def test_badtype_parsing(self):
"""Test parsing wrong type of attribute from their value."""
ofile = open(test3)
rel, attrs = read_header(ofile)
for name, value in attrs:
assert_raises(ParseArffError, parse_type, value)
def test_badtype_parsing(self):
"""Test parsing wrong type of attribute from their value."""
ofile = open(test3)
rel, attrs = read_header(ofile)
for name, value in attrs:
assert_raises(ParseArffError, parse_type, value)
def test_type_parsing(self):
# Test parsing type of attribute from their value.
with open(test2) as ofile:
rel, attrs = read_header(ofile)
expected = ['numeric', 'numeric', 'numeric', 'numeric', 'numeric',
'numeric', 'string', 'string', 'nominal', 'nominal']
for i in range(len(attrs)):
assert_(attrs[i].type_name == expected[i])
def test_type_parsing(self):
"""Test parsing type of attribute from their value."""
ofile = open(test2)
rel, attrs = read_header(ofile)
expected = ['numeric', 'numeric', 'numeric', 'numeric', 'numeric',
'numeric', 'string', 'string', 'nominal', 'nominal']
for i in range(len(attrs)):
assert_(parse_type(attrs[i][1]) == expected[i])
def test_type_parsing(self):
"""Test parsing type of attribute from their value."""
ofile = open(test2)
rel, attrs = read_header(ofile)
expected = ['numeric', 'numeric', 'numeric', 'numeric', 'numeric',
'numeric', 'string', 'string', 'nominal', 'nominal']
for i in range(len(attrs)):
assert parse_type(attrs[i][1]) == expected[i]
def test_badtype_parsing(self):
"""Test parsing wrong type of attribute from their value."""
ofile = open(test3)
rel, attrs = read_header(ofile)
for name, value in attrs:
try:
parse_type(value)
raise Error("Could parse type of crap, should not happen.")
except ParseArffError:
pass
def test_badtype_parsing(self):
"""Test parsing wrong type of attribute from their value."""
ofile = open(test3)
rel, attrs = read_header(ofile)
for name, value in attrs:
try:
parse_type(value)
raise Error("Could parse type of crap, should not happen.")
except ParseArffError:
pass
def test_type_parsing(self):
# Test parsing type of attribute from their value.
ofile = open(test2)
rel, attrs = read_header(ofile)
ofile.close()
expected = ['numeric', 'numeric', 'numeric', 'numeric', 'numeric',
'numeric', 'string', 'string', 'nominal', 'nominal']
for i in range(len(attrs)):
assert_(attrs[i].type_name == expected[i])
def iter_arff(filepath_or_buffer,
target: str = None,
compression="infer") -> base.typing.Stream:
"""Iterates over rows from an ARFF file.
Parameters
----------
filepath_or_buffer
Either a string indicating the location of a CSV file, or a buffer object that has a
`read` method.
target
Name of the target field.
compression
For on-the-fly decompression of on-disk data. If this is set to 'infer' and
`filepath_or_buffer` is a path, then the decompression method is inferred for the
following extensions: '.gz', '.zip'.
"""
# If a file is not opened, then we open it
buffer = filepath_or_buffer
if not hasattr(buffer, "read"):
buffer = utils.open_filepath(buffer, compression)
try:
rel, attrs = arffread.read_header(buffer)
except ValueError as e:
msg = f"Error while parsing header, error was: {e}"
raise arffread.ParseArffError(msg)
names = [attr.name for attr in attrs]
types = [
float if isinstance(attr, arffread.NumericAttribute) else None
for attr in attrs
]
for r in buffer:
x = {
name: typ(val) if typ else val
for name, typ, val in zip(names, types,
r.rstrip().split(","))
}
try:
y = x.pop(target) if target else None
except KeyError as e:
print(r)
raise e
yield x, y
# Close the file if we opened it
if buffer is not filepath_or_buffer:
buffer.close()
def test_dateheader_unsupported(self):
ofile = open(test8)
rel, attrs = read_header(ofile)
ofile.close()
assert_(rel == 'test8')
assert_(len(attrs) == 2)
assert_(attrs[0][0] == 'attr_datetime_utc')
assert_(attrs[0][1] == 'DATE "yyyy-MM-dd HH:mm Z"')
assert_(attrs[1][0] == 'attr_datetime_full')
assert_(attrs[1][1] == 'DATE "yy-MM-dd HH:mm:ss z"')
def test_dateheader_unsupported(self):
ofile = open(test8)
rel, attrs = read_header(ofile)
ofile.close()
assert_(rel == 'test8')
assert_(len(attrs) == 2)
assert_(attrs[0][0] == 'attr_datetime_utc')
assert_(attrs[0][1] == 'DATE "yyyy-MM-dd HH:mm Z"')
assert_(attrs[1][0] == 'attr_datetime_full')
assert_(attrs[1][1] == 'DATE "yy-MM-dd HH:mm:ss z"')
def test_fullheader1(self):
