translate_pattern incorrectly handles \s when preceded by \- inside a character class

[daniel55411]

Description

When \s appears after \- inside a character class, translate_pattern fails to translate
\s into whitespace codepoints and leaves it as a literal \\s in the resulting Python pattern.

This causes XSD pattern validation via xmlschema to silently fail for strings containing whitespace.

Steps to Reproduce

from elementpath.regex import translate_pattern

# \s is NOT translated — stays as literal \\s
print(translate_pattern(r"[\-\s',]{1,255}"))
# Output: [',\-\\s]{1,255}  ← \s not expanded

# \s IS translated correctly
print(translate_pattern(r"[\s\-',]{1,255}"))
# Output:
# [
#  ',\-]{1,255}
# \s expanded to whitespace codepoints

Expected Behavior

\s should be translated to the corresponding whitespace codepoints regardless of its position
within the character class.

Actual Behavior

When \- appears before \s in the character class, \s is not translated and remains as \\s
in the output Python pattern.

Workaround

Moving \s to the beginning of the character class fixes the issue:

translate_pattern(r"[\s\-',]{1,255}")

Environment

Name: elementpath
Version: 5.1.1
Python: 3.11
Required-by: xmlschema

[brunato]

Hi,
thank you for detailed description of this failing case, I'm confident that it will be fixed with a proper change of this regex pattern:

 _re_char_set = re.compile(r'(?<!.-)(\\[nrt|.\-^?*+{}()\]sSdDiIcCwW]|\\[pP]{[a-zA-Z\-0-9]+})')

That pattern is part of CharacterClass, used to split character class content, in order to apply substitutions.