Hide keyboard shortcuts

Hot-keys on this page

r m x p   toggle line displays

j k   next/prev highlighted chunk

0   (zero) top of page

1   (one) first highlighted chunk

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

1001

1002

1003

1004

1005

1006

1007

1008

1009

1010

1011

1012

1013

1014

1015

1016

1017

1018

1019

1020

1021

1022

1023

1024

1025

1026

1027

1028

1029

1030

1031

1032

1033

1034

1035

1036

1037

1038

1039

1040

1041

1042

1043

1044

1045

1046

1047

1048

1049

1050

1051

1052

1053

1054

1055

1056

1057

1058

1059

1060

1061

1062

1063

1064

1065

1066

1067

1068

1069

1070

1071

1072

1073

1074

1075

1076

1077

1078

1079

1080

1081

1082

1083

1084

1085

1086

1087

1088

1089

1090

1091

1092

1093

1094

1095

1096

1097

1098

1099

1100

1101

1102

1103

1104

1105

1106

1107

1108

1109

1110

1111

1112

1113

1114

1115

1116

1117

1118

1119

1120

1121

1122

1123

1124

1125

1126

1127

1128

1129

1130

1131

1132

1133

1134

1135

1136

1137

1138

1139

1140

1141

1142

1143

1144

1145

1146

1147

1148

1149

1150

1151

1152

1153

1154

1155

1156

1157

1158

1159

1160

1161

1162

1163

1164

1165

1166

1167

1168

1169

1170

1171

1172

1173

1174

1175

# This file is part of daf_butler. 

# 

# Developed for the LSST Data Management System. 

# This product includes software developed by the LSST Project 

# (http://www.lsst.org). 

# See the COPYRIGHT file at the top-level directory of this distribution 

# for details of code ownership. 

# 

# This program is free software: you can redistribute it and/or modify 

# it under the terms of the GNU General Public License as published by 

# the Free Software Foundation, either version 3 of the License, or 

# (at your option) any later version. 

# 

# This program is distributed in the hope that it will be useful, 

# but WITHOUT ANY WARRANTY; without even the implied warranty of 

# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 

# GNU General Public License for more details. 

# 

# You should have received a copy of the GNU General Public License 

# along with this program. If not, see <http://www.gnu.org/licenses/>. 

 

import itertools 

import contextlib 

 

from sqlalchemy import create_engine, text 

from sqlalchemy.sql import select, and_, exists 

from sqlalchemy.exc import IntegrityError 

 

from lsst.sphgeom import ConvexPolygon 

 

from ..core.utils import transactional 

 

from ..core.datasets import DatasetType, DatasetRef 

from ..core.registry import RegistryConfig, Registry 

from ..core.schema import Schema 

from ..core.execution import Execution 

from ..core.run import Run 

from ..core.quantum import Quantum 

from ..core.storageClass import StorageClassFactory 

from ..core.config import Config 

from ..core.sqlRegistryDatabaseDict import SqlRegistryDatabaseDict 

from .sqlPreFlight import SqlPreFlight 

 

__all__ = ("SqlRegistryConfig", "SqlRegistry") 

 

 

class SqlRegistryConfig(RegistryConfig): 

pass 

 

 

class SqlRegistry(Registry): 

"""Registry backed by a SQL database. 

 

Parameters 

---------- 

registryConfig : `SqlRegistryConfig` or `str` 

Load configuration 

schemaConfig : `SchemaConfig` or `str` 

Definition of the schema to use. 

create : `bool` 

Assume registry is empty and create a new one. 

""" 

 

defaultConfigFile = None 

"""Path to configuration defaults. Relative to $DAF_BUTLER_DIR/config or 

absolute path. Can be None if no defaults specified. 

""" 

 

def __init__(self, registryConfig, schemaConfig, create=False): 

super().__init__(registryConfig) 

 

self.config = SqlRegistryConfig(registryConfig) 

self.storageClasses = StorageClassFactory() 

self._schema = Schema(schemaConfig) 

self._engine = create_engine(self.config["db"]) 

self._datasetTypes = {} 

self._connection = self._engine.connect() 

self._preFlight = SqlPreFlight(self._schema, self._connection) 

if create: 

self._createTables() 

 

def __str__(self): 

return self.config["db"] 

 

@contextlib.contextmanager 

def transaction(self): 

"""Context manager that implements SQL transactions. 

 

Will roll back any changes to the `SqlRegistry` database 

in case an exception is raised in the enclosed block. 

 

This context manager may be nested. 

""" 

trans = self._connection.begin() 

try: 

yield 

trans.commit() 

except BaseException: 

trans.rollback() 

raise 

 

def _createTables(self): 

self._schema._metadata.create_all(self._engine) 

 

def _isValidDatasetType(self, datasetType): 

"""Check if given `DatasetType` instance is valid for this `Registry`. 

 

.. todo:: 

 

Insert checks for `storageClass`, `dataUnits` and `template`. 

""" 

return isinstance(datasetType, DatasetType) 

 

def _validateDataId(self, datasetType, dataId): 

"""Check if a dataId is valid for a particular `DatasetType`. 

 

.. todo:: 

 

Move this function to some other place once DataUnit relations are 

implemented. 