# Parsing trivial header with nothing.
with open(test1) as ofile:
rel, attrs = read_header(ofile)
# Test relation
assert_(rel == 'test1')
# Test numerical attributes
assert_(len(attrs) == 5)
for i in range(4):
assert_(attrs[i].name == 'attr%d' % i)
assert_(attrs[i].type_name == 'numeric')
# Test nominal attribute
assert_(attrs[4].name == 'class')
assert_(attrs[4].values == ('class0', 'class1', 'class2', 'class3'))
def test_fullheader1(self):
# Parsing trivial header with nothing.
ofile = open(test1)
rel, attrs = read_header(ofile)
ofile.close()
# Test relation
assert_(rel == 'test1')
# Test numerical attributes
assert_(len(attrs) == 5)
for i in range(4):
assert_(attrs[i][0] == 'attr%d' % i)
assert_(attrs[i][1] == 'REAL')
# Test nominal attribute
assert_(attrs[4][0] == 'class')
assert_(attrs[4][1] == '{class0, class1, class2, class3}')
def test_fullheader1(self):
"""Parsing trivial header with nothing."""
ofile = open(test1)
rel, attrs = read_header(ofile)
# Test relation
assert rel == 'test1'
# Test numerical attributes
assert len(attrs) == 5
for i in range(4):
assert attrs[i][0] == 'attr%d' % i
assert attrs[i][1] == 'REAL'
classes = attrs[4][1]
# Test nominal attribute
assert attrs[4][0] == 'class'
assert attrs[4][1] == '{class0, class1, class2, class3}'
def test_fullheader1(self):
"""Parsing trivial header with nothing."""
ofile = open(test1)
rel, attrs = read_header(ofile)
# Test relation
assert_(rel == "test1")
# Test numerical attributes
assert_(len(attrs) == 5)
for i in range(4):
assert_(attrs[i][0] == "attr%d" % i)
assert_(attrs[i][1] == "REAL")
classes = attrs[4][1]
# Test nominal attribute
assert_(attrs[4][0] == "class")
assert_(attrs[4][1] == "{class0, class1, class2, class3}")
def test_fullheader1(self):
"""Parsing trivial header with nothing."""
ofile = open(test1)
rel, attrs = read_header(ofile)
# Test relation
assert rel == 'test1'
# Test numerical attributes
assert len(attrs) == 5
for i in range(4):
assert attrs[i][0] == 'attr%d' % i
assert attrs[i][1] == 'REAL'
classes = attrs[4][1]
# Test nominal attribute
assert attrs[4][0] == 'class'
assert attrs[4][1] == '{class0, class1, class2, class3}'
def test_fullheader1(self):
# Parsing trivial header with nothing.
ofile = open(test1)
rel, attrs = read_header(ofile)
ofile.close()
# Test relation
assert_(rel == 'test1')
# Test numerical attributes
assert_(len(attrs) == 5)
for i in range(4):
assert_(attrs[i].name == 'attr%d' % i)
assert_(attrs[i].type_name == 'numeric')
# Test nominal attribute
assert_(attrs[4].name == 'class')
assert_(attrs[4].values == ('class0', 'class1', 'class2', 'class3'))
def test_fullheader1(self):
# Parsing trivial header with nothing.
ofile = open(test1)
rel, attrs = read_header(ofile)
ofile.close()
# Test relation
assert_(rel == 'test1')
# Test numerical attributes
assert_(len(attrs) == 5)
for i in range(4):
assert_(attrs[i][0] == 'attr%d' % i)
assert_(attrs[i][1] == 'REAL')
# Test nominal attribute
assert_(attrs[4][0] == 'class')
assert_(attrs[4][1] == '{class0, class1, class2, class3}')
def test_type_parsing(self):
