feast/sdk/python/feast/demos.py at codex/flink-doc-fix · feast-dev/feast · GitHub

History

851 lines (774 loc) · 34.8 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

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

# 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.

"""Demo notebook generation for Feast projects.

Usage::

from feast import copy_demo_notebooks

copy_demo_notebooks()

This will search for ``feature_store.yaml`` in the current directory and every

file inside the ``feast-config/`` directory, then write tailored Jupyter

notebooks into a ``./feast-demo-notebooks/<project>/`` directory for each

project found.

"""

import json

import logging

import os

import pathlib

from typing import Any, Optional

import yaml

_logger = logging.getLogger(__name__)

# ---------------------------------------------------------------------------

# Discovery helpers

# ---------------------------------------------------------------------------

def _find_feature_store_yamls(repo_path: pathlib.Path) -> list[pathlib.Path]:

"""Return all feature-store config paths found under *repo_path*.

Searches:

1. ``repo_path/feature_store.yaml``

2. Every file directly inside ``repo_path/feast-config/``

— each file is treated as a separate project config.

"""

found: list[pathlib.Path] = []

direct = repo_path / "feature_store.yaml"

if direct.exists():

found.append(direct)

feast_config_dir = repo_path / "feast-config"

if feast_config_dir.is_dir():

for entry in sorted(feast_config_dir.iterdir()):

if entry.is_file():

found.append(entry)

return found

def _parse_yaml(yaml_path: pathlib.Path) -> dict[str, Any]:

with open(yaml_path) as fh:

return yaml.safe_load(os.path.expandvars(fh.read())) or {}

def _extract_store_info(config: dict[str, Any]) -> dict[str, Any]:

"""Summarise the key fields from a raw ``feature_store.yaml`` dict."""

info: dict[str, Any] = {

"project": config.get("project", "my_feast_project"),

"provider": config.get("provider", "local"),

"online_store_type": "sqlite",

"offline_store_type": "file",

"registry_type": "file",

"auth_type": "no_auth",

"vector_enabled": False,

"embedding_dim": None,

}

online = config.get("online_store", {})

if isinstance(online, dict):

info["online_store_type"] = online.get("type", "sqlite").lower()

info["vector_enabled"] = bool(online.get("vector_enabled", False))

if online.get("embedding_dim"):

info["embedding_dim"] = online["embedding_dim"]

elif isinstance(online, str):

info["online_store_type"] = online.lower()

offline = config.get("offline_store", {})

if isinstance(offline, dict):

info["offline_store_type"] = offline.get("type", "file").lower()

elif isinstance(offline, str):

info["offline_store_type"] = offline.lower()

registry = config.get("registry", {})

if isinstance(registry, dict):

# Operator client YAML uses "registry_type" key; standard Feast uses "type"

info["registry_type"] = (

registry.get("registry_type") or registry.get("type", "file")

).lower()

# string registry value is a plain file path — keep default "file"

auth = config.get("auth", {})

if isinstance(auth, dict):

info["auth_type"] = auth.get("type", "no_auth").lower()

return info

# ---------------------------------------------------------------------------

# Notebook cell builders

# ---------------------------------------------------------------------------

def _md(source: str) -> dict[str, Any]:

return {

"cell_type": "markdown",

"metadata": {},

"source": source,

}

def _code(source: str, tags: Optional[list[str]] = None) -> dict[str, Any]:

meta: dict[str, Any] = {}

if tags:

meta["tags"] = tags

return {

"cell_type": "code",

"execution_count": None,

"metadata": meta,

"outputs": [],

"source": source,

}

def _notebook(cells: list[dict[str, Any]]) -> dict[str, Any]:

return {

"nbformat": 4,

"nbformat_minor": 5,

"metadata": {

"kernelspec": {

"display_name": "Python 3",

"language": "python",

"name": "python3",

"cells": cells,

}

# ---------------------------------------------------------------------------

# Per-store setup snippets

# ---------------------------------------------------------------------------

def _is_operator_client(info: dict[str, Any]) -> bool:

"""Return True when the feature_store.yaml was generated by the Feast operator.

The operator sets provider=local with registry_type=remote, online_store.type=remote,

and offline_store.type=remote.