 

datasetType : `DatasetType` 

The `DatasetType`. 

dataId : `dict` 

A `dict` of `DataUnit` link name, value pairs that label the 

`DatasetRef` within a collection. 

 

Raises 

------ 

ValueError 

If the dataId is invalid for the given datasetType. 

""" 

for name in datasetType.dataUnits: 

try: 

self._schema.dataUnits[name].validateId(dataId) 

except ValueError as err: 

raise ValueError("Error validating {}".format(datasetType.name)) from err 

 

def makeDatabaseDict(self, table, types, key, value): 

"""Construct a DatabaseDict backed by a table in the same database as 

this Registry. 

 

Parameters 

---------- 

table : `table` 

Name of the table that backs the returned DatabaseDict. If this 

table already exists, its schema must include at least everything 

in `types`. 

types : `dict` 

A dictionary mapping `str` field names to type objects, containing 

all fields to be held in the database. 

key : `str` 

The name of the field to be used as the dictionary key. Must not 

be present in ``value._fields``. 

value : `type` 

The type used for the dictionary's values, typically a 

`~collections.namedtuple`. Must have a ``_fields`` class 

attribute that is a tuple of field names (i.e. as defined by 

`~collections.namedtuple`); these field names must also appear 

in the ``types`` arg, and a `_make` attribute to construct it 

from a sequence of values (again, as defined by 

`~collections.namedtuple`). 

""" 

# We need to construct a temporary config for the table value because 

# SqlRegistryDatabaseDict.__init__ is required to take a config so it 

# can be called by DatabaseDict.fromConfig as well. 

# I suppose we could have Registry.makeDatabaseDict take a config as 

# well, since it"ll also usually be called by DatabaseDict.fromConfig, 

# but I strongly believe in having signatures that only take what they 

# really need. 

config = Config() 

config["table"] = table 

return SqlRegistryDatabaseDict(config, types=types, key=key, value=value, registry=self) 

 

def find(self, collection, datasetType, dataId): 

"""Lookup a dataset. 

 

This can be used to obtain a `DatasetRef` that permits the dataset to 

be read from a `Datastore`. 

 

Parameters 

---------- 

collection : `str` 

Identifies the collection to search. 

datasetType : `DatasetType` 

The `DatasetType`. 

dataId : `dict` 

A `dict` of `DataUnit` link name, value pairs that label the 

`DatasetRef` within a collection. 

 

Returns 

------- 

ref : `DatasetRef` 

A ref to the Dataset, or `None` if no matching Dataset 

was found. 

 

Raises 

------ 

ValueError 

If dataId is invalid. 

""" 

self._validateDataId(datasetType, dataId) 

datasetTable = self._schema.tables["Dataset"] 

datasetCollectionTable = self._schema.tables["DatasetCollection"] 

dataIdExpression = and_((self._schema.dataUnits.links[name] == dataId[name] 

for name in self._schema.dataUnits.getPrimaryKeyNames( 

datasetType.dataUnits))) 

result = self._connection.execute(select([datasetTable.c.dataset_id]).select_from( 

datasetTable.join(datasetCollectionTable)).where(and_( 

datasetTable.c.dataset_type_name == datasetType.name, 

datasetCollectionTable.c.collection == collection, 

dataIdExpression))).fetchone() 

# TODO update unit values and add Run, Quantum and assembler? 

if result is not None: 

return self.getDataset(result["dataset_id"]) 

else: 

return None 

 

def query(self, sql, **params): 

"""Execute a SQL SELECT statement directly. 

 

Named parameters are specified in the SQL query string by preceeding 

them with a colon. Parameter values are provided as additional 

keyword arguments. For example: 

 

registry.query("SELECT * FROM Camera WHERE camera=:name", name="HSC") 

 

Parameters 

---------- 

sql : `str` 

SQL query string. Must be a SELECT statement. 

**params 

Parameter name-value pairs to insert into the query. 

 

Yields 

------- 

row : `dict` 

The next row result from executing the query. 

 

""" 

# TODO: make this guard against non-SELECT queries. 

t = text(sql) 

for row in self._connection.execute(t, **params): 

yield dict(row) 

 

@transactional 

def registerDatasetType(self, datasetType): 

""" 

Add a new `DatasetType` to the SqlRegistry. 

 

It is not an error to register the same `DatasetType` twice. 

 

Parameters 

---------- 

datasetType : `DatasetType` 

The `DatasetType` to be added. 

 

Raises 

------ 

ValueError 

DatasetType is not valid for this registry or is already registered 

but not identical. 

 

Returns 

------- 

inserted : `bool` 

``True`` if ``datasetType`` was inserted, ``False`` if an identical 

existing `DatsetType` was found. 

