def write_dataset_example():
"""
An example for writing dataset.
:return: None
"""
gold = [2, 3, 6]
h = [5, 6]
gold_dataset = Dataset({
'test1': {
'golden': gold,
},
'test2': {
'golden': gold,
},
})
test_dataset = Dataset({
'test1': {
'test': h,
},
'test2': {
'test': h,
},
})
write_dataset(gold_dataset, "./data/golden.json")
write_dataset(test_dataset, "./data/test.json")
def test_dataset(self):
'''
Test dataset property creation and independence.
'''
prop = 'test'
dataset_a = Dataset()
dataset_a.properties[prop] = False
dataset_b = Dataset()
self.assertFalse(prop in dataset_b.properties)
def test_add_duplicate_codings(self):
'''
Test dataset property creation and independence.
'''
dataset_a = Dataset({'item1': {'a': [2]}})
dataset_b = Dataset({'item1': {'a': [2]}})
exception = False
try:
dataset_a + dataset_b
except DataIOError:
exception = True
self.assertTrue(exception, 'Did not throw DataIOError')
def test_add_coders(self):
'''
Test dataset property creation and independence.
'''
dataset_a = Dataset({'item1': {'a': [2]}})
dataset_b = Dataset({'item1': {'b': [2]}})
dataset_c = Dataset({'item1': {'a': [2], 'b': [2]}})
self.assertNotEqual(dataset_a, dataset_b)
self.assertEqual(dataset_a + dataset_b, dataset_c)
self.assertNotEqual(dataset_a, dataset_b)
self.assertNotEqual(dataset_a, dataset_c)
self.assertNotEqual(dataset_b, dataset_c)
def input_linear_mass_json(filepath):
'''
Reads a file path. Returns segmentation mass codings as a :class:`Dataset`.
:param filepath: Path to the mass file containing segment position codings.
:type filepath: :func:`str`
'''
from segeval.data import Dataset, DataIOError
dataset = Dataset()
data = dict()
# Open file
json_file = open(filepath, 'rU')
# Read in file
try:
data = json.load(json_file)
except Exception as exception:
raise DataIOError('Error occurred processing file: ' + filepath,
exception)
# Check type
if Field.segmentation_type in data:
if data[Field.segmentation_type] != SegmentationType.linear:
raise DataIOError(
'Segmentation type \'{0}\' expected, but encountered \'{1}\' for file: {2}'
.format(SegmentationType.linear, data[Field.segmentation_type],
filepath))
else:
raise DataIOError(
'The entry \'segmentation_type\' was expected in JSON for file:' +
filepath)
# Duplicate to store other properties
dataset.properties = data
# If separated into multiple items for one coder per file
if Field.items in data:
data = data[Field.items]
# Convert item labels into strings
for item, coder_masses in data.items():
dataset[item] = dict()
for coder, masses in coder_masses.items():
dataset[item][coder] = tuple(masses)
# Remove from properties
del dataset.properties[Field.items]
else:
raise DataIOError(
'Expected an entry \'{0}\' that contained segmentation codings for specific individual texts (i.e., items) in file: {1}'
.format(Field.items, filepath))
return dataset
def test_dataset_property(self):
'''
Test dataset property creation and independence.
'''
prop = 'test'
dataset = Dataset(properties={prop: True})
self.assertTrue(prop in dataset.properties)
self.assertTrue(dataset.properties[prop])
def input_linear_mass_json(filepath):
'''
Reads a file path. Returns segmentation mass codings as a :class:`Dataset`.
:param filepath: Path to the mass file containing segment position codings.
:type filepath: :func:`str`
'''
from segeval.data import Dataset, DataIOError
dataset = Dataset()
data = dict()
# Open file
json_file = open(filepath, 'rU')
# Read in file
try:
data = json.load(json_file)
except Exception as exception:
raise DataIOError(
'Error occurred processing file: ' + filepath, exception)
# Check type
if Field.segmentation_type in data:
if data[Field.segmentation_type] != SegmentationType.linear:
raise DataIOError('Segmentation type \'{0}\' expected, but encountered \'{1}\' for file: {2}'
.format(SegmentationType.linear, data[Field.segmentation_type], filepath))
else:
raise DataIOError(
'The entry \'segmentation_type\' was expected in JSON for file:' + filepath)
# Duplicate to store other properties
dataset.properties = data
# If separated into multiple items for one coder per file
if Field.items in data:
data = data[Field.items]
# Convert item labels into strings
for item, coder_masses in data.items():
dataset[item] = dict()
for coder, masses in coder_masses.items():
dataset[item][coder] = tuple(masses)
# Remove from properties
del dataset.properties[Field.items]
else:
raise DataIOError('Expected an entry \'{0}\' that contained segmentation codings for specific individual texts (i.e., items) in file: {1}'
.format(Field.items, filepath))
return dataset
def input_linear_mass_tsv(filepath, delimiter=DEFAULT_DELIMITER):
'''
Takes a file path. Returns segmentation mass codings as a :class:`Dataset`.
:param filepath: path to the mass file containing segment mass codings.
:param delimiter: the delimiter used when reading a TSV file (by default,
a tab, but it can also be a comma, whitespace, etc.
:type filepath: str
:type delimiter: str
'''
from segeval.data import Dataset, name_from_filepath
# List version of file
header = []
dataset = Dataset()
item = name_from_filepath(filepath)
dataset[item] = dict()
# Open file
with open(filepath, 'rU') as csv_file:
# Read in file
reader = csv.reader(csv_file, delimiter=delimiter)
for i, row in enumerate(reader):
# Read annotators from header
if i is 0:
for col_name in row[1:]:
header.append(col_name)
# Read data
else:
coder = None
for j, col in enumerate(row):
# Skip the first col
if j is 0:
coder = str(col)
dataset[item][coder] = list()
else:
dataset[item][coder].append(int(col))
dataset[item][coder] = tuple(dataset[item][coder])
return dataset
'''
from __future__ import absolute_import
from segeval.data import Dataset
from segeval.format import BoundaryFormat
KAZANTSEVA2012_G5 = Dataset(
{'ch1': {'an4': (2, 8, 2, 1),
'an1': (11, 2),
'an2': (2, 1, 7, 2, 1),
'an3': (9, 4)},
'ch11':{'an4': (10, 4, 3, 2, 1, 8, 3, 2, 6, 1, 2, 8, 10, 9, 4, 10, 4,
8, 4, 3, 5, 4),
'an1': (20, 22, 8, 11, 11, 11, 13, 11, 4),
'an2': (1, 7, 2, 4, 3, 3, 10, 1, 1, 5, 3, 2, 8, 3, 3, 3, 14,
4, 1, 1, 4, 4, 2, 7, 3, 2, 3, 3, 2, 1, 1),
'an3': (10, 10, 15, 11, 4, 10, 13, 5, 4, 23, 6)},
'ch4': {'an4': (2, 9, 2, 5, 2, 19, 1, 6),
'an1': (17, 25, 4),
'an2': (1, 10, 2, 4, 1, 2, 3, 10, 6, 6, 1),
'an3': (12, 5, 29)},
'ch3': {'an4': (2, 3, 4, 2, 5, 17, 4, 1),
'an1': (6, 5, 27),
'an2': (3, 8, 2, 3, 17, 2, 2, 1),
'an3': (3, 15, 15, 5)}})
'''
Segmentations provided by 4 coders (labeled group 5) of 4 chapters of
"The Moonstone" (Collins1868)_ collected by (KazantsevaSzpakowicz2012)_::
KAZANTSEVA2012_G5 = Dataset(
{'ch1': {'an4': (2, 8, 2, 1),