-
Notifications
You must be signed in to change notification settings - Fork 1.4k
Expand file tree
/
Copy pathfeature_server.py
More file actions
1528 lines (1327 loc) · 55.6 KB
/
Copy pathfeature_server.py
File metadata and controls
1528 lines (1327 loc) · 55.6 KB
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
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
# Copyright 2025 The Feast Authors
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# https://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import asyncio
import functools
import os
import sys
import threading
import time
import traceback
from collections import defaultdict
from concurrent.futures import ThreadPoolExecutor
from contextlib import asynccontextmanager
from datetime import datetime
from importlib import resources as importlib_resources
from types import SimpleNamespace
from typing import Any, DefaultDict, Dict, List, NamedTuple, Optional, Set, Union
import pandas as pd
from dateutil import parser
from fastapi import (
Depends,
FastAPI,
Query,
Request,
Response,
WebSocket,
WebSocketDisconnect,
status,
)
from fastapi.concurrency import run_in_threadpool
from fastapi.logger import logger
from fastapi.responses import JSONResponse
from fastapi.staticfiles import StaticFiles
from pydantic import BaseModel, Field, field_validator
import feast
from feast import metrics as feast_metrics
from feast import proto_json, utils
from feast.constants import DEFAULT_FEATURE_SERVER_REGISTRY_TTL
from feast.data_source import PushMode
from feast.errors import (
FeastError,
FeatureViewNotFoundException,
)
from feast.feast_object import FeastObject
from feast.feature_server_utils import convert_response_to_dict
from feast.feature_view import FeatureViewState
from feast.feature_view_utils import get_feature_view_from_feature_store
from feast.filter_models import ComparisonFilter, CompoundFilter
from feast.permissions.action import WRITE, AuthzedAction
from feast.permissions.security_manager import (
assert_permissions,
get_security_manager,
is_auth_necessary,
)
from feast.permissions.server.rest import inject_user_details
from feast.permissions.server.utils import (
ServerType,
init_auth_manager,
init_security_manager,
str_to_auth_manager_type,
)
from feast.vector_store_utils import VectorStoreRegistry, build_vector_store_object
# TODO: deprecate this in favor of push features
class WriteToFeatureStoreRequest(BaseModel):
feature_view_name: str
df: dict
allow_registry_cache: bool = True
transform_on_write: bool = True
class PushFeaturesRequest(BaseModel):
push_source_name: str
df: dict
allow_registry_cache: bool = True
to: str = "online"
transform_on_write: bool = True
class MaterializeRequest(BaseModel):
start_ts: Optional[str] = None
end_ts: Optional[str] = None
feature_views: Optional[List[str]] = None
disable_event_timestamp: bool = False
full_feature_names: bool = False
version: Optional[str] = Field(
None,
description=(
"Optional version to materialize (e.g. 'v2'). Requires feature_views "
"with exactly one entry and registry.enable_online_feature_view_versioning."
),
)
class MaterializeIncrementalRequest(BaseModel):
end_ts: str
feature_views: Optional[List[str]] = None
full_feature_names: bool = False
version: Optional[str] = Field(
None,
description=(
"Optional version to materialize (e.g. 'v2'). Requires feature_views "
"with exactly one entry and registry.enable_online_feature_view_versioning."
),
)
class GetOnlineFeaturesRequest(BaseModel):
entities: Dict[str, List[Any]]
feature_service: Optional[str] = None
features: List[str] = []
full_feature_names: bool = False
include_feature_view_version_metadata: bool = False
class GetOnlineDocumentsRequest(BaseModel):
feature_service: Optional[str] = None
features: List[str] = []
full_feature_names: bool = False
include_feature_view_version_metadata: bool = False
top_k: Optional[int] = None
query: Optional[List[float]] = None
query_string: Optional[str] = None
distance_metric: Optional[str] = None
api_version: Optional[int] = 1
filters: Optional[Union[ComparisonFilter, CompoundFilter]] = None
class OpenAISearchMetadata(BaseModel):
features_to_retrieve: Optional[List[str]] = None
content_field: Optional[str] = None
class OpenAIRankingOptions(BaseModel):
ranker: Optional[str] = None
score_threshold: Optional[float] = None
class OpenAISearchRequest(BaseModel):
query: Union[str, List[str]]
filters: Optional[Union[ComparisonFilter, CompoundFilter]] = None
max_num_results: Optional[int] = 10
ranking_options: Optional[OpenAIRankingOptions] = None
rewrite_query: Optional[bool] = None
metadata: Optional[OpenAISearchMetadata] = None
class OpenAIVectorStoreObject(BaseModel):
id: str
object: str = "vector_store"
name: str
status: str = "completed"
created_at: int = 0
class FeatureVectorResponse(BaseModel):
values: List[Any] = []
statuses: List[str] = []
event_timestamps: List[str] = []
class OnlineFeaturesMetadataResponse(BaseModel):
feature_names: List[str] = []
@field_validator("feature_names", mode="before")
@classmethod
def _unwrap_feature_list(cls, v: Any) -> Any:
"""Accept both the proto_json-patched flat list and the raw
protobuf ``{"val": [...]}`` dict produced by ``MessageToDict``
when the monkey-patch is absent or ineffective."""
