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

# -*- coding: utf-8 -*- 

# 

# 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 

# 

# http://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. 

# 

 

from __future__ import print_function 

from builtins import zip 

from past.builtins import basestring 

 

import unicodecsv as csv 

import logging 

import re 

import subprocess 

from tempfile import NamedTemporaryFile 

 

from airflow.exceptions import AirflowException 

from airflow.hooks.base_hook import BaseHook 

from airflow.utils.file import TemporaryDirectory 

from airflow import configuration 

import airflow.security.utils as utils 

 

 

class HiveCliHook(BaseHook): 

 

"""Simple wrapper around the hive CLI. 

 

It also supports the ``beeline`` 

a lighter CLI that runs JDBC and is replacing the heavier 

traditional CLI. To enable ``beeline``, set the use_beeline param in the 

extra field of your connection as in ``{ "use_beeline": true }`` 

 

Note that you can also set default hive CLI parameters using the 

``hive_cli_params`` to be used in your connection as in 

``{"hive_cli_params": "-hiveconf mapred.job.tracker=some.jobtracker:444"}`` 

Parameters passed here can be overridden by run_cli's hive_conf param 

 

The extra connection parameter ``auth`` gets passed as in the ``jdbc`` 

connection string as is. 

""" 

 

def __init__( 

self, 

hive_cli_conn_id="hive_cli_default", 

run_as=None): 

conn = self.get_connection(hive_cli_conn_id) 

self.hive_cli_params = conn.extra_dejson.get('hive_cli_params', '') 

self.use_beeline = conn.extra_dejson.get('use_beeline', False) 

self.auth = conn.extra_dejson.get('auth', 'noSasl') 

self.conn = conn 

self.run_as = run_as 

 

def run_cli(self, hql, schema=None, verbose=True, hive_conf=None): 

""" 

Run an hql statement using the hive cli. If hive_conf is specified it should be a 

dict and the entries will be set as key/value pairs in HiveConf 

 

 

:param hive_conf: if specified these key value pairs will be passed to hive as 

``-hiveconf "key"="value"``. Note that they will be passed after the 

``hive_cli_params`` and thus will override whatever values are specified in 

the database. 

:type hive_conf: dict 

 

>>> hh = HiveCliHook() 

>>> result = hh.run_cli("USE airflow;") 

>>> ("OK" in result) 

True 

""" 

conn = self.conn 

schema = schema or conn.schema 

if schema: 

hql = "USE {schema};\n{hql}".format(**locals()) 

 

with TemporaryDirectory(prefix='airflow_hiveop_') as tmp_dir: 

with NamedTemporaryFile(dir=tmp_dir) as f: 

f.write(hql.encode('UTF-8')) 

f.flush() 

fname = f.name 

hive_bin = 'hive' 

cmd_extra = [] 

 

if self.use_beeline: 

hive_bin = 'beeline' 

jdbc_url = "jdbc:hive2://{conn.host}:{conn.port}/{conn.schema}" 

if configuration.get('core', 'security') == 'kerberos': 

template = conn.extra_dejson.get( 

'principal', "hive/_HOST@EXAMPLE.COM") 

if "_HOST" in template: 

template = utils.replace_hostname_pattern( 

utils.get_components(template)) 

 

proxy_user = "" # noqa 

if conn.extra_dejson.get('proxy_user') == "login" and conn.login: 

proxy_user = "hive.server2.proxy.user={0}".format(conn.login) 

elif conn.extra_dejson.get('proxy_user') == "owner" and self.run_as: 

proxy_user = "hive.server2.proxy.user={0}".format(self.run_as) 

 

jdbc_url += ";principal={template};{proxy_user}" 

elif self.auth: 

jdbc_url += ";auth=" + self.auth 

 

jdbc_url = jdbc_url.format(**locals()) 

 

cmd_extra += ['-u', jdbc_url] 

if conn.login: 

cmd_extra += ['-n', conn.login] 

if conn.password: 

cmd_extra += ['-p', conn.password] 

 

hive_conf = hive_conf or {} 

for key, value in hive_conf.items(): 

cmd_extra += ['-hiveconf', '{0}={1}'.format(key, value)] 

 

hive_cmd = [hive_bin, '-f', fname] + cmd_extra 

 

if self.hive_cli_params: 

hive_params_list = self.hive_cli_params.split() 

hive_cmd.extend(hive_params_list) 

if verbose: 

logging.info(" ".join(hive_cmd)) 

sp = subprocess.Popen( 

hive_cmd, 

stdout=subprocess.PIPE, 

stderr=subprocess.STDOUT, 

cwd=tmp_dir) 

self.sp = sp 

stdout = '' 

while True: 

line = sp.stdout.readline() 

if not line: 

break 

stdout += line.decode('UTF-8') 

if verbose: 

logging.info(line.decode('UTF-8').strip()) 

sp.wait() 

