def test_parquet(s3):
dd = pytest.importorskip('dask.dataframe')
pytest.importorskip('fastparquet')
from dask.dataframe.io.parquet import to_parquet, read_parquet
import pandas as pd
import numpy as np
url = 's3://%s/test.parquet' % test_bucket_name
data = pd.DataFrame({'i32': np.arange(1000, dtype=np.int32),
'i64': np.arange(1000, dtype=np.int64),
'f': np.arange(1000, dtype=np.float64),
'bhello': np.random.choice(
['hello', 'you', 'people'],
size=1000).astype("O")},
index=pd.Index(np.arange(1000), name='foo'))
df = dd.from_pandas(data, chunksize=500)
to_parquet(url, df, object_encoding='utf8')
files = [f.split('/')[-1] for f in s3.ls(url)]
assert '_metadata' in files
assert 'part.0.parquet' in files
df2 = read_parquet(url, index='foo')
assert len(df2.divisions) > 1
pd.util.testing.assert_frame_equal(data, df2.compute())
def test_local():
with tmpdir() as tmp:
tmp = str(tmp)
data = pd.DataFrame({
'i32':
np.arange(1000, dtype=np.int32),
'i64':
np.arange(1000, dtype=np.int64),
'f':
np.arange(1000, dtype=np.float64),
'bhello':
np.random.choice(['hello', 'you', 'people'], size=1000).astype("O")
})
df = dd.from_pandas(data, chunksize=500)
to_parquet(tmp, df, write_index=False)
files = os.listdir(tmp)
assert '_metadata' in files
assert 'part.0.parquet' in files
df2 = read_parquet(tmp, index=False)
assert len(df2.divisions) > 1
out, out2 = df.compute(), df2.compute().reset_index()
for column in df.columns:
assert (out[column] == out2[column]).all()
def test_parquet(s3):
dd = pytest.importorskip('dask.dataframe')
pytest.importorskip('fastparquet')
from dask.dataframe.io.parquet import to_parquet, read_parquet
import pandas as pd
import numpy as np
url = 's3://%s/test.parquet' % test_bucket_name
data = pd.DataFrame(
{
'i32':
np.arange(1000, dtype=np.int32),
'i64':
np.arange(1000, dtype=np.int64),
'f':
np.arange(1000, dtype=np.float64),
'bhello':
np.random.choice(['hello', 'you', 'people'], size=1000).astype("O")
},
index=pd.Index(np.arange(1000), name='foo'))
df = dd.from_pandas(data, chunksize=500)
to_parquet(url, df, object_encoding='utf8')
files = [f.split('/')[-1] for f in s3.ls(url)]
assert '_metadata' in files
assert 'part.0.parquet' in files
df2 = read_parquet(url, index='foo')
assert len(df2.divisions) > 1
pd.util.testing.assert_frame_equal(data, df2.compute())
def test_roundtrip(df, write_kwargs, read_kwargs):
with tmpdir() as tmp:
tmp = str(tmp)
if df.index.name is None:
df.index.name = 'index'
ddf = dd.from_pandas(df, npartitions=2)
to_parquet(tmp, ddf, **write_kwargs)
ddf2 = read_parquet(tmp, index=df.index.name, **read_kwargs)
assert_eq(ddf, ddf2)
def test_roundtrip(df, write_kwargs, read_kwargs):
with tmpdir() as tmp:
tmp = str(tmp)
if df.index.name is None:
df.index.name = 'index'
ddf = dd.from_pandas(df, npartitions=2)
to_parquet(tmp, ddf, **write_kwargs)
ddf2 = read_parquet(tmp, index=df.index.name, **read_kwargs)
assert_eq(ddf, ddf2)
def test_categorical():
with tmpdir() as tmp:
df = pd.DataFrame({'x': ['a', 'b', 'c'] * 100}, dtype='category')
ddf = dd.from_pandas(df, npartitions=3)
to_parquet(tmp, ddf)
ddf2 = read_parquet(tmp, categories=['x'])
assert ddf2.compute().x.cat.categories.tolist() == ['a', 'b', 'c']
ddf2.loc[:1000].compute()
df.index.name = 'index' # defaults to 'index' in this case
assert assert_eq(df, ddf2)
def test_categorical():
with tmpdir() as tmp:
df = pd.DataFrame({'x': ['a', 'b', 'c'] * 100},
dtype='category')
ddf = dd.from_pandas(df, npartitions=3)
to_parquet(tmp, ddf)
ddf2 = read_parquet(tmp, categories=['x'])
assert ddf2.x.cat.categories.tolist() == ['a', 'b', 'c']
ddf2.loc[:1000].compute()
df.index.name = 'index' # defaults to 'index' in this case
assert assert_eq(df, ddf2)
def test_ordering():
with tmpdir() as tmp:
tmp = str(tmp)
df = pd.DataFrame({'a': [1, 2, 3],
'b': [10, 20, 30],