""" 

270 ↛ 271line 270 didn't jump to line 271, because the condition on line 270 was never true if not self._isValidDatasetType(datasetType): 

raise ValueError("DatasetType is not valid for this registry") 

# If a DatasetType is already registered it must be identical 

try: 

# A DatasetType entry with this name may exist, get it first. 

# Note that we can't just look in the cache, because it may not be there yet. 

existingDatasetType = self.getDatasetType(datasetType.name) 

except KeyError: 

# No registered DatasetType with this name exists, move on to inserting it 

pass 

else: 

# A DatasetType with this name exists, check if is equal 

if datasetType == existingDatasetType: 

return False 

else: 

raise ValueError("DatasetType: {} != existing {}".format(datasetType, existingDatasetType)) 

# Insert it 

datasetTypeTable = self._schema.tables["DatasetType"] 

datasetTypeUnitsTable = self._schema.tables["DatasetTypeUnits"] 

values = {"dataset_type_name": datasetType.name, 

"storage_class": datasetType.storageClass.name} 

self._connection.execute(datasetTypeTable.insert().values(**values)) 

292 ↛ 296line 292 didn't jump to line 296, because the condition on line 292 was never false if datasetType.dataUnits: 

self._connection.execute(datasetTypeUnitsTable.insert(), 

[{"dataset_type_name": datasetType.name, "unit_name": dataUnitName} 

for dataUnitName in datasetType.dataUnits]) 

self._datasetTypes[datasetType.name] = datasetType 

# Also register component DatasetTypes (if any) 

for compName, compStorageClass in datasetType.storageClass.components.items(): 

compType = DatasetType(datasetType.componentTypeName(compName), 

datasetType.dataUnits, 

compStorageClass) 

self.registerDatasetType(compType) 

return True 

 

def getDatasetType(self, name): 

"""Get the `DatasetType`. 

 

Parameters 

---------- 

name : `str` 

Name of the type. 

 

Returns 

------- 

type : `DatasetType` 

The `DatasetType` associated with the given name. 

 

Raises 

------ 

KeyError 

Requested named DatasetType could not be found in registry. 

""" 

datasetTypeTable = self._schema.tables["DatasetType"] 

datasetTypeUnitsTable = self._schema.tables["DatasetTypeUnits"] 

# Get StorageClass from DatasetType table 

result = self._connection.execute(select([datasetTypeTable.c.storage_class]).where( 

datasetTypeTable.c.dataset_type_name == name)).fetchone() 

 

if result is None: 

raise KeyError("Could not find entry for datasetType {}".format(name)) 

 

storageClass = self.storageClasses.getStorageClass(result["storage_class"]) 

# Get DataUnits (if any) from DatasetTypeUnits table 

result = self._connection.execute(select([datasetTypeUnitsTable.c.unit_name]).where( 

datasetTypeUnitsTable.c.dataset_type_name == name)).fetchall() 

dataUnits = (r[0] for r in result) if result else () 

datasetType = DatasetType(name=name, 

storageClass=storageClass, 

dataUnits=dataUnits) 

return datasetType 

 

@transactional 

def addDataset(self, datasetType, dataId, run, producer=None, recursive=False): 

"""Adds a Dataset entry to the `Registry` 

 

This always adds a new Dataset; to associate an existing Dataset with 

a new collection, use ``associate``. 

 

Parameters 

---------- 

datasetType : `DatasetType` 

Type of the Dataset. 

dataId : `dict` 

A `dict` of `DataUnit` link name, value pairs that label the 

`DatasetRef` within a collection. 

run : `Run` 

The `Run` instance that produced the Dataset. Ignored if 

``producer`` is passed (`producer.run` is then used instead). 

A Run must be provided by one of the two arguments. 

producer : `Quantum` 

Unit of work that produced the Dataset. May be ``None`` to store 

no provenance information, but if present the `Quantum` must 

already have been added to the SqlRegistry. 

recursive : `bool` 

If True, recursively add Dataset and attach entries for component 

Datasets as well. 

 

Returns 

------- 

ref : `DatasetRef` 

A newly-created `DatasetRef` instance. 

 

Raises 

------ 

ValueError 

If a Dataset with the given `DatasetRef` already exists in the 

given collection. 

 

Exception 

If ``dataId`` contains unknown or invalid `DataUnit` entries. 

""" 

# TODO this is obviously not the most efficient way to check 

# for existence. 

# TODO also note that this check is not safe 

# in the presence of concurrent calls to addDataset. 

# Then again, it is undoubtedly not the only place where 

# this problem occurs. Needs some serious thought. 

if self.find(run.collection, datasetType, dataId) is not None: 

raise ValueError("A dataset of type {} with id: {} already exists in collection {}".format( 

datasetType, dataId, run.collection)) 

datasetTable = self._schema.tables["Dataset"] 

datasetRef = None 

# TODO add producer 

result = self._connection.execute(datasetTable.insert().values(dataset_type_name=datasetType.name, 

run_id=run.id, 

quantum_id=None, 

**dataId)) 

datasetRef = DatasetRef(datasetType=datasetType, dataId=dataId, id=result.inserted_primary_key[0], 

run=run) 

# A dataset is always associated with its Run collection 

self.associate(run.collection, [datasetRef, ]) 

 

if recursive: 

for component in datasetType.storageClass.components: 

compTypeName = datasetType.componentTypeName(component) 

compDatasetType = self.getDatasetType(compTypeName) 

compRef = self.addDataset(compDatasetType, dataId, run=run, producer=producer, 

recursive=True, transactional=False) 

self.attachComponent(component, datasetRef, compRef) 

return datasetRef 

 

def getDataset(self, id): 

"""Retrieve a Dataset entry. 

