-
Notifications
You must be signed in to change notification settings - Fork 1.3k
Expand file tree
/
Copy pathmapper.py
More file actions
480 lines (417 loc) · 15.3 KB
/
mapper.py
File metadata and controls
480 lines (417 loc) · 15.3 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
"""
dbt to Feast type and object mapper.
This module provides functionality to map dbt model metadata to Feast objects
including DataSource, Entity, and FeatureView.
"""
from datetime import timedelta
from typing import Any, Dict, List, Optional, Union
from feast.dbt.parser import DbtModel
from feast.entity import Entity
from feast.feature_view import FeatureView
from feast.field import Field
from feast.types import (
Array,
Bool,
Bytes,
FeastType,
Float32,
Float64,
Int32,
Int64,
String,
UnixTimestamp,
)
from feast.value_type import ValueType
# Mapping from FeastType to ValueType for entity value inference
FEAST_TYPE_TO_VALUE_TYPE: Dict[FeastType, ValueType] = {
String: ValueType.STRING,
Int32: ValueType.INT32,
Int64: ValueType.INT64,
Float32: ValueType.FLOAT,
Float64: ValueType.DOUBLE,
Bool: ValueType.BOOL,
Bytes: ValueType.BYTES,
UnixTimestamp: ValueType.UNIX_TIMESTAMP,
}
def feast_type_to_value_type(feast_type: FeastType) -> ValueType:
"""Convert a FeastType to its corresponding ValueType for entities."""
return FEAST_TYPE_TO_VALUE_TYPE.get(feast_type, ValueType.STRING)
# Comprehensive mapping from dbt/warehouse types to Feast types
# Covers BigQuery, Snowflake, Redshift, PostgreSQL, and common SQL types
DBT_TO_FEAST_TYPE_MAP: Dict[str, FeastType] = {
# String types
"STRING": String,
"TEXT": String,
"VARCHAR": String,
"CHAR": String,
"CHARACTER": String,
"NVARCHAR": String,
"NCHAR": String,
"CHARACTER VARYING": String,
# Integer types
"INT": Int64,
"INT32": Int32,
"INT64": Int64,
"INTEGER": Int64,
"BIGINT": Int64,
"SMALLINT": Int32,
"TINYINT": Int32,
"BYTEINT": Int32,
"NUMBER": Int64, # Snowflake - default to Int64, precision handling below
"NUMERIC": Int64,
"DECIMAL": Int64,
# Float types
"FLOAT": Float32,
"FLOAT32": Float32,
"FLOAT64": Float64,
"DOUBLE": Float64,
"DOUBLE PRECISION": Float64,
"REAL": Float32,
# Boolean types
"BOOL": Bool,
"BOOLEAN": Bool,
# Timestamp types
"TIMESTAMP": UnixTimestamp,
"TIMESTAMP_NTZ": UnixTimestamp,
"TIMESTAMP_LTZ": UnixTimestamp,
"TIMESTAMP_TZ": UnixTimestamp,
"DATETIME": UnixTimestamp,
"DATE": UnixTimestamp,
"TIME": UnixTimestamp,
# Binary types
"BYTES": Bytes,
"BINARY": Bytes,
"VARBINARY": Bytes,
"BLOB": Bytes,
}
def map_dbt_type_to_feast_type(dbt_type: str) -> FeastType:
"""
Map a dbt data type to a Feast type.
Handles various database type formats including:
- Simple types: STRING, INT64, FLOAT
- Parameterized types: VARCHAR(255), NUMBER(10,2), DECIMAL(18,0)
- Array types: ARRAY<STRING>, ARRAY<INT64>
Args:
dbt_type: The dbt/database data type string
Returns:
The corresponding Feast type
"""
if not dbt_type:
return String
# Normalize the type string
normalized = dbt_type.upper().strip()
# Handle ARRAY types: ARRAY<element_type>
if normalized.startswith("ARRAY<") and normalized.endswith(">"):
element_type_str = normalized[6:-1].strip()
element_type = map_dbt_type_to_feast_type(element_type_str)
# Array only supports primitive types
valid_array_types = {
String,
Int32,
Int64,
Float32,
Float64,
Bool,
Bytes,
UnixTimestamp,
}
if element_type in valid_array_types:
return Array(element_type)
