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

""" 

Functions for changing global ufunc configuration 

 

This provides helpers which wrap `umath.geterrobj` and `umath.seterrobj` 

""" 

import collections.abc 

import contextlib 

 

from .overrides import set_module 

from .umath import ( 

UFUNC_BUFSIZE_DEFAULT, 

ERR_IGNORE, ERR_WARN, ERR_RAISE, ERR_CALL, ERR_PRINT, ERR_LOG, ERR_DEFAULT, 

SHIFT_DIVIDEBYZERO, SHIFT_OVERFLOW, SHIFT_UNDERFLOW, SHIFT_INVALID, 

) 

from . import umath 

 

__all__ = [ 

"seterr", "geterr", "setbufsize", "getbufsize", "seterrcall", "geterrcall", 

"errstate", 

] 

 

_errdict = {"ignore": ERR_IGNORE, 

"warn": ERR_WARN, 

"raise": ERR_RAISE, 

"call": ERR_CALL, 

"print": ERR_PRINT, 

"log": ERR_LOG} 

 

_errdict_rev = {value: key for key, value in _errdict.items()} 

 

 

@set_module('numpy') 

def seterr(all=None, divide=None, over=None, under=None, invalid=None): 

""" 

Set how floating-point errors are handled. 

 

Note that operations on integer scalar types (such as `int16`) are 

handled like floating point, and are affected by these settings. 

 

Parameters 

---------- 

all : {'ignore', 'warn', 'raise', 'call', 'print', 'log'}, optional 

Set treatment for all types of floating-point errors at once: 

 

- ignore: Take no action when the exception occurs. 

- warn: Print a `RuntimeWarning` (via the Python `warnings` module). 

- raise: Raise a `FloatingPointError`. 

- call: Call a function specified using the `seterrcall` function. 

- print: Print a warning directly to ``stdout``. 

- log: Record error in a Log object specified by `seterrcall`. 

 

The default is not to change the current behavior. 

divide : {'ignore', 'warn', 'raise', 'call', 'print', 'log'}, optional 

Treatment for division by zero. 

over : {'ignore', 'warn', 'raise', 'call', 'print', 'log'}, optional 

Treatment for floating-point overflow. 

under : {'ignore', 'warn', 'raise', 'call', 'print', 'log'}, optional 

Treatment for floating-point underflow. 

invalid : {'ignore', 'warn', 'raise', 'call', 'print', 'log'}, optional 

Treatment for invalid floating-point operation. 

 

Returns 

------- 

old_settings : dict 

Dictionary containing the old settings. 

 

See also 

-------- 

seterrcall : Set a callback function for the 'call' mode. 

geterr, geterrcall, errstate 

 

Notes 

----- 

The floating-point exceptions are defined in the IEEE 754 standard [1]_: 

 

- Division by zero: infinite result obtained from finite numbers. 

- Overflow: result too large to be expressed. 

- Underflow: result so close to zero that some precision 

was lost. 

- Invalid operation: result is not an expressible number, typically 

indicates that a NaN was produced. 

 

.. [1] https://en.wikipedia.org/wiki/IEEE_754 

 

Examples 

-------- 

>>> old_settings = np.seterr(all='ignore') #seterr to known value 

>>> np.seterr(over='raise') 

{'divide': 'ignore', 'over': 'ignore', 'under': 'ignore', 'invalid': 'ignore'} 

>>> np.seterr(**old_settings) # reset to default 

{'divide': 'ignore', 'over': 'raise', 'under': 'ignore', 'invalid': 'ignore'} 

 

>>> np.int16(32000) * np.int16(3) 

30464 

>>> old_settings = np.seterr(all='warn', over='raise') 

>>> np.int16(32000) * np.int16(3) 

Traceback (most recent call last): 

File "<stdin>", line 1, in <module> 

FloatingPointError: overflow encountered in short_scalars 

 

>>> from collections import OrderedDict 

>>> old_settings = np.seterr(all='print') 

>>> OrderedDict(np.geterr()) 

OrderedDict([('divide', 'print'), ('over', 'print'), ('under', 'print'), ('invalid', 'print')]) 

>>> np.int16(32000) * np.int16(3) 

30464 

 

""" 

 

pyvals = umath.geterrobj() 

old = geterr() 

 

if divide is None: 

divide = all or old['divide'] 

if over is None: 

over = all or old['over'] 

if under is None: 

under = all or old['under'] 

if invalid is None: 

invalid = all or old['invalid'] 

 

maskvalue = ((_errdict[divide] << SHIFT_DIVIDEBYZERO) + 

(_errdict[over] << SHIFT_OVERFLOW) + 

(_errdict[under] << SHIFT_UNDERFLOW) + 

(_errdict[invalid] << SHIFT_INVALID)) 

 

pyvals[1] = maskvalue 

umath.seterrobj(pyvals) 

return old 

 

 

@set_module('numpy') 

def geterr(): 

""" 

Get the current way of handling floating-point errors. 

