test_tsio.py 37.7 KB
Newer Older
1
from datetime import datetime, timedelta
2
3
from pathlib import Path
import pytz
4

5
from dateutil import parser
6
import pytest
7
8
import numpy as np
import pandas as pd
9

10
from tshistory.snapshot import Snapshot
11
from tshistory.util import rename_series
12
13
14
15
16
17
from tshistory.testutil import (
    assert_df,
    assert_group_equals,
    genserie,
    tempattr
)
18

19
DATADIR = Path(__file__).parent / 'data'
20

Aurélien Campéas's avatar
Aurélien Campéas committed
21

22
23
24
25
def utcdt(*dt):
    return pd.Timestamp(datetime(*dt), tz='UTC')


26
def test_tstamp_roundtrip(engine, tsh):
27
28
    ts = genserie(datetime(2017, 10, 28, 23),
                  'H', 4, tz='UTC')
29
30
31
32
33
34
35
36
37
38
    ts.index = ts.index.tz_convert('Europe/Paris')

    assert_df("""
2017-10-29 01:00:00+02:00    0
2017-10-29 02:00:00+02:00    1
2017-10-29 02:00:00+01:00    2
2017-10-29 03:00:00+01:00    3
Freq: H
    """, ts)

39
40
    tsh.insert(engine, ts, 'tztest', 'Babar')
    back = tsh.get(engine, 'tztest')
41
42
43

    # though un localized we understand it's been normalized to utc
    assert_df("""
44
45
46
47
2017-10-28 23:00:00+00:00    0.0
2017-10-29 00:00:00+00:00    1.0
2017-10-29 01:00:00+00:00    2.0
2017-10-29 02:00:00+00:00    3.0
48
49
50
""", back)

    assert (ts.index == back.index).all()
51
    assert str(back.index.dtype) == 'datetime64[ns, UTC]'
52
53


54
def test_differential(engine, tsh):
55
    ts_begin = genserie(datetime(2010, 1, 1), 'D', 10)
56
    tsh.insert(engine, ts_begin, 'ts_test', 'test')
57

58
    id1 = tsh.last_id(engine, 'ts_test')
59
    assert tsh._previous_cset(engine, 'ts_test', id1) is None
60

61
62
    assert tsh.exists(engine, 'ts_test')
    assert not tsh.exists(engine, 'this_does_not_exist')
63

64
    assert_df("""
65
66
67
68
69
70
71
72
73
74
2010-01-01    0.0
2010-01-02    1.0
2010-01-03    2.0
2010-01-04    3.0
2010-01-05    4.0
2010-01-06    5.0
2010-01-07    6.0
2010-01-08    7.0
2010-01-09    8.0
2010-01-10    9.0
75
""", tsh.get(engine, 'ts_test'))
76
77

    # we should detect the emission of a message
78
    tsh.insert(engine, ts_begin, 'ts_test', 'babar')
79

80
    assert_df("""
81
82
83
84
85
86
87
88
89
90
2010-01-01    0.0
2010-01-02    1.0
2010-01-03    2.0
2010-01-04    3.0
2010-01-05    4.0
2010-01-06    5.0
2010-01-07    6.0
2010-01-08    7.0
2010-01-09    8.0
2010-01-10    9.0
91
""", tsh.get(engine, 'ts_test'))
92
93
94
95

    ts_slight_variation = ts_begin.copy()
    ts_slight_variation.iloc[3] = 0
    ts_slight_variation.iloc[6] = 0
96
    tsh.insert(engine, ts_slight_variation, 'ts_test', 'celeste')
97
    id2 = tsh.last_id(engine, 'ts_test')
98
    assert tsh._previous_cset(engine, 'ts_test', id2) == id1
99

100
    assert_df("""
101
102
103
104
105
106
107
108
109
110
2010-01-01    0.0
2010-01-02    1.0
2010-01-03    2.0
2010-01-04    0.0
2010-01-05    4.0
2010-01-06    5.0
2010-01-07    0.0
2010-01-08    7.0
2010-01-09    8.0
2010-01-10    9.0
111
""", tsh.get(engine, 'ts_test'))
112

113
    ts_longer = genserie(datetime(2010, 1, 3), 'D', 15)
114
115
116
117
    ts_longer.iloc[1] = 2.48
    ts_longer.iloc[3] = 3.14
    ts_longer.iloc[5] = ts_begin.iloc[7]

118
    tsh.insert(engine, ts_longer, 'ts_test', 'test')
119
120
121
    id3 = tsh.last_id(engine, 'ts_test')

    assert id1 < id2 < id3
122

123
    assert_df("""
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
2010-01-01     0.00
2010-01-02     1.00
2010-01-03     0.00
2010-01-04     2.48
2010-01-05     2.00
2010-01-06     3.14
2010-01-07     4.00
2010-01-08     7.00
2010-01-09     6.00
2010-01-10     7.00
2010-01-11     8.00
2010-01-12     9.00
2010-01-13    10.00
2010-01-14    11.00
2010-01-15    12.00
2010-01-16    13.00
2010-01-17    14.00
141
""", tsh.get(engine, 'ts_test'))
142
143

    # start testing manual overrides
144
    ts_begin = genserie(datetime(2010, 1, 1), 'D', 5, initval=[2])
145
    ts_begin.loc['2010-01-04'] = -1
146
    tsh.insert(engine, ts_begin, 'ts_mixte', 'test')
147
148

    # -1 represents bogus upstream data
149
    assert_df("""
150
151
152
153
154
2010-01-01    2.0
2010-01-02    2.0
2010-01-03    2.0
2010-01-04   -1.0
2010-01-05    2.0
155
""", tsh.get(engine, 'ts_mixte'))
156
157

    # refresh all the period + 1 extra data point
158
    ts_more = genserie(datetime(2010, 1, 2), 'D', 5, [2])
159
    ts_more.loc['2010-01-04'] = -1
160
    tsh.insert(engine, ts_more, 'ts_mixte', 'test')
161