if isinstance(v, dict) and "val" in v:
return v["val"]
return v
class OnlineFeaturesResponse(BaseModel):
metadata: Optional[OnlineFeaturesMetadataResponse] = None
results: List[FeatureVectorResponse] = []
status: Optional[bool] = None
class ChatMessage(BaseModel):
role: str
content: str
class ChatRequest(BaseModel):
messages: List[ChatMessage]
def _parse_feature_info(
features: Union[List[str], "feast.FeatureService"],
) -> tuple:
"""Return ``(feature_view_names, feature_count)`` from resolved features.
``features`` is either a list of ``"feature_view:feature"`` strings or
a ``FeatureService`` with ``feature_view_projections``.
Returns:
(fv_names, feat_count) where fv_names is a list of unique feature
view name strings and feat_count is the total number of features.
"""
from feast.feature_service import FeatureService
from feast.utils import _parse_feature_ref
if isinstance(features, FeatureService):
projections = features.feature_view_projections
fv_names = [p.name for p in projections]
feat_count = sum(len(p.features) for p in projections)
elif isinstance(features, list):
feat_count = len(features)
fv_names = list({_parse_feature_ref(ref)[0] for ref in features if ":" in ref})
else:
fv_names = []
feat_count = 0
return fv_names, feat_count
def _resolve_feature_counts(
features: Union[List[str], "feast.FeatureService"],
) -> tuple:
"""Return ``(feature_count_str, feature_view_count_str)`` for Prometheus labels."""
fv_names, feat_count = _parse_feature_info(features)
return str(feat_count), str(len(fv_names))
def _emit_online_audit(
request: GetOnlineFeaturesRequest,
features: Union[List[str], "feast.FeatureService"],
entity_count: int,
status: str,
latency_ms: float,
):
"""Best-effort audit log emission for online feature requests."""
try:
from feast.permissions.security_manager import get_security_manager
requestor_id = "anonymous"
sm = get_security_manager()
if sm and sm.current_user:
requestor_id = sm.current_user.username or "anonymous"
fv_names, feat_count = _parse_feature_info(features)
feast_metrics.emit_online_audit_log(
requestor_id=requestor_id,
entity_keys=list(request.entities.keys()),
entity_count=entity_count,
feature_views=fv_names,
feature_count=feat_count,
status=status,
latency_ms=latency_ms,
)
except Exception:
logger.warning("Failed to emit online audit log", exc_info=True)
async def _get_features(
request: Union[GetOnlineFeaturesRequest, GetOnlineDocumentsRequest],
store: "feast.FeatureStore",
):
if request.feature_service:
feature_service = await run_in_threadpool(
store.get_feature_service, request.feature_service, allow_cache=True
)
assert_permissions(
resource=feature_service, actions=[AuthzedAction.READ_ONLINE]
)
features = feature_service # type: ignore
elif is_auth_necessary(get_security_manager()):
all_feature_views, all_on_demand_feature_views = await run_in_threadpool(
utils._get_feature_views_to_use,
store.registry,
store.project,
request.features,
allow_cache=True,
hide_dummy_entity=False,
)
for feature_view in all_feature_views:
assert_permissions(
resource=feature_view, actions=[AuthzedAction.READ_ONLINE]
)
for od_feature_view in all_on_demand_feature_views:
assert_permissions(
resource=od_feature_view, actions=[AuthzedAction.READ_ONLINE]
)
features = request.features # type: ignore
else:
features = request.features # type: ignore
return features
async def load_static_artifacts(app: FastAPI, store):
"""
Load static artifacts (models, lookup tables, etc.) into app.state.