 

if sp.returncode: 

raise AirflowException(stdout) 

 

return stdout 

 

def test_hql(self, hql): 

""" 

Test an hql statement using the hive cli and EXPLAIN 

 

""" 

create, insert, other = [], [], [] 

for query in hql.split(';'): # naive 

query_original = query 

query = query.lower().strip() 

 

if query.startswith('create table'): 

create.append(query_original) 

elif query.startswith(('set ', 

'add jar ', 

'create temporary function')): 

other.append(query_original) 

elif query.startswith('insert'): 

insert.append(query_original) 

other = ';'.join(other) 

for query_set in [create, insert]: 

for query in query_set: 

 

query_preview = ' '.join(query.split())[:50] 

logging.info("Testing HQL [{0} (...)]".format(query_preview)) 

if query_set == insert: 

query = other + '; explain ' + query 

else: 

query = 'explain ' + query 

try: 

self.run_cli(query, verbose=False) 

except AirflowException as e: 

message = e.args[0].split('\n')[-2] 

logging.info(message) 

error_loc = re.search('(\d+):(\d+)', message) 

if error_loc and error_loc.group(1).isdigit(): 

l = int(error_loc.group(1)) 

begin = max(l-2, 0) 

end = min(l+3, len(query.split('\n'))) 

context = '\n'.join(query.split('\n')[begin:end]) 

logging.info("Context :\n {0}".format(context)) 

else: 

logging.info("SUCCESS") 

 

def load_file( 

self, 

filepath, 

table, 

delimiter=",", 

field_dict=None, 

create=True, 

overwrite=True, 

partition=None, 

recreate=False): 

""" 

Loads a local file into Hive 

 

Note that the table generated in Hive uses ``STORED AS textfile`` 

which isn't the most efficient serialization format. If a 

large amount of data is loaded and/or if the tables gets 

queried considerably, you may want to use this operator only to 

stage the data into a temporary table before loading it into its 

final destination using a ``HiveOperator``. 

 

:param table: target Hive table, use dot notation to target a 

specific database 

:type table: str 

:param create: whether to create the table if it doesn't exist 

:type create: bool 

:param recreate: whether to drop and recreate the table at every 

execution 

:type recreate: bool 

:param partition: target partition as a dict of partition columns 

and values 

:type partition: dict 

:param delimiter: field delimiter in the file 

:type delimiter: str 

""" 

hql = '' 

if recreate: 

hql += "DROP TABLE IF EXISTS {table};\n" 

if create or recreate: 

fields = ",\n ".join( 

[k + ' ' + v for k, v in field_dict.items()]) 

hql += "CREATE TABLE IF NOT EXISTS {table} (\n{fields})\n" 

if partition: 

pfields = ",\n ".join( 

[p + " STRING" for p in partition]) 

hql += "PARTITIONED BY ({pfields})\n" 

hql += "ROW FORMAT DELIMITED\n" 

hql += "FIELDS TERMINATED BY '{delimiter}'\n" 

hql += "STORED AS textfile;" 

hql = hql.format(**locals()) 

logging.info(hql) 

self.run_cli(hql) 

hql = "LOAD DATA LOCAL INPATH '{filepath}' " 

if overwrite: 

hql += "OVERWRITE " 

hql += "INTO TABLE {table} " 

if partition: 

pvals = ", ".join( 

["{0}='{1}'".format(k, v) for k, v in partition.items()]) 

hql += "PARTITION ({pvals});" 

hql = hql.format(**locals()) 

logging.info(hql) 

self.run_cli(hql) 

 

def kill(self): 

if hasattr(self, 'sp'): 

if self.sp.poll() is None: 

print("Killing the Hive job") 

self.sp.kill() 

 

 

class HiveMetastoreHook(BaseHook): 

 

""" Wrapper to interact with the Hive Metastore""" 

 

def __init__(self, metastore_conn_id='metastore_default'): 

self.metastore_conn = self.get_connection(metastore_conn_id) 

self.metastore = self.get_metastore_client() 

 

def __getstate__(self): 

# This is for pickling to work despite the thirft hive client not 

# being pickable 

d = dict(self.__dict__) 

del d['metastore'] 

return d 

 

def __setstate__(self, d): 

self.__dict__.update(d) 

self.__dict__['metastore'] = self.get_metastore_client() 

 

def get_metastore_client(self): 

""" 

Returns a Hive thrift client. 

