleanup()
logger.info("Cleanup complete")
@pytest.mark.asyncio
async def test_github_documents_load():
"""Tests FR-002: GitHub document loading with source metadata."""
config = Config() # type: ignore
# Load GitHub documents directly
github_config = DataManagerConfig(
doc_load_local=[]
)
dm = DataManager(config=github_config)
vs = dm.setup_vectorstore(github_repos=["byoung/ai-me"])
agent = AIMeAgent(
bot_full_name=config.bot_full_name,
model=config.model,
vectorstore=vs,
github_token=config.github_token,
session_id="test-session"
)
await agent.create_ai_me_agent()
response = await agent.run("Do you have python experience?")
assert "yes" in response.lower(), (
f"yes' in response but got: {response}"
)
@pytest.mark.asyncio
async def test_rear_knowledge_contains_it245(ai_me_agent):
"""Tests REQ-001: Knowledge base retrieval of personal documentation."""
response = await ai_me_agent.run("What is IT-245?")
assert "IT-245" in response or "It-245" in response or "it-245" in response
logger.info("β IT-245 found in response")
@pytest.mark.asyncio
async def test_github_commits_contains_shas(ai_me_agent):
"""Tests REQ-002: MCP GitHub integration - retrieve commit history."""
response = await ai_me_agent.run("What are some recent commits I've made?")
assert response, "Response is empty"
assert len(response) > 10, "Response is too short"
logger.info("β Response contains commit information")
@pytest.mark.asyncio
async def test_unknown_person_contains_negative_response(ai_me_agent):
"""Tests REQ-003: Graceful handling of out-of-scope requests."""
response = await ai_me_agent.run(
"Do you know Slartibartfast?" # Presumed unknown person
)
assert response, "Response is empty"
assert (
"don't know" in response.lower()
or "not familiar" in response.lower()
or "no information" in response.lower()
or "don't have any information" in response.lower()
), f"Response doesn't indicate lack of knowledge: {response}"
logger.info(f"β Test passed - correctly handled out-of-scope query")
@pytest.mark.asyncio
async def test_carol_knowledge_contains_product(ai_me_agent):
"""Tests FR-002, FR-003: Verify asking about Carol returns 'product'."""
response_raw = await ai_me_agent.run("Do you know Carol?")
response = response_raw.lower() # Convert to lowercase for matching
# Assert that 'product' appears in the response (Carol is Product Owner)
assert "product" in response, (
f"Expected 'product' in response but got: {response}"
)
logger.info("β Test passed: Response contains 'product'")
@pytest.mark.asyncio
async def test_mcp_time_server_returns_current_date(ai_me_agent):
"""Tests FR-009, NFR-001: Verify that the MCP time server returns the current date."""
response = await ai_me_agent.run("What is today's date?")
# Check for current date in various formats (ISO or natural language)
now = datetime.now()
expected_date, current_year, current_month, current_day = (
now.strftime("%Y-%m-%d"),
str(now.year),
now.strftime("%B"),
str(now.day),
)
# Accept either ISO format or natural language date
has_date = (
expected_date in response
or (
current_year in response
and current_month in response
and current_day in response
)
)
assert has_date, (
f"Expected response to contain current date "
f"({expected_date} or {current_month} {current_day}, {current_year}) "
f"but got: {response}"
)
logger.info(f"β Test passed: Response contains current date")
@pytest.mark.asyncio
async def test_mcp_memory_server_remembers_favorite_color(ai_me_agent):
"""Tests FR-013, NFR-002:
Verify that the MCP memory server persists information across interactions.
"""
await ai_me_agent.run("My favorite color is chartreuse.")
response2 = await ai_me_agent.run("What's my favorite color?")
# Check that the agent remembers the color
assert "chartreuse" in response2.lower(), (
f"Expected agent to remember favorite color 'chartreuse' "
f"but got: {response2}"
)
msg = (
"β Test passed: Agent remembered favorite color 'chartreuse' "
"across interactions"
)
logger.info(msg)
@pytest.mark.asyncio
async def test_github_relative_links_converted_to_absolute_urls():
"""Tests FR-004: Document processing converts relative GitHub links to absolute URLs.