This function can be extended to load various types of static artifacts:
- Small ML models (scikit-learn, small neural networks)
- Lookup tables and reference data
- Configuration parameters
- Pre-computed embeddings
Note: Not recommended for large language models - use dedicated
model serving solutions (vLLM, TGI, etc.) for those.
"""
try:
# Import here to avoid loading heavy dependencies unless needed
import importlib.util
import inspect
from pathlib import Path
# Look for static artifacts loading in the feature repository
# This allows templates and users to define their own artifact loading
repo_path = Path(store.repo_path) if store.repo_path else Path.cwd()
artifacts_file = repo_path / "static_artifacts.py"
if artifacts_file.exists():
# Load and execute custom static artifacts loading
spec = importlib.util.spec_from_file_location(
"static_artifacts", artifacts_file
)
if spec and spec.loader:
artifacts_module = importlib.util.module_from_spec(spec)
spec.loader.exec_module(artifacts_module)
# Look for load_artifacts function
if hasattr(artifacts_module, "load_artifacts"):
load_func = artifacts_module.load_artifacts
if inspect.iscoroutinefunction(load_func):
await load_func(app)
else:
load_func(app)
logger.info("Loaded static artifacts from static_artifacts.py")
except Exception as e:
# Non-fatal error - feature server should still start
logger.warning(f"Failed to load static artifacts: {e}")
def _authorize_materialize_views(
store: "feast.FeatureStore",
feature_view_names: Optional[List[str]],
version: Optional[str] = None,
) -> List[str]:
"""Resolve + authorize feature views for materialization.
Returns the resolved list of FV names (all eligible FVs when
feature_view_names is None).
"""
parsed_version = store._validate_materialize_version(version, feature_view_names)
feature_views_to_materialize = store._get_feature_views_to_materialize(
feature_view_names, version=parsed_version
)
for fv in feature_views_to_materialize:
assert_permissions(
resource=fv,
actions=[AuthzedAction.WRITE_ONLINE],
)
return [fv.name for fv in feature_views_to_materialize]
def _check_already_materializing(
store: "feast.FeatureStore",
fv_names: List[str],
) -> Optional[JSONResponse]:
"""Return a 409 JSONResponse if any requested FV is already MATERIALIZING."""
conflicting: List[str] = []
for fv_name in fv_names:
try:
fv = store.registry.get_feature_view(
fv_name, store.project, allow_cache=False
)
if getattr(fv, "state", None) == FeatureViewState.MATERIALIZING:
conflicting.append(fv_name)
except (FeatureViewNotFoundException, KeyError):
pass
except Exception as e:
logger.warning(
f"Unexpected error checking MATERIALIZING state for {fv_name}: {e}"
)
if conflicting:
return JSONResponse(
status_code=409,
content={
"error": (
f"Cannot start async materialization — the following feature "
f"views are already in MATERIALIZING state: {conflicting}. "
f"Use ?force=true to override."
),
"feature_views": conflicting,
},
)
return None
def _update_fv_state(
store: "feast.FeatureStore",
fv_names: List[str],
state: FeatureViewState,
) -> None:
"""Set FV state in the registry for each named feature view."""
for fv_name in fv_names:
try:
fv = store.registry.get_feature_view(
fv_name, store.project, allow_cache=False
)
fv.state = state
store.registry.apply_feature_view(fv, store.project)
except (FeatureViewNotFoundException, KeyError):
logger.warning(f"Feature view {fv_name} not found; skip state={state}")
except Exception as e:
logger.warning(f"Failed to set state={state} for {fv_name}: {e}")
def _reset_stuck_materializing_to_generated(
store: "feast.FeatureStore",
fv_names: List[str],
) -> None:
"""Reset FVs currently in MATERIALIZING to GENERATED (force override)."""
stuck: List[str] = []
for fv_name in fv_names:
try:
fv = store.registry.get_feature_view(
fv_name, store.project, allow_cache=False
)
if getattr(fv, "state", None) == FeatureViewState.MATERIALIZING:
stuck.append(fv_name)
except (FeatureViewNotFoundException, KeyError):
pass
except Exception as e:
logger.warning(f"Unexpected error while force-resetting {fv_name}: {e}")
if stuck:
_update_fv_state(store, stuck, FeatureViewState.GENERATED)
logger.info(
"Force reset MATERIALIZING → GENERATED for feature views: %s", stuck
)
def _parse_materialize_timestamps(
request: "MaterializeRequest",
) -> tuple:
"""Parse and validate start/end timestamps from a MaterializeRequest."""