""" 

from thrift.transport import TSocket, TTransport 

from thrift.protocol import TBinaryProtocol 

from hive_service import ThriftHive 

ms = self.metastore_conn 

auth_mechanism = ms.extra_dejson.get('authMechanism', 'NOSASL') 

if configuration.get('core', 'security') == 'kerberos': 

auth_mechanism = ms.extra_dejson.get('authMechanism', 'GSSAPI') 

kerberos_service_name = ms.extra_dejson.get('kerberos_service_name', 'hive') 

 

socket = TSocket.TSocket(ms.host, ms.port) 

if configuration.get('core', 'security') == 'kerberos' and auth_mechanism == 'GSSAPI': 

try: 

import saslwrapper as sasl 

except ImportError: 

import sasl 

 

def sasl_factory(): 

sasl_client = sasl.Client() 

sasl_client.setAttr("host", ms.host) 

sasl_client.setAttr("service", kerberos_service_name) 

sasl_client.init() 

return sasl_client 

 

from thrift_sasl import TSaslClientTransport 

transport = TSaslClientTransport(sasl_factory, "GSSAPI", socket) 

else: 

transport = TTransport.TBufferedTransport(socket) 

 

protocol = TBinaryProtocol.TBinaryProtocol(transport) 

 

return ThriftHive.Client(protocol) 

 

def get_conn(self): 

return self.metastore 

 

def check_for_partition(self, schema, table, partition): 

"""Checks whether a partition exists 

 

>>> hh = HiveMetastoreHook() 

>>> t = 'static_babynames_partitioned' 

>>> hh.check_for_partition('airflow', t, "ds='2015-01-01'") 

True 

""" 

self.metastore._oprot.trans.open() 

partitions = self.metastore.get_partitions_by_filter( 

schema, table, partition, 1) 

self.metastore._oprot.trans.close() 

if partitions: 

return True 

else: 

return False 

 

def get_table(self, table_name, db='default'): 

"""Get a metastore table object 

 

>>> hh = HiveMetastoreHook() 

>>> t = hh.get_table(db='airflow', table_name='static_babynames') 

>>> t.tableName 

'static_babynames' 

>>> [col.name for col in t.sd.cols] 

['state', 'year', 'name', 'gender', 'num'] 

""" 

self.metastore._oprot.trans.open() 

if db == 'default' and '.' in table_name: 

db, table_name = table_name.split('.')[:2] 

table = self.metastore.get_table(dbname=db, tbl_name=table_name) 

self.metastore._oprot.trans.close() 

return table 

 

def get_tables(self, db, pattern='*'): 

""" 

Get a metastore table object 

""" 

self.metastore._oprot.trans.open() 

tables = self.metastore.get_tables(db_name=db, pattern=pattern) 

objs = self.metastore.get_table_objects_by_name(db, tables) 

self.metastore._oprot.trans.close() 

return objs 

 

def get_databases(self, pattern='*'): 

""" 

Get a metastore table object 

""" 

self.metastore._oprot.trans.open() 

dbs = self.metastore.get_databases(pattern) 

self.metastore._oprot.trans.close() 

return dbs 

 

def get_partitions( 

self, schema, table_name, filter=None): 

""" 

Returns a list of all partitions in a table. Works only 

for tables with less than 32767 (java short max val). 

For subpartitioned table, the number might easily exceed this. 

 

>>> hh = HiveMetastoreHook() 

>>> t = 'static_babynames_partitioned' 

>>> parts = hh.get_partitions(schema='airflow', table_name=t) 

>>> len(parts) 

1 

>>> parts 

[{'ds': '2015-01-01'}] 

""" 

self.metastore._oprot.trans.open() 

table = self.metastore.get_table(dbname=schema, tbl_name=table_name) 

if len(table.partitionKeys) == 0: 

raise AirflowException("The table isn't partitioned") 

else: 

if filter: 

parts = self.metastore.get_partitions_by_filter( 

db_name=schema, tbl_name=table_name, 

filter=filter, max_parts=32767) 

else: 

parts = self.metastore.get_partitions( 

db_name=schema, tbl_name=table_name, max_parts=32767) 

 

self.metastore._oprot.trans.close() 

pnames = [p.name for p in table.partitionKeys] 

return [dict(zip(pnames, p.values)) for p in parts] 

 

def max_partition(self, schema, table_name, field=None, filter=None): 

""" 

Returns the maximum value for all partitions in a table. Works only 

for tables that have a single partition key. For subpartitioned 

table, we recommend using signal tables. 

