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

""" 

NetCDF reader/writer module. 

 

This module is used to read and create NetCDF files. NetCDF files are 

accessed through the `netcdf_file` object. Data written to and from NetCDF 

files are contained in `netcdf_variable` objects. Attributes are given 

as member variables of the `netcdf_file` and `netcdf_variable` objects. 

 

This module implements the Scientific.IO.NetCDF API to read and create 

NetCDF files. The same API is also used in the PyNIO and pynetcdf 

modules, allowing these modules to be used interchangeably when working 

with NetCDF files. 

 

Only NetCDF3 is supported here; for NetCDF4 see 

`netCDF4-python <http://unidata.github.io/netcdf4-python/>`__, 

which has a similar API. 

 

""" 

 

from __future__ import division, print_function, absolute_import 

 

# TODO: 

# * properly implement ``_FillValue``. 

# * fix character variables. 

# * implement PAGESIZE for Python 2.6? 

 

# The Scientific.IO.NetCDF API allows attributes to be added directly to 

# instances of ``netcdf_file`` and ``netcdf_variable``. To differentiate 

# between user-set attributes and instance attributes, user-set attributes 

# are automatically stored in the ``_attributes`` attribute by overloading 

#``__setattr__``. This is the reason why the code sometimes uses 

#``obj.__dict__['key'] = value``, instead of simply ``obj.key = value``; 

# otherwise the key would be inserted into userspace attributes. 

 

 

__all__ = ['netcdf_file'] 

 

 

import sys 

import warnings 

import weakref 

from operator import mul 

from collections import OrderedDict 

 

import mmap as mm 

 

import numpy as np 

from numpy.compat import asbytes, asstr 

from numpy import frombuffer, dtype, empty, array, asarray 

from numpy import little_endian as LITTLE_ENDIAN 

from functools import reduce 

 

from scipy._lib.six import integer_types, text_type, binary_type 

 

IS_PYPY = ('__pypy__' in sys.modules) 

 

ABSENT = b'\x00\x00\x00\x00\x00\x00\x00\x00' 

ZERO = b'\x00\x00\x00\x00' 

NC_BYTE = b'\x00\x00\x00\x01' 

NC_CHAR = b'\x00\x00\x00\x02' 

NC_SHORT = b'\x00\x00\x00\x03' 

NC_INT = b'\x00\x00\x00\x04' 

NC_FLOAT = b'\x00\x00\x00\x05' 

NC_DOUBLE = b'\x00\x00\x00\x06' 

NC_DIMENSION = b'\x00\x00\x00\n' 

NC_VARIABLE = b'\x00\x00\x00\x0b' 

NC_ATTRIBUTE = b'\x00\x00\x00\x0c' 

FILL_BYTE = b'\x81' 

FILL_CHAR = b'\x00' 

FILL_SHORT = b'\x80\x01' 

FILL_INT = b'\x80\x00\x00\x01' 

FILL_FLOAT = b'\x7C\xF0\x00\x00' 

FILL_DOUBLE = b'\x47\x9E\x00\x00\x00\x00\x00\x00' 

 

TYPEMAP = {NC_BYTE: ('b', 1), 

NC_CHAR: ('c', 1), 

NC_SHORT: ('h', 2), 

NC_INT: ('i', 4), 

NC_FLOAT: ('f', 4), 

NC_DOUBLE: ('d', 8)} 

 

FILLMAP = {NC_BYTE: FILL_BYTE, 

NC_CHAR: FILL_CHAR, 

NC_SHORT: FILL_SHORT, 

NC_INT: FILL_INT, 

NC_FLOAT: FILL_FLOAT, 

NC_DOUBLE: FILL_DOUBLE} 

 

REVERSE = {('b', 1): NC_BYTE, 

('B', 1): NC_CHAR, 

('c', 1): NC_CHAR, 

('h', 2): NC_SHORT, 

('i', 4): NC_INT, 

('f', 4): NC_FLOAT, 

('d', 8): NC_DOUBLE, 

 

# these come from asarray(1).dtype.char and asarray('foo').dtype.char, 

# used when getting the types from generic attributes. 

('l', 4): NC_INT, 

('S', 1): NC_CHAR} 

 

 

class netcdf_file(object): 

""" 

A file object for NetCDF data. 

 

A `netcdf_file` object has two standard attributes: `dimensions` and 

`variables`. The values of both are dictionaries, mapping dimension 

names to their associated lengths and variable names to variables, 

respectively. Application programs should never modify these 

dictionaries. 

