/
snowflake.py
759 lines (632 loc) · 26.4 KB
/
snowflake.py
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
import contextlib
import os
import uuid
from datetime import datetime
from pathlib import Path
from typing import (
Any,
Callable,
ContextManager,
Dict,
Iterator,
List,
Optional,
Tuple,
Union,
cast,
)
import numpy as np
import pandas as pd
import pyarrow
import pyarrow as pa
from pydantic import Field
from pydantic.typing import Literal
from pytz import utc
from feast import OnDemandFeatureView
from feast.data_source import DataSource
from feast.errors import InvalidEntityType
from feast.feature_logging import LoggingConfig, LoggingSource
from feast.feature_view import DUMMY_ENTITY_ID, DUMMY_ENTITY_VAL, FeatureView
from feast.infra.offline_stores import offline_utils
from feast.infra.offline_stores.offline_store import (
OfflineStore,
RetrievalJob,
RetrievalMetadata,
)
from feast.infra.offline_stores.snowflake_source import (
SavedDatasetSnowflakeStorage,
SnowflakeLoggingDestination,
SnowflakeSource,
)
from feast.infra.utils.snowflake_utils import (
execute_snowflake_statement,
get_snowflake_conn,
write_pandas,
write_parquet,
)
from feast.registry import BaseRegistry
from feast.repo_config import FeastConfigBaseModel, RepoConfig
from feast.saved_dataset import SavedDatasetStorage
from feast.usage import log_exceptions_and_usage
try:
from snowflake.connector import SnowflakeConnection
except ImportError as e:
from feast.errors import FeastExtrasDependencyImportError
raise FeastExtrasDependencyImportError("snowflake", str(e))
class SnowflakeOfflineStoreConfig(FeastConfigBaseModel):
"""Offline store config for Snowflake"""
type: Literal["snowflake.offline"] = "snowflake.offline"
""" Offline store type selector"""
config_path: Optional[str] = (
Path(os.environ["HOME"]) / ".snowsql/config"
).__str__()
""" Snowflake config path -- absolute path required (Cant use ~)"""
account: Optional[str] = None
""" Snowflake deployment identifier -- drop .snowflakecomputing.com"""
user: Optional[str] = None
""" Snowflake user name """
password: Optional[str] = None
""" Snowflake password """
role: Optional[str] = None
""" Snowflake role name"""
warehouse: Optional[str] = None
""" Snowflake warehouse name """
database: Optional[str] = None
""" Snowflake database name """
schema_: Optional[str] = Field(None, alias="schema")
""" Snowflake schema name """
storage_integration_name: Optional[str] = None
""" Storage integration name in snowflake """
blob_export_location: Optional[str] = None
""" Location (in S3, Google storage or Azure storage) where data is offloaded """
class Config:
allow_population_by_field_name = True
class SnowflakeOfflineStore(OfflineStore):
@staticmethod
@log_exceptions_and_usage(offline_store="snowflake")
def pull_latest_from_table_or_query(
config: RepoConfig,
data_source: DataSource,
join_key_columns: List[str],
feature_name_columns: List[str],
timestamp_field: str,
created_timestamp_column: Optional[str],
start_date: datetime,
end_date: datetime,
) -> RetrievalJob:
assert isinstance(data_source, SnowflakeSource)
assert isinstance(config.offline_store, SnowflakeOfflineStoreConfig)
from_expression = (
data_source.get_table_query_string()
) # returns schema.table as a string
if join_key_columns:
partition_by_join_key_string = '"' + '", "'.join(join_key_columns) + '"'
partition_by_join_key_string = (
"PARTITION BY " + partition_by_join_key_string
)
else:
partition_by_join_key_string = ""
timestamp_columns = [timestamp_field]
if created_timestamp_column:
timestamp_columns.append(created_timestamp_column)
timestamp_desc_string = '"' + '" DESC, "'.join(timestamp_columns) + '" DESC'
field_string = (
'"'
+ '", "'.join(join_key_columns + feature_name_columns + timestamp_columns)
+ '"'
)
if data_source.snowflake_options.warehouse:
config.offline_store.warehouse = data_source.snowflake_options.warehouse
snowflake_conn = get_snowflake_conn(config.offline_store)
query = f"""
SELECT
{field_string}
