def test_jay_empty_frame(tempfile):
d0 = dt.Frame()
d0.save(tempfile, format="jay")
assert os.path.isfile(tempfile)
d1 = dt.open(tempfile)
assert d1.shape == (0, 0)
assert d1.names == tuple()
def test_jay_unicode_names(tempfile):
dt0 = dt.Frame({"py": [1], "ру": [2], "рy": [3], "pу": [4]})
assert len(set(dt0.names)) == 4
dt0.save(tempfile, format="jay")
assert os.path.isfile(tempfile)
dt1 = dt.open(tempfile)
assert dt0.names == dt1.names
assert_equals(dt0, dt1)
def test_key_save(tempfile):
dt0 = dt.Frame(D=range(6), A=[3, 7, 5, 2, 2, 3], B=[1, 2, 2, 3, 4, 4])
dt0.key = ["A", "B"]
dt0.internal.check()
dt0.save(tempfile, format="jay")
dt1 = dt.open(tempfile)
assert dt1.key == ("A", "B")
dt1.internal.check()
def test_key_save(tempfile):
dt0 = dt.Frame(D=range(6), A=[3, 7, 5, 2, 2, 3], B=[1, 2, 2, 3, 4, 4])
dt0.key = ["A", "B"]
frame_integrity_check(dt0)
dt0.to_jay(tempfile)
dt1 = dt.open(tempfile)
assert dt1.key == ("A", "B")
frame_integrity_check(dt1)
def test_open(tempfile_jay):
DT = dt.Frame(A=range(5))
DT.to_jay(tempfile_jay)
with pytest.warns(FutureWarning):
DT2 = dt.open(tempfile_jay)
assert DT.source is None
assert DT2.source == tempfile_jay
assert_equals(DT, DT2)
def test_issue627():
"""Test saving Frame with unicode file names"""
dir1 = tempfile.mkdtemp()
dt0 = dt.Frame({"py": [1], "ру": [2], "рy": [3], "pу": [4]})
assert dt0.shape == (1, 4)
dt0.save(dir1)
dt1 = dt.open(dir1)
assert_equals(dt0, dt1)
shutil.rmtree(dir1)
def test_save_view(tempdir):
dt0 = dt.Frame([4, 0, -2, 3, 17, 2, 0, 1, 5], names=["fancy"])
dt1 = dt0.sort(0)
assert dt1.internal.isview
dt.save(dt1, tempdir)
dt2 = dt.open(tempdir)
assert not dt2.internal.isview
assert dt2.names == dt1.names
assert dt2.topython() == dt1.topython()
def test_rbind_mmapped(tempfile):
dt0 = dt.Frame({"A": [1, 5, 7], "B": ["one", "two", None]})
dt0.to_jay(tempfile)
del dt0
dt1 = dt.open(tempfile)
dt2 = dt.Frame({"A": [-1], "B": ["zero"]})
dt1.rbind(dt2)
dtr = dt.Frame({"A": [1, 5, 7, -1], "B": ["one", "two", None, "zero"]})
assert_equals(dt1, dtr)
def test_h2oai7014(tempfile):
data = dt.Frame([[None, 't'], [3580, 1047]], names=["ID", "count"])
data.to_jay(tempfile)
# The data has to be opened from file
counts = dt.open(tempfile)
counts = counts[1:, :]
counts = counts[:, :, sort("count")]
counts.materialize()
assert counts.to_list() == [['t'], [1047]]
def test_save_and_load():
dir1 = tempfile.mkdtemp()
dt0 = dt.Frame({"A": [1, 7, 100, 12],
"B": [True, None, False, None],
"C": ["alpha", "beta", None, "delta"]})
dt0.save(dir1)
dt1 = dt.open(dir1)
assert_equals(dt0, dt1)
shutil.rmtree(dir1)
def test_save_and_load():
dir1 = tempfile.mkdtemp()
dt0 = dt.Frame({"A": [1, 7, 100, 12],
"B": [True, None, False, None],
"C": ["alpha", "beta", None, "delta"]})
with pytest.warns(FutureWarning):
dt0.save(dir1, format="nff")
dt1 = dt.open(dir1)
assert_equals(dt0, dt1)
shutil.rmtree(dir1)
def test_jay_simple(tempfile):