 

All other attributes correspond to global attributes defined in the 

NetCDF file. Global file attributes are created by assigning to an 

attribute of the `netcdf_file` object. 

 

Parameters 

---------- 

filename : string or file-like 

string -> filename 

mode : {'r', 'w', 'a'}, optional 

read-write-append mode, default is 'r' 

mmap : None or bool, optional 

Whether to mmap `filename` when reading. Default is True 

when `filename` is a file name, False when `filename` is a 

file-like object. Note that when mmap is in use, data arrays 

returned refer directly to the mmapped data on disk, and the 

file cannot be closed as long as references to it exist. 

version : {1, 2}, optional 

version of netcdf to read / write, where 1 means *Classic 

format* and 2 means *64-bit offset format*. Default is 1. See 

`here <https://www.unidata.ucar.edu/software/netcdf/docs/netcdf_introduction.html#select_format>`__ 

for more info. 

maskandscale : bool, optional 

Whether to automatically scale and/or mask data based on attributes. 

Default is False. 

 

Notes 

----- 

The major advantage of this module over other modules is that it doesn't 

require the code to be linked to the NetCDF libraries. This module is 

derived from `pupynere <https://bitbucket.org/robertodealmeida/pupynere/>`_. 

 

NetCDF files are a self-describing binary data format. The file contains 

metadata that describes the dimensions and variables in the file. More 

details about NetCDF files can be found `here 

<https://www.unidata.ucar.edu/software/netcdf/docs/user_guide.html>`__. There 

are three main sections to a NetCDF data structure: 

 

1. Dimensions 

2. Variables 

3. Attributes 

 

The dimensions section records the name and length of each dimension used 

by the variables. The variables would then indicate which dimensions it 

uses and any attributes such as data units, along with containing the data 

values for the variable. It is good practice to include a 

variable that is the same name as a dimension to provide the values for 

that axes. Lastly, the attributes section would contain additional 

information such as the name of the file creator or the instrument used to 

collect the data. 

 

When writing data to a NetCDF file, there is often the need to indicate the 

'record dimension'. A record dimension is the unbounded dimension for a 

variable. For example, a temperature variable may have dimensions of 

latitude, longitude and time. If one wants to add more temperature data to 

the NetCDF file as time progresses, then the temperature variable should 

have the time dimension flagged as the record dimension. 

 

In addition, the NetCDF file header contains the position of the data in 

the file, so access can be done in an efficient manner without loading 

unnecessary data into memory. It uses the ``mmap`` module to create 

Numpy arrays mapped to the data on disk, for the same purpose. 

 

Note that when `netcdf_file` is used to open a file with mmap=True 

(default for read-only), arrays returned by it refer to data 

directly on the disk. The file should not be closed, and cannot be cleanly 

closed when asked, if such arrays are alive. You may want to copy data arrays 

obtained from mmapped Netcdf file if they are to be processed after the file 

is closed, see the example below. 

 

Examples 

-------- 

To create a NetCDF file: 

 

>>> from scipy.io import netcdf 

>>> f = netcdf.netcdf_file('simple.nc', 'w') 

>>> f.history = 'Created for a test' 

>>> f.createDimension('time', 10) 

>>> time = f.createVariable('time', 'i', ('time',)) 

>>> time[:] = np.arange(10) 

>>> time.units = 'days since 2008-01-01' 

>>> f.close() 

 

Note the assignment of ``arange(10)`` to ``time[:]``. Exposing the slice 

of the time variable allows for the data to be set in the object, rather 

than letting ``arange(10)`` overwrite the ``time`` variable. 

 

To read the NetCDF file we just created: 

 

>>> from scipy.io import netcdf 

>>> f = netcdf.netcdf_file('simple.nc', 'r') 

>>> print(f.history) 

b'Created for a test' 

>>> time = f.variables['time'] 

>>> print(time.units) 

b'days since 2008-01-01' 

>>> print(time.shape) 

(10,) 

>>> print(time[-1]) 

9 

 

NetCDF files, when opened read-only, return arrays that refer 

directly to memory-mapped data on disk: 

 

>>> data = time[:] 

>>> data.base.base 

<mmap.mmap object at 0x7fe753763180> 

 

If the data is to be processed after the file is closed, it needs 

to be copied to main memory: 

 

>>> data = time[:].copy() 

>>> f.close() 

>>> data.mean() 

4.5 

 

A NetCDF file can also be used as context manager: 

 

>>> from scipy.io import netcdf 

>>> with netcdf.netcdf_file('simple.nc', 'r') as f: 

... print(f.history) 

b'Created for a test' 

 

""" 

def __init__(self, filename, mode='r', mmap=None, version=1, 

maskandscale=False): 

"""Initialize netcdf_file from fileobj (str or file-like).""" 

if mode not in 'rwa': 

raise ValueError("Mode must be either 'r', 'w' or 'a'.") 

