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
|
"""
test_overwrite.py
Tests for overwriting parquet files.
"""
import os
import pytest
import pandas as pd
from fastparquet import write, ParquetFile
from fastparquet.writer import overwrite
from .util import tempdir
def test_single_part_in_partitions(tempdir):
# Step 1 - Writing of a 1st df, with `row_group_offsets=0`,
# `file_scheme=hive` and `partition_on=['location', 'color`].
# 'location/color' keys (used in this test for partitioning) are
# intentionally in sorted order, as partitioning in fastparquet rely on
# pandas groupby with std settings, which result in sorting keys for
# grouping. This way, that this setting changes or not, this test case is
# not impacted anyway.
df1 = pd.DataFrame({'humidity': [0.9, 0.8, 0.93],
'pressure': [0.95e5, 1.1e5, 1e5],
'location': ['Milan', 'Paris', 'Paris'],
'color': ['blue', 'black', 'red']})
write(tempdir, df1, row_group_offsets=0, file_scheme='hive',
partition_on=['location', 'color'])
# Step 2 - Overwriting with a 2nd df having overlapping data.
df2 = pd.DataFrame({
'humidity': [0.5, 0.3, 0.4, 0.8, 1.1],
'pressure': [9e4, 1e5, 1.1e5, 1.1e5, 0.95e5],
'location': ['Milan', 'Paris', 'Paris', 'Paris', 'Paris'],
'color': ['red', 'black', 'black', 'green', 'green' ]})
overwrite(tempdir, df2, row_group_offsets=0)
recorded = ParquetFile(tempdir).to_pandas()
expected = pd.DataFrame({
'humidity': [0.9, 0.3, 0.4, 0.93, 0.5, 0.8, 1.1],
'pressure': [9.5e4, 1e5, 1.1e5, 1e5, 9e4, 1.1e5, 9.5e4],
'location': ['Milan', 'Paris', 'Paris', 'Paris', 'Milan', 'Paris', 'Paris'],
'color': ['blue', 'black', 'black', 'red', 'red', 'green', 'green']
})
expected = expected.astype({'location': 'category', 'color': 'category'})
# df1 is 3 rows, df2 is 5 rows. Because of overlapping data with keys
# 'location' = 'Paris' & 'color' = 'black' (1 row in df1, 2 rows in df2)
# resulting df contains for this combination:
# - values from df2
# - and not that of df1.
# Total resulting number of rows is 7.
assert expected.equals(recorded)
def test_multiple_parts_in_partitions(tempdir):
# Several existing parts in partition 'Paris/yes'.
df1 = pd.DataFrame({'humidity': [0.3, 0.8, 0.9, 0.7],
'pressure': [1e5, 1.1e5, 0.95e5, 1e5],
'location': ['Paris', 'Paris', 'Milan', 'Paris'],
'exterior': ['yes', 'no', 'yes', 'yes']})
write(tempdir, df1, row_group_offsets=1, file_scheme='hive',
write_index=False, partition_on=['location', 'exterior'])
df2 = pd.DataFrame({'humidity': [0.4, 0.8, 0.9, 0.7],
'pressure': [1.1e5, 1.1e5, 0.95e5, 1e5],
'location': ['Paris', 'Tokyo', 'Milan', 'Paris'],
'exterior': ['yes', 'no', 'no', 'yes']})
overwrite(tempdir, df2, row_group_offsets=1)
expected = pd.DataFrame(
{'humidity': [0.4, 0.7, 0.8, 0.9, 0.8, 0.9],
'pressure': [1.1e5, 1e5, 1.1e5, 9.5e4, 1.1e5, 9.5e4],
'location': ['Paris', 'Paris', 'Paris', 'Milan', 'Tokyo', 'Milan'],
'exterior': ['yes', 'yes', 'no', 'yes', 'no', 'no']})\
.astype({'location': 'category', 'exterior': 'category'})
recorded = ParquetFile(tempdir).to_pandas()
assert expected.equals(recorded)
def test_with_actually_no_rg_to_overwrite(tempdir):
# Step 1 - Writing of a 1st df, with `row_group_offsets=0`,
# `file_scheme=hive` and `partition_on=['location', 'color`].
df1 = pd.DataFrame({'humidity': [0.9, 0.8, 0.93],
'pressure': [0.95e5, 1.1e5, 1e5],
'location': ['Milan', 'Paris', 'Paris'],
'color': ['blue', 'black', 'red']})
write(tempdir, df1, row_group_offsets=0, file_scheme='hive',
partition_on=['location', 'color'])
# Step 2 - 'Overwriting' with a 2nd df having actually no overlapping data.
df2 = pd.DataFrame({
'humidity': [0.5, 0.3],
'pressure': [9e4, 1e5],
'location': ['Milan', 'Paris'],
'color': ['red', 'green']})
overwrite(tempdir, df2, row_group_offsets=0)
expected = pd.DataFrame({
'humidity': [0.9, 0.8, 0.93, 0.5, 0.3],
'pressure': [9.5e4, 1.1e5, 1e5, 9e4, 1e5],
'location': ['Milan', 'Paris', 'Paris', 'Milan', 'Paris'],
'color': ['blue', 'black', 'red', 'red', 'green']})
expected = expected.astype({'location': 'category', 'color': 'category'})
recorded = ParquetFile(tempdir).to_pandas()
assert expected.equals(recorded)
def test_multiple_parts_in_partitions_thru_write(tempdir):
