e7f2b240
 tangwang
first commit
|
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
|
"""
Milvus Service for Vector Storage and Similarity Search
Manages text and image embeddings in separate collections
"""
import logging
from typing import Any, Dict, List, Optional
from pymilvus import (
DataType,
MilvusClient,
)
from app.config import settings
logger = logging.getLogger(__name__)
class MilvusService:
"""Service for managing vector embeddings in Milvus"""
def __init__(self, uri: Optional[str] = None):
"""Initialize Milvus service
Args:
uri: Milvus connection URI. If None, uses settings.milvus_uri
"""
if uri:
self.uri = uri
else:
# Use absolute path for Milvus Lite
self.uri = settings.milvus_uri_absolute
self.text_collection_name = settings.text_collection_name
self.image_collection_name = settings.image_collection_name
self.text_dim = settings.text_dim
self.image_dim = settings.image_dim
# Use MilvusClient for simplified operations
self._client: Optional[MilvusClient] = None
logger.info(f"Initializing Milvus service with URI: {self.uri}")
def is_connected(self) -> bool:
"""Check if connected to Milvus"""
return self._client is not None
def connect(self) -> None:
"""Connect to Milvus"""
if self.is_connected():
return
try:
self._client = MilvusClient(uri=self.uri)
logger.info(f"Connected to Milvus at {self.uri}")
except Exception as e:
logger.error(f"Failed to connect to Milvus: {e}")
raise
def disconnect(self) -> None:
"""Disconnect from Milvus"""
if self._client:
self._client.close()
self._client = None
logger.info("Disconnected from Milvus")
@property
def client(self) -> MilvusClient:
"""Get the Milvus client"""
if not self._client:
raise RuntimeError("Milvus not connected. Call connect() first.")
return self._client
def create_text_collection(self, recreate: bool = False) -> None:
"""Create collection for text embeddings with product metadata
Args:
recreate: If True, drop existing collection and recreate
"""
if recreate and self.client.has_collection(self.text_collection_name):
self.client.drop_collection(self.text_collection_name)
logger.info(f"Dropped existing collection: {self.text_collection_name}")
if self.client.has_collection(self.text_collection_name):
logger.info(f"Text collection already exists: {self.text_collection_name}")
return
# Create collection with schema (includes metadata fields)
schema = MilvusClient.create_schema(
auto_id=False,
enable_dynamic_field=True, # Allow additional metadata fields
)
# Core fields
schema.add_field(field_name="id", datatype=DataType.INT64, is_primary=True)
schema.add_field(field_name="text", datatype=DataType.VARCHAR, max_length=2000)
schema.add_field(
field_name="embedding", datatype=DataType.FLOAT_VECTOR, dim=self.text_dim
)
# Product metadata fields
schema.add_field(
field_name="productDisplayName", datatype=DataType.VARCHAR, max_length=500
)
schema.add_field(field_name="gender", datatype=DataType.VARCHAR, max_length=50)
schema.add_field(
field_name="masterCategory", datatype=DataType.VARCHAR, max_length=100
)
schema.add_field(
field_name="subCategory", datatype=DataType.VARCHAR, max_length=100
)
schema.add_field(
field_name="articleType", datatype=DataType.VARCHAR, max_length=100
)
schema.add_field(
field_name="baseColour", datatype=DataType.VARCHAR, max_length=50
)
schema.add_field(field_name="season", datatype=DataType.VARCHAR, max_length=50)
schema.add_field(field_name="usage", datatype=DataType.VARCHAR, max_length=50)
# Create index parameters
index_params = self.client.prepare_index_params()
index_params.add_index(
field_name="embedding",
index_type="AUTOINDEX",
metric_type="COSINE",
)
# Create collection
self.client.create_collection(
collection_name=self.text_collection_name,
schema=schema,
index_params=index_params,
)
logger.info(
f"Created text collection with metadata: {self.text_collection_name}"
)
def create_image_collection(self, recreate: bool = False) -> None:
"""Create collection for image embeddings with product metadata
Args:
recreate: If True, drop existing collection and recreate
"""
