feast/sdk/python/feast/dbt/codegen.py at v0.60.0 · feast-dev/feast · GitHub

History

423 lines (359 loc) · 13.6 KB

Raw

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

"""

Code generator for dbt to Feast imports.

This module generates Python code files containing Feast object definitions

(Entity, DataSource, FeatureView) from dbt model metadata.

"""

import logging

from typing import Any, List, Optional, Set, Union

from jinja2 import BaseLoader, Environment

from feast.dbt.mapper import map_dbt_type_to_feast_type

from feast.dbt.parser import DbtModel

from feast.types import (

Array,

Bool,

Bytes,

Float32,

Float64,

Int32,

Int64,

String,

UnixTimestamp,

)

logger = logging.getLogger(__name__)

# Template for generating a complete Feast definitions file

FEAST_FILE_TEMPLATE = '''"""

Feast feature definitions generated from dbt models.

Source: {{ manifest_path }}

Project: {{ project_name }}

Generated by: feast dbt import

"""

from datetime import timedelta

from feast import Entity, FeatureView, Field

{% if type_imports %}

from feast.types import {{ type_imports | join(', ') }}

{% endif %}

{% if data_source_type == 'bigquery' %}

from feast.infra.offline_stores.bigquery_source import BigQuerySource

{% elif data_source_type == 'snowflake' %}

from feast.infra.offline_stores.snowflake_source import SnowflakeSource

{% elif data_source_type == 'file' %}

from feast.infra.offline_stores.file_source import FileSource

{% endif %}

# =============================================================================

# Entities

# =============================================================================

{% for entity in entities %}

{{ entity.var_name }} = Entity(

name="{{ entity.name }}",

join_keys=["{{ entity.join_key }}"],

description="{{ entity.description }}",

tags={{ entity.tags }},

)

{% endfor %}

# =============================================================================

# Data Sources

# =============================================================================

{% for source in data_sources %}

{% if data_source_type == 'bigquery' %}

{{ source.var_name }} = BigQuerySource(

name="{{ source.name }}",

table="{{ source.table }}",

timestamp_field="{{ source.timestamp_field }}",

description="{{ source.description }}",

tags={{ source.tags }},

)

{% elif data_source_type == 'snowflake' %}

{{ source.var_name }} = SnowflakeSource(

name="{{ source.name }}",

database="{{ source.database }}",

schema="{{ source.schema }}",

table="{{ source.table }}",

timestamp_field="{{ source.timestamp_field }}",

description="{{ source.description }}",

tags={{ source.tags }},

)

{% elif data_source_type == 'file' %}

{{ source.var_name }} = FileSource(

name="{{ source.name }}",

path="{{ source.path }}",

timestamp_field="{{ source.timestamp_field }}",

description="{{ source.description }}",

tags={{ source.tags }},

)

{% endif %}

{% endfor %}

# =============================================================================

# Feature Views

# =============================================================================

{% for fv in feature_views %}

{{ fv.var_name }} = FeatureView(

name="{{ fv.name }}",

entities=[{{ fv.entity_vars | join(', ') }}],

ttl=timedelta(days={{ fv.ttl_days }}),

schema=[

{% for field in fv.fields %}

Field(name="{{ field.name }}", dtype={{ field.dtype }}{% if field.description %}, description="{{ field.description }}"{% endif %}),

{% endfor %}

online={{ fv.online }},

source={{ fv.source_var }},

description="{{ fv.description }}",

tags={{ fv.tags }},

)

{% endfor %}

'''

def _get_feast_type_name(feast_type: Any) -> str:

"""Get the string name of a Feast type for code generation."""

if isinstance(feast_type, Array):

# Safely get base_type. Should always exist since Array.__init__ sets it.

# Example: Array(String) -> base_type = String

base_type = getattr(feast_type, "base_type", None)

if base_type is None:

logger.warning(

"Array type missing 'base_type' attribute. "

"This indicates a bug in Array initialization. Falling back to String."

)

base_type = String

base_type_name = _get_feast_type_name(base_type)

return f"Array({base_type_name})"

# Map type objects to their names.

# Note: ImageBytes and PdfBytes are excluded since dbt manifests only expose

# generic BYTES type without semantic information about binary content.

type_map = {

String: "String",

Int32: "Int32",

Int64: "Int64",

Float32: "Float32",

Float64: "Float64",

Bool: "Bool",

UnixTimestamp: "UnixTimestamp",

Bytes: "Bytes",

}

return type_map.get(feast_type, "String")

def _make_var_name(name: str) -> str:

"""Convert a name to a valid Python variable name."""

# Replace hyphens and spaces with underscores

var_name = name.replace("-", "_").replace(" ", "_")

# Ensure it starts with a letter or underscore

if var_name and var_name[0].isdigit():

var_name = f"_{var_name}"

return var_name

def _escape_description(desc: Optional[str]) -> str:

"""Escape a description string for use in Python code."""

if not desc:

return ""

# Escape quotes and newlines

return desc.replace("\\", "\\\\").replace('"', '\\"').replace("\n", " ")

class DbtCodeGenerator:

"""

Generates Python code for Feast objects from dbt models.

This class creates complete, importable Python files containing

Entity, DataSource, and FeatureView definitions.

