def _append_text(element: etree.ElementBase, text: Optional[str]) -> None:
if not text:
return
last_child = _get_last_child_or_none(element)
if last_child is not None and last_child.tail:
last_child.tail = last_child.tail + '' + text
elif last_child is not None:
last_child.tail = text
elif element.text:
element.text = element.text + '' + text
else:
element.text = text
def extend_element(
element: etree.ElementBase,
children_or_attributes: Iterable[etree.ElementBase]
):
for item in children_or_attributes:
if isinstance(item, dict):
element.attrib.update(item)
continue
if isinstance(item, str):
try:
previous_element = element[-1]
except IndexError:
previous_element = None
if previous_element is not None:
previous_element.tail = (
(previous_element.tail or '')
+ item
)
else:
element.text = (
(element.text or '')
+ item
)
continue
element.append(item)
def append_text(element: etree.ElementBase, text: str):
try:
if element[-1].tail:
element[-1].tail += text
else:
element[-1].tail = text
except IndexError:
if element.text:
element.text += text
else:
element.text = text
def _transform(
self,
token_generator: str,
operation: Operation,
element: etree.ElementBase,
):
value = ''.join(element.itertext())
if not value:
return
element.clear()
element.text = transform(value, operation, token_generator)
def replace_notelabel(el: ElementBase, text: str) -> None:
'修改脚注的标签'
while el.tag != 'a':
el = el.getparent()
if el.tag != 'a':
a = el.find('.//a')
if a is not None:
el = a
while len(el):
el = el[0]
el.text = text
def transform_context(context: etree.ElementBase) -> etree.ElementBase:
sent: List[str] = []
before = context.text
head_tag = context[0]
head = head_tag.text
after = head_tag.tail
before_tok = space_tokenize(before)
head_tok = space_tokenize(head)
after_tok = space_tokenize(after)
sent = before_tok + head_tok + after_tok
new_sent = transform_tokens(sent)
new_before = new_sent[:len(before_tok)]
new_head = new_sent[len(before_tok):len(before_tok) + len(head_tok)]
new_after = new_sent[len(before_tok) + len(head_tok):]
context.text = "\n" + "".join(tok + " " for tok in new_before)
head_tag.text = " ".join(tok for tok in new_head)
head_tag.tail = "".join(" " + tok for tok in new_after) + "\n"
return context
def _transform(self, operation: Callable, element: etree.ElementBase):
value = ''.join(element.itertext())
if not value:
return
element.clear()
element.text = operation(value)
def try_fix_namespace_path(node: etree.ElementBase):
if is_namespace_path(node):
node.text = node.text.replace(':', os.sep)