"""

return (

info["registry_type"] == "remote"

and info["online_store_type"] == "remote"

and info["offline_store_type"] == "remote"

)

# ---------------------------------------------------------------------------

# Notebook generators

# ---------------------------------------------------------------------------

def _apply_md(info: dict[str, Any]) -> dict[str, Any]:

"""Return the markdown cell that introduces the apply / registry-sync section."""

if info["registry_type"] == "remote":

return _md(

"## 4. Registry Sync\n\nRefresh the registry cache to load the latest feature definitions."

)

return _md(

"## 4. Apply Feature Definitions\n\n"

"Register entities, feature views, and services into the registry. "

"Skip if already applied."

)

def _apply_code(info: dict[str, Any]) -> dict[str, Any]:

"""Return the code cell that applies (local) or refreshes (remote) the registry."""

if info["registry_type"] == "remote":

return _code(

"store.refresh_registry()\n"

"fvs = store.list_feature_views()\n"

"print(f'Registry synced — {len(fvs)} feature view(s) available.')"

)

# Local file registry — auto-apply if empty, then refresh.

return _code(

"fvs = store.list_feature_views()\n"

"entities = store.list_entities()\n"

"\n"

"if fvs or entities:\n"

" print(f'Registry ready: {len(entities)} entity/entities, {len(fvs)} feature view(s)')\n"

"else:\n"

" print('Registry is empty — running feast apply ...')\n"

" !feast -f {FEAST_FS_YAML} apply\n"

" store.refresh_registry()\n"

" print('Apply complete.')"

)

def _path_setup_cell(yaml_abs: str) -> dict[str, Any]:

"""Return a code cell that sets ``FEAST_FS_YAML`` to the absolute path of

the feature-store config resolved at generation time."""

return _code(

"import os\n"

"\n"

f"FEAST_FS_YAML = r{repr(yaml_abs)}\n"

"\n"

"assert os.path.exists(FEAST_FS_YAML), (\n"

" f'Config not found at {FEAST_FS_YAML!r}. '\n"

" 'Update FEAST_FS_YAML to the correct path.'\n"

")\n"

"print(f'Using feature_store.yaml: {FEAST_FS_YAML}')",

tags=["parameters"],

)

def _nb_overview(info: dict[str, Any], yaml_abs: str) -> dict[str, Any]:

project = info["project"]

ost = info["online_store_type"]

offst = info["offline_store_type"]

auth = info["auth_type"]

provider = info["provider"]

vector_enabled = info["vector_enabled"]

cells: list[dict[str, Any]] = [

_md(

f"# Feature Store Overview — `{project}`\n\n"

"Explore the entities, feature views, feature services, and data sources "

"registered in this project."

_md("## 1. Prerequisites"),

_code(

"# Verify feast installation\nimport feast\nprint(f'Feast version: {feast.__version__}')"

_md("## 2. Feature Store Path"),

_path_setup_cell(yaml_abs),

_md(

f"## 3. Connect to the Feature Store\n"

f"The feature store for project **`{project}`** is configured with:\n\n"

f"| Setting | Value |\n"

f"|---------|-------|\n"

f"| Provider | `{provider}` |\n"

f"| Online store | `{ost}` |\n"

f"| Offline store | `{offst}` |\n"

f"| Auth | `{auth}` |\n"

+ (

f"| Vector search | enabled (embedding dim: {info['embedding_dim']}) |\n"

if vector_enabled

else ""

)

_code(

"from feast import FeatureStore\n"

"\n"

"store = FeatureStore(fs_yaml_file=FEAST_FS_YAML)\n"

"print(f'Connected to project: {store.project}')"

_apply_md(info),

_apply_code(info),

_md("## 5. List Entities"),

_code(

"entities = store.list_entities()\n"

"print(f'Found {len(entities)} entity/entities\\n')\n"

"for e in entities:\n"

" print(f' • {e.name} (join_key={e.join_key}, type={e.value_type})')"

_md("## 6. List Feature Views"),

_code(

"feature_views = store.list_feature_views()\n"

"print(f'Found {len(feature_views)} batch feature view(s)\\n')\n"

"for fv in feature_views:\n"

" feature_names = [f.name for f in fv.features]\n"

" print(f' • {fv.name}')\n"

" print(f' Features : {feature_names}')\n"

" print(f' Entities : {fv.entities}')\n"

" print(f' TTL : {fv.ttl}')\n"

_md("## 7. List On-Demand Feature Views"),

_code(

"odfvs = store.list_on_demand_feature_views()\n"

"if odfvs:\n"

" print(f'Found {len(odfvs)} on-demand feature view(s)\\n')\n"

" for odfv in odfvs:\n"

" print(f' • {odfv.name}')\n"

"else:\n"

" print('No on-demand feature views defined.')"