# Test parsing type of attribute from their value.
ofile = open(test2)
rel, attrs = read_header(ofile)
ofile.close()
expected = [
"numeric",
"numeric",
"numeric",
"numeric",
"numeric",
"numeric",
"string",
"string",
"nominal",
"nominal",
]
for i in range(len(attrs)):
assert_(parse_type(attrs[i][1]) == expected[i])
def test_dateheader(self):
with open(test7) as ofile:
rel, attrs = read_header(ofile)
assert_(rel == 'test7')
assert_(len(attrs) == 5)
assert_(attrs[0].name == 'attr_year')
assert_(attrs[0].date_format == '%Y')
assert_(attrs[1].name == 'attr_month')
assert_(attrs[1].date_format == '%Y-%m')
assert_(attrs[2].name == 'attr_date')
assert_(attrs[2].date_format == '%Y-%m-%d')
assert_(attrs[3].name == 'attr_datetime_local')
assert_(attrs[3].date_format == '%Y-%m-%d %H:%M')
assert_(attrs[4].name == 'attr_datetime_missing')
assert_(attrs[4].date_format == '%Y-%m-%d %H:%M')
def test_dateheader(self):
ofile = open(test7)
rel, attrs = read_header(ofile)
ofile.close()
assert_(rel == "test7")
assert_(len(attrs) == 5)
assert_(attrs[0][0] == "attr_year")
assert_(attrs[0][1] == "DATE yyyy")
assert_(attrs[1][0] == "attr_month")
assert_(attrs[1][1] == "DATE yyyy-MM")
assert_(attrs[2][0] == "attr_date")
assert_(attrs[2][1] == "DATE yyyy-MM-dd")
assert_(attrs[3][0] == "attr_datetime_local")
assert_(attrs[3][1] == 'DATE "yyyy-MM-dd HH:mm"')
assert_(attrs[4][0] == "attr_datetime_missing")
assert_(attrs[4][1] == 'DATE "yyyy-MM-dd HH:mm"')
def test_dateheader(self):
ofile = open(test7)
rel, attrs = read_header(ofile)
ofile.close()
assert_(rel == 'test7')
assert_(len(attrs) == 5)
assert_(attrs[0].name == 'attr_year')
assert_(attrs[0].date_format == '%Y')
assert_(attrs[1].name == 'attr_month')
assert_(attrs[1].date_format == '%Y-%m')
assert_(attrs[2].name == 'attr_date')
assert_(attrs[2].date_format == '%Y-%m-%d')
assert_(attrs[3].name == 'attr_datetime_local')
assert_(attrs[3].date_format == '%Y-%m-%d %H:%M')
assert_(attrs[4].name == 'attr_datetime_missing')
assert_(attrs[4].date_format == '%Y-%m-%d %H:%M')
def test_dateheader(self):
ofile = open(test7)
rel, attrs = read_header(ofile)
ofile.close()
assert_(rel == 'test7')
assert_(len(attrs) == 5)
assert_(attrs[0][0] == 'attr_year')
assert_(attrs[0][1] == 'DATE yyyy')
assert_(attrs[1][0] == 'attr_month')
assert_(attrs[1][1] == 'DATE yyyy-MM')
assert_(attrs[2][0] == 'attr_date')
assert_(attrs[2][1] == 'DATE yyyy-MM-dd')
assert_(attrs[3][0] == 'attr_datetime_local')
assert_(attrs[3][1] == 'DATE "yyyy-MM-dd HH:mm"')
assert_(attrs[4][0] == 'attr_datetime_missing')
assert_(attrs[4][1] == 'DATE "yyyy-MM-dd HH:mm"')
def test_dateheader(self):
ofile = open(test7)
rel, attrs = read_header(ofile)
ofile.close()
assert_(rel == 'test7')
assert_(len(attrs) == 5)
assert_(attrs[0][0] == 'attr_year')
assert_(attrs[0][1] == 'DATE yyyy')
assert_(attrs[1][0] == 'attr_month')
assert_(attrs[1][1] == 'DATE yyyy-MM')
assert_(attrs[2][0] == 'attr_date')
assert_(attrs[2][1] == 'DATE yyyy-MM-dd')
assert_(attrs[3][0] == 'attr_datetime_local')
assert_(attrs[3][1] == 'DATE "yyyy-MM-dd HH:mm"')
assert_(attrs[4][0] == 'attr_datetime_missing')
assert_(attrs[4][1] == 'DATE "yyyy-MM-dd HH:mm"')
def read_dateheader_unsupported():
with open(test8) as ofile:
_, _ = read_header(ofile)
def badtype_read():
with open(test3) as ofile:
_, _ = read_header(ofile)
def read_dateheader_unsupported():
ofile = open(test8)
rel, attrs = read_header(ofile)
ofile.close()
def badtype_read():
ofile = open(test3)
rel, attrs = read_header(ofile)
ofile.close()
def read_dateheader_unsupported():
ofile = open(test8)
rel, attrs = read_header(ofile)
ofile.close()
def badtype_read():
ofile = open(test3)
rel, attrs = read_header(ofile)
ofile.close()
