COMPAT: avoid calling getsizeof() on PyPy · rs2/pandas@d2623e4 (original) (raw)

`@@ -11,7 +11,7 @@

11

`import pytest

12

13

`from pandas import (DataFrame, compat, option_context)

14

from pandas.compat import StringIO, lrange, u

14

from pandas.compat import StringIO, lrange, u, PYPY

15

`import pandas.io.formats.format as fmt

16

`import pandas as pd

17

`@@ -323,23 +323,6 @@ def test_info_memory_usage(self):

323

`# excluded column with object dtype, so estimate is accurate

324

`assert not re.match(r"memory usage: [^+]++", res[-1])

325

326

df_with_object_index = pd.DataFrame({'a': [1]}, index=['foo'])

327

df_with_object_index.info(buf=buf, memory_usage=True)

328

res = buf.getvalue().splitlines()

329

assert re.match(r"memory usage: [^+]++", res[-1])

330

-

331

df_with_object_index.info(buf=buf, memory_usage='deep')

332

res = buf.getvalue().splitlines()

333

assert re.match(r"memory usage: [^+]+$", res[-1])

334

-

335

assert (df_with_object_index.memory_usage(

336

index=True, deep=True).sum() > df_with_object_index.memory_usage(

337

index=True).sum())

338

-

339

df_object = pd.DataFrame({'a': ['a']})

340

assert (df_object.memory_usage(deep=True).sum() >

341

df_object.memory_usage().sum())

342

-

343

326

`# Test a DataFrame with duplicate columns

344

327

`dtypes = ['int64', 'int64', 'int64', 'float64']

345

328

`data = {}

`@@ -349,6 +332,15 @@ def test_info_memory_usage(self):

349

332

`df = DataFrame(data)

350

333

`df.columns = dtypes

351

334

335

df_with_object_index = pd.DataFrame({'a': [1]}, index=['foo'])

336

df_with_object_index.info(buf=buf, memory_usage=True)

337

res = buf.getvalue().splitlines()

338

assert re.match(r"memory usage: [^+]++", res[-1])

339

+

340

df_with_object_index.info(buf=buf, memory_usage='deep')

341

res = buf.getvalue().splitlines()

342

assert re.match(r"memory usage: [^+]+$", res[-1])

343

+

352

344

`# Ensure df size is as expected

353

345

`# (cols * rows * bytes) + index size

354

346

`df_size = df.memory_usage().sum()

`@@ -377,9 +369,47 @@ def test_info_memory_usage(self):

377

369

`df.memory_usage(index=True)

378

370

`df.index.values.nbytes

379

371

372

mem = df.memory_usage(deep=True).sum()

373

assert mem > 0

374

+

375

@pytest.mark.skipif(PYPY,

376

reason="on PyPy deep=True doesn't change result")

377

def test_info_memory_usage_deep_not_pypy(self):

378

df_with_object_index = pd.DataFrame({'a': [1]}, index=['foo'])

379

assert (df_with_object_index.memory_usage(

380

index=True, deep=True).sum() >

381

df_with_object_index.memory_usage(

382

index=True).sum())

383

+

384

df_object = pd.DataFrame({'a': ['a']})

385

assert (df_object.memory_usage(deep=True).sum() >

386

df_object.memory_usage().sum())

387

+

388

@pytest.mark.skipif(not PYPY,

389

reason="on PyPy deep=True does not change result")

390

def test_info_memory_usage_deep_pypy(self):

391

df_with_object_index = pd.DataFrame({'a': [1]}, index=['foo'])

392

assert (df_with_object_index.memory_usage(

393

index=True, deep=True).sum() ==

394

df_with_object_index.memory_usage(

395

index=True).sum())

396

+

397

df_object = pd.DataFrame({'a': ['a']})

398

assert (df_object.memory_usage(deep=True).sum() ==

399

df_object.memory_usage().sum())

400

+

401

@pytest.mark.skipif(PYPY, reason="PyPy getsizeof() fails by design")

402

def test_usage_via_getsizeof(self):

403

df = DataFrame(

404

data=1,

405

index=pd.MultiIndex.from_product(

406

[['a'], range(1000)]),

407

columns=['A']

408

)

409

mem = df.memory_usage(deep=True).sum()

380

410

`# sys.getsizeof will call the .memory_usage with

381

411

`# deep=True, and add on some GC overhead

382

diff = df.memory_usage(deep=True).sum() - sys.getsizeof(df)

412

diff = mem - sys.getsizeof(df)

383

413

`assert abs(diff) < 100

384

414

385

415

`def test_info_memory_usage_qualified(self):