# Several existing parts in folder 'Paris/yes'.
df1 = pd.DataFrame({'humidity': [0.3, 0.8, 0.9, 0.7],
'pressure': [1e5, 1.1e5, 0.95e5, 1e5],
'location': ['Paris', 'Paris', 'Milan', 'Paris'],
'exterior': ['yes', 'no', 'yes', 'yes']})
write(tempdir, df1, row_group_offsets=1, file_scheme='hive',
write_index=False, partition_on=['location', 'exterior'])
df2 = pd.DataFrame({'humidity': [0.4, 0.8, 0.9, 0.7],
'pressure': [1.1e5, 1.1e5, 0.95e5, 1e5],
'location': ['Paris', 'Tokyo', 'Milan', 'Paris'],
'exterior': ['yes', 'no', 'no', 'yes']})
write(tempdir, df2, row_group_offsets=1, append='overwrite')
expected = pd.DataFrame(
{'humidity': [0.4, 0.7, 0.8, 0.9, 0.8, 0.9],
'pressure': [1.1e5, 1e5, 1.1e5, 9.5e4, 1.1e5, 9.5e4],
'location': ['Paris', 'Paris', 'Paris', 'Milan', 'Tokyo', 'Milan'],
'exterior': ['yes', 'yes', 'no', 'yes', 'no', 'no']})\
.astype({'location': 'category', 'exterior': 'category'})
recorded = ParquetFile(tempdir).to_pandas()
assert expected.equals(recorded)
def test_no_partitioning_exception(tempdir):
df1 = pd.DataFrame({'humidity': [0.3, 0.8, 0.9, 0.7],
'pressure': [1e5, 1.1e5, 0.95e5, 1e5],
'location': ['Paris', 'Paris', 'Milan', 'Paris'],
'exterior': ['yes', 'no', 'yes', 'yes']})
# No partitions.
write(tempdir, df1, row_group_offsets=1, file_scheme='hive',
write_index=False)
with pytest.raises(ValueError, match="^No partitioning"):
overwrite(tempdir, df1, row_group_offsets=0)
def test_simple_scheme_exception(tempdir):
df1 = pd.DataFrame({'humidity': [0.3, 0.8, 0.9, 0.7],
'pressure': [1e5, 1.1e5, 0.95e5, 1e5],
'location': ['Paris', 'Paris', 'Milan', 'Paris'],
'exterior': ['yes', 'no', 'yes', 'yes']})
# Simple file scheme.
fn = os.path.join(tempdir, 'foo.parquet')
write(fn, df1, row_group_offsets=1, file_scheme='simple',
write_index=False)
with pytest.raises(ValueError, match="^Not possible to overwrite"):
overwrite(fn, df1, row_group_offsets=0)
def test_multiple_parts_in_partitions_with_renaming(tempdir):
# Several existing parts in partition 'Paris/yes'.
df1 = pd.DataFrame({'humidity': [0.3, 0.8, 0.9, 0.7],
'pressure': [1e5, 1.1e5, 0.95e5, 1e5],
'location': ['Paris', 'Paris', 'Milan', 'Paris'],
'exterior': ['yes', 'no', 'yes', 'yes']})
write(tempdir, df1, row_group_offsets=1, file_scheme='hive',
write_index=False, partition_on=['location', 'exterior'])
df2 = pd.DataFrame({'humidity': [0.4, 0.8, 0.9, 0.7],
'pressure': [1.1e5, 1.1e5, 0.95e5, 1e5],
'location': ['Paris', 'Tokyo', 'Milan', 'Paris'],
'exterior': ['yes', 'no', 'no', 'yes']})
# 'overwrite' without file shuffling.
# Because we 1st add new data, then remove old ones, a hole in file ids
# appears.
overwrite(tempdir, df2, row_group_offsets=1, sort_pnames=False)
recorded = ParquetFile(tempdir)
pnames_rec = [rg.columns[0].file_path for rg in recorded.row_groups]
pnames_ref = ['location=Paris/exterior=yes/part.4.parquet',
'location=Paris/exterior=yes/part.7.parquet',
'location=Paris/exterior=no/part.1.parquet',
'location=Milan/exterior=yes/part.2.parquet',
'location=Tokyo/exterior=no/part.5.parquet',
'location=Milan/exterior=no/part.6.parquet']
assert pnames_rec == pnames_ref
# overwrite' again with file shuffling.
overwrite(tempdir, df2, row_group_offsets=1, sort_pnames=True)
recorded = ParquetFile(tempdir)
pnames_rec = [rg.columns[0].file_path for rg in recorded.row_groups]
pnames_ref = ['location=Paris/exterior=yes/part.0.parquet',
'location=Paris/exterior=yes/part.1.parquet',
'location=Paris/exterior=no/part.2.parquet',
'location=Milan/exterior=yes/part.3.parquet',
'location=Tokyo/exterior=no/part.4.parquet',
'location=Milan/exterior=no/part.5.parquet']
assert pnames_rec == pnames_ref
|