162
    assert_df("""
163
164
165
166
167
168
2010-01-01    2.0
2010-01-02    2.0
2010-01-03    2.0
2010-01-04   -1.0
2010-01-05    2.0
2010-01-06    2.0
169
""", tsh.get(engine, 'ts_mixte'))
170
171

    # just append an extra data point
172
173
    # with no intersection with the previous ts
    ts_one_more = genserie(datetime(2010, 1, 7), 'D', 1, [3])
174
    tsh.insert(engine, ts_one_more, 'ts_mixte', 'test')
175

176
    assert_df("""
177
178
179
180
181
182
183
2010-01-01    2.0
2010-01-02    2.0
2010-01-03    2.0
2010-01-04   -1.0
2010-01-05    2.0
2010-01-06    2.0
2010-01-07    3.0
184
""", tsh.get(engine, 'ts_mixte'))
185

186
    with engine.connect() as cn:
187
        cn.execute('set search_path to "{0}.timeserie", {0}, public'.format(tsh.namespace))
188
189
        allts = pd.read_sql("select seriename, table_name from registry "
                            "where seriename in ('ts_test', 'ts_mixte')",
190
                            cn)
191

192
        assert_df("""
193
194
195
seriename table_name
0   ts_test    ts_test
1  ts_mixte   ts_mixte
196
""".format(tsh.namespace), allts)
197

198
        assert_df("""
199
200
201
202
203
204
205
2010-01-01    2.0
2010-01-02    2.0
2010-01-03    2.0
2010-01-04   -1.0
2010-01-05    2.0
2010-01-06    2.0
2010-01-07    3.0
206
""", tsh.get(cn, 'ts_mixte',
207
             revision_date=datetime.now()))
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
def test_serie_metadata(engine, tsh):
    serie = genserie(datetime(2010, 1, 1), 'D', 1, initval=[1])
    tsh.insert(engine, serie, 'ts-metadata', 'babar')

    initialmeta = tsh.metadata(engine, 'ts-metadata')
    assert initialmeta == {
        'tzaware': False,
        'index_type': 'datetime64[ns]',
        'value_type': 'float64',
        'index_names': []
    }

    tsh.update_metadata(engine, 'ts-metadata',
                        {'topic': 'banana spot price'}
    )
    assert tsh.metadata(engine, 'ts-metadata')['topic'] == 'banana spot price'

    with pytest.raises(AssertionError):
        tsh.update_metadata(engine, 'ts-metadata', {'tzaware': True})

    tsh.update_metadata(engine, 'ts-metadata', {'tzaware': True}, internal=True)
    assert tsh.metadata(engine, 'ts-metadata') == {
        'tzaware': True,
        'index_type': 'datetime64[ns]',
        'value_type': 'float64',
        'index_names': [],
        'topic': 'banana spot price'
    }
    # unbreak the serie for the second test pass :o
    tsh.update_metadata(engine, 'ts-metadata', initialmeta, internal=True)


242
243
def test_changeset_metadata(engine, tsh):
    serie = genserie(datetime(2010, 1, 1), 'D', 1, initval=[1])
244
    tsh.insert(engine, serie, 'ts-cs-metadata', 'babar',
245
246
               {'foo': 'A', 'bar': 42})

247
    log = tsh.log(engine, names=['ts-cs-metadata'])
248
249
250
251
    meta = tsh.changeset_metadata(engine, log[0]['rev'])
    assert meta == {'foo': 'A', 'bar': 42}


252
def test_bad_import(engine, tsh):
253
    # the data were parsed as date by pd.read_json()
Aurélien Campéas's avatar
Aurélien Campéas committed
254
    df_result = pd.read_csv(str(DATADIR / 'test_data.csv'))
255
256
257
    df_result['Gas Day'] = df_result['Gas Day'].apply(parser.parse, dayfirst=True, yearfirst=False)
    df_result.set_index('Gas Day', inplace=True)
    ts = df_result['SC']
258

259
260
    tsh.insert(engine, ts, 'SND_SC', 'test')
    result = tsh.get(engine, 'SND_SC')
261
    assert result.dtype == 'float64'
262
263
264

    # insertion of empty ts
    ts = pd.Series(name='truc', dtype='object')
265
266
    tsh.insert(engine, ts, 'empty_ts', 'test')
    assert tsh.get(engine, 'empty_ts') is None
267
268
269

    # nan in ts
    # all na
270
    ts = genserie(datetime(2010, 1, 10), 'D', 10, [np.nan], name='truc')
271
    tsh.insert(engine, ts, 'test_nan', 'test')
272
273
    assert len(tsh.get(engine, 'test_nan')) == 0
    assert len(tsh.get(engine, 'test_nan', _keep_nans=True)) == 10
274
275
276
277
278

    # mixe na
    ts = pd.Series([np.nan] * 5 + [3] * 5,
                   index=pd.date_range(start=datetime(2010, 1, 10),
                                       freq='D', periods=10), name='truc')
279
280
    tsh.insert(engine, ts, 'test_nan', 'test')
    result = tsh.get(engine, 'test_nan')
281

282
283
    tsh.insert(engine, ts, 'test_nan', 'test')
    result = tsh.get(engine, 'test_nan')
284
    assert_df("""
285
286
287
288
289
2010-01-15    3.0
2010-01-16    3.0
2010-01-17    3.0
2010-01-18    3.0
2010-01-19    3.0
290
""", result)
291
292

    # get_ts with name not in database
293
    assert tsh.get(engine, 'inexisting_name', 'test') is None
294
295


296
def test_revision_date(engine, tsh):
297
    # we prepare a good joke for the end of the test
298
299
    # ival = Snapshot._interval
    # Snapshot._interval = 3
300
301
302

    for i in range(1, 5):
        with engine.connect() as cn:
303
304
            tsh.insert(cn, genserie(datetime(2017, 1, i), 'D', 3, [i]), 'revdate',
                       'test', _insertion_date=utcdt(2016, 1, i))
305
306

    # end of prologue, now some real meat
307
    idate0 = pd.Timestamp('2015-1-1 00:00:00', tz='UTC')
308
309
310
311
    ts = genserie(datetime(2010, 1, 4), 'D', 4, [0], name='truc')
    tsh.insert(engine, ts, 'ts_through_time',
               'test', _insertion_date=idate0)
    assert idate0 == tsh.latest_insertion_date(engine, 'ts_through_time')
312