{f''', TRIM({repr(DUMMY_ENTITY_VAL)}::VARIANT,'"') AS "{DUMMY_ENTITY_ID}"''' if not join_key_columns else ""}
FROM (
SELECT {field_string},
ROW_NUMBER() OVER({partition_by_join_key_string} ORDER BY {timestamp_desc_string}) AS "_feast_row"
FROM {from_expression}
WHERE "{timestamp_field}" BETWEEN TO_TIMESTAMP_NTZ({start_date.timestamp()}) AND TO_TIMESTAMP_NTZ({end_date.timestamp()})
)
WHERE "_feast_row" = 1
"""
return SnowflakeRetrievalJob(
query=query,
snowflake_conn=snowflake_conn,
config=config,
full_feature_names=False,
on_demand_feature_views=None,
)
@staticmethod
@log_exceptions_and_usage(offline_store="snowflake")
def pull_all_from_table_or_query(
config: RepoConfig,
data_source: DataSource,
join_key_columns: List[str],
feature_name_columns: List[str],
timestamp_field: str,
start_date: datetime,
end_date: datetime,
) -> RetrievalJob:
assert isinstance(data_source, SnowflakeSource)
from_expression = data_source.get_table_query_string()
field_string = (
'"'
+ '", "'.join(join_key_columns + feature_name_columns + [timestamp_field])
+ '"'
)
if data_source.snowflake_options.warehouse:
config.offline_store.warehouse = data_source.snowflake_options.warehouse
snowflake_conn = get_snowflake_conn(config.offline_store)
start_date = start_date.astimezone(tz=utc)
end_date = end_date.astimezone(tz=utc)
query = f"""
SELECT {field_string}
FROM {from_expression}
WHERE "{timestamp_field}" BETWEEN TIMESTAMP '{start_date}' AND TIMESTAMP '{end_date}'
"""
return SnowflakeRetrievalJob(
query=query,
snowflake_conn=snowflake_conn,
config=config,
full_feature_names=False,
)
@staticmethod
@log_exceptions_and_usage(offline_store="snowflake")
def get_historical_features(
config: RepoConfig,
feature_views: List[FeatureView],
feature_refs: List[str],
entity_df: Union[pd.DataFrame, str],
registry: BaseRegistry,
project: str,
full_feature_names: bool = False,
) -> RetrievalJob:
assert isinstance(config.offline_store, SnowflakeOfflineStoreConfig)
snowflake_conn = get_snowflake_conn(config.offline_store)
entity_schema = _get_entity_schema(entity_df, snowflake_conn, config)
entity_df_event_timestamp_col = (
offline_utils.infer_event_timestamp_from_entity_df(entity_schema)
)
entity_df_event_timestamp_range = _get_entity_df_event_timestamp_range(
entity_df,
entity_df_event_timestamp_col,
snowflake_conn,
)
@contextlib.contextmanager
def query_generator() -> Iterator[str]:
table_name = offline_utils.get_temp_entity_table_name()
_upload_entity_df(entity_df, snowflake_conn, config, table_name)
expected_join_keys = offline_utils.get_expected_join_keys(
project, feature_views, registry
)
offline_utils.assert_expected_columns_in_entity_df(
entity_schema, expected_join_keys, entity_df_event_timestamp_col
)
# Build a query context containing all information required to template the Snowflake SQL query
query_context = offline_utils.get_feature_view_query_context(
feature_refs,
feature_views,
registry,
project,
entity_df_event_timestamp_range,
)
query_context = _fix_entity_selections_identifiers(query_context)
# Generate the Snowflake SQL query from the query context
query = offline_utils.build_point_in_time_query(
query_context,
left_table_query_string=table_name,
entity_df_event_timestamp_col=entity_df_event_timestamp_col,
entity_df_columns=entity_schema.keys(),
query_template=MULTIPLE_FEATURE_VIEW_POINT_IN_TIME_JOIN,
full_feature_names=full_feature_names,
)
yield query
return SnowflakeRetrievalJob(
query=query_generator,
snowflake_conn=snowflake_conn,
config=config,
full_feature_names=full_feature_names,
on_demand_feature_views=OnDemandFeatureView.get_requested_odfvs(
feature_refs, project, registry
),
metadata=RetrievalMetadata(
features=feature_refs,
keys=list(entity_schema.keys() - {entity_df_event_timestamp_col}),
min_event_timestamp=entity_df_event_timestamp_range[0],
max_event_timestamp=entity_df_event_timestamp_range[1],
),
)
@staticmethod
def write_logged_features(
config: RepoConfig,
data: Union[pyarrow.Table, Path],
source: LoggingSource,
logging_config: LoggingConfig,
registry: BaseRegistry,