Example::

generator = DbtCodeGenerator(

data_source_type="bigquery",

timestamp_field="event_timestamp",

ttl_days=7

)

code = generator.generate(

models=models,

entity_column="user_id",

manifest_path="target/manifest.json",

project_name="my_project"

)

with open("features.py", "w") as f:

f.write(code)

"""

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

# Set up Jinja2 environment

self.env = Environment(

loader=BaseLoader(),

trim_blocks=True,

lstrip_blocks=True,

)

self.template = self.env.from_string(FEAST_FILE_TEMPLATE)

def generate(

self,

models: List[DbtModel],

entity_columns: Union[str, List[str]],

manifest_path: str = "",

project_name: str = "",

exclude_columns: Optional[List[str]] = None,

online: bool = True,

) -> str:

"""

Generate Python code for Feast objects from dbt models.

Args:

models: List of DbtModel objects to generate code for

entity_columns: Entity column name(s) - single string or list of strings

manifest_path: Path to the dbt manifest (for documentation)

project_name: dbt project name (for documentation)

exclude_columns: Columns to exclude from features

online: Whether to enable online serving

Returns:

Generated Python code as a string

"""

# Normalize entity_columns to list

entity_cols: List[str] = (

[entity_columns] if isinstance(entity_columns, str) else entity_columns

)

if not entity_cols:

raise ValueError("At least one entity column must be specified")

# Note: entity columns should NOT be excluded - FeatureView.__init__

# expects entity columns to be in the schema and will extract them

excluded = {self.timestamp_field}

if exclude_columns:

excluded.update(exclude_columns)

# Collect all Feast types used for imports

type_imports: Set[str] = set()

# Prepare entity data - create one entity per entity column

entities = []

entity_vars = [] # Track variable names for feature views

for entity_col in entity_cols:

entity_var = _make_var_name(entity_col)

entity_vars.append(entity_var)

entities.append(

{

"var_name": entity_var,

"name": entity_col,

"join_key": entity_col,

"description": "Entity key for dbt models",

"tags": {"source": "dbt"},

}

)

# Prepare data sources and feature views

data_sources = []

feature_views = []

for model in models:

# Check required columns exist

column_names = [c.name for c in model.columns]

if self.timestamp_field not in column_names:

continue

# Skip if ANY entity column is missing

if not all(e in column_names for e in entity_cols):

continue

# Build tags

tags = {"dbt.model": model.name}

for tag in model.tags:

tags[f"dbt.tag.{tag}"] = "true"

# Data source

source_var = _make_var_name(f"{model.name}_source")

source_data = {

"var_name": source_var,

"name": f"{model.name}_source",

"timestamp_field": self.timestamp_field,

"description": _escape_description(model.description),

"tags": tags,

}

if self.data_source_type == "bigquery":

source_data["table"] = model.full_table_name

elif self.data_source_type == "snowflake":

source_data["database"] = model.database

source_data["schema"] = model.schema

source_data["table"] = model.alias

elif self.data_source_type == "file":

source_data["path"] = f"/data/{model.name}.parquet"

data_sources.append(source_data)

# Feature view fields

fields = []

for column in model.columns:

if column.name in excluded:

continue

feast_type = map_dbt_type_to_feast_type(column.data_type)

type_name = _get_feast_type_name(feast_type)

# Track base type for imports. For Array types, import both Array and base type.

# Example: Array(Int64) requires imports: Array, Int64

if isinstance(feast_type, Array):

type_imports.add("Array")

base_type = getattr(feast_type, "base_type", None)

if base_type is None:

logger.warning(

"Array type missing 'base_type' attribute while generating imports. "

"This indicates a bug in Array initialization. Falling back to String."

)

base_type = String

base_type_name = _get_feast_type_name(base_type)

type_imports.add(base_type_name)

else:

type_imports.add(type_name)

fields.append(

{

"name": column.name,

"dtype": type_name,

"description": _escape_description(column.description),

}

)

# Feature view

fv_var = _make_var_name(f"{model.name}_fv")

feature_views.append(

{

"var_name": fv_var,

"name": model.name,

"entity_vars": entity_vars,

"source_var": source_var,

"ttl_days": self.ttl_days,

"fields": fields,

"online": online,

"description": _escape_description(model.description),

"tags": tags,

}

)

# Sort type imports for consistent output

sorted_types = sorted(type_imports)

# Render template

return self.template.render(

manifest_path=manifest_path,

project_name=project_name,

data_source_type=self.data_source_type,

type_imports=sorted_types,

entities=entities,

data_sources=data_sources,

feature_views=feature_views,

)

def generate_feast_code(

models: List[DbtModel],

entity_columns: Union[str, List[str]],

data_source_type: str = "bigquery",

timestamp_field: str = "event_timestamp",

ttl_days: int = 1,

manifest_path: str = "",

project_name: str = "",

exclude_columns: Optional[List[str]] = None,

online: bool = True,

) -> str:

"""

Convenience function to generate Feast code from dbt models.

Args:

models: List of DbtModel objects

entity_columns: Entity column name(s) - single string or list of strings

data_source_type: Type of data source (bigquery, snowflake, file)

timestamp_field: Timestamp column name

ttl_days: TTL in days for feature views

manifest_path: Path to manifest for documentation

project_name: Project name for documentation

exclude_columns: Columns to exclude from features

online: Whether to enable online serving

Returns:

Generated Python code as a string

"""

generator = DbtCodeGenerator(

data_source_type=data_source_type,

timestamp_field=timestamp_field,

ttl_days=ttl_days,

)

return generator.generate(

models=models,

entity_columns=entity_columns,

manifest_path=manifest_path,

project_name=project_name,

exclude_columns=exclude_columns,

online=online,

)

feast/sdk/python/feast/dbt/codegen.py at v0.60.0 · feast-dev/feast · GitHub

Uh oh!

Expand file tree

Breadcrumbs

codegen.py

Latest commit

History

Breadcrumbs

codegen.py

File metadata and controls