Spaces:
Build error
Build error
File size: 16,287 Bytes
cb782c0 2190b45 cb782c0 2190b45 | 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 | """Tests for SQLiteVectorIndex using sqlite-vec.
Tests verify:
- Vector indexing and search
- True CRUD operations (real deletes)
- Filtering by user_id, session_id, etc.
- Persistence across instances
- Memory stats and bounding
"""
from __future__ import annotations
import os
import tempfile
import numpy as np
import pytest
from headroom.memory.models import Memory
from headroom.memory.ports import VectorFilter
# Check if sqlite-vec is available
try:
from headroom.memory.adapters.sqlite_vector import is_sqlite_vec_available
SQLITE_VEC_AVAILABLE = is_sqlite_vec_available()
except ImportError:
SQLITE_VEC_AVAILABLE = False
@pytest.mark.skipif(not SQLITE_VEC_AVAILABLE, reason="sqlite-vec not available")
class TestSQLiteVectorIndex:
"""Tests for SQLiteVectorIndex."""
@pytest.fixture
def index(self):
"""Create a temporary SQLite vector index."""
with tempfile.NamedTemporaryFile(suffix=".db", delete=False) as f:
db_path = f.name
from headroom.memory.adapters.sqlite_vector import SQLiteVectorIndex
index = SQLiteVectorIndex(dimension=384, db_path=db_path)
yield index
if os.path.exists(db_path):
os.unlink(db_path)
@pytest.mark.asyncio
async def test_index_and_search(self, index):
"""Test basic indexing and search."""
np.random.seed(42)
# Create memories with random embeddings
memories = []
for i in range(10):
embedding = np.random.randn(384).astype(np.float32)
memory = Memory(
content=f"Test content {i}",
user_id="alice",
embedding=embedding,
)
await index.index(memory)
memories.append(memory)
assert index.size == 10
# Search with first memory's embedding - should find itself
filter = VectorFilter(
query_vector=memories[0].embedding,
top_k=3,
user_id="alice",
)
results = await index.search(filter)
assert len(results) == 3
assert results[0].memory.id == memories[0].id
assert results[0].similarity > 0.99 # Should be ~1.0 for exact match
@pytest.mark.asyncio
async def test_true_delete(self, index):
"""Test that delete actually removes entries."""
np.random.seed(42)
# Add memories
memories = []
for i in range(5):
embedding = np.random.randn(384).astype(np.float32)
memory = Memory(
content=f"Content {i}",
user_id="alice",
embedding=embedding,
)
await index.index(memory)
memories.append(memory)
assert index.size == 5
# Delete one
result = await index.remove(memories[0].id)
assert result is True
assert index.size == 4
# Search should not find deleted memory
filter = VectorFilter(
query_vector=memories[0].embedding,
top_k=10,
user_id="alice",
)
results = await index.search(filter)
result_ids = {r.memory.id for r in results}
assert memories[0].id not in result_ids
@pytest.mark.asyncio
async def test_update_embedding(self, index):
"""Test updating an existing entry."""
np.random.seed(42)
embedding1 = np.random.randn(384).astype(np.float32)
memory = Memory(
content="Original content",
user_id="alice",
embedding=embedding1,
)
await index.index(memory)
# Update with new embedding
embedding2 = np.random.randn(384).astype(np.float32)
memory.embedding = embedding2
memory.content = "Updated content"
await index.index(memory)
# Should still be only 1 entry
assert index.size == 1
# Get stored embedding
stored = await index.get_embedding(memory.id)
assert stored is not None
np.testing.assert_array_almost_equal(stored, embedding2)
@pytest.mark.asyncio
async def test_filter_by_user(self, index):
"""Test filtering search results by user_id."""
np.random.seed(42)
# Create memories for different users with similar embeddings
base_embedding = np.random.randn(384).astype(np.float32)
for user in ["alice", "bob", "charlie"]:
# Slightly perturb embedding for each user
embedding = base_embedding + np.random.randn(384).astype(np.float32) * 0.1
memory = Memory(
content=f"Content for {user}",
user_id=user,
embedding=embedding,
)
await index.index(memory)
# Search filtered by user
filter = VectorFilter(
query_vector=base_embedding,
top_k=10,
user_id="alice",
)
results = await index.search(filter)
assert len(results) == 1
assert results[0].memory.user_id == "alice"
@pytest.mark.asyncio
async def test_filter_by_session(self, index):
"""Test filtering by session_id."""