dt0 = dt.Frame({"A": [-1, 7, 10000, 12],
"B": [True, None, False, None],
"C": ["alpha", "beta", None, "delta"]})
dt0.save(tempfile, format="jay")
assert os.path.isfile(tempfile)
with open(tempfile, "rb") as inp:
assert inp.read(8) == b"JAY1\x00\x00\x00\x00"
dt1 = dt.open(tempfile)
assert_equals(dt0, dt1)
def test_jay_keys(tempfile):
d0 = dt.Frame([["ab", "cd", "eee", "coo", "aop"], [1, 2, 3, 4, 5]],
names=("x", "y"))
d0.key = "x"
assert len(d0.key) == 1
assert d0.to_list() == [["ab", "aop", "cd", "coo", "eee"], [1, 5, 2, 4, 3]]
d0.to_jay(tempfile)
d1 = dt.open(tempfile)
assert d1.key == ("x", )
assert_equals(d0, d1)
def test_issue1728(tempfile):
data = dt.Frame({'department1': [None, 't'], 'C0': [3580, 1047]})
data.to_jay(tempfile)
del data
counts = dt.open(tempfile)
counts = counts[1:, :]
counts = counts[:, :, dt.sort(-1)]
counts.materialize()
frame_integrity_check(counts)
assert counts.to_dict() == {'department1': ['t'], 'C0': [1047]}
def test_issue627():
"""Test saving Frame with unicode file names"""
dir1 = tempfile.mkdtemp()
dt0 = dt.Frame({"py": [1], "ру": [2], "рy": [3], "pу": [4]})
assert dt0.shape == (1, 4)
with pytest.warns(FutureWarning):
dt0.save(dir1, format="nff")
dt1 = dt.open(dir1)
assert_equals(dt0, dt1)
shutil.rmtree(dir1)
def test_rbind_mmapped():
dir0 = tempfile.mkdtemp()
dt0 = dt.Frame({"A": [1, 5, 7], "B": ["one", "two", None]})
dt.save(dt0, dir0)
del dt0
dt1 = dt.open(dir0)
dt2 = dt.Frame({"A": [-1], "B": ["zero"]})
dt1.rbind(dt2)
dtr = dt.Frame({"A": [1, 5, 7, -1], "B": ["one", "two", None, "zero"]})
assert_equals(dt1, dtr)
shutil.rmtree(dir0)
def test_issue689(tempdir):
n = 300000 # Must be > 65536
data = [i % 8 for i in range(n)]
d0 = dt.Frame(data, names=["A"])
dt.save(d0, tempdir)
del d0
d1 = dt.open(tempdir)
# Do not check d1! we want it to be lazy at this point
d2 = d1(rows=lambda g: g[0] == 1)
assert d2.internal.check()
assert d2.shape == (n / 8, 1)
def test_issue689(tempfile):
n = 300000 # Must be > 65536
data = [i % 8 for i in range(n)]
d0 = dt.Frame(data, names=["A"])
d0.to_jay(tempfile)
del d0
d1 = dt.open(tempfile)
# Do not check d1! we want it to be lazy at this point
d2 = d1[f[0] == 1, :]
frame_integrity_check(d2)
assert d2.shape == (n / 8, 1)
def test_empty_string_col():
"""
Test that Frame with an empty string column can be saved/opened.
See #604
"""
dir1 = tempfile.mkdtemp()
dt0 = dt.Frame([[1, 2, 3], ["", "", ""]])
dt0.save(dir1)
dt1 = dt.open(dir1)
assert_equals(dt0, dt1)
shutil.rmtree(dir1)
def test_save_view(tempdir):
dt0 = dt.Frame([4, 0, -2, 3, 17, 2, 0, 1, 5], names=["fancy"])
dt1 = dt0.sort(0)
assert dt1.internal.isview
dt1.internal.check()
dt1.save(tempdir, format="nff")
dt2 = dt.open(tempdir)
assert not dt2.internal.isview
dt2.internal.check()
assert dt2.names == dt1.names
assert dt2.to_list() == dt1.to_list()
def test_empty_string_col():
"""
Test that Frame with an empty string column can be saved/opened.