):
assert isinstance(logging_config.destination, SnowflakeLoggingDestination)
snowflake_conn = get_snowflake_conn(config.offline_store)
if isinstance(data, Path):
write_parquet(
snowflake_conn,
data,
source.get_schema(registry),
table_name=logging_config.destination.table_name,
auto_create_table=True,
)
else:
write_pandas(
snowflake_conn,
data.to_pandas(),
table_name=logging_config.destination.table_name,
auto_create_table=True,
)
@staticmethod
def offline_write_batch(
config: RepoConfig,
feature_view: FeatureView,
table: pyarrow.Table,
progress: Optional[Callable[[int], Any]],
):
if not feature_view.batch_source:
raise ValueError(
"feature view does not have a batch source to persist offline data"
)
if not isinstance(config.offline_store, SnowflakeOfflineStoreConfig):
raise ValueError(
f"offline store config is of type {type(config.offline_store)} when snowflake type required"
)
if not isinstance(feature_view.batch_source, SnowflakeSource):
raise ValueError(
f"feature view batch source is {type(feature_view.batch_source)} not snowflake source"
)
pa_schema, column_names = offline_utils.get_pyarrow_schema_from_batch_source(
config, feature_view.batch_source
)
if column_names != table.column_names:
raise ValueError(
f"The input pyarrow table has schema {table.schema} with the incorrect columns {table.column_names}. "
f"The schema is expected to be {pa_schema} with the columns (in this exact order) to be {column_names}."
)
if table.schema != pa_schema:
table = table.cast(pa_schema)
snowflake_conn = get_snowflake_conn(config.offline_store)
write_pandas(
snowflake_conn,
table.to_pandas(),
table_name=feature_view.batch_source.table,
auto_create_table=True,
)
class SnowflakeRetrievalJob(RetrievalJob):
def __init__(
self,
query: Union[str, Callable[[], ContextManager[str]]],
snowflake_conn: SnowflakeConnection,
config: RepoConfig,
full_feature_names: bool,
on_demand_feature_views: Optional[List[OnDemandFeatureView]] = None,
metadata: Optional[RetrievalMetadata] = None,
):
if not isinstance(query, str):
self._query_generator = query
else:
@contextlib.contextmanager
def query_generator() -> Iterator[str]:
assert isinstance(query, str)
yield query
self._query_generator = query_generator
self.snowflake_conn = snowflake_conn
self.config = config
self._full_feature_names = full_feature_names
self._on_demand_feature_views = (
on_demand_feature_views if on_demand_feature_views else []
)
self._metadata = metadata
self.export_path: Optional[str]
if self.config.offline_store.blob_export_location:
self.export_path = f"{self.config.offline_store.blob_export_location}/{self.config.project}/{uuid.uuid4()}"
else:
self.export_path = None
@property
def full_feature_names(self) -> bool:
return self._full_feature_names
@property
def on_demand_feature_views(self) -> Optional[List[OnDemandFeatureView]]:
return self._on_demand_feature_views
def _to_df_internal(self) -> pd.DataFrame:
with self._query_generator() as query:
df = execute_snowflake_statement(
self.snowflake_conn, query
).fetch_pandas_all()
return df
def _to_arrow_internal(self) -> pa.Table:
with self._query_generator() as query:
pa_table = execute_snowflake_statement(
self.snowflake_conn, query
).fetch_arrow_all()
if pa_table:
return pa_table
else:
empty_result = execute_snowflake_statement(self.snowflake_conn, query)
return pa.Table.from_pandas(
pd.DataFrame(columns=[md.name for md in empty_result.description])
)
def to_snowflake(self, table_name: str, temporary=False) -> None:
"""Save dataset as a new Snowflake table"""
if self.on_demand_feature_views:
transformed_df = self.to_df()
write_pandas(
self.snowflake_conn, transformed_df, table_name, auto_create_table=True
)
return None
with self._query_generator() as query:
query = f'CREATE {"TEMPORARY" if temporary else ""} TABLE IF NOT EXISTS "{table_name}" AS ({query});\n'
execute_snowflake_statement(self.snowflake_conn, query)
def to_sql(self) -> str:
"""
Returns the SQL query that will be executed in Snowflake to build the historical feature table.