Validates that when documents are loaded from GitHub with relative links
(e.g., /resume.md), they are rewritten to full GitHub URLs
(e.g., https://github.com/owner/repo/blob/main/resume.md).
This is a unit-level test of the DataManager.process_documents() method.
"""
from langchain_core.documents import Document
sample_doc = Document(
page_content=(
"Check out [my resume](/resume.md) and "
"[projects](/projects.md) for more info."
),
metadata={
"source": "github://byoung/ai-me/docs/about.md",
"github_repo": "byoung/ai-me"
}
)
# Verify metadata is set correctly before processing
assert sample_doc.metadata["github_repo"] == "byoung/ai-me", (
"Sample doc metadata should have github_repo"
)
data_config = DataManagerConfig()
data_manager = DataManager(config=data_config)
processed_docs = data_manager.process_documents([sample_doc])
assert len(processed_docs) == 1, "Expected 1 processed document"
processed_content = processed_docs[0].page_content
# Check that relative links have been converted to absolute GitHub URLs
assert "https://github.com/byoung/ai-me/blob/main/resume.md" in processed_content, (
f"Expected absolute GitHub URL for /resume.md in processed content, "
f"but got: {processed_content}"
)
assert "https://github.com/byoung/ai-me/blob/main/projects.md" in processed_content, (
f"Expected absolute GitHub URL for /projects.md in processed content, "
f"but got: {processed_content}"
)
logger.info("β Test passed: Relative GitHub links converted to absolute URLs")
logger.info(f" Original: [my resume](/resume.md)")
logger.info(f" Converted: [my resume](https://github.com/byoung/ai-me/blob/main/resume.md)")
@pytest.mark.asyncio
async def test_agent_responses_cite_sources(ai_me_agent):
"""Tests FR-004, FR-011: Agent responses include source citations.
Validates that agent responses include proper source attribution,
which could be GitHub URLs, local paths, or explicit source references.
"""
questions = [
"What do you know about ReaR?",
"Tell me about your experience in technology",
]
for question in questions:
logger.info(f"\n{'='*60}\nSource citation test: {question}\n{'='*60}")
response = await ai_me_agent.run(question)
# Check that response includes some form of source attribution
# Could be: GitHub URL, local path, "Sources" section, etc.
has_source = (
"https://github.com/" in response or
".md" in response or # Local markdown file reference
"source" in response.lower() or
"documentation" in response.lower()
)
assert has_source, (
f"Expected source attribution in response to '{question}' "
f"but found none. Response: {response}"
)
# Verify response is substantive (not just metadata)
min_length = 50
assert len(response) > min_length, (
f"Response to '{question}' was too short: {response}"
)
logger.info(f"β Source citation found for: {question[:40]}...")
logger.info("\nβ Test passed: Agent responses cite sources (FR-004, FR-011)")
@pytest.mark.asyncio
async def test_user_story_2_multi_topic_consistency(ai_me_agent):
"""
Tests FR-001, FR-003, FR-005, NFR-002: User Story 2 - Multi-Topic Consistency
Verify that the agent maintains consistent first-person perspective
across multiple conversation topics.
This tests that the agent:
- Uses first-person perspective (I, my, me) consistently
- Maintains professional tone across different topic switches
- Shows context awareness of different topics
- Remains in-character as the personified individual
"""
# Ask 3 questions about different topics
topics = [
("What is your background in technology?", "background|experience|technology"),
("What programming languages are you skilled in?", "programming|language|skilled"),
]
first_person_patterns = [
r"\bi\b", r"\bme\b", r"\bmy\b", r"\bmyself\b",
r"\bI['m]", r"\bI['ve]", r"\bI['ll]"
]
for question, topic_keywords in topics:
logger.info(f"\n{'='*60}\nMulti-topic test question: {question}\n{'='*60}")
response = await ai_me_agent.run(question)
response_lower = response.lower()
# Check for first-person usage
first_person_found = any(
re.search(pattern, response, re.IGNORECASE)
for pattern in first_person_patterns
)
assert first_person_found, (
f"Expected first-person perspective in response to '{question}' "
f"but got: {response}"
)
# Verify response is substantive (not just "I don't know")
min_length = 50 # Substantive responses should be > 50 chars
assert len(response) > min_length, (
f"Response to '{question}' was too short (likely not substantive): {response}"
)
logger.info(f"β First-person perspective maintained for: {question[:40]}...")
logger.info(f" Response preview: {response[:100]}...")
logger.info("\nβ Test passed: Consistent first-person perspective across 3+ topics")
@pytest.mark.asyncio
async def test_tool_failure_error_messages_are_friendly(caplog, ai_me_agent):
"""
Tests FR-012, NFR-003: Error Message Quality (FR-012)
Verify that tool failures return user-friendly messages without Python tracebacks.