 

Parameters 

---------- 

id : `int` 

The unique identifier for the Dataset. 

 

Returns 

------- 

ref : `DatasetRef` 

A ref to the Dataset, or `None` if no matching Dataset 

was found. 

""" 

datasetTable = self._schema.tables["Dataset"] 

with self._connection.begin(): 

result = self._connection.execute( 

select([datasetTable]).where(datasetTable.c.dataset_id == id)).fetchone() 

if result is not None: 

datasetType = self.getDatasetType(result["dataset_type_name"]) 

run = self.getRun(id=result.run_id) 

# dataUnitName gives a `str` key which which is used to lookup 

# the corresponding sqlalchemy.core.Column entry to index the result 

# because the name of the key may not be the name of the name of the 

# DataUnit link. 

dataId = {dataUnitName: result[self._schema.dataUnits.links[dataUnitName]] 

for dataUnitName in self._schema.dataUnits.getPrimaryKeyNames(datasetType.dataUnits)} 

# Get components (if present) 

# TODO check against expected components 

components = {} 

datasetCompositionTable = self._schema.tables["DatasetComposition"] 

results = self._connection.execute( 

select([datasetCompositionTable.c.component_name, 

datasetCompositionTable.c.component_dataset_id]).where( 

datasetCompositionTable.c.parent_dataset_id == id)).fetchall() 

447 ↛ 450line 447 didn't jump to line 450, because the condition on line 447 was never false if results is not None: 

for result in results: 

components[result["component_name"]] = self.getDataset(result["component_dataset_id"]) 

ref = DatasetRef(datasetType=datasetType, dataId=dataId, id=id, run=run) 

ref._components = components 

return ref 

else: 

return None 

 

@transactional 

def attachComponent(self, name, parent, component): 

"""Attach a component to a dataset. 

 

Parameters 

---------- 

name : `str` 

Name of the component. 

parent : `DatasetRef` 

A reference to the parent dataset. Will be updated to reference 

the component. 

component : `DatasetRef` 

A reference to the component dataset. 

""" 

# TODO Insert check for component name and type against parent.storageClass specified components 

datasetCompositionTable = self._schema.tables["DatasetComposition"] 

values = dict(component_name=name, 

parent_dataset_id=parent.id, 

component_dataset_id=component.id) 

self._connection.execute(datasetCompositionTable.insert().values(**values)) 

parent._components[name] = component 

 

@transactional 

def associate(self, collection, refs): 

"""Add existing Datasets to a collection, possibly creating the 

collection in the process. 

 

Parameters 

---------- 

collection : `str` 

Indicates the collection the Datasets should be associated with. 

refs : `list` of `DatasetRef` 

A `list` of `DatasetRef` instances that already exist in this 

`SqlRegistry`. 

""" 

datasetCollectionTable = self._schema.tables["DatasetCollection"] 

self._connection.execute(datasetCollectionTable.insert(), 

[{"dataset_id": ref.id, "collection": collection} for ref in refs]) 

 

@transactional 

def disassociate(self, collection, refs, remove=True): 

r"""Remove existing Datasets from a collection. 

 

``collection`` and ``ref`` combinations that are not currently 

associated are silently ignored. 

 

Parameters 

---------- 

collection : `str` 

The collection the Datasets should no longer be associated with. 

refs : `list` of `DatasetRef` 

A `list` of `DatasetRef` instances that already exist in this 

`SqlRegistry`. 

remove : `bool` 

If `True`, remove Datasets from the `SqlRegistry` if they are not 

associated with any collection (including via any composites). 

 

Returns 

------- 

removed : `list` of `DatasetRef` 

If `remove` is `True`, the `list` of `DatasetRef`\ s that were 

removed. 

""" 

519 ↛ 520line 519 didn't jump to line 520, because the condition on line 519 was never true if remove: 

raise NotImplementedError("Cleanup of datasets not yet implemented") 

datasetCollectionTable = self._schema.tables["DatasetCollection"] 

for ref in refs: 

self._connection.execute(datasetCollectionTable.delete().where( 

and_(datasetCollectionTable.c.dataset_id == ref.id, 

datasetCollectionTable.c.collection == collection))) 

return [] 

 

@transactional 

def addDatasetLocation(self, ref, datastoreName): 

"""Add datastore name locating a given dataset. 

 

Typically used by `Datastore`. 

 

Parameters 

---------- 

ref : `DatasetRef` 

A reference to the dataset for which to add storage information. 

datastoreName : `str` 

Name of the datastore holding this dataset. 

""" 

datasetStorageTable = self._schema.tables["DatasetStorage"] 

values = dict(dataset_id=ref.id, 

datastore_name=datastoreName) 

self._connection.execute(datasetStorageTable.insert().values(**values)) 

 

def getDatasetLocations(self, ref): 

"""Retrieve datastore locations for a given dataset. 

 

Typically used by `Datastore`. 

 

Parameters 

---------- 

ref : `DatasetRef` 

A reference to the dataset for which to retrieve storage 

information. 

 

Returns 

------- 

datastores : `set` of `str` 

All the matching datastores holding this dataset. Empty set 

if the dataset does not exist anywhere. 