313
    idate1 = pd.Timestamp('2015-1-1 15:45:23', tz='UTC')
314
315
316
317
    ts = genserie(datetime(2010, 1, 4), 'D', 4, [1], name='truc')
    tsh.insert(engine, ts, 'ts_through_time',
               'test', _insertion_date=idate1)
    assert idate1 == tsh.latest_insertion_date(engine, 'ts_through_time')
318

319
    idate2 = pd.Timestamp('2015-1-2 15:43:23', tz='UTC')
320
321
322
323
    ts = genserie(datetime(2010, 1, 4), 'D', 4, [2], name='truc')
    tsh.insert(engine, ts, 'ts_through_time',
               'test', _insertion_date=idate2)
    assert idate2 == tsh.latest_insertion_date(engine, 'ts_through_time')
324

325
    idate3 = pd.Timestamp('2015-1-3', tz='UTC')
326
327
328
329
    ts = genserie(datetime(2010, 1, 4), 'D', 4, [3], name='truc')
    tsh.insert(engine, ts, 'ts_through_time',
               'test', _insertion_date=idate3)
    assert idate3 == tsh.latest_insertion_date(engine, 'ts_through_time')
330

331
    ts = tsh.get(engine, 'ts_through_time')
332

333
    assert_df("""
334
335
336
337
2010-01-04    3.0
2010-01-05    3.0
2010-01-06    3.0
2010-01-07    3.0
338
""", ts)
339

340
    ts = tsh.get(engine, 'ts_through_time',
Aurélien Campéas's avatar
Aurélien Campéas committed
341
                 revision_date=datetime(2015, 1, 2, 18, 43, 23))
342

343
    assert_df("""
344
345
346
347
2010-01-04    2.0
2010-01-05    2.0
2010-01-06    2.0
2010-01-07    2.0
348
""", ts)
349

350
    ts = tsh.get(engine, 'ts_through_time',
351
                 revision_date=datetime(2015, 1, 1, 18, 43, 23))
352

353
    assert_df("""
354
355
356
357
2010-01-04    1.0
2010-01-05    1.0
2010-01-06    1.0
2010-01-07    1.0
358
""", ts)
359

360
    ts = tsh.get(engine, 'ts_through_time',
361
                 revision_date=datetime(2014, 1, 1, 18, 43, 23))
362
363
364

    assert ts is None

365
366
367
368
369
370
371
372
373
374
    # epilogue: back to the revdate issue
    assert_df("""
2017-01-01    1.0
2017-01-02    2.0
2017-01-03    3.0
2017-01-04    4.0
2017-01-05    4.0
2017-01-06    4.0
""", tsh.get(engine, 'revdate'))

375
    oldstate = tsh.get(engine, 'revdate', revision_date=datetime(2016, 1, 2))
376
377
378
    assert_df("""
2017-01-01    1.0
2017-01-02    2.0
379
380
381
2017-01-03    2.0
2017-01-04    2.0
""", oldstate)
382

383

384
def test_deletion(engine, tsh):
385
386
    ts_begin = genserie(datetime(2010, 1, 1), 'D', 11)
    ts_begin.iloc[-1] = np.nan
387
    tsh.insert(engine, ts_begin, 'ts_del', 'test')
388

389
    _, ts = Snapshot(engine, tsh, 'ts_del').find()
390
    assert ts.iloc[-2] == 9.0
391

392
    ts_begin.iloc[0] = np.nan
393
    ts_begin.iloc[3] = np.nan
394

395
    tsh.insert(engine, ts_begin, 'ts_del', 'test')
396

397
    assert_df("""
398
399
400
401
402
403
404
2010-01-02    1.0
2010-01-03    2.0
2010-01-05    4.0
2010-01-06    5.0
2010-01-07    6.0
2010-01-08    7.0
2010-01-09    8.0
405
2010-01-10    9.0
406
""", tsh.get(engine, 'ts_del'))
407

408
    ts2 = tsh.get(engine, 'ts_del',
409
410
                 # force snapshot reconstruction feature
                 revision_date=datetime(2038, 1, 1))
411
    assert (tsh.get(engine, 'ts_del') == ts2).all()
412

413
414
415
    ts_begin.iloc[0] = 42
    ts_begin.iloc[3] = 23

416
    tsh.insert(engine, ts_begin, 'ts_del', 'test')
417

418
    assert_df("""
419
420
421
422
423
424
425
426
427
2010-01-01    42.0
2010-01-02     1.0
2010-01-03     2.0
2010-01-04    23.0
2010-01-05     4.0
2010-01-06     5.0
2010-01-07     6.0
2010-01-08     7.0
2010-01-09     8.0
428
2010-01-10     9.0
429
""", tsh.get(engine, 'ts_del'))
430
431
432

    # now with string!

433
    ts_string = genserie(datetime(2010, 1, 1), 'D', 10, ['machin'])
434
    tsh.insert(engine, ts_string, 'ts_string_del', 'test')
435
436
437
438

    ts_string[4] = None
    ts_string[5] = None

439
    tsh.insert(engine, ts_string, 'ts_string_del', 'test')
440
    assert_df("""
441
442
443
444
445
446
447
448
2010-01-01    machin
2010-01-02    machin
2010-01-03    machin
2010-01-04    machin
2010-01-07    machin
2010-01-08    machin
2010-01-09    machin
2010-01-10    machin
449
""", tsh.get(engine, 'ts_string_del'))
450
451
452
453

    ts_string[4] = 'truc'
    ts_string[6] = 'truc'

454
    tsh.insert(engine, ts_string, 'ts_string_del', 'test')
455
    assert_df("""
456
457
458
459
460
461
462
463
2010-01-01    machin
2010-01-02    machin
2010-01-03    machin
2010-01-04    machin
2010-01-05      truc
2010-01-07      truc
2010-01-08    machin
2010-01-09    machin
464
2010-01-10    machin
465
""", tsh.get(engine, 'ts_string_del'))
466

