Merge pull request #29 from RockyCYG/master

feat: Remove possible incompatible characters for XML compatibility

Author: codematrixer

hmdriver2/_xpath.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 
+import re
 from typing import Dict
 from lxml import etree
 from functools import cached_property
@@ -33,15 +34,26 @@ def __call__(self, xpath: str) -> '_XMLElement':
 
  return _XMLElement(None, self._d)
 
+ @staticmethod
+ def _sanitize_text(text: str) -> str:
+ """Remove XML-incompatible control characters."""
+ return re.sub(r'[\x00-\x1F\x7F]', '', text)
+
  @staticmethod
  def _json2xml(hierarchy: Dict) -> etree.Element:
+ """Convert JSON-like hierarchy to XML."""
  attributes = hierarchy.get("attributes", {})
- tag = attributes.get("type", "orgRoot") or "orgRoot"
- xml = etree.Element(tag, attrib=attributes)
+
+ # 过滤所有属性的值，确保无非法字符
+ cleaned_attributes = {k: _XPath._sanitize_text(str(v)) for k, v in attributes.items()}
+
+ tag = cleaned_attributes.get("type", "orgRoot") or "orgRoot"
+ xml = etree.Element(tag, attrib=cleaned_attributes)
 
  children = hierarchy.get("children", [])
  for item in children:
  xml.append(_XPath._json2xml(item))
+
  return xml