 

Returns 

------- 

res : dict 

A dictionary with keys "divide", "over", "under", and "invalid", 

whose values are from the strings "ignore", "print", "log", "warn", 

"raise", and "call". The keys represent possible floating-point 

exceptions, and the values define how these exceptions are handled. 

 

See Also 

-------- 

geterrcall, seterr, seterrcall 

 

Notes 

----- 

For complete documentation of the types of floating-point exceptions and 

treatment options, see `seterr`. 

 

Examples 

-------- 

>>> from collections import OrderedDict 

>>> sorted(np.geterr().items()) 

[('divide', 'warn'), ('invalid', 'warn'), ('over', 'warn'), ('under', 'ignore')] 

>>> np.arange(3.) / np.arange(3.) 

array([nan, 1., 1.]) 

 

>>> oldsettings = np.seterr(all='warn', over='raise') 

>>> OrderedDict(sorted(np.geterr().items())) 

OrderedDict([('divide', 'warn'), ('invalid', 'warn'), ('over', 'raise'), ('under', 'warn')]) 

>>> np.arange(3.) / np.arange(3.) 

array([nan, 1., 1.]) 

 

""" 

maskvalue = umath.geterrobj()[1] 

mask = 7 

res = {} 

val = (maskvalue >> SHIFT_DIVIDEBYZERO) & mask 

res['divide'] = _errdict_rev[val] 

val = (maskvalue >> SHIFT_OVERFLOW) & mask 

res['over'] = _errdict_rev[val] 

val = (maskvalue >> SHIFT_UNDERFLOW) & mask 

res['under'] = _errdict_rev[val] 

val = (maskvalue >> SHIFT_INVALID) & mask 

res['invalid'] = _errdict_rev[val] 

return res 

 

 

@set_module('numpy') 

def setbufsize(size): 

""" 

Set the size of the buffer used in ufuncs. 

 

Parameters 

---------- 

size : int 

Size of buffer. 

 

""" 

if size > 10e6: 

raise ValueError("Buffer size, %s, is too big." % size) 

if size < 5: 

raise ValueError("Buffer size, %s, is too small." % size) 

if size % 16 != 0: 

raise ValueError("Buffer size, %s, is not a multiple of 16." % size) 

 

pyvals = umath.geterrobj() 

old = getbufsize() 

pyvals[0] = size 

umath.seterrobj(pyvals) 

return old 

 

 

@set_module('numpy') 

def getbufsize(): 

""" 

Return the size of the buffer used in ufuncs. 

 

Returns 

------- 

getbufsize : int 

Size of ufunc buffer in bytes. 

 

""" 

return umath.geterrobj()[0] 

 

 

@set_module('numpy') 

def seterrcall(func): 

""" 

Set the floating-point error callback function or log object. 

 

There are two ways to capture floating-point error messages. The first 

is to set the error-handler to 'call', using `seterr`. Then, set 

the function to call using this function. 

 

The second is to set the error-handler to 'log', using `seterr`. 

Floating-point errors then trigger a call to the 'write' method of 

the provided object. 

 

Parameters 

---------- 

func : callable f(err, flag) or object with write method 

Function to call upon floating-point errors ('call'-mode) or 

object whose 'write' method is used to log such message ('log'-mode). 

 

The call function takes two arguments. The first is a string describing 

the type of error (such as "divide by zero", "overflow", "underflow", 

or "invalid value"), and the second is the status flag. The flag is a 

byte, whose four least-significant bits indicate the type of error, one 

of "divide", "over", "under", "invalid":: 

 

[0 0 0 0 divide over under invalid] 

 

In other words, ``flags = divide + 2*over + 4*under + 8*invalid``. 

 

If an object is provided, its write method should take one argument, 

a string. 

 

Returns 

------- 

h : callable, log instance or None 

The old error handler. 

 

See Also 

-------- 

seterr, geterr, geterrcall 

 

Examples 

-------- 

Callback upon error: 

 

>>> def err_handler(type, flag): 

... print("Floating point error (%s), with flag %s" % (type, flag)) 

... 

 

>>> saved_handler = np.seterrcall(err_handler) 

>>> save_err = np.seterr(all='call') 

>>> from collections import OrderedDict 

 

>>> np.array([1, 2, 3]) / 0.0 

Floating point error (divide by zero), with flag 1 

array([inf, inf, inf]) 

 

>>> np.seterrcall(saved_handler) 

<function err_handler at 0x...> 

>>> OrderedDict(sorted(np.seterr(**save_err).items())) 

OrderedDict([('divide', 'call'), ('invalid', 'call'), ('over', 'call'), ('under', 'call')]) 

 

Log error message: 

 

>>> class Log: 

... def write(self, msg): 

... print("LOG: %s" % msg) 

... 

