A base class for any node in an HTML tree.
An HTML element without content is called an empty node. It has a start tag but neither a content nor an end tag. It has only attributes.
The BaseNode() class is a base class for any of these HTML elements. It is intended to be overridden.
Create a new base node.
def __init__(self, parent: str=None, identifier: str=None, **kwargs):
"""Create a new base node.
:param parent: (str) Parent identifier
:param identifier: (str) This node identifier
:raises: NodeInvalidIdentifierError: if 'identifier' contains invalid characters or if invalid length
"""
if identifier is not None:
ident = BaseNode.validate_identifier(identifier)
self.__identifier = ident
else:
self.__identifier = str(uuid.uuid1())
self._parent = None
self.set_parent(parent)
Return the given identifier if it matches the requirements.
An identifier should contain at least 1 character and no whitespace.
@staticmethod
def validate_identifier(identifier: str) -> str:
"""Return the given identifier if it matches the requirements.
An identifier should contain at least 1 character and no whitespace.
:param identifier: (str) Key to be validated
:raises: NodeInvalidIdentifierError: if it contains invalid characters
:raises: NodeInvalidIdentifierError: if invalid length
:return: (str) Validated identifier
"""
entry = BaseNode.full_strip(identifier)
if len(entry) != len(identifier):
raise NodeInvalidIdentifierError(identifier)
if len(identifier) == 0:
raise NodeInvalidIdentifierError(identifier)
return identifier
Fully strip the string: multiple whitespace, tab and CR/LF.
@staticmethod
def full_strip(entry):
"""Fully strip the string: multiple whitespace, tab and CR/LF.
:return: (str) Cleaned string
"""
e = re.sub('[\\s]+', '', entry)
e = re.sub('[\t]+', '', e)
e = re.sub('[\n]+', '', e)
e = re.sub('[\r]+', '', e)
if '\ufeff' in e:
e = re.sub('\ufeff', '', e)
return e
Return the (supposed-) unique ID of the node within the scope of a tree.
@property
def identifier(self) -> str:
"""Return the (supposed-) unique ID of the node within the scope of a tree. """
return self.__identifier
Return true if node has no children.
def is_leaf(self) -> bool:
"""Return true if node has no children."""
return True
Return true if node has no parent, i.e. like root.
def is_root(self) -> bool:
"""Return true if node has no parent, i.e. like root."""
return self._parent is None
The parent identifier.
def get_parent(self) -> str:
"""The parent identifier.
:return: (str) node identifier
"""
return self._parent
Set the parent identifier.
def set_parent(self, node_id: str) -> None:
"""Set the parent identifier.
:param node_id: (str) Identifier of the parent
"""
if self.__identifier == node_id:
raise NodeKeyError(self.__identifier, node_id)
self._parent = node_id
To be overriden. Return True if the given node ID is a direct child.
def has_child(self, node_id: str) -> bool:
"""To be overriden. Return True if the given node ID is a direct child.
:param node_id: (str) Identifier of the node
:return: (bool) True if given identifier is a direct child.
"""
return not self.is_leaf()
To be overriden. Serialize the node into HTML.
def serialize(self, nbs: int=4) -> str:
"""To be overriden. Serialize the node into HTML.
:param nbs: (int) Number of spaces for the indentation
:return: (str)
"""
return ''
def __repr__(self):
return self.serialize()
def __str__(self):
return 'Node ({:s})'.format(self.identifier)