467
    ts_string[ts_string.index] = np.nan
468
    tsh.insert(engine, ts_string, 'ts_string_del', 'test')
469

470
    erased = tsh.get(engine, 'ts_string_del')
471
472
    assert len(erased) == 0

473
474
    # first insertion with only nan

475
    ts_begin = genserie(datetime(2010, 1, 1), 'D', 10, [np.nan])
476
    tsh.insert(engine, ts_begin, 'ts_null', 'test')
477

478
    assert len(tsh.get(engine, 'ts_null')) == 0
479

480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
    # exhibit issue with nans handling
    ts_repushed = genserie(datetime(2010, 1, 1), 'D', 11)
    ts_repushed[0:3] = np.nan

    assert_df("""
2010-01-01     NaN
2010-01-02     NaN
2010-01-03     NaN
2010-01-04     3.0
2010-01-05     4.0
2010-01-06     5.0
2010-01-07     6.0
2010-01-08     7.0
2010-01-09     8.0
2010-01-10     9.0
2010-01-11    10.0
Freq: D
""", ts_repushed)

499
500
    tsh.insert(engine, ts_repushed, 'ts_repushed', 'test')
    diff = tsh.insert(engine, ts_repushed, 'ts_repushed', 'test')
501
502
    assert diff is None

503
    # there is no difference
504
    assert 0 == len(tsh.diff(ts_repushed, ts_repushed))
505
506
507
508
509

    ts_add = genserie(datetime(2010, 1, 1), 'D', 15)
    ts_add.iloc[0] = np.nan
    ts_add.iloc[13:] = np.nan
    ts_add.iloc[8] = np.nan
510
    diff = tsh.diff(ts_repushed, ts_add)
511
512
513
514
515
516

    assert_df("""
2010-01-02     1.0
2010-01-03     2.0
2010-01-09     NaN
2010-01-12    11.0
517
2010-01-13    12.0""", diff.sort_index())
518
519
520
521
    # value on nan => value
    # nan on value => nan
    # nan on nan => Nothing
    # nan on nothing=> Nothing
522

Aurélien Campéas's avatar
Aurélien Campéas committed
523
    # full erasing
524
525
    # numeric
    ts_begin = genserie(datetime(2010, 1, 1), 'D', 4)
526
    tsh.insert(engine, ts_begin, 'ts_full_del', 'test')
527

Aurélien Campéas's avatar
Aurélien Campéas committed
528
    ts_begin.iloc[:] = np.nan
529
    tsh.insert(engine, ts_begin, 'ts_full_del', 'test')
530
531

    ts_end = genserie(datetime(2010, 1, 1), 'D', 4)
532
    tsh.insert(engine, ts_end, 'ts_full_del', 'test')
533
534
535
536

    # string

    ts_begin = genserie(datetime(2010, 1, 1), 'D', 4, ['text'])
537
    tsh.insert(engine, ts_begin, 'ts_full_del_str', 'test')
538
539

    ts_begin.iloc[:] = np.nan
540
    tsh.insert(engine, ts_begin, 'ts_full_del_str', 'test')
541
542

    ts_end = genserie(datetime(2010, 1, 1), 'D', 4, ['text'])
543
    tsh.insert(engine, ts_end, 'ts_full_del_str', 'test')
544

Aurélien Campéas's avatar
Aurélien Campéas committed
545

546
def test_get_history(engine, tsh):
547
548
    for numserie in (1, 2, 3):
        with engine.connect() as cn:
549
550
551
            tsh.insert(cn, genserie(datetime(2017, 1, 1), 'D', numserie), 'smallserie',
                       'aurelien.campeas@pythonian.fr',
                       _insertion_date=utcdt(2017, 2, numserie))
552
553
554
555
556
557
558
559
560
561
562

    ts = tsh.get(engine, 'smallserie')
    assert_df("""
2017-01-01    0.0
2017-01-02    1.0
2017-01-03    2.0
""", ts)

    logs = tsh.log(engine, names=['smallserie'])
    assert [
        {'author': 'aurelien.campeas@pythonian.fr',
563
         'meta': {},
564
         'date': pd.Timestamp('2017-02-01 00:00:00+0000', tz='UTC'),
565
566
567
         'names': ['smallserie']
        },
        {'author': 'aurelien.campeas@pythonian.fr',
568
         'meta': {},
569
         'date': pd.Timestamp('2017-02-02 00:00:00+0000', tz='UTC'),
570
571
572
         'names': ['smallserie']
        },
        {'author': 'aurelien.campeas@pythonian.fr',
573
         'meta': {},
574
         'date': pd.Timestamp('2017-02-03 00:00:00+0000', tz='UTC'),
575
576
577
578
579
         'names': ['smallserie']
        }
    ] == [{k: v for k, v in log.items() if k != 'rev'}
          for log in logs]
    histts = tsh.get_history(engine, 'smallserie')
580
    assert histts.name == 'smallserie'
581
582

    assert_df("""
583
584
585
586
587
588
589
insertion_date             value_date
2017-02-01 00:00:00+00:00  2017-01-01    0.0
2017-02-02 00:00:00+00:00  2017-01-01    0.0
                           2017-01-02    1.0
2017-02-03 00:00:00+00:00  2017-01-01    0.0
                           2017-01-02    1.0
                           2017-01-03    2.0
590
591
""", histts)

592
593
594
595
596
597
598
599
    diffs = tsh.get_history(engine, 'smallserie', diffmode=True)
    assert_df("""
insertion_date             value_date
2017-02-01 00:00:00+00:00  2017-01-01    0.0
2017-02-02 00:00:00+00:00  2017-01-02    1.0
2017-02-03 00:00:00+00:00  2017-01-03    2.0
""", diffs)

600
    for idate in histts.index.get_level_values('insertion_date').unique():
601
        with engine.connect() as cn:
602
            idate = idate.replace(tzinfo=pytz.timezone('UTC'))
603
604
            tsh.insert(cn, histts[idate], 'smallserie2',
                       'aurelien.campeas@pythonian.f', _insertion_date=idate)
605
606
607
608
609
610
611
612
613