 

if hasattr(filename, 'seek'): # file-like 

self.fp = filename 

self.filename = 'None' 

if mmap is None: 

mmap = False 

elif mmap and not hasattr(filename, 'fileno'): 

raise ValueError('Cannot use file object for mmap') 

else: # maybe it's a string 

self.filename = filename 

omode = 'r+' if mode == 'a' else mode 

self.fp = open(self.filename, '%sb' % omode) 

if mmap is None: 

# Mmapped files on PyPy cannot be usually closed 

# before the GC runs, so it's better to use mmap=False 

# as the default. 

mmap = (not IS_PYPY) 

 

if mode != 'r': 

# Cannot read write-only files 

mmap = False 

 

self.use_mmap = mmap 

self.mode = mode 

self.version_byte = version 

self.maskandscale = maskandscale 

 

self.dimensions = OrderedDict() 

self.variables = OrderedDict() 

 

self._dims = [] 

self._recs = 0 

self._recsize = 0 

 

self._mm = None 

self._mm_buf = None 

if self.use_mmap: 

self._mm = mm.mmap(self.fp.fileno(), 0, access=mm.ACCESS_READ) 

self._mm_buf = np.frombuffer(self._mm, dtype=np.int8) 

 

self._attributes = OrderedDict() 

 

if mode in 'ra': 

self._read() 

 

def __setattr__(self, attr, value): 

# Store user defined attributes in a separate dict, 

# so we can save them to file later. 

try: 

self._attributes[attr] = value 

except AttributeError: 

pass 

self.__dict__[attr] = value 

 

def close(self): 

"""Closes the NetCDF file.""" 

if hasattr(self, 'fp') and not self.fp.closed: 

try: 

self.flush() 

finally: 

self.variables = OrderedDict() 

if self._mm_buf is not None: 

ref = weakref.ref(self._mm_buf) 

self._mm_buf = None 

if ref() is None: 

# self._mm_buf is gc'd, and we can close the mmap 

self._mm.close() 

else: 

# we cannot close self._mm, since self._mm_buf is 

# alive and there may still be arrays referring to it 

warnings.warn(( 

"Cannot close a netcdf_file opened with mmap=True, when " 

"netcdf_variables or arrays referring to its data still exist. " 

"All data arrays obtained from such files refer directly to " 

"data on disk, and must be copied before the file can be cleanly " 

"closed. (See netcdf_file docstring for more information on mmap.)" 

), category=RuntimeWarning) 

self._mm = None 

self.fp.close() 

__del__ = close 

 

def __enter__(self): 

return self 

 

def __exit__(self, type, value, traceback): 

self.close() 

 

def createDimension(self, name, length): 

""" 

Adds a dimension to the Dimension section of the NetCDF data structure. 

 

Note that this function merely adds a new dimension that the variables can 

reference. The values for the dimension, if desired, should be added as 

a variable using `createVariable`, referring to this dimension. 

 

Parameters 

---------- 

name : str 

Name of the dimension (Eg, 'lat' or 'time'). 

length : int 

Length of the dimension. 

 

See Also 

-------- 

createVariable 

 

""" 

if length is None and self._dims: 

raise ValueError("Only first dimension may be unlimited!") 

 

self.dimensions[name] = length 

self._dims.append(name) 

 

def createVariable(self, name, type, dimensions): 

""" 

Create an empty variable for the `netcdf_file` object, specifying its data 

type and the dimensions it uses. 

 

Parameters 

---------- 

name : str 

Name of the new variable. 

type : dtype or str 

Data type of the variable. 

dimensions : sequence of str 

List of the dimension names used by the variable, in the desired order. 

 

Returns 

------- 

variable : netcdf_variable 

The newly created ``netcdf_variable`` object. 

This object has also been added to the `netcdf_file` object as well. 

 

See Also 

-------- 

createDimension 

 

Notes 

----- 

Any dimensions to be used by the variable should already exist in the 

NetCDF data structure or should be created by `createDimension` prior to 

creating the NetCDF variable. 

