Airbnb 的上下文感知身份模型：如何让社交功能与隐私共存

# context_identity.py — 最小上下文感知身份模型示例
# 运行: python context_identity.py

from dataclasses import dataclass, field
from enum import Enum
from typing import Optional
import hashlib

class Context(Enum):
    STAY = "stay"
    EXPERIENCE = "experience"
    COMMUNITY = "community"

@dataclass
class GlobalIdentity:
    """内部唯一身份，仅服务端可见"""
    uid: str
    real_name: str
    email: str
    phone: str
    compliance_id: str  # 合规审计用

@dataclass
class ScopedProfile:
    """上下文隔离的外部档案"""
    scoped_id: str          # 上下文内唯一标识，非全局 uid
    display_name: str
    avatar_url: str
    bio: Optional[str] = None
    # 不同上下文有不同字段白名单
    visible_fields: list = field(default_factory=list)

def _derive_scoped_id(uid: str, context: Context) -> str:
    """从全局 uid + 上下文派生不可逆的 scoped_id，防止跨上下文反查"""
    raw = f"{uid}|{context.value}|SALT_REPLACE_WITH_SECRET"
    return hashlib.sha256(raw.encode()).hexdigest()[:16]

# 字段白名单：每个上下文只暴露必要字段
CONTEXT_FIELD_POLICY = {
    Context.STAY: ["display_name", "avatar_url", "bio"],
    Context.EXPERIENCE: ["display_name", "avatar_url", "bio"],
    Context.COMMUNITY: ["display_name", "avatar_url"],
}

# 显示名策略：不同上下文使用不同命名规则
def _derive_display_name(real_name: str, context: Context) -> str:
    if context == Context.STAY:
        return real_name  # 住宿场景需要真实姓名
    if context == Context.EXPERIENCE:
        # 体验场景只显示姓氏首字 + 随机标签
        return f"{real_name[0]}·爱好者"
    return f"{real_name[0]}·用户"

def build_scoped_profile(identity: GlobalIdentity, context: Context) -> ScopedProfile:
    """根据上下文策略构建隔离档案"""
    scoped_id = _derive_scoped_id(identity.uid, context)
    display_name = _derive_display_name(identity.real_name, context)
    avatar_seed = scoped_id  # 不同上下文用不同头像种子
    avatar_url = f"https://avatar.example.com/{avatar_seed}.png"
    visible_fields = CONTEXT_FIELD_POLICY[context]
    bio = f"{context.value} 场景的公开简介" if "bio" in visible_fields else None

    return ScopedProfile(
        scoped_id=scoped_id,
        display_name=display_name,
        avatar_url=avatar_url,
        bio=bio,
        visible_fields=visible_fields,
    )

# ---- 演示 ----
if __name__ == "__main__":
    user = GlobalIdentity(
        uid="u_7f3a9c",
        real_name="李明",
        email="liming@example.com",
        phone="+86-138-xxxx-xxxx",
        compliance_id="c_001",
    )

    for ctx in Context:
        profile = build_scoped_profile(user, ctx)
        print(f"\n=== {ctx.value} 上下文档案 ===")
        print(f"  scoped_id:    {profile.scoped_id}")
        print(f"  display_name: {profile.display_name}")
        print(f"  avatar_url:   {profile.avatar_url}")
        print(f"  bio:          {profile.bio}")
        print(f"  visible:      {profile.visible_fields}")
        # 关键：全局 uid、真实姓名、电话、邮箱绝不出现

    # 跨上下文关联测试
    stay_id = build_scoped_profile(user, Context.STAY).scoped_id
    exp_id = build_scoped_profile(user, Context.EXPERIENCE).scoped_id
    print(f"\n跨上下文 scoped_id 是否相同: {stay_id == exp_id}")
    print("（不同上下文的 scoped_id 完全不同，无法通过 ID 反查同一用户）")

=== stay 上下文档案 ===
  scoped_id:    5e8f3a1b9c2d4e6f
  display_name: 李明
  avatar_url:   https://avatar.example.com/5e8f3a1b9c2d4e6f.png
  bio:          stay 场景的公开简介
  visible:      ['display_name', 'avatar_url', 'bio']

=== experience 上下文档案 ===
  scoped_id:    a1b2c3d4e5f67890
  display_name: 李·爱好者
  avatar_url:   https://avatar.example.com/a1b2c3d4e5f67890.png
  bio:          experience 场景的公开简介
  visible:      ['display_name', 'avatar_url', 'bio']

=== community 上下文档案 ===
  scoped_id:    f1e2d3c4b5a69780
  display_name: 李·用户
  avatar_url:   https://avatar.example.com/f1e2d3c4b5a69780.png
  bio:          None
  visible:      ['display_name', 'avatar_url']

跨上下文 scoped_id 是否相同: False