'c': [100, 200, 300]},
index=pd.Index([-1, -2, -3], name='myindex'),
columns=['c', 'a', 'b'])
ddf = dd.from_pandas(df, npartitions=2)
to_parquet(tmp, ddf)
pf = fastparquet.ParquetFile(tmp)
assert pf.columns == ['myindex', 'c', 'a', 'b']
ddf2 = read_parquet(tmp, index='myindex')
assert_eq(ddf, ddf2)
def test_index():
with tmpdir() as tmp:
tmp = str(tmp)
df = pd.DataFrame(
{
'x': [6, 2, 3, 4, 5],
'y': [1.0, 2.0, 1.0, 2.0, 1.0]
},
index=pd.Index([10, 20, 30, 40, 50], name='myindex'))
ddf = dd.from_pandas(df, npartitions=3)
to_parquet(tmp, ddf)
ddf2 = read_parquet(tmp)
assert_eq(ddf, ddf2)
def test_ordering():
with tmpdir() as tmp:
tmp = str(tmp)
df = pd.DataFrame({'a': [1, 2, 3],
'b': [10, 20, 30],
'c': [100, 200, 300]},
index=pd.Index([-1, -2, -3], name='myindex'),
columns=['c', 'a', 'b'])
ddf = dd.from_pandas(df, npartitions=2)
to_parquet(tmp, ddf)
pf = fastparquet.ParquetFile(tmp)
assert pf.columns == ['myindex', 'c', 'a', 'b']
ddf2 = read_parquet(tmp, index='myindex')
assert_eq(ddf, ddf2)
def test_parquet_wstoragepars(s3):
dd = pytest.importorskip('dask.dataframe')
pytest.importorskip('fastparquet')
from dask.dataframe.io.parquet import to_parquet, read_parquet
import pandas as pd
import numpy as np
url = 's3://%s/test.parquet' % test_bucket_name
data = pd.DataFrame({'i32': np.array([0, 5, 2, 5])})
df = dd.from_pandas(data, chunksize=500)
to_parquet(url, df, write_index=False)
read_parquet(url, storage_options={'default_fill_cache': False})
assert s3.current().default_fill_cache is False
read_parquet(url, storage_options={'default_fill_cache': True})
assert s3.current().default_fill_cache is True
read_parquet(url, storage_options={'default_block_size': 2**20})
assert s3.current().default_block_size == 2**20
with s3.current().open(url + '/_metadata') as f:
assert f.blocksize == 2**20
def test_parquet_wstoragepars(s3):
dd = pytest.importorskip('dask.dataframe')
pytest.importorskip('fastparquet')
from dask.dataframe.io.parquet import to_parquet, read_parquet
import pandas as pd
import numpy as np
url = 's3://%s/test.parquet' % test_bucket_name
data = pd.DataFrame({'i32': np.array([0, 5, 2, 5])})
df = dd.from_pandas(data, chunksize=500)
to_parquet(url, df, write_index=False)
read_parquet(url, storage_options={'default_fill_cache': False})
assert s3.current().default_fill_cache is False
read_parquet(url, storage_options={'default_fill_cache': True})
assert s3.current().default_fill_cache is True
read_parquet(url, storage_options={'default_block_size': 2**20})
assert s3.current().default_block_size == 2**20
with s3.current().open(url + '/_metadata') as f:
assert f.blocksize == 2**20
def fn(tmpdir):
ddf = dd.from_pandas(df, npartitions=3)
to_parquet(str(tmpdir), ddf)
return str(tmpdir)
def fn(tmpdir):
ddf = dd.from_pandas(df, npartitions=3)
to_parquet(str(tmpdir), ddf)
return str(tmpdir)
# Example : Country,Year
tableName = fileName+'.table/'+fileName+'.parquet/'
parquetDataPath = parquetfilePath+tableName
print('Source = '+csvfilePath)
print('Output = '+parquetDataPath)
########## Parquet Convertor ################################
fileArray = os.listdir(csvfilePath)
parquetDataPath = parquetfilePath+tableName
# Browse csv files
for filename in fileArray:
if filename.endswith(".csv"):
print('file = '+filename)
# Create parquet Folder if needed
if not(os.path.isdir(parquetDataPath)):
os.makedirs(parquetDataPath)
print('Create = '+parquetDataPath)
print('Read file = '+csvfilePath+filename)
# Read CSV
df = pd.read_csv(csvfilePath+filename,sep=',', quotechar='"')
print('to parquet = '+parquetDataPath+filename[:-4]+'.parquet')
# Parquet partition
df = dd.from_pandas(df, npartitions=1)
# Parquet conversion
ddp.to_parquet(df,parquetDataPath, write_index=False, partition_on=parquetPartition)
##############################################################