 

""" 

shape = tuple([self.dimensions[dim] for dim in dimensions]) 

shape_ = tuple([dim or 0 for dim in shape]) # replace None with 0 for numpy 

 

type = dtype(type) 

typecode, size = type.char, type.itemsize 

if (typecode, size) not in REVERSE: 

raise ValueError("NetCDF 3 does not support type %s" % type) 

 

data = empty(shape_, dtype=type.newbyteorder("B")) # convert to big endian always for NetCDF 3 

self.variables[name] = netcdf_variable( 

data, typecode, size, shape, dimensions, 

maskandscale=self.maskandscale) 

return self.variables[name] 

 

def flush(self): 

""" 

Perform a sync-to-disk flush if the `netcdf_file` object is in write mode. 

 

See Also 

-------- 

sync : Identical function 

 

""" 

if hasattr(self, 'mode') and self.mode in 'wa': 

self._write() 

sync = flush 

 

def _write(self): 

self.fp.seek(0) 

self.fp.write(b'CDF') 

self.fp.write(array(self.version_byte, '>b').tostring()) 

 

# Write headers and data. 

self._write_numrecs() 

self._write_dim_array() 

self._write_gatt_array() 

self._write_var_array() 

 

def _write_numrecs(self): 

# Get highest record count from all record variables. 

for var in self.variables.values(): 

if var.isrec and len(var.data) > self._recs: 

self.__dict__['_recs'] = len(var.data) 

self._pack_int(self._recs) 

 

def _write_dim_array(self): 

if self.dimensions: 

self.fp.write(NC_DIMENSION) 

self._pack_int(len(self.dimensions)) 

for name in self._dims: 

self._pack_string(name) 

length = self.dimensions[name] 

self._pack_int(length or 0) # replace None with 0 for record dimension 

else: 

self.fp.write(ABSENT) 

 

def _write_gatt_array(self): 

self._write_att_array(self._attributes) 

 

def _write_att_array(self, attributes): 

if attributes: 

self.fp.write(NC_ATTRIBUTE) 

self._pack_int(len(attributes)) 

for name, values in attributes.items(): 

self._pack_string(name) 

self._write_att_values(values) 

else: 

self.fp.write(ABSENT) 

 

def _write_var_array(self): 

if self.variables: 

self.fp.write(NC_VARIABLE) 

self._pack_int(len(self.variables)) 

 

# Sort variable names non-recs first, then recs. 

def sortkey(n): 

v = self.variables[n] 

if v.isrec: 

return (-1,) 

return v._shape 

variables = sorted(self.variables, key=sortkey, reverse=True) 

 

# Set the metadata for all variables. 

for name in variables: 

self._write_var_metadata(name) 

# Now that we have the metadata, we know the vsize of 

# each record variable, so we can calculate recsize. 

self.__dict__['_recsize'] = sum([ 

var._vsize for var in self.variables.values() 

if var.isrec]) 

# Set the data for all variables. 

for name in variables: 

self._write_var_data(name) 

else: 

self.fp.write(ABSENT) 

 

def _write_var_metadata(self, name): 

var = self.variables[name] 

 

self._pack_string(name) 

self._pack_int(len(var.dimensions)) 

for dimname in var.dimensions: 

dimid = self._dims.index(dimname) 

self._pack_int(dimid) 

 

self._write_att_array(var._attributes) 

 

nc_type = REVERSE[var.typecode(), var.itemsize()] 

self.fp.write(asbytes(nc_type)) 

 

if not var.isrec: 

vsize = var.data.size * var.data.itemsize 

vsize += -vsize % 4 

else: # record variable 

try: 

vsize = var.data[0].size * var.data.itemsize 

except IndexError: 

vsize = 0 

rec_vars = len([v for v in self.variables.values() 

if v.isrec]) 

if rec_vars > 1: 

vsize += -vsize % 4 

self.variables[name].__dict__['_vsize'] = vsize 

self._pack_int(vsize) 

 

# Pack a bogus begin, and set the real value later. 

self.variables[name].__dict__['_begin'] = self.fp.tell() 

self._pack_begin(0) 

 

def _write_var_data(self, name): 

var = self.variables[name] 

 

# Set begin in file header. 

the_beguine = self.fp.tell() 

self.fp.seek(var._begin) 

self._pack_begin(the_beguine) 

self.fp.seek(the_beguine) 

 

# Write data. 

if not var.isrec: 

self.fp.write(var.data.tostring()) 

count = var.data.size * var.data.itemsize 

self._write_var_padding(var, var._vsize - count) 

else: # record variable 

# Handle rec vars with shape[0] < nrecs. 

if self._recs > len(var.data): 

shape = (self._recs,) + var.data.shape[1:] 