if request.disable_event_timestamp:
now = datetime.now()
return datetime(1970, 1, 1), now
if not request.start_ts or not request.end_ts:
raise ValueError(
"start_ts and end_ts are required when disable_event_timestamp is False"
)
try:
start_date = utils.make_tzaware(parser.parse(request.start_ts))
end_date = utils.make_tzaware(parser.parse(request.end_ts))
except (ValueError, TypeError) as e:
raise ValueError(f"Invalid timestamp format: {e}") from e
if start_date >= end_date:
raise ValueError(f"start_ts ({start_date}) must be before end_ts ({end_date})")
return start_date, end_date
def get_app(
store: "feast.FeatureStore",
registry_ttl_sec: int = DEFAULT_FEATURE_SERVER_REGISTRY_TTL,
):
"""
Creates a FastAPI app that can be used to start a feature server.
Args:
store: The FeatureStore to use for serving features
registry_ttl_sec: The TTL in seconds for the registry cache
Returns:
A FastAPI app
Example:
```python
from feast import FeatureStore
store = FeatureStore(repo_path="feature_repo")
app = get_app(store)
```
The app provides the following endpoints:
- `/get-online-features`: Get online features
- `/search`: Vector similarity search (RAG)
- `/retrieve-online-documents`: Deprecated alias for `/search`
- `/v1/vector_stores`: List vector stores (GET)
- `/v1/vector_stores/{vector_store_id}`: Get a vector store (GET)
- `/v1/vector_stores/{vector_store_id}/search`: OpenAI-compatible vector search
- `/push`: Push features to the feature store
- `/write-to-online-store`: Write to the online store
- `/health`: Health check
- `/materialize`: Materialize features
- `/materialize-incremental`: Materialize features incrementally
- `/chat`: Chat UI
- `/ws/chat`: WebSocket endpoint for chat
MCP Support:
- If MCP is enabled in feature server configuration, MCP endpoints will be added automatically
"""
proto_json.patch()
# Asynchronously refresh registry, notifying shutdown and canceling the active timer if the app is shutting down
shutting_down = False
active_timer: Optional[threading.Timer] = None
# --- Offline write batching config and batcher ---
fs_cfg = getattr(store.config, "feature_server", None)
batching_cfg = None
if fs_cfg is not None:
enabled = getattr(fs_cfg, "offline_push_batching_enabled", False)
batch_size = getattr(fs_cfg, "offline_push_batching_batch_size", None)
batch_interval_seconds = getattr(
fs_cfg, "offline_push_batching_batch_interval_seconds", None
)
if enabled is True:
size_ok = isinstance(batch_size, int) and not isinstance(batch_size, bool)
interval_ok = isinstance(batch_interval_seconds, int) and not isinstance(
batch_interval_seconds, bool
)
if size_ok and interval_ok:
batching_cfg = SimpleNamespace(
enabled=True,
batch_size=batch_size,
batch_interval_seconds=batch_interval_seconds,
)
else:
logger.warning(
"Offline write batching enabled but missing or invalid numeric values; "
"disabling batching (batch_size=%r, batch_interval_seconds=%r)",
batch_size,
batch_interval_seconds,
)
offline_batcher: Optional[OfflineWriteBatcher] = None
if batching_cfg is not None and batching_cfg.enabled is True:
offline_batcher = OfflineWriteBatcher(store=store, cfg=batching_cfg)
logger.debug("Offline write batching is ENABLED")
else:
logger.debug("Offline write batching is DISABLED")