if recreate and self.client.has_collection(self.image_collection_name):
self.client.drop_collection(self.image_collection_name)
logger.info(f"Dropped existing collection: {self.image_collection_name}")
if self.client.has_collection(self.image_collection_name):
logger.info(
f"Image collection already exists: {self.image_collection_name}"
)
return
# Create collection with schema (includes metadata fields)
schema = MilvusClient.create_schema(
auto_id=False,
enable_dynamic_field=True, # Allow additional metadata fields
)
# Core fields
schema.add_field(field_name="id", datatype=DataType.INT64, is_primary=True)
schema.add_field(
field_name="image_path", datatype=DataType.VARCHAR, max_length=500
)
schema.add_field(
field_name="embedding", datatype=DataType.FLOAT_VECTOR, dim=self.image_dim
)
# Product metadata fields
schema.add_field(
field_name="productDisplayName", datatype=DataType.VARCHAR, max_length=500
)
schema.add_field(field_name="gender", datatype=DataType.VARCHAR, max_length=50)
schema.add_field(
field_name="masterCategory", datatype=DataType.VARCHAR, max_length=100
)
schema.add_field(
field_name="subCategory", datatype=DataType.VARCHAR, max_length=100
)
schema.add_field(
field_name="articleType", datatype=DataType.VARCHAR, max_length=100
)
schema.add_field(
field_name="baseColour", datatype=DataType.VARCHAR, max_length=50
)
schema.add_field(field_name="season", datatype=DataType.VARCHAR, max_length=50)
schema.add_field(field_name="usage", datatype=DataType.VARCHAR, max_length=50)
# Create index parameters
index_params = self.client.prepare_index_params()
index_params.add_index(
field_name="embedding",
index_type="AUTOINDEX",
metric_type="COSINE",
)
# Create collection
self.client.create_collection(
collection_name=self.image_collection_name,
schema=schema,
index_params=index_params,
)
logger.info(
f"Created image collection with metadata: {self.image_collection_name}"
)
def insert_text_embeddings(
self,
embeddings: List[Dict[str, Any]],
) -> int:
"""Insert text embeddings with metadata into collection
Args:
embeddings: List of dictionaries with keys:
- id: unique ID (product ID)
- text: the text that was embedded
- embedding: the embedding vector
- productDisplayName, gender, masterCategory, etc. (metadata)
Returns:
Number of inserted embeddings
"""
if not embeddings:
return 0
try:
# Insert data directly (all fields including metadata)
# Milvus will accept all fields defined in schema + dynamic fields
data = embeddings
# Insert data
result = self.client.insert(
collection_name=self.text_collection_name,
data=data,
)
logger.info(f"Inserted {len(data)} text embeddings")
return len(data)
except Exception as e:
logger.error(f"Failed to insert text embeddings: {e}")
raise
def insert_image_embeddings(
self,
embeddings: List[Dict[str, Any]],
) -> int:
"""Insert image embeddings with metadata into collection
Args:
embeddings: List of dictionaries with keys:
- id: unique ID (product ID)
- image_path: path to the image file
- embedding: the embedding vector
- productDisplayName, gender, masterCategory, etc. (metadata)
Returns:
Number of inserted embeddings
"""
if not embeddings:
return 0
try:
# Insert data directly (all fields including metadata)
# Milvus will accept all fields defined in schema + dynamic fields
data = embeddings
# Insert data
result = self.client.insert(
collection_name=self.image_collection_name,
data=data,
)
logger.info(f"Inserted {len(data)} image embeddings")
return len(data)
except Exception as e:
logger.error(f"Failed to insert image embeddings: {e}")
raise
def search_similar_text(
self,
query_embedding: List[float],
limit: int = 10,
filters: Optional[str] = None,
output_fields: Optional[List[str]] = None,
) -> List[Dict[str, Any]]:
"""Search for similar text embeddings
Args:
query_embedding: Query embedding vector
limit: Maximum number of results
filters: Filter expression (e.g., "product_id in [1, 2, 3]")
output_fields: List of fields to return
Returns:
List of search results with fields:
- id: embedding ID
- distance: similarity distance