This tests that the agent:
- Returns human-readable error messages
- logs an error that can be reviewed in our dashboard/logs
Uses mocking to simulate tool failures without adding test-specific code to agent.py
"""
logger.info(f"\n{'='*60}\nError Handling Test\n{'='*60}")
# Mock the Runner.run method to simulate a tool failure
# This tests the catch-all exception handler without adding test code to production
test_scenarios = [
RuntimeError("Simulated tool timeout"),
ValueError("Invalid tool parameters"),
]
for error in test_scenarios:
logger.info(f"\nTesting error scenario: {error.__class__.__name__}: {error}")
# Clear previous log records for this iteration
caplog.clear()
# Mock Runner.run to raise an exception
with patch('agent.Runner.run', new_callable=AsyncMock) as mock_run:
mock_run.side_effect = error
response = await ai_me_agent.run("Any user question")
logger.info(f"Response: {response[:100]}...")
# PRIMARY CHECK: Verify "I encountered an unexpected error" is in response
assert "I encountered an unexpected error" in response, (
f"Response must contain 'I encountered an unexpected error'. Got: {response}"
)
# SECONDARY CHECK: Verify error was logged by agent.py
error_logs = [record for record in caplog.records if record.levelname == "ERROR"]
assert len(error_logs) > 0, "Expected at least one ERROR log record from agent.py"
# Find the agent.py error log (contains "Unexpected error:")
agent_error_logged = any(
"Unexpected error:" in record.message for record in error_logs
)
assert agent_error_logged, (
f"Expected ERROR log with 'Unexpected error:' from agent.py. "
f"Got: {[r.message for r in error_logs]}"
)
error_messages = [
r.message for r in error_logs
if "Unexpected error:" in r.message
]
logger.info(
f"β Error properly logged to logger: {error_messages}"
)
logger.info("\nβ Test passed: Error messages are friendly (FR-012) + properly logged")
@pytest.mark.asyncio
async def test_logger_setup_format(caplog):
"""Tests NFR-003 (Structured Logging): Verify setup_logger creates structured logging.
Tests that setup_logger() configures syslog-style format with JSON support for
structured logging of user/agent interactions.
This validates the logger configuration that our production app relies on
for analytics and debugging.
"""
# Force logger setup to run by clearing handlers so setup_logger reconfigures
root_logger = logging.getLogger()
original_handlers = root_logger.handlers[:]
for handler in root_logger.handlers[:]:
root_logger.removeHandler(handler)
try:
# Now call setup_logger with no handlers - should trigger full setup
test_logger = setup_logger("test.structured_logging")
# Verify logger was created
assert test_logger.name == "test.structured_logging"
# Verify root logger now has handlers (setup_logger should have added them)
assert len(root_logger.handlers) > 0, (
"Root logger should have handlers after setup_logger"
)
# Verify we have a StreamHandler (console output)
has_stream_handler = any(
isinstance(handler, logging.StreamHandler)
for handler in root_logger.handlers
)
assert has_stream_handler, "Should have StreamHandler for console output"
# Test that logging works with structured JSON format
# The formatters should support JSON logging for analytics
test_logger.info(
'{"session_id": "test-session", "user_input": "test message"}'
)
logger.info(
"β Test passed: Logger setup configures structured logging (NFR-003)"
)
finally:
# Restore original handlers
for handler in root_logger.handlers[:]:
root_logger.removeHandler(handler)
for handler in original_handlers:
root_logger.addHandler(handler)
if __name__ == "__main__":
# Allow running tests directly with python test.py
pytest.main([__file__, "-v", "-s"])
|