return Array(String) # Fallback for complex nested types
# Handle parameterized types: VARCHAR(255), NUMBER(10,2), etc.
# Extract base type by removing parentheses and parameters
base_type = normalized.split("(")[0].strip()
# Handle Snowflake NUMBER with precision
if base_type == "NUMBER" and "(" in normalized:
try:
# Parse precision and scale: NUMBER(precision, scale)
params = normalized.split("(")[1].rstrip(")").split(",")
precision = int(params[0].strip())
scale = int(params[1].strip()) if len(params) > 1 else 0
if scale > 0:
# Has decimal places, use Float64
return Float64
elif precision <= 9:
return Int32
elif precision <= 18:
return Int64
else:
# Precision > 18, may exceed Int64 range
return Float64
except (ValueError, IndexError):
return Int64
# Look up in mapping table
if base_type in DBT_TO_FEAST_TYPE_MAP:
return DBT_TO_FEAST_TYPE_MAP[base_type]
# Default to String for unknown types
return String
class DbtToFeastMapper:
"""
Maps dbt models to Feast objects.
Supports creating DataSource, Entity, and FeatureView objects from
dbt model metadata.
Example::
mapper = DbtToFeastMapper(data_source_type="bigquery")
data_source = mapper.create_data_source(model)
feature_view = mapper.create_feature_view(
model, data_source, entity_column="driver_id"
)
Args:
data_source_type: Type of data source ('bigquery', 'snowflake', 'file')
timestamp_field: Default timestamp field name
ttl_days: Default TTL in days for feature views
"""
def __init__(
self,
data_source_type: str = "bigquery",
timestamp_field: str = "event_timestamp",
ttl_days: int = 1,
):
self.data_source_type = data_source_type.lower()
self.timestamp_field = timestamp_field
self.ttl_days = ttl_days
def _infer_entity_value_type(self, model: DbtModel, entity_col: str) -> ValueType:
"""Infer entity ValueType from dbt model column type."""
for column in model.columns:
if column.name == entity_col:
feast_type = map_dbt_type_to_feast_type(column.data_type)
return feast_type_to_value_type(feast_type)
return ValueType.UNKNOWN
def create_data_source(
self,
model: DbtModel,
timestamp_field: Optional[str] = None,
created_timestamp_column: Optional[str] = None,
) -> Any:
"""
Create a Feast DataSource from a dbt model.
Args:
model: The DbtModel to create a DataSource from
timestamp_field: Override the default timestamp field
created_timestamp_column: Column for created timestamp (dedup)
Returns:
A Feast DataSource (BigQuerySource, SnowflakeSource, or FileSource)
Raises:
ValueError: If data_source_type is not supported
"""
ts_field = timestamp_field or self.timestamp_field
# Build tags from dbt metadata
tags = {"dbt.model": model.name}
for tag in model.tags:
tags[f"dbt.tag.{tag}"] = "true"
if self.data_source_type == "bigquery":
from feast.infra.offline_stores.bigquery_source import BigQuerySource
return BigQuerySource(
name=f"{model.name}_source",
table=model.full_table_name,
timestamp_field=ts_field,
created_timestamp_column=created_timestamp_column or "",
description=model.description,
tags=tags,
)
elif self.data_source_type == "snowflake":
from feast.infra.offline_stores.snowflake_source import SnowflakeSource
return SnowflakeSource(
name=f"{model.name}_source",
database=model.database,
schema=model.schema,
table=model.alias,
timestamp_field=ts_field,
created_timestamp_column=created_timestamp_column or "",
description=model.description,
tags=tags,
)
elif self.data_source_type == "file":
from feast.infra.offline_stores.file_source import FileSource
# For file sources, use the model name as a placeholder path
return FileSource(
name=f"{model.name}_source",
path=f"/data/{model.name}.parquet",
timestamp_field=ts_field,
created_timestamp_column=created_timestamp_column or "",
description=model.description,
tags=tags,
)
else:
raise ValueError(
f"Unsupported data_source_type: {self.data_source_type}. "
f"Supported types: bigquery, snowflake, file"
)
def create_entity(
self,
name: str,
join_keys: Optional[List[str]] = None,
description: str = "",
tags: Optional[Dict[str, str]] = None,
value_type: ValueType = ValueType.STRING,
) -> Entity:
"""
Create a Feast Entity.
Args:
name: Entity name
join_keys: List of join key column names (defaults to [name])
description: Entity description
tags: Optional tags
value_type: Value type for the entity (default: STRING)
Returns:
A Feast Entity
"""
return Entity(
name=name,
join_keys=join_keys or [name],
value_type=value_type,
description=description,
tags=tags or {},
)
def create_feature_view(
self,
model: DbtModel,
source: Any,
entity_columns: Union[str, List[str]],
entities: Optional[Union[Entity, List[Entity]]] = None,
timestamp_field: Optional[str] = None,
ttl_days: Optional[int] = None,
exclude_columns: Optional[List[str]] = None,
online: bool = True,
) -> FeatureView:
"""
Create a Feast FeatureView from a dbt model.