""" 

datasetStorageTable = self._schema.tables["DatasetStorage"] 

result = self._connection.execute( 

select([datasetStorageTable.c.datastore_name]).where( 

and_(datasetStorageTable.c.dataset_id == ref.id))).fetchall() 

 

return {r["datastore_name"] for r in result} 

 

@transactional 

def removeDatasetLocation(self, datastoreName, ref): 

"""Remove datastore location associated with this dataset. 

 

Typically used by `Datastore` when a dataset is removed. 

 

Parameters 

---------- 

datastoreName : `str` 

Name of this `Datastore`. 

ref : `DatasetRef` 

A reference to the dataset for which information is to be removed. 

""" 

datasetStorageTable = self._schema.tables["DatasetStorage"] 

self._connection.execute(datasetStorageTable.delete().where( 

and_(datasetStorageTable.c.dataset_id == ref.id, 

datasetStorageTable.c.datastore_name == datastoreName))) 

 

@transactional 

def addExecution(self, execution): 

"""Add a new `Execution` to the `SqlRegistry`. 

 

If ``execution.id`` is `None` the `SqlRegistry` will update it to 

that of the newly inserted entry. 

 

Parameters 

---------- 

execution : `Execution` 

Instance to add to the `SqlRegistry`. 

The given `Execution` must not already be present in the 

`SqlRegistry`. 

 

Raises 

------ 

Exception 

If `Execution` is already present in the `SqlRegistry`. 

""" 

executionTable = self._schema.tables["Execution"] 

result = self._connection.execute(executionTable.insert().values(execution_id=execution.id, 

start_time=execution.startTime, 

end_time=execution.endTime, 

host=execution.host)) 

# Reassign id, may have been `None` 

execution._id = result.inserted_primary_key[0] 

 

def getExecution(self, id): 

"""Retrieve an Execution. 

 

Parameters 

---------- 

id : `int` 

The unique identifier for the Execution. 

""" 

executionTable = self._schema.tables["Execution"] 

result = self._connection.execute( 

select([executionTable.c.start_time, 

executionTable.c.end_time, 

executionTable.c.host]).where(executionTable.c.execution_id == id)).fetchone() 

628 ↛ 634line 628 didn't jump to line 634, because the condition on line 628 was never false if result is not None: 

return Execution(startTime=result["start_time"], 

endTime=result["end_time"], 

host=result["host"], 

id=id) 

else: 

return None 

 

@transactional 

def makeRun(self, collection): 

"""Create a new `Run` in the `SqlRegistry` and return it. 

 

If a run with this collection already exists, return that instead. 

 

Parameters 

---------- 

collection : `str` 

The collection used to identify all inputs and outputs 

of the `Run`. 

 

Returns 

------- 

run : `Run` 

A new `Run` instance. 

""" 

run = Run(collection=collection) 

self.addRun(run) 

return run 

 

@transactional 

def ensureRun(self, run): 

"""Conditionally add a new `Run` to the `SqlRegistry`. 

 

If the ``run.id`` is ``None`` or a `Run` with this `id` doesn't exist 

in the `Registry` yet, add it. Otherwise, ensure the provided run is 

identical to the one already in the registry. 

 

Parameters 

---------- 

run : `Run` 

Instance to add to the `SqlRegistry`. 

 

Raises 

------ 

ValueError 

If ``run`` already exists, but is not identical. 

""" 

if run.id is not None: 

existingRun = self.getRun(id=run.id) 

if run != existingRun: 

raise ValueError("{} != existing: {}".format(run, existingRun)) 

return 

self.addRun(run) 

 

@transactional 

def addRun(self, run): 

"""Add a new `Run` to the `SqlRegistry`. 

 

Parameters 

---------- 

run : `Run` 

Instance to add to the `SqlRegistry`. 

The given `Run` must not already be present in the `SqlRegistry` 

(or any other). Therefore its `id` must be `None` and its 

`collection` must not be associated with any existing `Run`. 

 

Raises 

------ 

ValueError 

If a run already exists with this collection. 

""" 

runTable = self._schema.tables["Run"] 

# TODO: this check is probably undesirable, as we may want to have multiple Runs output 

# to the same collection. Fixing this requires (at least) modifying getRun() accordingly. 

selection = select([exists().where(runTable.c.collection == run.collection)]) 

if self._connection.execute(selection).scalar(): 

raise ValueError("A run already exists with this collection: {}".format(run.collection)) 

# First add the Execution part 

self.addExecution(run) 

# Then the Run specific part 

self._connection.execute(runTable.insert().values(execution_id=run.id, 

collection=run.collection, 

environment_id=None, # TODO add environment 

pipeline_id=None)) # TODO add pipeline 

# TODO: set given Run's "id" attribute. 

 

def getRun(self, id=None, collection=None): 

""" 

Get a `Run` corresponding to its collection or id 

 

Parameters 

---------- 

id : `int`, optional 

Lookup by run `id`, or: 

collection : `str` 

If given, lookup by `collection` name instead. 

 

Returns 

------- 

run : `Run` 

The `Run` instance. 

 

Raises 

------ 

ValueError 

Must supply one of ``collection`` or ``id``. 