_md("## 8. List Feature Services"),

_code(

"services = store.list_feature_services()\n"

"if services:\n"

" print(f'Found {len(services)} feature service(s)\\n')\n"

" for svc in services:\n"

" views = [p.name for p in svc.feature_view_projections]\n"

" print(f' • {svc.name} -> views: {views}')\n"

"else:\n"

" print('No feature services defined.')"

_md("## 9. List Data Sources"),

_code(

"sources = store.list_data_sources()\n"

"print(f'Found {len(sources)} data source(s)\\n')\n"

"for src in sources:\n"

" print(f' • {src.name} ({type(src).__name__})')"

_md(

"## Next Steps\n\n"

"- **`02_historical_features_training.ipynb`** — retrieve historical features for training.\n"

"- **`03_online_features_serving.ipynb`** — materialize and serve online features."

]

return _notebook(cells)

def _nb_historical(info: dict[str, Any], yaml_abs: str) -> dict[str, Any]:

project = info["project"]

cells: list[dict[str, Any]] = [

_md(

f"# Historical Features & Training Datasets — `{project}`\n\n"

"Retrieve point-in-time correct feature values to build ML training datasets."

_md("## 1. Feature Store Path"),

_path_setup_cell(yaml_abs),

_md("## 2. Connect to the Feature Store"),

_code(

"from feast import FeatureStore\n"

"\n"

"store = FeatureStore(fs_yaml_file=FEAST_FS_YAML)\n"

"print(f'Project : {store.project}')\n"

"print('Feature views:', [fv.name for fv in store.list_feature_views()])"

_md(

"## 3. Discover Available Features\n\nList feature views and read a sample of entity data."

_code(

"import pandas as pd\n"

"from datetime import datetime, timedelta, timezone\n"

"\n"

"fvs = store.list_feature_views()\n"

"entities = store.list_entities()\n"

"\n"

"if not fvs:\n"

" print('No feature views found — run `feast apply` first.')\n"

"else:\n"

" first_fv = fvs[0]\n"

"\n"

" # Identify the entity join key.\n"

" entity_name = entities[0].join_key if entities else 'entity_id'\n"

" if first_fv.entities:\n"

" fv_entity = next(\n"

" (e for e in entities if e.name in set(first_fv.entities)),\n"

" entities[0] if entities else None,\n"

" )\n"

" if fv_entity:\n"

" entity_name = fv_entity.join_key\n"

"\n"

" # Read latest entity values from the offline store.\n"

" # This uses the same mechanism Feast uses for materialization.\n"

" source = first_fv.batch_source\n"

" provider = store._get_provider()\n"

" sample_df = provider.offline_store.pull_latest_from_table_or_query(\n"

" config=store.config,\n"

" data_source=source,\n"

" join_key_columns=[entity_name],\n"

" feature_name_columns=[f.name for f in first_fv.features],\n"

" timestamp_field=source.timestamp_field,\n"

" created_timestamp_column=source.created_timestamp_column or '',\n"

" start_date=datetime(2000, 1, 1, tzinfo=timezone.utc),\n"

" end_date=datetime.now(tz=timezone.utc),\n"

" ).to_df()\n"

"\n"

" print(f'Feature view : {first_fv.name}')\n"

" print(f'Entity join key : {entity_name!r}')\n"

" print(f'Rows in source : {len(sample_df):,}')\n"

" print(f'Columns : {list(sample_df.columns)}')\n"

" if len(sample_df) > 0:\n"

" display(sample_df.head())\n"

" else:\n"

" print('No data found — check that your data source has been populated.')"

_md(

"## 4. Build an Entity DataFrame\n\n"

"Specify which entity IDs and at what timestamps you want features for."