# Resize in-place does not always work since 

# the array might not be single-segment 

try: 

var.data.resize(shape) 

except ValueError: 

var.__dict__['data'] = np.resize(var.data, shape).astype(var.data.dtype) 

 

pos0 = pos = self.fp.tell() 

for rec in var.data: 

# Apparently scalars cannot be converted to big endian. If we 

# try to convert a ``=i4`` scalar to, say, '>i4' the dtype 

# will remain as ``=i4``. 

if not rec.shape and (rec.dtype.byteorder == '<' or 

(rec.dtype.byteorder == '=' and LITTLE_ENDIAN)): 

rec = rec.byteswap() 

self.fp.write(rec.tostring()) 

# Padding 

count = rec.size * rec.itemsize 

self._write_var_padding(var, var._vsize - count) 

pos += self._recsize 

self.fp.seek(pos) 

self.fp.seek(pos0 + var._vsize) 

 

def _write_var_padding(self, var, size): 

encoded_fill_value = var._get_encoded_fill_value() 

num_fills = size // len(encoded_fill_value) 

self.fp.write(encoded_fill_value * num_fills) 

 

def _write_att_values(self, values): 

if hasattr(values, 'dtype'): 

nc_type = REVERSE[values.dtype.char, values.dtype.itemsize] 

else: 

types = [(t, NC_INT) for t in integer_types] 

types += [ 

(float, NC_FLOAT), 

(str, NC_CHAR) 

] 

# bytes index into scalars in py3k. Check for "string" types 

if isinstance(values, text_type) or isinstance(values, binary_type): 

sample = values 

else: 

try: 

sample = values[0] # subscriptable? 

except TypeError: 

sample = values # scalar 

 

for class_, nc_type in types: 

if isinstance(sample, class_): 

break 

 

typecode, size = TYPEMAP[nc_type] 

dtype_ = '>%s' % typecode 

# asarray() dies with bytes and '>c' in py3k. Change to 'S' 

dtype_ = 'S' if dtype_ == '>c' else dtype_ 

 

values = asarray(values, dtype=dtype_) 

 

self.fp.write(asbytes(nc_type)) 

 

if values.dtype.char == 'S': 

nelems = values.itemsize 

else: 

nelems = values.size 

self._pack_int(nelems) 

 

if not values.shape and (values.dtype.byteorder == '<' or 

(values.dtype.byteorder == '=' and LITTLE_ENDIAN)): 

values = values.byteswap() 

self.fp.write(values.tostring()) 

count = values.size * values.itemsize 

self.fp.write(b'\x00' * (-count % 4)) # pad 

 

def _read(self): 

# Check magic bytes and version 

magic = self.fp.read(3) 

if not magic == b'CDF': 

raise TypeError("Error: %s is not a valid NetCDF 3 file" % 

self.filename) 

self.__dict__['version_byte'] = frombuffer(self.fp.read(1), '>b')[0] 

 

# Read file headers and set data. 

self._read_numrecs() 

self._read_dim_array() 

self._read_gatt_array() 

self._read_var_array() 

 

def _read_numrecs(self): 

self.__dict__['_recs'] = self._unpack_int() 

 

def _read_dim_array(self): 

header = self.fp.read(4) 

if header not in [ZERO, NC_DIMENSION]: 

raise ValueError("Unexpected header.") 

count = self._unpack_int() 

 

for dim in range(count): 

name = asstr(self._unpack_string()) 

length = self._unpack_int() or None # None for record dimension 

self.dimensions[name] = length 

self._dims.append(name) # preserve order 

 

def _read_gatt_array(self): 

for k, v in self._read_att_array().items(): 

self.__setattr__(k, v) 

 

def _read_att_array(self): 

header = self.fp.read(4) 

if header not in [ZERO, NC_ATTRIBUTE]: 

raise ValueError("Unexpected header.") 

count = self._unpack_int() 

 

attributes = OrderedDict() 

for attr in range(count): 

name = asstr(self._unpack_string()) 

attributes[name] = self._read_att_values() 

return attributes 

 

def _read_var_array(self): 

header = self.fp.read(4) 

if header not in [ZERO, NC_VARIABLE]: 

raise ValueError("Unexpected header.") 