"""
with self._query_generator() as query:
return query
def to_arrow_chunks(self, arrow_options: Optional[Dict] = None) -> Optional[List]:
with self._query_generator() as query:
arrow_batches = execute_snowflake_statement(
self.snowflake_conn, query
).get_result_batches()
return arrow_batches
def persist(self, storage: SavedDatasetStorage):
assert isinstance(storage, SavedDatasetSnowflakeStorage)
self.to_snowflake(table_name=storage.snowflake_options.table)
@property
def metadata(self) -> Optional[RetrievalMetadata]:
return self._metadata
def supports_remote_storage_export(self) -> bool:
return (
self.config.offline_store.storage_integration_name
and self.config.offline_store.blob_export_location
)
def to_remote_storage(self) -> List[str]:
if not self.export_path:
raise ValueError(
"to_remote_storage() requires `blob_export_location` to be specified in config"
)
if not self.config.offline_store.storage_integration_name:
raise ValueError(
"to_remote_storage() requires `storage_integration_name` to be specified in config"
)
table = f"temporary_{uuid.uuid4().hex}"
self.to_snowflake(table)
copy_into_query = f"""copy into '{self.config.offline_store.blob_export_location}/{table}' from "{self.config.offline_store.database}"."{self.config.offline_store.schema_}"."{table}"\n
storage_integration = {self.config.offline_store.storage_integration_name}\n
file_format = (TYPE = PARQUET)\n
DETAILED_OUTPUT = TRUE\n
HEADER = TRUE;\n
"""
cursor = execute_snowflake_statement(self.snowflake_conn, copy_into_query)
all_rows = (
cursor.fetchall()
) # This may be need pagination at some point in the future.
file_name_column_index = [
idx for idx, rm in enumerate(cursor.description) if rm.name == "FILE_NAME"
][0]
return [f"{self.export_path}/{row[file_name_column_index]}" for row in all_rows]
def _get_entity_schema(
entity_df: Union[pd.DataFrame, str],
snowflake_conn: SnowflakeConnection,
config: RepoConfig,
) -> Dict[str, np.dtype]:
if isinstance(entity_df, pd.DataFrame):
return dict(zip(entity_df.columns, entity_df.dtypes))
else:
query = f"SELECT * FROM ({entity_df}) LIMIT 1"
limited_entity_df = execute_snowflake_statement(
snowflake_conn, query
).fetch_pandas_all()
return dict(zip(limited_entity_df.columns, limited_entity_df.dtypes))
def _upload_entity_df(
entity_df: Union[pd.DataFrame, str],
snowflake_conn: SnowflakeConnection,
config: RepoConfig,
table_name: str,
) -> None:
if isinstance(entity_df, pd.DataFrame):
# Write the data from the DataFrame to the table
write_pandas(
snowflake_conn,
entity_df,
table_name,
auto_create_table=True,
create_temp_table=True,
)
return None
elif isinstance(entity_df, str):
# If the entity_df is a string (SQL query), create a Snowflake table out of it,
query = f'CREATE TEMPORARY TABLE "{table_name}" AS ({entity_df})'
execute_snowflake_statement(snowflake_conn, query)
return None
else:
raise InvalidEntityType(type(entity_df))
def _fix_entity_selections_identifiers(query_context) -> list:
for i, qc in enumerate(query_context):
for j, es in enumerate(qc.entity_selections):
query_context[i].entity_selections[j] = f'"{es}"'.replace(" AS ", '" AS "')