_code(

"if not fvs:\n"

" raise SystemExit('No feature views — run feast apply first.')\n"

"\n"

"# Use real entity IDs and timestamps from the sample.\n"

"if entity_name in sample_df.columns and len(sample_df) > 0:\n"

" entity_ids = sample_df[entity_name].dropna().unique()[:5].tolist()\n"

" # Detect the timestamp column from the source's configuration.\n"

" ts_col = source.timestamp_field if source.timestamp_field in sample_df.columns else None\n"

" if not ts_col:\n"

" ts_col = next((c for c in sample_df.columns if 'timestamp' in c.lower()), None)\n"

" if ts_col:\n"

" timestamps = (\n"

" sample_df[sample_df[entity_name].isin(entity_ids)]\n"

" .sort_values(ts_col, ascending=False)\n"

" .drop_duplicates(subset=[entity_name])[ts_col]\n"

" .tolist()\n"

" )\n"

" else:\n"

" timestamps = [datetime.now() - timedelta(hours=i) for i in range(len(entity_ids))]\n"

"else:\n"

" entity_ids = [1001, 1002, 1003]\n"

" timestamps = [datetime.now() - timedelta(hours=i) for i in range(len(entity_ids))]\n"

" print('Using placeholder entity IDs — replace with real values from your data.')\n"

"\n"

"entity_df = pd.DataFrame(\n"

" {\n"

" entity_name: entity_ids[:len(timestamps)],\n"

" 'event_timestamp': timestamps[:len(entity_ids)],\n"

" }\n"

")\n"

"print(f'Entity IDs : {entity_ids}')\n"

"print(f'Rows : {len(entity_df)}')\n"

"entity_df"

_md("## 5. Choose Features to Retrieve"),

_code(

"# List all available feature views and their features.\n"

"print('Available feature views:')\n"

"for fv in fvs:\n"

" features = [f.name for f in fv.features]\n"

" print(f' {fv.name}: {features}')\n"

"\n"

"# Select features from the first feature view.\n"

"# Using a single view avoids name collisions across views with identical column names.\n"

"feature_refs = [f'{first_fv.name}:{f.name}' for f in first_fv.features]\n"

"print('\\nWill retrieve:', feature_refs)"

_md("## 6. Retrieve Historical Features"),

_code(

"if feature_refs:\n"

" training_df = store.get_historical_features(\n"

" entity_df=entity_df,\n"

" features=feature_refs,\n"

" ).to_df()\n"

" print(f'Training dataset shape: {training_df.shape}')\n"

" training_df.head()\n"

"else:\n"

" print('No feature views found — run `feast apply` first.')"

_md(

"## 7. (Optional) Retrieve via FeatureService\n\nRetrieve features using a versioned FeatureService instead of individual feature references."

_code(

"services = store.list_feature_services()\n"

"if not services:\n"

" print('No feature services found — define one in your feature repo.')\n"

"else:\n"

" svc = services[0]\n"

"\n"

" # Detect extra request-data columns required by ODFVs in this service.\n"

" odfv_map = {v.name: v for v in store.list_on_demand_feature_views()}\n"

" missing_cols = {\n"

" field.name: field.dtype\n"

" for proj in svc.feature_view_projections\n"

" if proj.name in odfv_map\n"

" for rs in odfv_map[proj.name].source_request_sources.values()\n"

" for field in rs.schema\n"

" if field.name not in entity_df.columns\n"

" }\n"

"\n"

" if missing_cols:\n"

" print('This service requires the following extra columns in entity_df:')\n"

" for col, dtype in missing_cols.items():\n"

" print(f' entity_df[{col!r}] = <your {dtype} values here>')\n"

" print('Add them to entity_df above and re-run this cell.')\n"

" else:\n"

" # Check if service needs entity keys not already in entity_df.\n"

" svc_entities = set()\n"

" for proj in svc.feature_view_projections:\n"

" fv_match = next((fv for fv in fvs if fv.name == proj.name), None)\n"

" if fv_match:\n"

" for ent_name in fv_match.entities:\n"

" ent_obj = next((e for e in entities if e.name == ent_name), None)\n"

" if ent_obj:\n"

" svc_entities.add(ent_obj.join_key)\n"

" missing_keys = svc_entities - set(entity_df.columns)\n"

" if missing_keys:\n"