    # this is perfectly round-tripable
    assert (tsh.get(engine, 'smallserie2') == ts).all()
    assert (tsh.get_history(engine, 'smallserie2') == histts).all()

    # get history ranges
    tsa = tsh.get_history(engine, 'smallserie',
                          from_insertion_date=datetime(2017, 2, 2))
    assert_df("""
614
615
616
617
618
619
insertion_date             value_date
2017-02-02 00:00:00+00:00  2017-01-01    0.0
                           2017-01-02    1.0
2017-02-03 00:00:00+00:00  2017-01-01    0.0
                           2017-01-02    1.0
                           2017-01-03    2.0
620
621
622
623
624
""", tsa)

    tsb = tsh.get_history(engine, 'smallserie',
                          to_insertion_date=datetime(2017, 2, 2))
    assert_df("""
625
626
627
628
insertion_date             value_date
2017-02-01 00:00:00+00:00  2017-01-01    0.0
2017-02-02 00:00:00+00:00  2017-01-01    0.0
                           2017-01-02    1.0
629
630
631
632
633
634
""", tsb)

    tsc = tsh.get_history(engine, 'smallserie',
                          from_insertion_date=datetime(2017, 2, 2),
                          to_insertion_date=datetime(2017, 2, 2))
    assert_df("""
635
636
637
insertion_date             value_date
2017-02-02 00:00:00+00:00  2017-01-01    0.0
                           2017-01-02    1.0
638
639
""", tsc)

640
641
642
643
644
645
646
647
648
    tsc = tsh.get_history(engine, 'smallserie',
                          from_insertion_date=datetime(2017, 2, 4),
                          to_insertion_date=datetime(2017, 2, 4))
    assert tsc is None

    tsc = tsh.get_history(engine, 'smallserie',
                          from_insertion_date=datetime(2016, 2, 1),
                          to_insertion_date=datetime(2017, 2, 2))
    assert_df("""
649
650
651
652
insertion_date             value_date
2017-02-01 00:00:00+00:00  2017-01-01    0.0
2017-02-02 00:00:00+00:00  2017-01-01    0.0
                           2017-01-02    1.0
653
654
655
656
657
658
659
""", tsc)

    tsc = tsh.get_history(engine, 'smallserie',
                          from_insertion_date=datetime(2016, 2, 1),
                          to_insertion_date=datetime(2016, 12, 31))
    assert tsc is None

660
661
662
663
664
    # restrictions on value dates
    tsc = tsh.get_history(engine, 'smallserie',
                          from_value_date=datetime(2017, 1, 1),
                          to_value_date=datetime(2017, 1, 2))
    assert_df("""
665
666
667
668
669
670
insertion_date             value_date
2017-02-01 00:00:00+00:00  2017-01-01    0.0
2017-02-02 00:00:00+00:00  2017-01-01    0.0
                           2017-01-02    1.0
2017-02-03 00:00:00+00:00  2017-01-01    0.0
                           2017-01-02    1.0
671
672
673
674
675
""", tsc)

    tsc = tsh.get_history(engine, 'smallserie',
                          from_value_date=datetime(2017, 1, 2))
    assert_df("""
676
677
678
679
insertion_date             value_date
2017-02-02 00:00:00+00:00  2017-01-02    1.0
2017-02-03 00:00:00+00:00  2017-01-02    1.0
                           2017-01-03    2.0
680
681
682
683
684
""", tsc)

    tsc = tsh.get_history(engine, 'smallserie',
                          to_value_date=datetime(2017, 1, 2))
    assert_df("""
685
686
687
688
689
690
insertion_date             value_date
2017-02-01 00:00:00+00:00  2017-01-01    0.0
2017-02-02 00:00:00+00:00  2017-01-01    0.0
                           2017-01-02    1.0
2017-02-03 00:00:00+00:00  2017-01-01    0.0
                           2017-01-02    1.0
691
692
""", tsc)

693

694
695
696
697
698
699
700
701
702
def test_history_delta(engine, tsh):
    for d in range(1, 3):
        idate = utcdt(2018, 1, d)
        serie = genserie(idate - timedelta(hours=1), 'H', 6, initval=[d])
        tsh.insert(engine, serie, 'hd', 'aurelien.campeas@pythonian.fr',
                   _insertion_date=idate)

    assert_df("""
insertion_date             value_date               
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
2018-01-01 00:00:00+00:00  2017-12-31 23:00:00+00:00    1.0
                           2018-01-01 00:00:00+00:00    1.0
                           2018-01-01 01:00:00+00:00    1.0
                           2018-01-01 02:00:00+00:00    1.0
                           2018-01-01 03:00:00+00:00    1.0
                           2018-01-01 04:00:00+00:00    1.0
2018-01-02 00:00:00+00:00  2017-12-31 23:00:00+00:00    1.0
                           2018-01-01 00:00:00+00:00    1.0
                           2018-01-01 01:00:00+00:00    1.0
                           2018-01-01 02:00:00+00:00    1.0
                           2018-01-01 03:00:00+00:00    1.0
                           2018-01-01 04:00:00+00:00    1.0
                           2018-01-01 23:00:00+00:00    2.0
                           2018-01-02 00:00:00+00:00    2.0
                           2018-01-02 01:00:00+00:00    2.0
                           2018-01-02 02:00:00+00:00    2.0
                           2018-01-02 03:00:00+00:00    2.0
                           2018-01-02 04:00:00+00:00    2.0
    """, tsh.get_history(engine, 'hd'))