return query_context
def _get_entity_df_event_timestamp_range(
entity_df: Union[pd.DataFrame, str],
entity_df_event_timestamp_col: str,
snowflake_conn: SnowflakeConnection,
) -> Tuple[datetime, datetime]:
if isinstance(entity_df, pd.DataFrame):
entity_df_event_timestamp = entity_df.loc[
:, entity_df_event_timestamp_col
].infer_objects()
if pd.api.types.is_string_dtype(entity_df_event_timestamp):
entity_df_event_timestamp = pd.to_datetime(
entity_df_event_timestamp, utc=True
)
entity_df_event_timestamp_range = (
entity_df_event_timestamp.min().to_pydatetime(),
entity_df_event_timestamp.max().to_pydatetime(),
)
elif isinstance(entity_df, str):
# If the entity_df is a string (SQL query), determine range
# from table
query = f'SELECT MIN("{entity_df_event_timestamp_col}") AS "min_value", MAX("{entity_df_event_timestamp_col}") AS "max_value" FROM ({entity_df})'
results = execute_snowflake_statement(snowflake_conn, query).fetchall()
entity_df_event_timestamp_range = cast(Tuple[datetime, datetime], results[0])
else:
raise InvalidEntityType(type(entity_df))
return entity_df_event_timestamp_range
MULTIPLE_FEATURE_VIEW_POINT_IN_TIME_JOIN = """
/*
Compute a deterministic hash for the `left_table_query_string` that will be used throughout
all the logic as the field to GROUP BY the data
*/
WITH "entity_dataframe" AS (
SELECT *,
"{{entity_df_event_timestamp_col}}" AS "entity_timestamp"
{% for featureview in featureviews %}
{% if featureview.entities %}
,(
{% for entity in featureview.entities %}
CAST("{{entity}}" AS VARCHAR) ||
{% endfor %}
CAST("{{entity_df_event_timestamp_col}}" AS VARCHAR)
) AS "{{featureview.name}}__entity_row_unique_id"
{% else %}
,CAST("{{entity_df_event_timestamp_col}}" AS VARCHAR) AS "{{featureview.name}}__entity_row_unique_id"
{% endif %}
{% endfor %}
FROM "{{ left_table_query_string }}"
),
{% for featureview in featureviews %}
"{{ featureview.name }}__entity_dataframe" AS (
SELECT
{{ featureview.entities | map('tojson') | join(', ')}}{% if featureview.entities %},{% else %}{% endif %}
"entity_timestamp",
"{{featureview.name}}__entity_row_unique_id"
FROM "entity_dataframe"
GROUP BY
{{ featureview.entities | map('tojson') | join(', ')}}{% if featureview.entities %},{% else %}{% endif %}
"entity_timestamp",
"{{featureview.name}}__entity_row_unique_id"
),
/*
This query template performs the point-in-time correctness join for a single feature set table
to the provided entity table.
1. We first join the current feature_view to the entity dataframe that has been passed.
This JOIN has the following logic:
- For each row of the entity dataframe, only keep the rows where the `timestamp_field`
is less than the one provided in the entity dataframe
- If there a TTL for the current feature_view, also keep the rows where the `timestamp_field`
is higher the the one provided minus the TTL
- For each row, Join on the entity key and retrieve the `entity_row_unique_id` that has been
computed previously
The output of this CTE will contain all the necessary information and already filtered out most
of the data that is not relevant.