" print(f'This service requires additional entity columns: {missing_keys}')\n"

" print('Add them to entity_df above and re-run this cell.')\n"

" else:\n"

" print(f'Using feature service: {svc.name}')\n"

" training_df_svc = store.get_historical_features(\n"

" entity_df=entity_df,\n"

" features=svc,\n"

" full_feature_names=True,\n"

" ).to_df()\n"

" print(f'Dataset shape: {training_df_svc.shape}')\n"

" training_df_svc.head()"

_md("## 8. Use the Training Dataset"),

_code(

"# Example: split into features (X) and labels (y)\n"

"# Adjust column names to match your actual feature names and label.\n"

"if feature_refs and 'training_df' in dir():\n"

" label_col = 'label' # TODO: replace with your label column\n"

" feature_cols = [c for c in training_df.columns\n"

" if c not in ('event_timestamp', entity_name, label_col)]\n"

" X = training_df[feature_cols]\n"

" print('Feature matrix shape:', X.shape)\n"

" print('Feature columns:', feature_cols)"

_md(

"## Next Steps\n\n"

"- **`03_online_features_serving.ipynb`** — materialize and serve online features."

]

return _notebook(cells)

def _nb_online(info: dict[str, Any], yaml_abs: str) -> dict[str, Any]:

project = info["project"]

auth = info["auth_type"]

vector_enabled = info["vector_enabled"]

cells: list[dict[str, Any]] = [

_md(

f"# Online Feature Serving — `{project}`\n\n"

"Materialize features and retrieve them at low latency for inference."

_md("## 1. Feature Store Path"),

_path_setup_cell(yaml_abs),

_md("## 2. Connect to the Feature Store"),

_code(

"from feast import FeatureStore\n"

"\n"

"store = FeatureStore(fs_yaml_file=FEAST_FS_YAML)\n"

"print(f'Project : {store.project}')"

]

# Materialization section.

materialize_md = (

"## 3. Materialize Features\n\n"

+ (

"> **Optional** — materialization is typically handled server-side.\n\n"

if _is_operator_client(info)

else ""

)

+ "Load feature values into the online store for low-latency serving.\n\n"

"| Method | When to use |\n"

"|--------|-------------|\n"

"| `materialize_incremental` | Regular runs — only new data since last run |\n"

"| `materialize` | First run or full refresh of a time window |"

)

cells += [

_md(materialize_md),

_code(

"from datetime import datetime, timedelta, timezone\n"

"\n"

"fvs = store.list_feature_views()\n"

"\n"

"if not fvs:\n"

" print('No feature views found — run feast apply first (see section 3).')\n"

"else:\n"

" # Check last materialization watermarks across all feature views.\n"

" last_written = [\n"

" fv.materialization_intervals[-1][1]\n"

" for fv in fvs\n"

" if fv.materialization_intervals\n"

" ]\n"

"\n"

" if not last_written:\n"

" # No materialization history — do a full initial load.\n"

" end_date = datetime.now(tz=timezone.utc)\n"

" start_date = end_date - timedelta(days=30)\n"

" print(f'First materialization: loading {start_date.date()} → {end_date.date()} ...')\n"

" store.materialize(start_date=start_date, end_date=end_date)\n"

" else:\n"

" # Incremental: only pick up data since the last run.\n"

" end_date = datetime.now(tz=timezone.utc)\n"

" print(f'Incremental materialization up to {end_date} ...')\n"

" store.materialize_incremental(end_date=end_date)\n"

"\n"

" print('Materialization complete.')"

_md("### 3b. Force a Full Refresh"),

_code(

"# from datetime import datetime, timedelta, timezone\n"

"# store.materialize(\n"

"# start_date=datetime.now(tz=timezone.utc) - timedelta(days=7),\n"

"# end_date=datetime.now(tz=timezone.utc),\n"

"# )"

]

cells += [

_md("## 4. Retrieve Online Features"),

_code(

"entities = store.list_entities()\n"

"fvs = store.list_feature_views()\n"

"\n"

"if not entities or not fvs:\n"

" print('No entities or feature views — run `feast apply` first.')\n"

"else:\n"

" first_fv = fvs[0]\n"

" feature_refs = [f'{first_fv.name}:{f.name}' for f in first_fv.features[:3]]\n"

"\n"