 

begin = 0 

dtypes = {'names': [], 'formats': []} 

rec_vars = [] 

count = self._unpack_int() 

for var in range(count): 

(name, dimensions, shape, attributes, 

typecode, size, dtype_, begin_, vsize) = self._read_var() 

# https://www.unidata.ucar.edu/software/netcdf/docs/user_guide.html 

# Note that vsize is the product of the dimension lengths 

# (omitting the record dimension) and the number of bytes 

# per value (determined from the type), increased to the 

# next multiple of 4, for each variable. If a record 

# variable, this is the amount of space per record. The 

# netCDF "record size" is calculated as the sum of the 

# vsize's of all the record variables. 

# 

# The vsize field is actually redundant, because its value 

# may be computed from other information in the header. The 

# 32-bit vsize field is not large enough to contain the size 

# of variables that require more than 2^32 - 4 bytes, so 

# 2^32 - 1 is used in the vsize field for such variables. 

if shape and shape[0] is None: # record variable 

rec_vars.append(name) 

# The netCDF "record size" is calculated as the sum of 

# the vsize's of all the record variables. 

self.__dict__['_recsize'] += vsize 

if begin == 0: 

begin = begin_ 

dtypes['names'].append(name) 

dtypes['formats'].append(str(shape[1:]) + dtype_) 

 

# Handle padding with a virtual variable. 

if typecode in 'bch': 

actual_size = reduce(mul, (1,) + shape[1:]) * size 

padding = -actual_size % 4 

if padding: 

dtypes['names'].append('_padding_%d' % var) 

dtypes['formats'].append('(%d,)>b' % padding) 

 

# Data will be set later. 

data = None 

else: # not a record variable 

# Calculate size to avoid problems with vsize (above) 

a_size = reduce(mul, shape, 1) * size 

if self.use_mmap: 

data = self._mm_buf[begin_:begin_+a_size].view(dtype=dtype_) 

data.shape = shape 

else: 

pos = self.fp.tell() 

self.fp.seek(begin_) 

data = frombuffer(self.fp.read(a_size), dtype=dtype_ 

).copy() 

data.shape = shape 

self.fp.seek(pos) 

 

# Add variable. 

self.variables[name] = netcdf_variable( 

data, typecode, size, shape, dimensions, attributes, 

maskandscale=self.maskandscale) 

 

if rec_vars: 

# Remove padding when only one record variable. 

if len(rec_vars) == 1: 

dtypes['names'] = dtypes['names'][:1] 

dtypes['formats'] = dtypes['formats'][:1] 

 

# Build rec array. 

if self.use_mmap: 

rec_array = self._mm_buf[begin:begin+self._recs*self._recsize].view(dtype=dtypes) 

rec_array.shape = (self._recs,) 

else: 

pos = self.fp.tell() 

self.fp.seek(begin) 

rec_array = frombuffer(self.fp.read(self._recs*self._recsize), 

dtype=dtypes).copy() 

rec_array.shape = (self._recs,) 

self.fp.seek(pos) 

 

for var in rec_vars: 

self.variables[var].__dict__['data'] = rec_array[var] 

 

def _read_var(self): 

name = asstr(self._unpack_string()) 

dimensions = [] 

shape = [] 

dims = self._unpack_int() 

 

for i in range(dims): 

dimid = self._unpack_int() 

dimname = self._dims[dimid] 

dimensions.append(dimname) 

dim = self.dimensions[dimname] 

shape.append(dim) 

dimensions = tuple(dimensions) 

shape = tuple(shape) 

 

attributes = self._read_att_array() 

nc_type = self.fp.read(4) 

vsize = self._unpack_int() 

begin = [self._unpack_int, self._unpack_int64][self.version_byte-1]() 

 

typecode, size = TYPEMAP[nc_type] 

dtype_ = '>%s' % typecode 

 

return name, dimensions, shape, attributes, typecode, size, dtype_, begin, vsize 

 

def _read_att_values(self): 

nc_type = self.fp.read(4) 

n = self._unpack_int() 

 

typecode, size = TYPEMAP[nc_type] 

 

count = n*size 

values = self.fp.read(int(count)) 

self.fp.read(-count % 4) # read padding 

 

if typecode is not 'c': 

values = frombuffer(values, dtype='>%s' % typecode).copy() 

if values.shape == (1,): 

values = values[0] 

else: 

values = values.rstrip(b'\x00') 

return values 

 

def _pack_begin(self, begin): 

if self.version_byte == 1: 

self._pack_int(begin) 

elif self.version_byte == 2: 

self._pack_int64(begin) 

 

def _pack_int(self, value): 

self.fp.write(array(value, '>i').tostring()) 

_pack_int32 = _pack_int 

 

def _unpack_int(self): 

return int(frombuffer(self.fp.read(4), '>i')[0]) 