    assert_df("""
insertion_date             value_date               
725
726
2018-01-01 00:00:00+00:00  2017-12-31 23:00:00+00:00    1.0
                           2018-01-01 00:00:00+00:00    1.0
727
728
                           2018-01-01 01:00:00+00:00    1.0
                           2018-01-01 02:00:00+00:00    1.0
729
730
731
732
733
734
735
736
2018-01-02 00:00:00+00:00  2017-12-31 23:00:00+00:00    1.0
                           2018-01-01 00:00:00+00:00    1.0
                           2018-01-01 01:00:00+00:00    1.0
                           2018-01-01 02:00:00+00:00    1.0
                           2018-01-01 03:00:00+00:00    1.0
                           2018-01-01 04:00:00+00:00    1.0
                           2018-01-01 23:00:00+00:00    2.0
                           2018-01-02 00:00:00+00:00    2.0
737
738
739
740
741
742
                           2018-01-02 01:00:00+00:00    2.0
                           2018-01-02 02:00:00+00:00    2.0
""",  tsh.get_history(engine, 'hd', deltaafter=timedelta(hours=2)))

    assert_df("""
insertion_date             value_date               
743
2018-01-01 00:00:00+00:00  2018-01-01 00:00:00+00:00    1.0
744
                           2018-01-01 01:00:00+00:00    1.0
745
2018-01-02 00:00:00+00:00  2018-01-02 00:00:00+00:00    2.0
746
747
                           2018-01-02 01:00:00+00:00    2.0
""",  tsh.get_history(engine, 'hd',
748
                      deltabefore=timedelta(hours=0),
749
750
751
                      deltaafter=timedelta(hours=1)))


752
753
754
755
756
757
758
759
760
761
762
def test_nr_gethistory(engine, tsh):
    s0 = pd.Series([-1, 0, 0, -1],
                   index=pd.DatetimeIndex(start=datetime(2016, 12, 29),
                                          end=datetime(2017, 1, 1),
                                          freq='D'))
    tsh.insert(engine, s0, 'foo', 'zogzog')

    s1 = pd.Series([1, 0, 0, 1],
                   index=pd.DatetimeIndex(start=datetime(2017, 1, 1),
                                          end=datetime(2017, 1, 4),
                                          freq='D'))
763
    idate = utcdt(2016, 1, 1)
764
765
    for i in range(5):
        with engine.connect() as cn:
766
767
768
            tsh.insert(cn, s1 * i, 'foo',
                       'aurelien.campeas@pythonian.f',
                       _insertion_date=idate + timedelta(days=i))
769
770
771
772
773
774
775
776

    df = tsh.get_history(engine, 'foo',
                         datetime(2016, 1, 3),
                         datetime(2016, 1, 4),
                         datetime(2017, 1, 1),
                         datetime(2017, 1, 4))

    assert_df("""
777
778
779
780
781
782
783
784
785
insertion_date             value_date
2016-01-03 00:00:00+00:00  2017-01-01    2.0
                           2017-01-02    0.0
                           2017-01-03    0.0
                           2017-01-04    2.0
2016-01-04 00:00:00+00:00  2017-01-01    3.0
                           2017-01-02    0.0
                           2017-01-03    0.0
                           2017-01-04    3.0
786
787
788
""", df)


789
def test_add_na(engine, tsh):
790
791
792
793
794
    # a serie of NaNs won't be insert in base
    # in case of first insertion
    ts_nan = genserie(datetime(2010, 1, 1), 'D', 5)
    ts_nan[[True] * len(ts_nan)] = np.nan

795
    diff = tsh.insert(engine, ts_nan, 'ts_add_na', 'test')
796
    assert len(diff) == 5
797
    result = tsh.get(engine, 'ts_add_na')
798
799
800
801
802
803
804
805
806
807
    assert len(result) == 0

    result = tsh.get(engine, 'ts_add_na', _keep_nans=True)
    assert_df("""
2010-01-01   NaN
2010-01-02   NaN
2010-01-03   NaN
2010-01-04   NaN
2010-01-05   NaN
""", result)
808
809
810

    # in case of insertion in existing data
    ts_begin = genserie(datetime(2010, 1, 1), 'D', 5)
811
    tsh.insert(engine, ts_begin, 'ts_add_na', 'test')
812
813
814
815
816

    ts_nan = genserie(datetime(2010, 1, 6), 'D', 5)
    ts_nan[[True] * len(ts_nan)] = np.nan
    ts_nan = pd.concat([ts_begin, ts_nan])

817
    diff = tsh.insert(engine, ts_nan, 'ts_add_na', 'test')
818
819
    assert diff is None

820
    result = tsh.get(engine, 'ts_add_na')
821
    assert len(result) == 5
822
823


824
def test_dtype_mismatch(engine, tsh):
825
    tsh.insert(engine,
826
827
828
829
830
               genserie(datetime(2015, 1, 1), 'D', 11).astype('str'),
               'error1',
               'test')

    with pytest.raises(Exception) as excinfo:
831
        tsh.insert(engine,
832
833
834
835
836
837
                   genserie(datetime(2015, 1, 1), 'D', 11),
                   'error1',
                   'test')

    assert 'Type error when inserting error1, new type is float64, type in base is object' == str(excinfo.value)

838
    tsh.insert(engine,
839
840
841
842
843
               genserie(datetime(2015, 1, 1), 'D', 11),
               'error2',
               'test')

    with pytest.raises(Exception) as excinfo:
844
        tsh.insert(engine,
845
846
847
848
849
                   genserie(datetime(2015, 1, 1), 'D', 11).astype('str'),
                   'error2',
                   'test')

    assert 'Type error when inserting error2, new type is object, type in base is float64' == str(excinfo.value)
850
851


852
853
854
855
856
857
858
859
860
def test_precision(engine, tsh):
    floaty = 0.123456789123456789
    ts = genserie(datetime(2015, 1, 1), 'D', 5, initval=[floaty])

    tsh.insert(engine, ts, 'precision', 'test')
    ts_round = tsh.get(engine, 'precision')
    assert 0.12345678912346 == ts_round.iloc[0]

    diff = tsh.insert(engine, ts_round, 'precision', 'test')
Aurélien Campéas's avatar
Aurélien Campéas committed
861
    assert diff is None  # the roundtriped series does not produce a diff when reinserted
862

Aurélien Campéas's avatar
Aurélien Campéas committed
863
    diff = tsh.insert(engine, ts, 'precision', 'test')  # neither does the original series
864
865
866
    assert diff is None