""" 

executionTable = self._schema.tables["Execution"] 

runTable = self._schema.tables["Run"] 

run = None 

# Retrieve by id 

if (id is not None) and (collection is None): 

result = self._connection.execute(select([executionTable.c.execution_id, 

executionTable.c.start_time, 

executionTable.c.end_time, 

executionTable.c.host, 

runTable.c.collection, 

runTable.c.environment_id, 

runTable.c.pipeline_id]).select_from( 

runTable.join(executionTable)).where( 

runTable.c.execution_id == id)).fetchone() 

# Retrieve by collection 

750 ↛ 761line 750 didn't jump to line 761, because the condition on line 750 was never false elif (collection is not None) and (id is None): 

result = self._connection.execute(select([executionTable.c.execution_id, 

executionTable.c.start_time, 

executionTable.c.end_time, 

executionTable.c.host, 

runTable.c.collection, 

runTable.c.environment_id, 

runTable.c.pipeline_id]).select_from( 

runTable.join(executionTable)).where( 

runTable.c.collection == collection)).fetchone() 

else: 

raise ValueError("Either collection or id must be given") 

if result is not None: 

run = Run(id=result["execution_id"], 

startTime=result["start_time"], 

endTime=result["end_time"], 

host=result["host"], 

collection=result["collection"], 

environment=None, # TODO add environment 

pipeline=None) # TODO add pipeline 

return run 

 

@transactional 

def addQuantum(self, quantum): 

r"""Add a new `Quantum` to the `SqlRegistry`. 

 

Parameters 

---------- 

quantum : `Quantum` 

Instance to add to the `SqlRegistry`. 

The given `Quantum` must not already be present in the 

`SqlRegistry` (or any other), therefore its: 

 

- `run` attribute must be set to an existing `Run`. 

- `predictedInputs` attribute must be fully populated with 

`DatasetRef`\ s, and its. 

- `actualInputs` and `outputs` will be ignored. 

""" 

quantumTable = self._schema.tables["Quantum"] 

datasetConsumersTable = self._schema.tables["DatasetConsumers"] 

# First add the Execution part 

self.addExecution(quantum) 

# Then the Quantum specific part 

self._connection.execute(quantumTable.insert().values(execution_id=quantum.id, 

task=quantum.task, 

run_id=quantum.run.id)) 

# Attach dataset consumers 

# We use itertools.chain here because quantum.predictedInputs is a 

# dict of ``name : [DatasetRef, ...]`` and we need to flatten it 

# for inserting. 

flatInputs = itertools.chain.from_iterable(quantum.predictedInputs.values()) 

self._connection.execute(datasetConsumersTable.insert(), 

[{"quantum_id": quantum.id, "dataset_id": ref.id, "actual": False} 

for ref in flatInputs]) 

 

def getQuantum(self, id): 

"""Retrieve an Quantum. 

 

Parameters 

---------- 

id : `int` 

The unique identifier for the Quantum. 

""" 

executionTable = self._schema.tables["Execution"] 

quantumTable = self._schema.tables["Quantum"] 

result = self._connection.execute( 

select([quantumTable.c.task, 

quantumTable.c.run_id, 

executionTable.c.start_time, 

executionTable.c.end_time, 

executionTable.c.host]).select_from(quantumTable.join(executionTable)).where( 

quantumTable.c.execution_id == id)).fetchone() 

822 ↛ 841line 822 didn't jump to line 841, because the condition on line 822 was never false if result is not None: 

run = self.getRun(id=result["run_id"]) 

quantum = Quantum(task=result["task"], 

run=run, 

startTime=result["start_time"], 

endTime=result["end_time"], 

host=result["host"], 

id=id) 

# Add predicted and actual inputs to quantum 

datasetConsumersTable = self._schema.tables["DatasetConsumers"] 

for result in self._connection.execute(select([datasetConsumersTable.c.dataset_id, 

datasetConsumersTable.c.actual]).where( 

datasetConsumersTable.c.quantum_id == id)): 

ref = self.getDataset(result["dataset_id"]) 

quantum.addPredictedInput(ref) 

if result["actual"]: 

quantum._markInputUsed(ref) 

return quantum 

else: 

return None 

 

@transactional 

def markInputUsed(self, quantum, ref): 

"""Record the given `DatasetRef` as an actual (not just predicted) 

input of the given `Quantum`. 

 

This updates both the `SqlRegistry`"s `Quantum` table and the Python 

`Quantum.actualInputs` attribute. 

 

Parameters 

---------- 

quantum : `Quantum` 

Producer to update. 

Will be updated in this call. 

ref : `DatasetRef` 

To set as actually used input. 

 

Raises 

------ 

KeyError 

If ``quantum`` is not a predicted consumer for ``ref``. 

""" 

datasetConsumersTable = self._schema.tables["DatasetConsumers"] 

result = self._connection.execute(datasetConsumersTable.update().where(and_( 

datasetConsumersTable.c.quantum_id == quantum.id, 

datasetConsumersTable.c.dataset_id == ref.id)).values(actual=True)) 

if result.rowcount != 1: 

raise KeyError("{} is not a predicted consumer for {}".format(ref, quantum)) 

quantum._markInputUsed(ref) 

 

def getDataUnitDefinition(self, dataUnitName): 

"""Return the definition of a DataUnit (an actual `DataUnit` object). 