np.random.seed(42)
embedding = np.random.randn(384).astype(np.float32)
# Same user, different sessions
for session in ["session1", "session2", None]:
memory = Memory(
content=f"Content for {session}",
user_id="alice",
session_id=session,
embedding=embedding + np.random.randn(384).astype(np.float32) * 0.01,
)
await index.index(memory)
filter = VectorFilter(
query_vector=embedding,
top_k=10,
user_id="alice",
session_id="session1",
)
results = await index.search(filter)
assert len(results) == 1
assert results[0].memory.session_id == "session1"
@pytest.mark.asyncio
async def test_min_similarity_filter(self, index):
"""Test minimum similarity threshold."""
np.random.seed(42)
# Create memories with varying similarity to query
query = np.random.randn(384).astype(np.float32)
query = query / np.linalg.norm(query) # Normalize
# Very similar
similar = query + np.random.randn(384).astype(np.float32) * 0.1
similar = similar / np.linalg.norm(similar)
# Less similar
less_similar = np.random.randn(384).astype(np.float32)
less_similar = less_similar / np.linalg.norm(less_similar)
await index.index(Memory(content="Similar", user_id="alice", embedding=similar))
await index.index(Memory(content="Less similar", user_id="alice", embedding=less_similar))
# High threshold should filter out dissimilar
filter = VectorFilter(
query_vector=query,
top_k=10,
min_similarity=0.8,
)
results = await index.search(filter)
# Only the similar one should pass
assert len(results) <= 1
if len(results) == 1:
assert results[0].similarity >= 0.8
@pytest.mark.skipif(not SQLITE_VEC_AVAILABLE, reason="sqlite-vec not available")
class TestSQLiteVectorIndexPersistence:
"""Tests for persistence across index instances."""
@pytest.mark.asyncio
async def test_data_persists_across_instances(self):
"""Test that data survives index restart."""
with tempfile.NamedTemporaryFile(suffix=".db", delete=False) as f:
db_path = f.name
try:
from headroom.memory.adapters.sqlite_vector import SQLiteVectorIndex
# Create index and add data
index1 = SQLiteVectorIndex(dimension=384, db_path=db_path)
np.random.seed(42)
embedding = np.random.randn(384).astype(np.float32)
memory = Memory(
content="Persistent content",
user_id="alice",
embedding=embedding,
)
await index1.index(memory)
memory_id = memory.id
# Create new index instance
index2 = SQLiteVectorIndex(dimension=384, db_path=db_path)
assert index2.size == 1
# Should find the memory
filter = VectorFilter(
query_vector=embedding,
top_k=1,
)
results = await index2.search(filter)
assert len(results) == 1
assert results[0].memory.id == memory_id
finally:
if os.path.exists(db_path):
os.unlink(db_path)
@pytest.mark.skipif(not SQLITE_VEC_AVAILABLE, reason="sqlite-vec not available")
class TestSQLiteVectorIndexMemoryStats:
"""Tests for memory statistics."""
@pytest.fixture
def index(self):
"""Create a temporary SQLite vector index."""
with tempfile.NamedTemporaryFile(suffix=".db", delete=False) as f:
db_path = f.name
from headroom.memory.adapters.sqlite_vector import SQLiteVectorIndex
index = SQLiteVectorIndex(dimension=384, db_path=db_path, page_cache_size_kb=4096)
yield index
if os.path.exists(db_path):
os.unlink(db_path)
@pytest.mark.asyncio
async def test_memory_stats(self, index):
"""Test memory statistics."""
stats = index.get_memory_stats()
assert stats.name == "sqlite_vector_index"
assert stats.entry_count == 0
assert stats.budget_bytes == 4096 * 1024 # 4MB cache
# Add some entries
np.random.seed(42)
for i in range(10):
embedding = np.random.randn(384).astype(np.float32)
memory = Memory(
content=f"Content {i}",
user_id="alice",
embedding=embedding,
)
await index.index(memory)
stats = index.get_memory_stats()
assert stats.entry_count == 10
assert stats.size_bytes > 0
@pytest.mark.asyncio
async def test_stats(self, index):
"""Test index statistics."""
np.random.seed(42)
for i in range(5):
embedding = np.random.randn(384).astype(np.float32)
memory = Memory(
content=f"Content {i}",
user_id="alice" if i < 3 else "bob",
embedding=embedding,
)
await index.index(memory)
stats = index.stats()
assert stats["size"] == 5
assert stats["dimension"] == 384
assert stats["users"] == 2
assert stats["page_cache_size_kb"] == 4096
assert stats["db_size_bytes"] > 0
@pytest.mark.skipif(not SQLITE_VEC_AVAILABLE, reason="sqlite-vec not available")
class TestSQLiteVectorIndexEdgeCases:
"""Tests for edge cases."""
