def test_register(self): registry = formats.get_registry() assert_false(CustomFormat in registry.values()) formats.register('trt', CustomFormat) assert_true(CustomFormat in registry.values()) assert_true('trt' in registry.keys())
def _read_data(self, dataset, format=None): """Reads a data file and returns an iterable that can be used as testing or training data. """ # Attempt to detect file format if "format" isn't specified if not format: format_class = formats.detect(dataset) if not format_class: raise FormatError("Could not automatically detect format for the given " "data source.") else: registry = formats.get_registry() if format not in registry.keys(): raise ValueError("'{0}' format not supported.".format(format)) format_class = registry[format] return format_class(dataset, **self.format_kwargs).to_iterable()
def _read_data(self, dataset, format=None): """Reads a data file and returns an iterable that can be used as testing or training data. """ # Attempt to detect file format if "format" isn't specified if not format: format_class = formats.detect(dataset) if not format_class: raise FormatError('Could not automatically detect format for the given ' 'data source.') else: registry = formats.get_registry() if format not in registry.keys(): raise ValueError("'{0}' format not supported.".format(format)) format_class = registry[format] return format_class(dataset, **self.format_kwargs).to_iterable()
def read_data(file, format=None, **kwargs): """Reads a data file and returns an iterable that can be used as testing or training data. Adapted from: https://github.com/sloria/TextBlob/blob/dev/textblob/classifiers.py#L128 """ # Attempt to detect file format if "format" isn't specified if not format: format_class = formats.detect(file) if not format_class: raise FormatError('Could not automatically detect format for the ' 'given data source.') else: registry = formats.get_registry() if format not in registry.keys(): raise ValueError("'{0}' format not supported.".format(format)) format_class = registry[format] return format_class(file, **kwargs).to_iterable()
def test_available(self): registry = formats.get_registry() assert_true('csv' in registry.keys()) assert_true('json' in registry.keys()) assert_true('tsv' in registry.keys())