 

Parameters 

---------- 

dataUnitName : `str` 

Name of the DataUnit, e.g. "Camera", "Tract", etc. 

""" 

return self._schema.dataUnits[dataUnitName] 

 

@transactional 

def addDataUnitEntry(self, dataUnitName, values): 

"""Add a new `DataUnit` entry. 

 

dataUnitName : `str` 

Name of the `DataUnit` (e.g. ``"Camera"``). 

values : `dict` 

Dictionary of ``columnName, columnValue`` pairs. 

 

If ``values`` includes a "region" key, `setDataUnitRegion` will 

automatically be called to set it any associated spatial join 

tables. 

Region fields associated with a combination of DataUnits must be 

explicitly set separately. 

 

Raises 

------ 

TypeError 

If the given `DataUnit` does not have explicit entries in the 

registry. 

ValueError 

If an entry with the primary-key defined in `values` is already 

present. 

""" 

dataUnit = self._schema.dataUnits[dataUnitName] 

dataUnit.validateId(values) 

dataUnitTable = dataUnit.table 

v = values.copy() 

region = v.pop("region", None) 

911 ↛ 912line 911 didn't jump to line 912, because the condition on line 911 was never true if dataUnitTable is None: 

raise TypeError("DataUnit '{}' has no table.".format(dataUnitName)) 

try: 

self._connection.execute(dataUnitTable.insert().values(**v)) 

except IntegrityError as err: 

raise ValueError(str(err)) # TODO this should do an explicit validity check instead 

if region is not None: 

self.setDataUnitRegion((dataUnitName,), v, region) 

 

def findDataUnitEntry(self, dataUnitName, value): 

"""Return a `DataUnit` entry corresponding to a `value`. 

 

Parameters 

---------- 

dataUnitName : `str` 

Name of a `DataUnit` 

value : `dict` 

A dictionary of values that uniquely identify the `DataUnit`. 

 

Returns 

------- 

dataUnitEntry : `dict` 

Dictionary with all `DataUnit` values, or `None` if no matching 

entry is found. 

""" 

dataUnit = self._schema.dataUnits[dataUnitName] 

dataUnit.validateId(value) 

dataUnitTable = dataUnit.table 

primaryKeyColumns = dataUnit.primaryKeyColumns 

result = self._connection.execute(select([dataUnitTable]).where( 

and_((primaryKeyColumns[name] == value[name] for name in primaryKeyColumns)))).fetchone() 

if result is not None: 

return dict(result.items()) 

else: 

return None 

 

@transactional 

def setDataUnitRegion(self, dataUnitNames, value, region, update=True): 

"""Set the region field for a DataUnit instance or a combination 

thereof and update associated spatial join tables. 

 

Parameters 

---------- 

dataUnitNames : sequence 

A sequence of DataUnit names whose instances are jointly associated 

with a region on the sky. This must not include dependencies that 

are implied, e.g. "Patch" must not include "Tract", but "Sensor" 

needs to add "Visit". 

value : `dict` 

A dictionary of values that uniquely identify the DataUnits. 

region : `sphgeom.ConvexPolygon` 

Region on the sky. 

update : `bool` 

If True, existing region information for these DataUnits is being 

replaced. This is usually required because DataUnit entries are 

assumed to be pre-inserted prior to calling this function. 

""" 

primaryKey = set() 

regionUnitNames = [] 

for dataUnitName in dataUnitNames: 

dataUnit = self._schema.dataUnits[dataUnitName] 

dataUnit.validateId(value) 

primaryKey.update(dataUnit.primaryKey) 

regionUnitNames.append(dataUnitName) 

regionUnitNames += [d.name for d in dataUnit.requiredDependencies] 

table = self._schema.dataUnits.getRegionHolder(*regionUnitNames).table 

977 ↛ 978line 977 didn't jump to line 978, because the condition on line 977 was never true if table is None: 

raise TypeError("No region table found for '{}'.".format(dataUnitNames)) 

# Update the region for an existing entry 

if update: 

result = self._connection.execute( 

table.update().where( 

and_((table.columns[name] == value[name] for name in primaryKey)) 

).values( 

region=region.encode() 

) 

) 

988 ↛ 989line 988 didn't jump to line 989, because the condition on line 988 was never true if result.rowcount == 0: 

raise ValueError("No records were updated when setting region, did you forget update=False?") 

else: # Insert rather than update. 

self._connection.execute( 

table.insert().values( 

region=region.encode(), 

**value 

) 

) 

assert "SkyPix" not in dataUnitNames 

join = self._schema.dataUnits.getJoin(dataUnitNames, "SkyPix") 

if join is None or join.isView: 

return 

if update: 

# Delete any old SkyPix join entries for this DataUnit 

self._connection.execute( 

join.table.delete().where( 

and_((join.table.columns[name] == value[name] for name in primaryKey)) 

) 

) 

parameters = [] 

for begin, end in self.pixelization.envelope(region).ranges(): 

for skypix in range(begin, end): 

parameters.append(dict(value, skypix=skypix)) 

self._connection.execute(join.table.insert(), parameters) 

 

def getRegion(self, dataId): 

"""Get region associated with a dataId. 