See #604
"""
dir1 = tempfile.mkdtemp()
dt0 = dt.Frame([[1, 2, 3], ["", "", ""]])
with pytest.warns(FutureWarning):
dt0.save(dir1, format="nff")
dt1 = dt.open(dir1)
assert_equals(dt0, dt1)
shutil.rmtree(dir1)
def test_save_view(tempdir):
dt0 = dt.Frame([4, 0, -2, 3, 17, 2, 0, 1, 5], names=["fancy"])
dt1 = dt0.sort(0)
assert isview(dt1)
frame_integrity_check(dt1)
with pytest.warns(FutureWarning):
dt1.save(tempdir, format="nff")
dt2 = dt.open(tempdir)
assert not isview(dt2)
frame_integrity_check(dt2)
assert dt2.names == dt1.names
assert dt2.to_list() == dt1.to_list()
def test_jay_object_columns(tempfile):
src1 = [1, 2, 3, 4]
src2 = [(2, 3), (5, 6, 7), 9, {"A": 3}]
d0 = dt.Frame([src1, src2], names=["A", "B"])
assert d0.stypes == (dt.int32, dt.obj64)
with pytest.warns(DatatableWarning) as ws:
d0.to_jay(tempfile)
assert len(ws) == 1
assert "Column `B` of type obj64 was not saved" in ws[0].message.args[0]
d1 = dt.open(tempfile)
frame_integrity_check(d1)
assert d1.names == ("A", )
assert d1.to_list() == [src1]
def test_jay_object_columns(tempfile):
src1 = [1, 2, 3, 4]
src2 = [(2, 3), (5, 6, 7), 9, {"A": 3}]
d0 = dt.Frame([src1, src2], names=["A", "B"])
assert d0.stypes == (dt.int8, dt.obj64)
with pytest.warns(DatatableWarning) as ws:
d0.save(tempfile, format="jay")
assert len(ws) == 1
assert "Column `B` of type obj64 was not saved" in ws[0].message.args[0]
d1 = dt.open(tempfile)
d1.internal.check()
assert d1.names == ("A",)
assert d1.topython() == [src1]
def test_jay_view(tempfile, seed):
random.seed(seed)
src = [random.normalvariate(0, 1) for n in range(1000)]
dt0 = dt.Frame({"values": src})
dt1 = dt0.sort(0)
assert isview(dt1)
dt1.to_jay(tempfile)
assert os.path.isfile(tempfile)
dt2 = dt.open(tempfile)
assert not isview(dt2)
frame_integrity_check(dt1)
frame_integrity_check(dt2)
assert dt1.names == dt2.names
assert dt1.stypes == dt2.stypes
assert dt1.to_list() == dt2.to_list()
def test_jay_view(tempfile, seed):
random.seed(seed)
src = [random.normalvariate(0, 1) for n in range(1000)]
dt0 = dt.Frame({"values": src})
dt1 = dt0.sort(0)
assert dt1.internal.isview
dt1.save(tempfile, format="jay")
assert os.path.isfile(tempfile)
dt2 = dt.open(tempfile)
assert not dt2.internal.isview
dt1.internal.check()
dt2.internal.check()
assert dt1.names == dt2.names
assert dt1.stypes == dt2.stypes
assert dt1.topython() == dt2.topython()
def test_obj_columns(tempdir):
src1 = [1, 2, 3, 4]
src2 = [(2, 3), (5, 6, 7), 9, {"A": 3}]
d0 = dt.Frame([src1, src2], names=["A", "B"])
d0.internal.check()
assert d0.ltypes == (dt.ltype.int, dt.ltype.obj)
assert d0.shape == (4, 2)
with pytest.warns(DatatableWarning) as ws:
d0.save(tempdir)
assert len(ws) == 1
assert "Column 'B' of type obj64 was not saved" in ws[0].message.args[0]
del d0
d1 = dt.open(tempdir)
d1.internal.check()
assert d1.shape == (4, 1)
assert d1.names == ("A", )
assert d1.topython() == [src1]
def test_obj_columns(tempdir):
src1 = [1, 2, 3, 4]
src2 = [(2, 3), (5, 6, 7), 9, {"A": 3}]
d0 = dt.Frame([src1, src2], names=["A", "B"])
frame_integrity_check(d0)
assert d0.ltypes == (dt.ltype.int, dt.ltype.obj)
assert d0.shape == (4, 2)