 

>>> log = Log() 

>>> saved_handler = np.seterrcall(log) 

>>> save_err = np.seterr(all='log') 

 

>>> np.array([1, 2, 3]) / 0.0 

LOG: Warning: divide by zero encountered in true_divide 

array([inf, inf, inf]) 

 

>>> np.seterrcall(saved_handler) 

<numpy.core.numeric.Log object at 0x...> 

>>> OrderedDict(sorted(np.seterr(**save_err).items())) 

OrderedDict([('divide', 'log'), ('invalid', 'log'), ('over', 'log'), ('under', 'log')]) 

 

""" 

if func is not None and not isinstance(func, collections.abc.Callable): 

if (not hasattr(func, 'write') or 

not isinstance(func.write, collections.abc.Callable)): 

raise ValueError("Only callable can be used as callback") 

pyvals = umath.geterrobj() 

old = geterrcall() 

pyvals[2] = func 

umath.seterrobj(pyvals) 

return old 

 

 

@set_module('numpy') 

def geterrcall(): 

""" 

Return the current callback function used on floating-point errors. 

 

When the error handling for a floating-point error (one of "divide", 

"over", "under", or "invalid") is set to 'call' or 'log', the function 

that is called or the log instance that is written to is returned by 

`geterrcall`. This function or log instance has been set with 

`seterrcall`. 

 

Returns 

------- 

errobj : callable, log instance or None 

The current error handler. If no handler was set through `seterrcall`, 

``None`` is returned. 

 

See Also 

-------- 

seterrcall, seterr, geterr 

 

Notes 

----- 

For complete documentation of the types of floating-point exceptions and 

treatment options, see `seterr`. 

 

Examples 

-------- 

>>> np.geterrcall() # we did not yet set a handler, returns None 

 

>>> oldsettings = np.seterr(all='call') 

>>> def err_handler(type, flag): 

... print("Floating point error (%s), with flag %s" % (type, flag)) 

>>> oldhandler = np.seterrcall(err_handler) 

>>> np.array([1, 2, 3]) / 0.0 

Floating point error (divide by zero), with flag 1 

array([inf, inf, inf]) 

 

>>> cur_handler = np.geterrcall() 

>>> cur_handler is err_handler 

True 

 

""" 

return umath.geterrobj()[2] 

 

 

class _unspecified: 

pass 

 

 

_Unspecified = _unspecified() 

 

 

@set_module('numpy') 

class errstate(contextlib.ContextDecorator): 

""" 

errstate(**kwargs) 

 

Context manager for floating-point error handling. 

 

Using an instance of `errstate` as a context manager allows statements in 

that context to execute with a known error handling behavior. Upon entering 

the context the error handling is set with `seterr` and `seterrcall`, and 

upon exiting it is reset to what it was before. 

 

.. versionchanged:: 1.17.0 

`errstate` is also usable as a function decorator, saving 

a level of indentation if an entire function is wrapped. 

See :py:class:`contextlib.ContextDecorator` for more information. 

 

Parameters 

---------- 

kwargs : {divide, over, under, invalid} 

Keyword arguments. The valid keywords are the possible floating-point 

exceptions. Each keyword should have a string value that defines the 

treatment for the particular error. Possible values are 

{'ignore', 'warn', 'raise', 'call', 'print', 'log'}. 

 

See Also 

-------- 

seterr, geterr, seterrcall, geterrcall 

 

Notes 

----- 

For complete documentation of the types of floating-point exceptions and 

treatment options, see `seterr`. 

 

Examples 

-------- 

>>> from collections import OrderedDict 

>>> olderr = np.seterr(all='ignore') # Set error handling to known state. 

 

>>> np.arange(3) / 0. 

array([nan, inf, inf]) 

>>> with np.errstate(divide='warn'): 

... np.arange(3) / 0. 

array([nan, inf, inf]) 

 

>>> np.sqrt(-1) 

nan 

>>> with np.errstate(invalid='raise'): 

... np.sqrt(-1) 

Traceback (most recent call last): 

File "<stdin>", line 2, in <module> 

FloatingPointError: invalid value encountered in sqrt 

 

Outside the context the error handling behavior has not changed: 

 

>>> OrderedDict(sorted(np.geterr().items())) 

OrderedDict([('divide', 'ignore'), ('invalid', 'ignore'), ('over', 'ignore'), ('under', 'ignore')]) 

 

""" 

 

def __init__(self, *, call=_Unspecified, **kwargs): 

self.call = call 

self.kwargs = kwargs 

 

def __enter__(self): 

self.oldstate = seterr(**self.kwargs) 

if self.call is not _Unspecified: 

self.oldcall = seterrcall(self.call) 

 

def __exit__(self, *exc_info): 

seterr(**self.oldstate) 

if self.call is not _Unspecified: 

seterrcall(self.oldcall) 

 

 

def _setdef(): 

defval = [UFUNC_BUFSIZE_DEFAULT, ERR_DEFAULT, None] 

umath.seterrobj(defval) 

 

 

# set the default values 

_setdef()