# Dedicated pool for async materialize so long Spark/offline waits do not
# starve the default executor used by online serving and run_in_threadpool.
_mat_workers_raw = os.environ.get("FEAST_MATERIALIZE_MAX_WORKERS", "2")
try:
materialize_max_workers = max(1, int(_mat_workers_raw))
except ValueError:
logger.warning(
"Invalid FEAST_MATERIALIZE_MAX_WORKERS=%r; using default 2",
_mat_workers_raw,
)
materialize_max_workers = 2
materialize_executor = ThreadPoolExecutor(
max_workers=materialize_max_workers,
thread_name_prefix="feast-materialize",
)
def stop_refresh():
nonlocal shutting_down
shutting_down = True
if active_timer:
active_timer.cancel()
vs_registry = VectorStoreRegistry(store)
def async_refresh():
if shutting_down:
return
store.refresh_registry()
vs_registry.refresh()
if registry_ttl_sec:
nonlocal active_timer
active_timer = threading.Timer(registry_ttl_sec, async_refresh)
active_timer.start()
@asynccontextmanager
async def lifespan(app: FastAPI):
# Load static artifacts before initializing store
await load_static_artifacts(app, store)
await store.initialize()
async_refresh()
try:
yield
finally:
stop_refresh()
if offline_batcher is not None:
offline_batcher.shutdown()
# wait=False: do not block process exit on in-flight materialize
# (same fire-and-forget contract as returning 202 mid-job).
materialize_executor.shutdown(wait=False)
await store.close()
app = FastAPI(lifespan=lifespan)
@app.post(
"/get-online-features",
dependencies=[Depends(inject_user_details)],
response_model=OnlineFeaturesResponse,
)
async def get_online_features(request: GetOnlineFeaturesRequest) -> Any:
with feast_metrics.track_request_latency(
"/get-online-features",
) as metrics_ctx:
features = await _get_features(request, store)
feat_count, fv_count = _resolve_feature_counts(features)
metrics_ctx.feature_count = feat_count
metrics_ctx.feature_view_count = fv_count
entity_count = len(next(iter(request.entities.values()), []))
feast_metrics.track_online_features_entities(entity_count)
read_params = dict(
features=features,
entity_rows=request.entities,
full_feature_names=request.full_feature_names,
include_feature_view_version_metadata=request.include_feature_view_version_metadata,
)
audit_start_ms = time.monotonic() * 1000
audit_status = "success"
try:
if store._get_provider().async_supported.online.read:
response = await store.get_online_features_async(**read_params) # type: ignore
else:
response = await run_in_threadpool(
lambda: store.get_online_features(**read_params) # type: ignore
)
except Exception:
audit_status = "error"
raise
finally:
audit_latency_ms = time.monotonic() * 1000 - audit_start_ms
_emit_online_audit(
request, features, entity_count, audit_status, audit_latency_ms
)
response_dict = await run_in_threadpool(
convert_response_to_dict, response.proto
)
return JSONResponse(content=response_dict)
async def _search_online_documents(
request: GetOnlineDocumentsRequest,
*,
metrics_path: str,
) -> JSONResponse:
with feast_metrics.track_request_latency(metrics_path):
features = await _get_features(request, store)
read_params = dict(
features=features,
query=request.query,
top_k=request.top_k,
)
if request.api_version == 2 and request.query_string is not None:
read_params["query_string"] = request.query_string
if request.api_version == 2 and request.distance_metric is not None:
read_params["distance_metric"] = request.distance_metric
if request.api_version == 2 and request.filters is not None:
read_params["filters"] = request.filters
if request.api_version == 2:
read_params["include_feature_view_version_metadata"] = (
request.include_feature_view_version_metadata
)
response = await run_in_threadpool(
lambda: store.retrieve_online_documents_v2(**read_params) # type: ignore
)
else:
response = await run_in_threadpool(
lambda: store.retrieve_online_documents(**read_params) # type: ignore
)
response_dict = await run_in_threadpool(
convert_response_to_dict, response.proto
)
return JSONResponse(content=response_dict)
@app.post(
"/search",
dependencies=[Depends(inject_user_details)],
response_model=OnlineFeaturesResponse,
)
async def search(request: GetOnlineDocumentsRequest) -> JSONResponse:
"""Vector similarity search against online document embeddings."""
return await _search_online_documents(request, metrics_path="/search")
@app.post(
"/retrieve-online-documents",
dependencies=[Depends(inject_user_details)],
response_model=OnlineFeaturesResponse,
include_in_schema=False,
)
async def retrieve_online_documents(
request: GetOnlineDocumentsRequest,
) -> JSONResponse:
logger.warning(
"POST /retrieve-online-documents is deprecated; use POST /search instead."