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
def test_strip(engine, tsh):
    for i in range(1, 5):
        pubdate = utcdt(2017, 1, i)
        ts = genserie(datetime(2017, 1, 10), 'H', 1 + i)
        tsh.insert(engine, ts, 'xserie', 'babar', _insertion_date=pubdate)
        # also insert something completely unrelated
        tsh.insert(engine, genserie(datetime(2018, 1, 1), 'D', 1 + i), 'yserie', 'celeste')

    csida = tsh.changeset_at(engine, 'xserie', datetime(2017, 1, 3))
    assert csida is not None
    csidb = tsh.changeset_at(engine, 'xserie', datetime(2017, 1, 3, 1), mode='before')
    csidc = tsh.changeset_at(engine, 'xserie', datetime(2017, 1, 3, 1), mode='after')
    assert csidb < csida < csidc

    log = tsh.log(engine, names=['xserie', 'yserie'])
    assert [(idx, l['author']) for idx, l in enumerate(log, start=1)
    ] == [
        (1, 'babar'),
        (2, 'celeste'),
        (3, 'babar'),
        (4, 'celeste'),
        (5, 'babar'),
        (6, 'celeste'),
        (7, 'babar'),
        (8, 'celeste')
    ]

    h = tsh.get_history(engine, 'xserie')
    assert_df("""
insertion_date             value_date         
2017-01-01 00:00:00+00:00  2017-01-10 00:00:00    0.0
                           2017-01-10 01:00:00    1.0
2017-01-02 00:00:00+00:00  2017-01-10 00:00:00    0.0
                           2017-01-10 01:00:00    1.0
                           2017-01-10 02:00:00    2.0
2017-01-03 00:00:00+00:00  2017-01-10 00:00:00    0.0
                           2017-01-10 01:00:00    1.0
                           2017-01-10 02:00:00    2.0
                           2017-01-10 03:00:00    3.0
2017-01-04 00:00:00+00:00  2017-01-10 00:00:00    0.0
                           2017-01-10 01:00:00    1.0
                           2017-01-10 02:00:00    2.0
                           2017-01-10 03:00:00    3.0
                           2017-01-10 04:00:00    4.0
""", h)

    csid = tsh.changeset_at(engine, 'xserie', datetime(2017, 1, 3))
    with engine.connect() as cn:
        tsh.strip(cn, 'xserie', csid)

    assert_df("""
insertion_date             value_date         
2017-01-01 00:00:00+00:00  2017-01-10 00:00:00    0.0
                           2017-01-10 01:00:00    1.0
2017-01-02 00:00:00+00:00  2017-01-10 00:00:00    0.0
                           2017-01-10 01:00:00    1.0
                           2017-01-10 02:00:00    2.0
""", tsh.get_history(engine, 'xserie'))

    assert_df("""
2017-01-10 00:00:00    0.0
2017-01-10 01:00:00    1.0
2017-01-10 02:00:00    2.0
""", tsh.get(engine, 'xserie'))

    log = tsh.log(engine, names=['xserie', 'yserie'])
    # 5 and 7 have disappeared
    assert [l['author'] for l in log
    ] == ['babar', 'celeste', 'babar', 'celeste', 'celeste', 'celeste']

    log = tsh.log(engine, stripped=True, names=['xserie', 'yserie'])
938
939
940
941
942
943
    for l in log:
        if l['meta']:
            meta = l['meta']
            stripinfo = meta.get('tshistory.info')
            if stripinfo:
                assert stripinfo.startswith('got stripped from')
944
945


946
947
948
def test_long_name(engine, tsh):
    serie = genserie(datetime(2010, 1, 1), 'D', 40)

949
950
951
    name = 'a' * 64
    tsh.insert(engine, serie, name, 'babar')
    assert tsh.get(engine, name) is not None
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


def test_get_delta(engine, tsh):
    for idate in pd.DatetimeIndex(start=utcdt(2015, 1, 1),
                                  end=utcdt(2015, 1, 1, 3),
                                  freq='H'):
        ts = genserie(start=idate, freq='H', repeat=7)
        tsh.insert(engine, ts, 'republication', 'test',
                   _insertion_date=idate)

    hist = tsh.get_history(engine, 'republication')
    assert_df("""
insertion_date             value_date               
2015-01-01 00:00:00+00:00  2015-01-01 00:00:00+00:00    0.0
                           2015-01-01 01:00:00+00:00    1.0
                           2015-01-01 02:00:00+00:00    2.0
                           2015-01-01 03:00:00+00:00    3.0
                           2015-01-01 04:00:00+00:00    4.0
                           2015-01-01 05:00:00+00:00    5.0
                           2015-01-01 06:00:00+00:00    6.0
2015-01-01 01:00:00+00:00  2015-01-01 00:00:00+00:00    0.0
                           2015-01-01 01:00:00+00:00    0.0
                           2015-01-01 02:00:00+00:00    1.0
                           2015-01-01 03:00:00+00:00    2.0
                           2015-01-01 04:00:00+00:00    3.0
                           2015-01-01 05:00:00+00:00    4.0
                           2015-01-01 06:00:00+00:00    5.0
                           2015-01-01 07:00:00+00:00    6.0
2015-01-01 02:00:00+00:00  2015-01-01 00:00:00+00:00    0.0
                           2015-01-01 01:00:00+00:00    0.0
                           2015-01-01 02:00:00+00:00    0.0
                           2015-01-01 03:00:00+00:00    1.0
                           2015-01-01 04:00:00+00:00    2.0
                           2015-01-01 05:00:00+00:00    3.0
                           2015-01-01 06:00:00+00:00    4.0
                           2015-01-01 07:00:00+00:00    5.0
                           2015-01-01 08:00:00+00:00    6.0
2015-01-01 03:00:00+00:00  2015-01-01 00:00:00+00:00    0.0
                           2015-01-01 01:00:00+00:00    0.0
                           2015-01-01 02:00:00+00:00    0.0
                           2015-01-01 03:00:00+00:00    0.0
                           2015-01-01 04:00:00+00:00    1.0
                           2015-01-01 05:00:00+00:00    2.0
                           2015-01-01 06:00:00+00:00    3.0
                           2015-01-01 07:00:00+00:00    4.0
                           2015-01-01 08:00:00+00:00    5.0
                           2015-01-01 09:00:00+00:00    6.0
""", hist)