" # Resolve the correct entity join key for the first feature view.\n"

" entity_name = entities[0].join_key\n"

" if first_fv.entities:\n"

" fv_entity = next(\n"

" (e for e in entities if e.name in set(first_fv.entities)),\n"

" entities[0],\n"

" )\n"

" entity_name = fv_entity.join_key\n"

"\n"

" # Discover real entity IDs from the offline source.\n"

" from datetime import timezone\n"

" source = first_fv.batch_source\n"

" provider = store._get_provider()\n"

" sample_df = provider.offline_store.pull_latest_from_table_or_query(\n"

" config=store.config,\n"

" data_source=source,\n"

" join_key_columns=[entity_name],\n"

" feature_name_columns=[f.name for f in first_fv.features],\n"

" timestamp_field=source.timestamp_field,\n"

" created_timestamp_column=source.created_timestamp_column or '',\n"

" start_date=datetime(2000, 1, 1, tzinfo=timezone.utc),\n"

" end_date=datetime.now(tz=timezone.utc),\n"

" ).to_df()\n"

"\n"

" if len(sample_df) > 0 and entity_name in sample_df.columns:\n"

" entity_ids = sample_df[entity_name].dropna().unique()[:5].tolist()\n"

" else:\n"

" entity_ids = [1001, 1002]\n"

" print('Using placeholder IDs — replace with real values.')\n"

"\n"

" entity_rows = [{entity_name: eid} for eid in entity_ids]\n"

"\n"

" response = store.get_online_features(\n"

" features=feature_refs,\n"

" entity_rows=entity_rows,\n"

" )\n"

" import pandas as pd\n"

" print(pd.DataFrame(response.to_dict()))"

_md(

"## 5. Online Features via FeatureService\n\nRetrieve features using a versioned FeatureService."

_code(

"services = store.list_feature_services()\n"

"if not services:\n"

" print('No feature services defined.')\n"

"else:\n"

" svc = services[0]\n"

"\n"

" # Detect extra request-data fields required by ODFVs in this service.\n"

" odfv_map = {v.name: v for v in store.list_on_demand_feature_views()}\n"

" current_keys = set(entity_rows[0].keys()) if entity_rows else set()\n"

" missing_fields = {\n"

" field.name: field.dtype\n"

" for proj in svc.feature_view_projections\n"

" if proj.name in odfv_map\n"

" for rs in odfv_map[proj.name].source_request_sources.values()\n"

" for field in rs.schema\n"

" if field.name not in current_keys\n"

" }\n"

"\n"

" if missing_fields:\n"

" print('This service requires the following extra fields in each entity row:')\n"

" for col, dtype in missing_fields.items():\n"

" print(f' {col!r}: <your {dtype} value here>')\n"

" print('Add them to entity_rows above and re-run this cell.')\n"

" else:\n"

" # Check if service needs extra entity keys beyond what we have.\n"

" svc_entities = set()\n"

" for proj in svc.feature_view_projections:\n"

" fv_match = next((fv for fv in fvs if fv.name == proj.name), None)\n"

" if fv_match:\n"

" for ent_name in fv_match.entities:\n"

" ent_obj = next((e for e in entities if e.name == ent_name), None)\n"

" if ent_obj:\n"

" svc_entities.add(ent_obj.join_key)\n"

" missing_keys = svc_entities - current_keys\n"

" if missing_keys:\n"

" print(f'This service requires additional entity keys: {missing_keys}')\n"

" print('Add them to entity_rows above and re-run this cell.')\n"

" else:\n"

" print(f'Using feature service: {svc.name}')\n"

" response = store.get_online_features(\n"

" features=svc,\n"

" entity_rows=entity_rows,\n"

" full_feature_names=True,\n"

" )\n"

" import pandas as pd\n"

" print(pd.DataFrame(response.to_dict()))"

]

if auth in ("kubernetes", "oidc"):

cells.append(_md(f"## 6. Authentication (`{auth}`)"))

cells.append(_code("print(store.config.auth)"))

if vector_enabled:

dim = info.get("embedding_dim") or 384

section = 7 if auth in ("kubernetes", "oidc") else 6

cells.append(

_md(

f"## {section}. Vector / RAG Feature Retrieval\n\nSearch stored embeddings (dim: {dim})."