)
return await _search_online_documents(
request, metrics_path="/retrieve-online-documents"
)
@app.get(
"/v1/vector_stores",
dependencies=[Depends(inject_user_details)],
)
async def list_vector_stores() -> JSONResponse:
permitted: list = []
for obj in vs_registry.list_vector_stores():
fv = vs_registry.resolve(obj["id"])
try:
assert_permissions(resource=fv, actions=[AuthzedAction.DESCRIBE])
permitted.append(obj)
except Exception:
pass
return JSONResponse(content={"object": "list", "data": permitted})
@app.get(
"/v1/vector_stores/{vector_store_id}",
dependencies=[Depends(inject_user_details)],
)
async def get_vector_store(vector_store_id: str) -> JSONResponse:
try:
fv = vs_registry.resolve(vector_store_id)
assert_permissions(resource=fv, actions=[AuthzedAction.DESCRIBE])
except FeatureViewNotFoundException:
return JSONResponse(
status_code=404,
content={
"error": {
"message": f"No vector store found with id '{vector_store_id}'",
"type": "not_found_error",
}
},
)
return JSONResponse(content=build_vector_store_object(store.project, fv))
@app.post(
"/v1/vector_stores/{vector_store_id}/search",
dependencies=[Depends(inject_user_details)],
)
async def vector_store_search(
vector_store_id: str,
request: OpenAISearchRequest,
) -> JSONResponse:
with feast_metrics.track_request_latency(
"/v1/vector_stores/{vector_store_id}/search"
):
try:
feature_view = vs_registry.resolve(vector_store_id)
assert_permissions(
resource=feature_view,
actions=[AuthzedAction.READ_ONLINE],
)
result = await store.openai_search(
vector_store_id=feature_view.name,
query=request.query,
vs_id=vector_store_id,
max_num_results=request.max_num_results or 10,
filters=request.filters,
ranking_options=(
request.ranking_options.model_dump()
if request.ranking_options
else None
),
rewrite_query=request.rewrite_query,
features_to_retrieve=(
request.metadata.features_to_retrieve
if request.metadata
else None
),
)
except FeatureViewNotFoundException:
return JSONResponse(
status_code=404,
content={
"error": {
"message": f"No vector store found with id '{vector_store_id}'",
"type": "not_found_error",
}
},
)
except ValueError as e:
return JSONResponse(
status_code=422,
content={
"error": {
"message": str(e),
"type": "invalid_request_error",
}
},
)
return JSONResponse(content=result)
@app.post("/push", dependencies=[Depends(inject_user_details)])
async def push(request: PushFeaturesRequest) -> Response:
with feast_metrics.track_request_latency("/push"):
df = pd.DataFrame(request.df)
actions = []
if request.to == "offline":
to = PushMode.OFFLINE
actions = [AuthzedAction.WRITE_OFFLINE]
elif request.to == "online":
to = PushMode.ONLINE
actions = [AuthzedAction.WRITE_ONLINE]
elif request.to == "online_and_offline":
to = PushMode.ONLINE_AND_OFFLINE
actions = WRITE
else:
raise ValueError(
f"{request.to} is not a supported push format. Please specify one of these ['online', 'offline', 'online_and_offline']."
)
from feast.data_source import PushSource
all_fvs = store.list_feature_views(
allow_cache=request.allow_registry_cache
) + store.list_stream_feature_views(
allow_cache=request.allow_registry_cache
)
fvs_with_push_sources = {
fv
for fv in all_fvs
if (
fv.stream_source is not None
and isinstance(fv.stream_source, PushSource)
and fv.stream_source.name == request.push_source_name
)
}
for feature_view in fvs_with_push_sources:
assert_permissions(resource=feature_view, actions=actions)
async def _push_with_to(push_to: PushMode) -> None:
"""
Helper for performing a single push operation.
NOTE:
- Feast providers **do not currently support async offline writes**.
- Therefore:
* ONLINE and ONLINE_AND_OFFLINE → may be async, depending on provider.async_supported.online.write
* OFFLINE → always synchronous, but executed via run_in_threadpool when called from HTTP handlers.
- The OfflineWriteBatcher handles offline writes directly in its own background thread, but the offline store writes are currently synchronous only.