_unpack_int32 = _unpack_int 

 

def _pack_int64(self, value): 

self.fp.write(array(value, '>q').tostring()) 

 

def _unpack_int64(self): 

return frombuffer(self.fp.read(8), '>q')[0] 

 

def _pack_string(self, s): 

count = len(s) 

self._pack_int(count) 

self.fp.write(asbytes(s)) 

self.fp.write(b'\x00' * (-count % 4)) # pad 

 

def _unpack_string(self): 

count = self._unpack_int() 

s = self.fp.read(count).rstrip(b'\x00') 

self.fp.read(-count % 4) # read padding 

return s 

 

 

class netcdf_variable(object): 

""" 

A data object for the `netcdf` module. 

 

`netcdf_variable` objects are constructed by calling the method 

`netcdf_file.createVariable` on the `netcdf_file` object. `netcdf_variable` 

objects behave much like array objects defined in numpy, except that their 

data resides in a file. Data is read by indexing and written by assigning 

to an indexed subset; the entire array can be accessed by the index ``[:]`` 

or (for scalars) by using the methods `getValue` and `assignValue`. 

`netcdf_variable` objects also have attribute `shape` with the same meaning 

as for arrays, but the shape cannot be modified. There is another read-only 

attribute `dimensions`, whose value is the tuple of dimension names. 

 

All other attributes correspond to variable attributes defined in 

the NetCDF file. Variable attributes are created by assigning to an 

attribute of the `netcdf_variable` object. 

 

Parameters 

---------- 

data : array_like 

The data array that holds the values for the variable. 

Typically, this is initialized as empty, but with the proper shape. 

typecode : dtype character code 

Desired data-type for the data array. 

size : int 

Desired element size for the data array. 

shape : sequence of ints 

The shape of the array. This should match the lengths of the 

variable's dimensions. 

dimensions : sequence of strings 

The names of the dimensions used by the variable. Must be in the 

same order of the dimension lengths given by `shape`. 

attributes : dict, optional 

Attribute values (any type) keyed by string names. These attributes 

become attributes for the netcdf_variable object. 

maskandscale : bool, optional 

Whether to automatically scale and/or mask data based on attributes. 

Default is False. 

 

 

Attributes 

---------- 

dimensions : list of str 

List of names of dimensions used by the variable object. 

isrec, shape 

Properties 

 

See also 

-------- 

isrec, shape 

 

""" 

def __init__(self, data, typecode, size, shape, dimensions, 

attributes=None, 

maskandscale=False): 

self.data = data 

self._typecode = typecode 

self._size = size 

self._shape = shape 

self.dimensions = dimensions 

self.maskandscale = maskandscale 

 

self._attributes = attributes or OrderedDict() 

for k, v in self._attributes.items(): 

self.__dict__[k] = v 

 

def __setattr__(self, attr, value): 

# Store user defined attributes in a separate dict, 

# so we can save them to file later. 

try: 

self._attributes[attr] = value 

except AttributeError: 

pass 

self.__dict__[attr] = value 

 

def isrec(self): 

"""Returns whether the variable has a record dimension or not. 

 

A record dimension is a dimension along which additional data could be 

easily appended in the netcdf data structure without much rewriting of 

the data file. This attribute is a read-only property of the 

`netcdf_variable`. 

 

""" 

return bool(self.data.shape) and not self._shape[0] 

isrec = property(isrec) 

 

def shape(self): 

"""Returns the shape tuple of the data variable. 

 

This is a read-only attribute and can not be modified in the 

same manner of other numpy arrays. 

""" 

return self.data.shape 

shape = property(shape) 

 

def getValue(self): 

""" 

Retrieve a scalar value from a `netcdf_variable` of length one. 

 

Raises 

------ 

ValueError 

If the netcdf variable is an array of length greater than one, 

this exception will be raised. 

 

""" 

return self.data.item() 

 

def assignValue(self, value): 

""" 

Assign a scalar value to a `netcdf_variable` of length one. 

 

Parameters 

---------- 

value : scalar 

Scalar value (of compatible type) to assign to a length-one netcdf 

variable. This value will be written to file. 

 

Raises 

------ 

ValueError 

If the input is not a scalar, or if the destination is not a length-one 

netcdf variable. 