)

cells.append(

_code(

"import numpy as np\n"

"\n"

"# TODO: replace with a real query embedding from your encoder model\n"

f"query_embedding = np.random.rand({dim}).tolist()\n"

"\n"

"# List feature views with vector features\n"

"fvs = store.list_feature_views()\n"

"vec_fvs = [\n"

" fv for fv in fvs\n"

" if any(getattr(f, 'vector_index', False) for f in fv.features)\n"

"]\n"

"\n"

"if vec_fvs:\n"

" fv = vec_fvs[0]\n"

" results = store.retrieve_online_documents(\n"

" feature=f'{fv.name}:{fv.features[0].name}',\n"

" query=query_embedding,\n"

" top_k=5,\n"

" )\n"

" import pandas as pd\n"

" print(pd.DataFrame(results.to_dict()))\n"

"else:\n"

" print('No vector feature views found.')"

)

cells.append(

_md(

"## Next Steps\n\n"

"- Schedule `materialize_incremental` to keep the online store fresh.\n"

)

return _notebook(cells)

# ---------------------------------------------------------------------------

# Public API

# ---------------------------------------------------------------------------

def copy_demo_notebooks(

output_dir: str = "./feast-demo-notebooks",

repo_path: str = ".",

overwrite: bool = False,

) -> None:

"""Generate tailored demo notebooks for each Feast project found nearby.

The function searches *repo_path* (default: current working directory) for

feature-store YAML files in:

* ``<repo_path>/feature_store.yaml``

* Every file inside ``<repo_path>/feast-config/``

For each project discovered a sub-directory is created under *output_dir*

and one or more notebooks are written (the exact set depends on the project

configuration and may grow in future releases).

Parameters

----------

output_dir:

Root directory where notebooks are written.

Defaults to ``./feast-demo-notebooks``.

repo_path:

Directory to search for ``feature_store.yaml`` files.

Defaults to the current working directory.

overwrite:

When *False* (default) raise :class:`FileExistsError` if *output_dir*

already exists. Set to *True* to update notebooks in place.

"""

out = pathlib.Path(output_dir).resolve()

if not overwrite and out.exists():

raise FileExistsError(

f"Directory '{out}' already exists. "

"Remove it or pass overwrite=True to update notebooks in place."

)

root = pathlib.Path(repo_path).absolute()

yaml_paths = _find_feature_store_yamls(root)

if not yaml_paths:

_logger.warning(

"No feature_store.yaml found under '%s'. "

"Make sure you run this from a directory that contains feature_store.yaml "

"or a feast-config/ subdirectory.",

root,

)

return

out.mkdir(parents=True, exist_ok=True)

print(f"Writing demo notebooks to: {out}\n")

for yaml_path in yaml_paths:

raw = _parse_yaml(yaml_path)

info = _extract_store_info(raw)

project = info["project"]

project_dir = out / project

project_dir.mkdir(parents=True, exist_ok=True)

# Absolute path — use absolute() instead of resolve() to preserve

# Kubernetes ConfigMap/Secret symlinks.

yaml_abs_str = str(yaml_path.absolute())

notebooks = {

"01_feature_store_overview.ipynb": _nb_overview(info, yaml_abs_str),

"02_historical_features_training.ipynb": _nb_historical(info, yaml_abs_str),

"03_online_features_serving.ipynb": _nb_online(info, yaml_abs_str),

}

for nb_name, nb_content in notebooks.items():

nb_path = project_dir / nb_name

with open(nb_path, "w") as fh:

json.dump(nb_content, fh, indent=1)

print(

f" [{project}]\n"

f" feature_store.yaml : {yaml_abs_str}\n"

f" online_store : {info['online_store_type']}\n"

f" offline_store : {info['offline_store_type']}\n"

f" auth : {info['auth_type']}\n"

+ (" vector search : enabled\n" if info["vector_enabled"] else "")

+ f" → {project_dir}/"

)

for nb_name in notebooks:

print(f" ✓ {nb_name}")

print()

feast/sdk/python/feast/demos.py at codex/flink-doc-fix · feast-dev/feast · GitHub

Uh oh!

Expand file tree

Breadcrumbs

demos.py

Latest commit

History

Breadcrumbs

demos.py

File metadata and controls