from requests_html import HTML
from lxml import html as lxml_html
class TTSHTMLParser:
html: HTML
def __init__(self, html: str):
# keep original string and also parse with lxml for reliable element API
self.raw_html = html
self.html = HTML(html=html)
self._root = lxml_html.fromstring(html)
# 解析为[音色,文本]的元组列表
# 从最里层节点遍历获取
def parse(
self,
strip: bool = True,
delete_eol: bool = True,
delete_space: bool = True,
strict: bool = False, # 只保留字母,数字,中文和部分常用标点
) -> tuple[str, str]:
result = []
def clean(s: str) -> str:
if s is None:
return ""
t = s if not strip else s.strip()
if delete_space:
t = t.replace(" ", "")
if strict:
allowed_punc = {
",",
".",
";",
"-",
"—",
"-",
"–",
",",
"。",
";",
"、",
"!",
"?",
"!",
"?",
"%",
"%",
"+",
"+",
}
def is_chinese(ch: str) -> bool:
return "\u4e00" <= ch <= "\u9fff"
t = "".join(
c for c in t if c.isalnum() or is_chinese(c) or c in allowed_punc
)
return t.replace("\n", ",") if delete_eol else t
def process(elem) -> None:
if elem.tag == "root":
for child in elem:
process(child)
return
texts = elem.xpath("text()")
before_raw = texts[0] if texts else ""
before = clean(before_raw)
if before != "":
result.append((elem.tag, before))
children = list(elem)
if not children:
return
for child in children:
process(child)
after_raw = "".join(texts[1:]) if len(texts) > 1 else ""
after = clean(after_raw)
if after != "":
result.append((elem.tag, after))
root = self._root
process(root)
return tuple(result)