@pytest.fixture
def index(self):
"""Create a temporary SQLite vector index."""
with tempfile.NamedTemporaryFile(suffix=".db", delete=False) as f:
db_path = f.name
from headroom.memory.adapters.sqlite_vector import SQLiteVectorIndex
index = SQLiteVectorIndex(dimension=384, db_path=db_path)
yield index
if os.path.exists(db_path):
os.unlink(db_path)
@pytest.mark.asyncio
async def test_search_empty_index(self, index):
"""Test searching an empty index."""
np.random.seed(42)
query = np.random.randn(384).astype(np.float32)
filter = VectorFilter(query_vector=query, top_k=10)
results = await index.search(filter)
assert len(results) == 0
@pytest.mark.asyncio
async def test_remove_nonexistent(self, index):
"""Test removing a nonexistent entry."""
result = await index.remove("nonexistent-id")
assert result is False
@pytest.mark.asyncio
async def test_wrong_dimension_raises(self, index):
"""Test that wrong embedding dimension raises error."""
wrong_embedding = np.random.randn(128).astype(np.float32) # Wrong dimension
memory = Memory(
content="Test",
user_id="alice",
embedding=wrong_embedding,
)
with pytest.raises(ValueError, match="dimension"):
await index.index(memory)
@pytest.mark.asyncio
async def test_no_embedding_raises(self, index):
"""Test that missing embedding raises error."""
memory = Memory(
content="Test",
user_id="alice",
embedding=None,
)
with pytest.raises(ValueError, match="no embedding"):
await index.index(memory)
@pytest.mark.asyncio
async def test_clear(self, index):
"""Test clearing all entries."""
np.random.seed(42)
for i in range(5):
embedding = np.random.randn(384).astype(np.float32)
memory = Memory(
content=f"Content {i}",
user_id="alice",
embedding=embedding,
)
await index.index(memory)
assert index.size == 5
index.clear()
assert index.size == 0
@pytest.mark.asyncio
async def test_batch_index(self, index):
"""Test batch indexing."""
np.random.seed(42)
memories = []
for i in range(10):
embedding = np.random.randn(384).astype(np.float32)
memory = Memory(
content=f"Content {i}",
user_id="alice",
embedding=embedding,
)
memories.append(memory)
# Add one without embedding
memories.append(Memory(content="No embedding", user_id="alice"))
indexed = await index.index_batch(memories)
assert indexed == 10
assert index.size == 10
@pytest.mark.asyncio
async def test_batch_index_uses_single_connection(self, index, monkeypatch):
"""Test batch indexing reuses a single sqlite-vec connection."""
np.random.seed(42)
memories = [
Memory(
content=f"Content {i}",
user_id="alice",
embedding=np.random.randn(384).astype(np.float32),
)
for i in range(10)
]
original_get_conn = index._get_conn
conn_calls = 0
def counting_get_conn():
nonlocal conn_calls
conn_calls += 1
return original_get_conn()
monkeypatch.setattr(index, "_get_conn", counting_get_conn)
indexed = await index.index_batch(memories)
assert indexed == 10
assert conn_calls == 1
@pytest.mark.asyncio
async def test_batch_remove(self, index):
"""Test batch removal."""
np.random.seed(42)
memories = []
for i in range(5):
embedding = np.random.randn(384).astype(np.float32)
memory = Memory(
content=f"Content {i}",
user_id="alice",
embedding=embedding,
)
await index.index(memory)
memories.append(memory)
# Remove some
ids_to_remove = [memories[0].id, memories[2].id, "nonexistent"]
removed = await index.remove_batch(ids_to_remove)
assert removed == 2
assert index.size == 3
@pytest.mark.asyncio
async def test_batch_remove_uses_single_connection(self, index, monkeypatch):
"""Test batch removal reuses a single sqlite-vec connection."""
np.random.seed(42)
memories = []
for i in range(5):
memory = Memory(
content=f"Content {i}",
user_id="alice",
embedding=np.random.randn(384).astype(np.float32),
)
await index.index(memory)
memories.append(memory)
original_get_conn = index._get_conn
conn_calls = 0
def counting_get_conn():
nonlocal conn_calls
conn_calls += 1
return original_get_conn()
monkeypatch.setattr(index, "_get_conn", counting_get_conn)
removed = await index.remove_batch([memories[0].id, memories[2].id, "nonexistent"])
assert removed == 2
assert conn_calls == 1
|