- entity: the matched entity with requested fields
"""
try:
if output_fields is None:
output_fields = [
"id",
"text",
"productDisplayName",
"gender",
"masterCategory",
"subCategory",
"articleType",
"baseColour",
]
search_params = {}
if filters:
search_params["expr"] = filters
results = self.client.search(
collection_name=self.text_collection_name,
data=[query_embedding],
limit=limit,
output_fields=output_fields,
search_params=search_params,
)
# Format results
formatted_results = []
if results and len(results) > 0:
for hit in results[0]:
result = {"id": hit.get("id"), "distance": hit.get("distance")}
# Extract fields from entity
entity = hit.get("entity", {})
for field in output_fields:
if field in entity:
result[field] = entity.get(field)
formatted_results.append(result)
logger.debug(f"Found {len(formatted_results)} similar text embeddings")
return formatted_results
except Exception as e:
logger.error(f"Failed to search similar text: {e}")
raise
def search_similar_images(
self,
query_embedding: List[float],
limit: int = 10,
filters: Optional[str] = None,
output_fields: Optional[List[str]] = None,
) -> List[Dict[str, Any]]:
"""Search for similar image embeddings
Args:
query_embedding: Query embedding vector
limit: Maximum number of results
filters: Filter expression (e.g., "product_id in [1, 2, 3]")
output_fields: List of fields to return
Returns:
List of search results with fields:
- id: embedding ID
- distance: similarity distance
- entity: the matched entity with requested fields
"""
try:
if output_fields is None:
output_fields = [
"id",
"image_path",
"productDisplayName",
"gender",
"masterCategory",
"subCategory",
"articleType",
"baseColour",
]
search_params = {}
if filters:
search_params["expr"] = filters
results = self.client.search(
collection_name=self.image_collection_name,
data=[query_embedding],
limit=limit,
output_fields=output_fields,
search_params=search_params,
)
# Format results
formatted_results = []
if results and len(results) > 0:
for hit in results[0]:
result = {"id": hit.get("id"), "distance": hit.get("distance")}
# Extract fields from entity
entity = hit.get("entity", {})
for field in output_fields:
if field in entity:
result[field] = entity.get(field)
formatted_results.append(result)
logger.debug(f"Found {len(formatted_results)} similar image embeddings")
return formatted_results
except Exception as e:
logger.error(f"Failed to search similar images: {e}")
raise
def get_collection_stats(self, collection_name: str) -> Dict[str, Any]:
"""Get statistics for a collection
Args:
collection_name: Name of the collection
Returns:
Dictionary with collection statistics
"""
try:
stats = self.client.get_collection_stats(collection_name)
return {
"collection_name": collection_name,
"row_count": stats.get("row_count", 0),
}
except Exception as e:
logger.error(f"Failed to get collection stats: {e}")
return {"collection_name": collection_name, "row_count": 0}
def delete_by_ids(self, collection_name: str, ids: List[int]) -> int:
"""Delete embeddings by IDs
Args:
collection_name: Name of the collection
ids: List of IDs to delete
Returns:
Number of deleted embeddings
"""
if not ids:
return 0
try:
self.client.delete(
collection_name=collection_name,
ids=ids,
)
logger.info(f"Deleted {len(ids)} embeddings from {collection_name}")
return len(ids)
except Exception as e:
logger.error(f"Failed to delete embeddings: {e}")
raise
def clear_collection(self, collection_name: str) -> None:
"""Clear all data from a collection
Args:
collection_name: Name of the collection
"""
try:
if self.client.has_collection(collection_name):
self.client.drop_collection(collection_name)
logger.info(f"Dropped collection: {collection_name}")
except Exception as e:
logger.error(f"Failed to clear collection: {e}")
raise
# Global instance
_milvus_service: Optional[MilvusService] = None
def get_milvus_service() -> MilvusService:
"""Get or create the global Milvus service instance"""
global _milvus_service
if _milvus_service is None:
_milvus_service = MilvusService()
_milvus_service.connect()
return _milvus_service
|