"""docx XML 清洗：去脏值（单位 / NaN / 浮点）以避开 docx4j 8.3.9 JAXB 解析失败。

参考 wordparse_server.py 的 _clean_docx_xml，独立成模块方便测试。
"""
import re
import os
import zipfile

XML_DECL = re.compile(r'<\?xml[^?]*\?>')


def _fix_text(txt: str) -> str:
    """修三类问题：
      1. 带单位的数字属性："210mm" / "25.4cm" → "0"
      2. NaN / Inf / -Inf / 空字符串 → "0"
      3. 浮点属性 → 整数（JAXB BigInteger 不接小数）
    """
    decl = XML_DECL.search(txt)
    decl_str = decl.group(0) if decl else '<?xml version="1.0" encoding="UTF-8" standalone="yes"?>'
    if decl:
        txt = XML_DECL.sub('__XMLDECL__', txt, count=1)

    # 1. 单位
    txt = re.sub(r'="(-?[0-9]*\.?[0-9]+)\s*(mm|cm|pt|px|in|em|%)"', r'="0"', txt, flags=re.I)
    # 2. NaN / Inf
    txt = re.sub(r'="(NaN|nan|Inf|-Inf|inf|-inf)"', '="0"', txt, flags=re.I)
    # 3. 空
    txt = re.sub(r'=""', '="0"', txt)
    # 4. 浮点 → 整数
    def _num(m):
        try:
            f = float(m.group(1))
            if f != f or f == float('inf') or f == float('-inf'):
                return '="0"'
            return f'="{int(f)}"'
        except Exception:
            return m.group(0)

    txt = re.sub(r'="(-?[0-9]+\.[0-9]+(?:[eE][-+]?[0-9]+)?)"', _num, txt)

    if decl:
        txt = txt.replace('__XMLDECL__', decl_str, 1)
    return txt


def clean_docx(path: str) -> str:
    """清洗 docx 里的 word/document.xml，幂等。失败回退原文件。"""
    marker = path + '.cleaned'
    if os.path.exists(marker):
        return path

    try:
        with zipfile.ZipFile(path, 'r') as zin:
            entries = []
            changed = False
            for item in zin.infolist():
                data = zin.read(item.filename)
                if item.filename == 'word/document.xml':
                    raw = data.decode('utf-8', errors='replace')
                    new = _fix_text(raw)
                    if new != raw:
                        data = new.encode('utf-8')
                        changed = True
                entries.append((item, data))
        if not changed:
            open(marker, 'w').close()
            return path
        tmp = path + '.tmp'
        with zipfile.ZipFile(tmp, 'w', zipfile.ZIP_DEFLATED) as zout:
            for item, data in entries:
                zout.writestr(item, data)
        os.replace(tmp, path)
        open(marker, 'w').close()
        return path
    except Exception:
        return path