    deltas = tsh.get_delta(engine,  'republication', delta=timedelta(hours=3))

    assert_df("""
2015-01-01 03:00:00+00:00    3.0
2015-01-01 04:00:00+00:00    3.0
2015-01-01 05:00:00+00:00    3.0
2015-01-01 06:00:00+00:00    3.0
2015-01-01 07:00:00+00:00    4.0
2015-01-01 08:00:00+00:00    5.0
2015-01-01 09:00:00+00:00    6.0
""", deltas)

    deltas = tsh.get_delta(engine,  'republication', delta=timedelta(hours=5))
    assert_df("""
2015-01-01 05:00:00+00:00    5.0
2015-01-01 06:00:00+00:00    5.0
2015-01-01 07:00:00+00:00    5.0
2015-01-01 08:00:00+00:00    5.0
2015-01-01 09:00:00+00:00    6.0
""", deltas)

    hist = tsh.get_history(engine, 'republication',
                           deltabefore=-timedelta(hours=3),
                           deltaafter=timedelta(hours=3))
    assert_df("""
insertion_date             value_date               
2015-01-01 00:00:00+00:00  2015-01-01 03:00:00+00:00    3.0
2015-01-01 01:00:00+00:00  2015-01-01 04:00:00+00:00    3.0
2015-01-01 02:00:00+00:00  2015-01-01 05:00:00+00:00    3.0
2015-01-01 03:00:00+00:00  2015-01-01 06:00:00+00:00    3.0
""", hist)

    hist = tsh.get_history(engine, 'republication',
                           deltabefore=-timedelta(hours=5),
                           deltaafter=timedelta(hours=5))

    assert_df("""
insertion_date             value_date               
2015-01-01 00:00:00+00:00  2015-01-01 05:00:00+00:00    5.0
2015-01-01 01:00:00+00:00  2015-01-01 06:00:00+00:00    5.0
2015-01-01 02:00:00+00:00  2015-01-01 07:00:00+00:00    5.0
2015-01-01 03:00:00+00:00  2015-01-01 08:00:00+00:00    5.0
""", hist)


    # maybe a more interesting example, each days we insert 7 data points
    for idx, idate in enumerate(pd.DatetimeIndex(start=utcdt(2015, 1, 1),
                                                 end=utcdt(2015, 1, 4),
                                                 freq='D')):
        ts = genserie(start=idate, freq='H', repeat=7)
        tsh.insert(engine, ts, 'repu2', 'test', _insertion_date=idate)

    deltas = tsh.get_delta(engine, 'repu2', delta=timedelta(hours=3))
    assert_df("""
2015-01-01 03:00:00+00:00    3.0
2015-01-01 04:00:00+00:00    4.0
2015-01-01 05:00:00+00:00    5.0
2015-01-01 06:00:00+00:00    6.0
2015-01-02 03:00:00+00:00    3.0
2015-01-02 04:00:00+00:00    4.0
2015-01-02 05:00:00+00:00    5.0
2015-01-02 06:00:00+00:00    6.0
2015-01-03 03:00:00+00:00    3.0
2015-01-03 04:00:00+00:00    4.0
2015-01-03 05:00:00+00:00    5.0
2015-01-03 06:00:00+00:00    6.0
2015-01-04 03:00:00+00:00    3.0
2015-01-04 04:00:00+00:00    4.0
2015-01-04 05:00:00+00:00    5.0
2015-01-04 06:00:00+00:00    6.0
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
""", deltas)

    deltas = tsh.get_delta(engine, 'repu2', delta=timedelta(hours=3),
                           from_value_date=datetime(2015,1,2),
                           to_value_date=datetime(2015, 1,3))
    assert_df("""
2015-01-02 03:00:00+00:00    3.0
2015-01-02 04:00:00+00:00    4.0
2015-01-02 05:00:00+00:00    5.0
2015-01-02 06:00:00+00:00    6.0
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
""", deltas)

    # which is basically the same as below
    hist = tsh.get_history(engine, 'repu2',
                           deltabefore=-timedelta(hours=3))
    assert_df("""
insertion_date             value_date               
2015-01-01 00:00:00+00:00  2015-01-01 03:00:00+00:00    3.0
                           2015-01-01 04:00:00+00:00    4.0
                           2015-01-01 05:00:00+00:00    5.0
                           2015-01-01 06:00:00+00:00    6.0
2015-01-02 00:00:00+00:00  2015-01-02 03:00:00+00:00    3.0
                           2015-01-02 04:00:00+00:00    4.0
                           2015-01-02 05:00:00+00:00    5.0
                           2015-01-02 06:00:00+00:00    6.0
2015-01-03 00:00:00+00:00  2015-01-03 03:00:00+00:00    3.0
                           2015-01-03 04:00:00+00:00    4.0
                           2015-01-03 05:00:00+00:00    5.0
                           2015-01-03 06:00:00+00:00    6.0
2015-01-04 00:00:00+00:00  2015-01-04 03:00:00+00:00    3.0
                           2015-01-04 04:00:00+00:00    4.0
                           2015-01-04 05:00:00+00:00    5.0
                           2015-01-04 06:00:00+00:00    6.0
""", hist)
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128


def test_rename(engine, tsh):
    if tsh.namespace == 'zzz':
        return  # this test can only run once

    serie = genserie(datetime(2020, 1, 1), 'D', 3)

    tsh.insert(engine, serie, 'foo', 'Babar')
    tsh.insert(engine, serie, 'bar', 'Babar')
    tsh.insert(engine, serie, 'quux', 'Babar')

    rename_series(engine, {
        'foo': 'new-foo',
        'bar': 'new-bar'
    })

    tsh.resetcaches()

    assert tsh.get(engine, 'foo') is None
    assert tsh.get(engine, 'bar') is None

    for name in ('quux', 'new-foo', 'new-bar'):
        assert tsh.get(engine, name) is not None