 

>>> hh = HiveMetastoreHook() 

>>> t = 'static_babynames_partitioned' 

>>> hh.max_partition(schema='airflow', table_name=t) 

'2015-01-01' 

""" 

parts = self.get_partitions(schema, table_name, filter) 

if not parts: 

return None 

elif len(parts[0]) == 1: 

field = list(parts[0].keys())[0] 

elif not field: 

raise AirflowException( 

"Please specify the field you want the max " 

"value for") 

 

return max([p[field] for p in parts]) 

 

def table_exists(self, table_name, db='default'): 

""" 

Check if table exists 

 

>>> hh = HiveMetastoreHook() 

>>> hh.table_exists(db='airflow', table_name='static_babynames') 

True 

>>> hh.table_exists(db='airflow', table_name='does_not_exist') 

False 

""" 

try: 

t = self.get_table(table_name, db) 

return True 

except Exception as e: 

return False 

 

 

class HiveServer2Hook(BaseHook): 

""" 

Wrapper around the impyla library 

 

Note that the default authMechanism is PLAIN, to override it you 

can specify it in the ``extra`` of your connection in the UI as in 

""" 

def __init__(self, hiveserver2_conn_id='hiveserver2_default'): 

self.hiveserver2_conn_id = hiveserver2_conn_id 

 

def get_conn(self): 

db = self.get_connection(self.hiveserver2_conn_id) 

auth_mechanism = db.extra_dejson.get('authMechanism', 'PLAIN') 

kerberos_service_name = None 

if configuration.get('core', 'security') == 'kerberos': 

auth_mechanism = db.extra_dejson.get('authMechanism', 'GSSAPI') 

kerberos_service_name = db.extra_dejson.get('kerberos_service_name', 'hive') 

 

# impyla uses GSSAPI instead of KERBEROS as a auth_mechanism identifier 

if auth_mechanism == 'KERBEROS': 

logging.warning("Detected deprecated 'KERBEROS' for authMechanism for %s. Please use 'GSSAPI' instead", 

self.hiveserver2_conn_id) 

auth_mechanism = 'GSSAPI' 

 

from impala.dbapi import connect 

return connect( 

host=db.host, 

port=db.port, 

auth_mechanism=auth_mechanism, 

kerberos_service_name=kerberos_service_name, 

user=db.login, 

database=db.schema or 'default') 

 

def get_results(self, hql, schema='default', arraysize=1000): 

from impala.error import ProgrammingError 

with self.get_conn() as conn: 

if isinstance(hql, basestring): 

hql = [hql] 

results = { 

'data': [], 

'header': [], 

} 

for statement in hql: 

with conn.cursor() as cur: 

cur.execute(statement) 

records = [] 

try: 

# impala Lib raises when no results are returned 

# we're silencing here as some statements in the list 

# may be `SET` or DDL 

records = cur.fetchall() 

except ProgrammingError: 

logging.debug("get_results returned no records") 

if records: 

results = { 

'data': records, 

'header': cur.description, 

} 

return results 

 

def to_csv( 

self, 

hql, 

csv_filepath, 

schema='default', 

delimiter=',', 

lineterminator='\r\n', 

output_header=True, 

fetch_size=1000): 

schema = schema or 'default' 

with self.get_conn() as conn: 

with conn.cursor() as cur: 

logging.info("Running query: " + hql) 

cur.execute(hql) 

schema = cur.description 

with open(csv_filepath, 'wb') as f: 

writer = csv.writer(f, delimiter=delimiter, 

lineterminator=lineterminator, encoding='utf-8') 

if output_header: 

writer.writerow([c[0] 

for c in cur.description]) 

i = 0 

while True: 

rows = [row for row in cur.fetchmany(fetch_size) if row] 

if not rows: 

break 

 

writer.writerows(rows) 

i += len(rows) 

logging.info("Written {0} rows so far.".format(i)) 

logging.info("Done. Loaded a total of {0} rows.".format(i)) 

 

def get_records(self, hql, schema='default'): 

""" 

Get a set of records from a Hive query. 

 

>>> hh = HiveServer2Hook() 

>>> sql = "SELECT * FROM airflow.static_babynames LIMIT 100" 

>>> len(hh.get_records(sql)) 

100 

""" 

return self.get_results(hql, schema=schema)['data'] 

 

def get_pandas_df(self, hql, schema='default'): 

""" 

Get a pandas dataframe from a Hive query 

 

>>> hh = HiveServer2Hook() 

>>> sql = "SELECT * FROM airflow.static_babynames LIMIT 100" 

>>> df = hh.get_pandas_df(sql) 

>>> len(df.index) 

100 

""" 

import pandas as pd 

res = self.get_results(hql, schema=schema) 

df = pd.DataFrame(res['data']) 

df.columns = [c[0] for c in res['header']] 

return df