Args:
model: The DbtModel to create a FeatureView from
source: The DataSource for this FeatureView
entity_columns: Entity column name(s) - single string or list of strings
entities: Optional pre-created Entity or list of Entities
timestamp_field: Override the default timestamp field
ttl_days: Override the default TTL in days
exclude_columns: Additional columns to exclude from features
online: Whether to enable online serving
Returns:
A Feast FeatureView
"""
# Normalize to lists
entity_cols: List[str] = (
[entity_columns]
if isinstance(entity_columns, str)
else list(entity_columns)
)
entity_objs: List[Entity] = []
if entities is not None:
entity_objs = [entities] if isinstance(entities, Entity) else list(entities)
# Validate
if not entity_cols:
raise ValueError("At least one entity column must be specified")
if entity_objs and len(entity_cols) != len(entity_objs):
raise ValueError(
f"Number of entity_columns ({len(entity_cols)}) must match "
f"number of entities ({len(entity_objs)})"
)
ts_field = timestamp_field or self.timestamp_field
ttl = timedelta(days=ttl_days if ttl_days is not None else self.ttl_days)
# Columns to exclude from schema (timestamp + any explicitly excluded)
# Note: entity columns should NOT be excluded - FeatureView.__init__
# expects entity columns to be in the schema and will extract them
excluded = {ts_field}
if exclude_columns:
excluded.update(exclude_columns)
# Create schema from model columns (includes entity columns)
schema: List[Field] = []
for column in model.columns:
if column.name not in excluded:
feast_type = map_dbt_type_to_feast_type(column.data_type)
schema.append(
Field(
name=column.name,
dtype=feast_type,
description=column.description,
)
)
# Create entities if not provided
if not entity_objs:
entity_objs = []
for entity_col in entity_cols:
# Infer entity value type from model column
entity_value_type = self._infer_entity_value_type(model, entity_col)
ent = self.create_entity(
name=entity_col,
description=f"Entity for {model.name}",
value_type=entity_value_type,
)
entity_objs.append(ent)
# Build tags from dbt metadata
tags = {
"dbt.model": model.name,
"dbt.unique_id": model.unique_id,
}
for tag in model.tags:
tags[f"dbt.tag.{tag}"] = "true"
return FeatureView(
name=model.name,
source=source,
schema=schema,
entities=entity_objs,
ttl=ttl,
online=online,
description=model.description,
tags=tags,
)
def create_all_from_model(
self,
model: DbtModel,
entity_columns: Union[str, List[str]],
timestamp_field: Optional[str] = None,
ttl_days: Optional[int] = None,
exclude_columns: Optional[List[str]] = None,
online: bool = True,
) -> Dict[str, Union[List[Entity], Any, FeatureView]]:
"""
Create all Feast objects (DataSource, Entity, FeatureView) from a dbt model.
This is a convenience method that creates all necessary Feast objects
in one call.
Args:
model: The DbtModel to create objects from
entity_columns: Entity column name(s) - single string or list of strings
timestamp_field: Override the default timestamp field
ttl_days: Override the default TTL in days
exclude_columns: Additional columns to exclude from features
online: Whether to enable online serving
Returns:
Dict with keys 'entities', 'data_source', 'feature_view'
"""
# Normalize to list
entity_cols: List[str] = (
[entity_columns]
if isinstance(entity_columns, str)
else list(entity_columns)
)
# Create entities (plural)
entities_list = []
for entity_col in entity_cols:
entity_value_type = self._infer_entity_value_type(model, entity_col)
entity = self.create_entity(
name=entity_col,
description=f"Entity for {model.name}",
tags={"dbt.model": model.name},
value_type=entity_value_type,
)
entities_list.append(entity)
# Create data source
data_source = self.create_data_source(
model=model,
timestamp_field=timestamp_field,
)
# Create feature view
feature_view = self.create_feature_view(
model=model,
source=data_source,
entity_columns=entity_cols,
entities=entities_list,
timestamp_field=timestamp_field,
ttl_days=ttl_days,
exclude_columns=exclude_columns,
online=online,
)
return {
"entities": entities_list,
"data_source": data_source,
"feature_view": feature_view,
}