*/
"{{ featureview.name }}__subquery" AS (
SELECT
"{{ featureview.timestamp_field }}" as "event_timestamp",
{{'"' ~ featureview.created_timestamp_column ~ '" as "created_timestamp",' if featureview.created_timestamp_column else '' }}
{{featureview.entity_selections | join(', ')}}{% if featureview.entity_selections %},{% else %}{% endif %}
{% for feature in featureview.features %}
"{{ feature }}" as {% if full_feature_names %}"{{ featureview.name }}__{{featureview.field_mapping.get(feature, feature)}}"{% else %}"{{ featureview.field_mapping.get(feature, feature) }}"{% endif %}{% if loop.last %}{% else %}, {% endif %}
{% endfor %}
FROM {{ featureview.table_subquery }}
WHERE "{{ featureview.timestamp_field }}" <= '{{ featureview.max_event_timestamp }}'
{% if featureview.ttl == 0 %}{% else %}
AND "{{ featureview.timestamp_field }}" >= '{{ featureview.min_event_timestamp }}'
{% endif %}
),
"{{ featureview.name }}__base" AS (
SELECT
"subquery".*,
"entity_dataframe"."entity_timestamp",
"entity_dataframe"."{{featureview.name}}__entity_row_unique_id"
FROM "{{ featureview.name }}__subquery" AS "subquery"
INNER JOIN "{{ featureview.name }}__entity_dataframe" AS "entity_dataframe"
ON TRUE
AND "subquery"."event_timestamp" <= "entity_dataframe"."entity_timestamp"
{% if featureview.ttl == 0 %}{% else %}
AND "subquery"."event_timestamp" >= TIMESTAMPADD(second,-{{ featureview.ttl }},"entity_dataframe"."entity_timestamp")
{% endif %}
{% for entity in featureview.entities %}
AND "subquery"."{{ entity }}" = "entity_dataframe"."{{ entity }}"
{% endfor %}
),
/*
2. If the `created_timestamp_column` has been set, we need to
deduplicate the data first. This is done by calculating the
`MAX(created_at_timestamp)` for each event_timestamp.
We then join the data on the next CTE
*/
{% if featureview.created_timestamp_column %}
"{{ featureview.name }}__dedup" AS (
SELECT
"{{featureview.name}}__entity_row_unique_id",
"event_timestamp",
MAX("created_timestamp") AS "created_timestamp"
FROM "{{ featureview.name }}__base"
GROUP BY "{{featureview.name}}__entity_row_unique_id", "event_timestamp"
),
{% endif %}
/*
3. The data has been filtered during the first CTE "*__base"
Thus we only need to compute the latest timestamp of each feature.
*/
"{{ featureview.name }}__latest" AS (
SELECT
"event_timestamp",
{% if featureview.created_timestamp_column %}"created_timestamp",{% endif %}
"{{featureview.name}}__entity_row_unique_id"
FROM
(
SELECT *,
ROW_NUMBER() OVER(
PARTITION BY "{{featureview.name}}__entity_row_unique_id"
ORDER BY "event_timestamp" DESC{% if featureview.created_timestamp_column %},"created_timestamp" DESC{% endif %}
) AS "row_number"
FROM "{{ featureview.name }}__base"
{% if featureview.created_timestamp_column %}
INNER JOIN "{{ featureview.name }}__dedup"
USING ("{{featureview.name}}__entity_row_unique_id", "event_timestamp", "created_timestamp")
{% endif %}
)
WHERE "row_number" = 1
),
/*
4. Once we know the latest value of each feature for a given timestamp,
we can join again the data back to the original "base" dataset
*/
"{{ featureview.name }}__cleaned" AS (
SELECT "base".*
FROM "{{ featureview.name }}__base" AS "base"
INNER JOIN "{{ featureview.name }}__latest"
USING(
"{{featureview.name}}__entity_row_unique_id",
"event_timestamp"
{% if featureview.created_timestamp_column %}
,"created_timestamp"
{% endif %}
)
){% if loop.last %}{% else %}, {% endif %}
{% endfor %}
/*
Joins the outputs of multiple time travel joins to a single table.
The entity_dataframe dataset being our source of truth here.
*/
SELECT "{{ final_output_feature_names | join('", "')}}"
FROM "entity_dataframe"
{% for featureview in featureviews %}
LEFT JOIN (
SELECT
"{{featureview.name}}__entity_row_unique_id"
{% for feature in featureview.features %}
,{% if full_feature_names %}"{{ featureview.name }}__{{featureview.field_mapping.get(feature, feature)}}"{% else %}"{{ featureview.field_mapping.get(feature, feature) }}"{% endif %}
{% endfor %}
FROM "{{ featureview.name }}__cleaned"
) "{{ featureview.name }}__cleaned" USING ("{{featureview.name}}__entity_row_unique_id")
{% endfor %}
"""