baodan/api/insurance/ppt/masking.py

"""脱敏工具模块。

提供名称脱敏功能，用于 PPT/海报导出时替换真实保司和产品名称。
"""
import re
import logging
from difflib import SequenceMatcher

logger = logging.getLogger(__name__)

# 保司/产品名称中应保留的常见后缀
_PRESERVE_SUFFIXES = [
    "保险", "人寿", "财险", "资产", "金融", "集团", "控股",
    "储蓄保险计划", "保险计划", "储蓄计划", "保障计划", "危疾保障",
    "终身寿险", "定期寿险", "万用寿险",
]


def fallback_mask_name(name: str) -> str:
    """兜底脱敏规则：对未配置脱敏字段的名称进行智能脱敏。

    规则：
    - 长度 <= 2：保留首字，后面用 X
    - 长度 3-5：保留首尾，中间用 X
    - 长度 > 5：保留前 1-2 个关键字，替换中间 1-2 字
    - 引号、括号、后缀（保险/计划等）尽量保留
    """
    if not name or len(name.strip()) <= 1:
        return name

    name = name.strip()

    # 处理带引号的产品名，如「财富盈活」储蓄保险计划
    quote_match = re.match(r'^([「『"\'（(].*?[」』"\'）)])\s*(.*)$', name)
    if quote_match:
        inner = quote_match.group(1)
        suffix = quote_match.group(2)
        # 对引号内部分脱敏
        inner_clean = inner[1:-1]  # 去掉引号
        masked_inner = _mask_core(inner_clean)
        return f"{inner[0]}{masked_inner}{inner[-1]}{suffix}"

    # 普通名称
    return _mask_core(name)


def _mask_core(text: str) -> str:
    """对核心文字进行脱敏。"""
    if len(text) <= 1:
        return text
    if len(text) == 2:
        return text[0] + "X"
    if len(text) <= 5:
        return text[0] + "X" + text[-1]
    # 长度 > 5：保留前 2 字和后 2 字，中间用 X 替换
    return text[:2] + "X" + text[-2:]


def apply_company_mask(company_dict: dict, use_masked: bool) -> dict:
    """对公司信息字典应用脱敏。

    优先使用 maskedDisplayName，否则使用兜底脱敏规则。
    """
    if not use_masked:
        return company_dict

    masked_name = company_dict.get("maskedDisplayName", "")
    if masked_name:
        company_dict["displayName"] = masked_name
    else:
        original = company_dict.get("displayName", "")
        if original:
            company_dict["displayName"] = fallback_mask_name(original)

    return company_dict


def apply_product_mask(product_dict: dict, use_masked: bool) -> dict:
    """对产品信息字典应用脱敏。"""
    if not use_masked:
        return product_dict

    masked_name = product_dict.get("maskedDisplayName", "")
    if masked_name:
        product_dict["displayName"] = masked_name
    else:
        original = product_dict.get("displayName", "")
        if original:
            product_dict["displayName"] = fallback_mask_name(original)

    return product_dict


def mask_text(text: str, replacements: dict[str, str]) -> str:
    """对文本中的名称进行替换。

    参数:
        text: 需要替换的文本
        replacements: {真实名: 脱敏名} 映射
    """
    if not text or not replacements:
        return text
    for real_name, masked_name in replacements.items():
        if real_name and masked_name and real_name != masked_name:
            text = text.replace(real_name, masked_name)
    return text


def build_name_replacements(companies: list[dict] = None, products: list[dict] = None,
                             use_masked: bool = False) -> dict[str, str]:
    """构建名称替换映射。

    返回: {真实名: 脱敏名} 字典
    """
    if not use_masked:
        return {}

    replacements = {}

    if companies:
        for c in companies:
            real = c.get("displayName", "")
            masked = c.get("maskedDisplayName", "")
            if not masked:
                masked = fallback_mask_name(real) if real else ""
            if real and masked and real != masked:
                replacements[real] = masked
            # 也处理中文名
            real_zh = c.get("nameZh", "")
            if real_zh and real_zh != real and masked:
                replacements[real_zh] = masked

    if products:
        for p in products:
            real = p.get("displayName", "")
            masked = p.get("maskedDisplayName", "")
            if not masked:
                masked = fallback_mask_name(real) if real else ""
            if real and masked and real != masked:
                replacements[real] = masked

    return replacements


def match_product_by_name(product_name: str, all_products: list[dict]) -> dict | None:
    """通过名称和别名匹配产品表中的产品。

    返回匹配到的产品 dict 或 None。
    """
    if not product_name or not all_products:
        return None

    product_name_lower = product_name.strip().lower()

    for p in all_products:
        # 精确匹配
        if p.get("displayName", "").strip().lower() == product_name_lower:
            return p
        # 别名匹配
        aliases = p.get("aliases", [])
        if isinstance(aliases, list):
            for alias in aliases:
                if isinstance(alias, str) and alias.strip().lower() == product_name_lower:
                    return p

    # 模糊匹配：相似度 > 0.7
    best_match = None
    best_score = 0.0
    for p in all_products:
        name = p.get("displayName", "")
        if not name:
            continue
        score = SequenceMatcher(None, product_name_lower, name.strip().lower()).ratio()
        if score > best_score and score > 0.7:
            best_score = score
            best_match = p

    return best_match