with pytest.warns(DatatableWarning) as ws:
d0.save(tempdir, format="nff")
assert len(ws) == 2
assert "Method `Frame.save()` is deprecated" in ws[0].message.args[0]
assert "Column 'B' of type obj64 was not saved" in ws[1].message.args[0]
del d0
d1 = dt.open(tempdir)
frame_integrity_check(d1)
assert d1.shape == (4, 1)
assert d1.names == ("A", )
assert d1.to_list() == [src1]
def test_jay_all_types(tempfile):
d0 = dt.Frame(
[[True, False, None, True, True], [None, 1, -9, 12, 3],
[4, 1346, 999, None, None], [591, 0, None, -395734, 19384709],
[None, 777, 1093487019384, -384, None],
[2.987, 3.45e-24, -0.189134e+12, 45982.1, None],
[39408.301, 9.459027045e-125, 4.4508e+222, None, 3.14159],
["Life", "Liberty", "and", "Pursuit of Happiness", None],
["кохайтеся", "чорнобриві", ",", "та", "не з москалями"]],
stypes=[
dt.bool8, dt.int8, dt.int16, dt.int32, dt.int64, dt.float32,
dt.float64, dt.str32, dt.str64
],
names=["b8", "i8", "i16", "i32", "i64", "f32", "f64", "s32", "s64"])
# Force calculation of mins and maxs, so that they get saved into Jay
d0.min(), d0.max()
assert len(set(d0.stypes)) == d0.ncols
d0.save(tempfile, format="jay")
assert os.path.isfile(tempfile)
d1 = dt.open(tempfile)
assert_equals(d0, d1)
git = pd.__git_version__
task = "groupby"
solution = "pandas"
fun = ".groupby"
cache = "TRUE"
data_name = os.environ['SRC_GRP_LOCAL']
src_grp = os.path.join("data", data_name + ".jay")
print("loading dataset %s" % data_name, flush=True)
## failed attempt to improve read_csv to read 1e9 rows #99
#dtype = {"id1": "category", "id2": "category", "id3": "category", "id4": "int32", "id5": "int32", "id6": "int32", "v1": "int32", "v2": "int32", "v3": "float64"}
#x = pd.read_csv(src_grp, dtype=dtype, engine="c", low_memory=True)
import datatable as dt # for loading data only, see #47
x = dt.open(src_grp).to_pandas()
x['id1'] = x['id1'].astype('category') # remove after datatable#1691
x['id2'] = x['id2'].astype('category')
x['id3'] = x['id3'].astype('category')
print(len(x.index), flush=True)
print("grouping...", flush=True)
question = "sum v1 by id1" # q1
gc.collect()
t_start = timeit.default_timer()
ans = x.groupby(['id1']).agg({'v1': 'sum'})
ans.reset_index(inplace=True) # #68
print(ans.shape, flush=True)
t = timeit.default_timer() - t_start
m = memory_usage()
exec(open("./helpers.py").read())
ver = pd.__version__
git = pd.__git_version__
task = "groupby"
solution = "pandas"
fun = ".groupby"
cache = "TRUE"
data_name = os.environ['SRC_GRP_LOCAL']
src_grp = os.path.join("data", data_name+".jay")
print("loading dataset %s" % data_name, flush=True)
import datatable as dt # for loading data only, see #47
x = dt.open(src_grp).to_pandas()
x['id1'] = x['id1'].astype('category')
x['id2'] = x['id2'].astype('category')
x['id3'] = x['id3'].astype('category')
print(len(x.index), flush=True)
print("grouping...", flush=True)
question = "sum v1 by id1" # q1
gc.collect()
t_start = timeit.default_timer()
ans = x.groupby(['id1']).agg({'v1':'sum'})
ans.reset_index(inplace=True) # #68
print(ans.shape, flush=True)
t = timeit.default_timer() - t_start
m = memory_usage()