 

Parameters 

---------- 

dataId : `dict` 

A `dict` of `DataUnit` link name, value pairs that label the 

`DatasetRef` within a collection. 

 

Returns 

------- 

region : `lsst.sphgeom.ConvexPolygon` 

The region associated with a ``dataId`` or ``None`` if not present. 

 

Raises 

------ 

KeyError 

If the set of dataunits for the ``dataId`` does not correspond to 

a unique spatial lookup. 

""" 

dataUnitNames = (self._schema.dataUnits.getByLinkName(linkName).name for linkName in dataId) 

regionHolder = self._schema.dataUnits.getRegionHolder(*tuple(dataUnitNames)) 

# Skypix does not have a table to lookup the region in, instead generate it 

if regionHolder == self._schema.dataUnits["SkyPix"]: 

return self.pixelization.pixel(dataId["skypix"]) 

# Lookup region 

primaryKeyColumns = regionHolder.primaryKeyColumns 

result = self._connection.execute(select([regionHolder.regionColumn]).where( 

and_((primaryKeyColumns[name] == dataId[name] for name in primaryKeyColumns)))).fetchone() 

if result is not None: 

return ConvexPolygon.decode(result[0]) 

else: 

return None 

 

def selectDataUnits(self, collections, expr, neededDatasetTypes, futureDatasetTypes): 

r"""Evaluate a filter expression and lists of `DatasetType`\ s and 

return a set of data unit values. 

 

Returned set consists of combinations of units participating in data 

transformation from ``neededDatasetTypes`` to ``futureDatasetTypes``, 

restricted by existing data and filter expression. 

 

Parameters 

---------- 

collections : `list` of `str` 

An ordered `list` of collections indicating the collections to 

search for Datasets. 

expr : `str` 

An expression that limits the `DataUnit`\ s and (indirectly) the 

Datasets returned. 

neededDatasetTypes : `list` of `DatasetType` 

The `list` of `DatasetType`\ s whose instances should be included 

in the graph and limit its extent. 

futureDatasetTypes : `list` of `DatasetType` 

The `list` of `DatasetType`\ s whose instances may be added to the 

graph later, which requires that their `DataUnit` types must be 

present in the graph. 

 

Returns 

------- 

header : `tuple` of `tuple` 

Length of tuple equals the number of columns in the returned 

result set. Each item is a tuple with two elements - DataUnit 

name (e.g. "Visit") and unit value name (e.g. "visit"). 

rows : sequence of `tuple` 

Result set, this can be a single-pass iterator. Each tuple 

contains unit values corresponding to units in a header. 

""" 

return self._preFlight.selectDataUnits(collections, expr, 

neededDatasetTypes, 

futureDatasetTypes) 

 

def makeProvenanceGraph(self, expr, types=None): 

"""Make a `QuantumGraph` that contains the full provenance of all 

Datasets matching an expression. 

 

Parameters 

---------- 

expr : `str` 

An expression (SQL query that evaluates to a list of Dataset 

primary keys) that selects the Datasets. 

 

Returns 

------- 

graph : `QuantumGraph` 

Instance (with `units` set to `None`). 

""" 

raise NotImplementedError("Must be implemented by subclass") 

 

def export(self, expr): 

"""Export contents of the `SqlRegistry`, limited to those reachable 

from the Datasets identified by the expression `expr`, into a 

`TableSet` format such that it can be imported into a different 

database. 

 

Parameters 

---------- 

expr : `str` 

An expression (SQL query that evaluates to a list of Dataset 

primary keys) that selects the `Datasets, or a `QuantumGraph` 

that can be similarly interpreted. 

 

Returns 

------- 

ts : `TableSet` 

Containing all rows, from all tables in the `SqlRegistry` that 

are reachable from the selected Datasets. 

""" 

raise NotImplementedError("Must be implemented by subclass") 

 

@transactional 

def import_(self, tables, collection): 

"""Import (previously exported) contents into the (possibly empty) 

`SqlRegistry`. 

 

Parameters 

---------- 

ts : `TableSet` 

Contains the previously exported content. 

collection : `str` 

An additional collection assigned to the newly 

imported Datasets. 

""" 

raise NotImplementedError("Must be implemented by subclass") 

 

@transactional 

def subset(self, collection, expr, datasetTypes): 

r"""Create a new collection by subsetting an existing one. 

 

Parameters 

---------- 

collection : `str` 

Indicates the input collection to subset. 

expr : `str` 

An expression that limits the `DataUnit`\ s and (indirectly) 

Datasets in the subset. 

datasetTypes : `list` of `DatasetType` 

The `list` of `DatasetType`\ s whose instances should be included 

in the subset. 

 

Returns 

------- 

collection : `str` 

The newly created collection. 

""" 

raise NotImplementedError("Must be implemented by subclass") 

 

@transactional 

def merge(self, outputCollection, inputCollections): 

"""Create a new collection from a series of existing ones. 

 

Entries earlier in the list will be used in preference to later 

entries when both contain Datasets with the same `DatasetRef`. 

 

Parameters 

---------- 

outputCollection : `str` 

collection to use for the new collection. 

inputCollections : `list` of `str` 

A `list` of collections to combine. 

""" 

raise NotImplementedError("Must be implemented by subclass")