 

""" 

if not self.data.flags.writeable: 

# Work-around for a bug in NumPy. Calling itemset() on a read-only 

# memory-mapped array causes a seg. fault. 

# See NumPy ticket #1622, and SciPy ticket #1202. 

# This check for `writeable` can be removed when the oldest version 

# of numpy still supported by scipy contains the fix for #1622. 

raise RuntimeError("variable is not writeable") 

 

self.data.itemset(value) 

 

def typecode(self): 

""" 

Return the typecode of the variable. 

 

Returns 

------- 

typecode : char 

The character typecode of the variable (eg, 'i' for int). 

 

""" 

return self._typecode 

 

def itemsize(self): 

""" 

Return the itemsize of the variable. 

 

Returns 

------- 

itemsize : int 

The element size of the variable (eg, 8 for float64). 

 

""" 

return self._size 

 

def __getitem__(self, index): 

if not self.maskandscale: 

return self.data[index] 

 

data = self.data[index].copy() 

missing_value = self._get_missing_value() 

data = self._apply_missing_value(data, missing_value) 

scale_factor = self._attributes.get('scale_factor') 

add_offset = self._attributes.get('add_offset') 

if add_offset is not None or scale_factor is not None: 

data = data.astype(np.float64) 

if scale_factor is not None: 

data = data * scale_factor 

if add_offset is not None: 

data += add_offset 

 

return data 

 

def __setitem__(self, index, data): 

if self.maskandscale: 

missing_value = ( 

self._get_missing_value() or 

getattr(data, 'fill_value', 999999)) 

self._attributes.setdefault('missing_value', missing_value) 

self._attributes.setdefault('_FillValue', missing_value) 

data = ((data - self._attributes.get('add_offset', 0.0)) / 

self._attributes.get('scale_factor', 1.0)) 

data = np.ma.asarray(data).filled(missing_value) 

if self._typecode not in 'fd' and data.dtype.kind == 'f': 

data = np.round(data) 

 

# Expand data for record vars? 

if self.isrec: 

if isinstance(index, tuple): 

rec_index = index[0] 

else: 

rec_index = index 

if isinstance(rec_index, slice): 

recs = (rec_index.start or 0) + len(data) 

else: 

recs = rec_index + 1 

if recs > len(self.data): 

shape = (recs,) + self._shape[1:] 

# Resize in-place does not always work since 

# the array might not be single-segment 

try: 

self.data.resize(shape) 

except ValueError: 

self.__dict__['data'] = np.resize(self.data, shape).astype(self.data.dtype) 

self.data[index] = data 

 

def _default_encoded_fill_value(self): 

""" 

The default encoded fill-value for this Variable's data type. 

""" 

nc_type = REVERSE[self.typecode(), self.itemsize()] 

return FILLMAP[nc_type] 

 

def _get_encoded_fill_value(self): 

""" 

Returns the encoded fill value for this variable as bytes. 

 

This is taken from either the _FillValue attribute, or the default fill 

value for this variable's data type. 

""" 

if '_FillValue' in self._attributes: 

fill_value = np.array(self._attributes['_FillValue'], 

dtype=self.data.dtype).tostring() 

if len(fill_value) == self.itemsize(): 

return fill_value 

else: 

return self._default_encoded_fill_value() 

else: 

return self._default_encoded_fill_value() 

 

def _get_missing_value(self): 

""" 

Returns the value denoting "no data" for this variable. 

 

If this variable does not have a missing/fill value, returns None. 

 

If both _FillValue and missing_value are given, give precedence to 

_FillValue. The netCDF standard gives special meaning to _FillValue; 

missing_value is just used for compatibility with old datasets. 

""" 

 

if '_FillValue' in self._attributes: 

missing_value = self._attributes['_FillValue'] 

elif 'missing_value' in self._attributes: 

missing_value = self._attributes['missing_value'] 

else: 

missing_value = None 

 

return missing_value 

 

@staticmethod 

def _apply_missing_value(data, missing_value): 

""" 

Applies the given missing value to the data array. 

 

Returns a numpy.ma array, with any value equal to missing_value masked 

out (unless missing_value is None, in which case the original array is 

returned). 

""" 

 

if missing_value is None: 

newdata = data 

else: 

try: 

missing_value_isnan = np.isnan(missing_value) 

except (TypeError, NotImplementedError): 

# some data types (e.g., characters) cannot be tested for NaN 

missing_value_isnan = False 

 

if missing_value_isnan: 

mymask = np.isnan(data) 

else: 

mymask = (data == missing_value) 

 

newdata = np.ma.masked_where(mymask, data) 

 

return newdata 

 

 

NetCDFFile = netcdf_file 

NetCDFVariable = netcdf_variable