def __init__(self, config):
self.config = config
Manipulation.__init__(self)
# Filter.__init__(self, config)
win_sizes = self.config.window if self.config.window is not None else [
1, 1
]
if not is_collection(win_sizes):
win_sizes = [win_sizes] * 2
self.before, self.after = win_sizes
def html_append_class(tag, classes):
appended = False
if 'class' not in tag.attrs:
tag['class'] = []
if not utils.is_collection(classes):
classes = (classes, )
for class_ in classes:
if class_ not in tag['class']:
tag['class'].append(class_)
appended = True
return appended
def html_find_parent(tag, names, cutoff=None):
if not utils.is_collection(names):
names = [names]
parent = tag.parent
while (parent is not None):
if (cutoff is not None and parent is cutoff):
return None
if parent.name in names:
return parent
parent = parent.parent
return parent
def get_data_class(inputs):
"""
Get datatype class of the input
"""
# input is already a datatype
if issubclass(type(inputs), Datatype):
return type(inputs)
# text info dict
if type(inputs) in [dict, OrderedDict] and "words" in inputs:
return Text
if is_collection(inputs[0]):
return get_data_class(inputs[0])
if type(inputs[0]) in (str, np.str_):
return Text
elif isinstance(inputs[0], Number) or type(inputs[0]) is np.ndarray:
return Numeric
else:
error(f"Unsupported type of data: {inputs[0]}")
def html_shallow_search(starting_tag, names, filter=None):
if isinstance(starting_tag, soup.NavigableString):
return []
if not utils.is_collection(names):
names = [names]
if starting_tag.name in names:
if filter is None or filter(starting_tag):
return [starting_tag]
results = []
for tag in starting_tag.children:
if isinstance(tag, soup.NavigableString):
continue
if tag.name in names:
if filter is None or filter(tag):
results.append(tag)
else:
results = results + html_shallow_search(tag, names, filter)
return results