"""
push_source_name = request.push_source_name
allow_registry_cache = request.allow_registry_cache
transform_on_write = request.transform_on_write
# Async currently only applies to online store writes (ONLINE / ONLINE_AND_OFFLINE paths) as theres no async for offline store
if push_to in (PushMode.ONLINE, PushMode.ONLINE_AND_OFFLINE) and (
store._get_provider().async_supported.online.write
):
await store.push_async(
push_source_name=push_source_name,
df=df,
allow_registry_cache=allow_registry_cache,
to=push_to,
transform_on_write=transform_on_write,
)
else:
await run_in_threadpool(
lambda: store.push(
push_source_name=push_source_name,
df=df,
allow_registry_cache=allow_registry_cache,
to=push_to,
transform_on_write=transform_on_write,
)
)
needs_online = to in (PushMode.ONLINE, PushMode.ONLINE_AND_OFFLINE)
needs_offline = to in (PushMode.OFFLINE, PushMode.ONLINE_AND_OFFLINE)
status_code = status.HTTP_200_OK
if offline_batcher is None or not needs_offline:
await _push_with_to(to)
else:
if needs_online:
await _push_with_to(PushMode.ONLINE)
offline_batcher.enqueue(
push_source_name=request.push_source_name,
df=df,
allow_registry_cache=request.allow_registry_cache,
transform_on_write=request.transform_on_write,
)
status_code = status.HTTP_202_ACCEPTED
feast_metrics.track_push(request.push_source_name, request.to)
return Response(status_code=status_code)
async def _get_feast_object(
feature_view_name: str, allow_registry_cache: bool
) -> FeastObject:
return await run_in_threadpool(
get_feature_view_from_feature_store,
store,
feature_view_name,
allow_registry_cache,
)
@app.post("/write-to-online-store", dependencies=[Depends(inject_user_details)])
async def write_to_online_store(request: WriteToFeatureStoreRequest) -> None:
df = pd.DataFrame(request.df)
feature_view_name = request.feature_view_name
allow_registry_cache = request.allow_registry_cache
resource = await _get_feast_object(feature_view_name, allow_registry_cache)
assert_permissions(resource=resource, actions=[AuthzedAction.WRITE_ONLINE])
await run_in_threadpool(
store.write_to_online_store,
feature_view_name=feature_view_name,
df=df,
allow_registry_cache=allow_registry_cache,
transform_on_write=request.transform_on_write,
)
@app.get("/health")
async def health():
try:
store.registry.list_projects(allow_cache=True)
return Response(status_code=status.HTTP_200_OK)
except Exception:
return Response(status_code=status.HTTP_503_SERVICE_UNAVAILABLE)
@app.post("/chat")
async def chat(request: ChatRequest):
# Process the chat request
# For now, just return dummy text
return {"response": "This is a dummy response from the Feast feature server."}
@app.get("/chat")
async def chat_ui():
# Serve the chat UI
static_dir_ref = importlib_resources.files(__spec__.parent) / "static/chat" # type: ignore[name-defined, arg-type]
with importlib_resources.as_file(static_dir_ref) as static_dir:
with open(os.path.join(static_dir, "index.html")) as f:
content = f.read()
return Response(content=content, media_type="text/html")
@app.post("/materialize", dependencies=[Depends(inject_user_details)])
async def materialize(
request: MaterializeRequest,
async_mode: bool = Query(False, alias="async"),
force: bool = Query(False),
):
with feast_metrics.track_request_latency("/materialize"):
fv_names = _authorize_materialize_views(
store, request.feature_views, version=request.version
)
start_date, end_date = _parse_materialize_timestamps(request)
if async_mode:
if force:
_reset_stuck_materializing_to_generated(store, fv_names)
else:
conflict = _check_already_materializing(store, fv_names)
if conflict:
return conflict
# Reserve MATERIALIZING before 202 so concurrent requests hit 409.
# store.materialize() treats already-MATERIALIZING as a no-op.
_update_fv_state(store, fv_names, FeatureViewState.MATERIALIZING)
def _run_materialize():
try:
store.materialize(
start_date,
end_date,
fv_names,
disable_event_timestamp=request.disable_event_timestamp,
full_feature_names=request.full_feature_names,
version=request.version,
)
except Exception as e:
logger.error(
f"Async materialization failed for {fv_names}: {e}",
exc_info=True,
)
_reset_stuck_materializing_to_generated(store, fv_names)
loop = asyncio.get_running_loop()
loop.run_in_executor(materialize_executor, _run_materialize)