group_xinghuo_jinrong/tests/test_wave2_profile_slots.py

"""Wave 2：画像槽位定义与归一化纯函数测试（不依赖 DB/LLM）。"""

from __future__ import annotations

import pytest

from app.config.profile_slots import (
    HIGH_SENSITIVITY_PATHS,
    SLOT_PATHS,
    SLOTS,
    build_extraction_prompt,
    get_slot,
    normalize_value,
)


def test_slots_integrity():
    """12 槽位、path 唯一、字段合法、与 style_tags 路径对齐。"""
    assert len(SLOTS) == 12
    assert len(SLOT_PATHS) == len(set(SLOT_PATHS)) == 12

    valid_merge = {"latest", "set_union"}
    valid_fmt = {"band", "enum", "list", "text"}
    valid_src = {"chat", "chat_or_questionnaire", "chat_or_consult_or_trade"}
    valid_sens = {"high", "medium", "low"}
    for s in SLOTS:
        assert s["merge_mode"] in valid_merge
        assert s["value_format"] in valid_fmt
        assert s["allowed_source"] in valid_src
        assert s["sensitivity"] in valid_sens
        assert s["name"] and s["examples"]
        # style_tags 顶层路径必须对齐 06 文档 §4.1
        top = s["path"].split(".")[0]
        assert top in {"basic", "financial", "investment", "lifecycle"} or s["path"] == "threshold_pref_summary"

    # 高敏槽位恰好 3 个（年龄/年收入/月可投）
    assert set(HIGH_SENSITIVITY_PATHS) == {
        "basic.age_band",
        "financial.income_band",
        "financial.monthly_investable",
    }
    # 集合累计槽位恰好 2 个（偏好/排除）
    set_union_paths = {s["path"] for s in SLOTS if s["merge_mode"] == "set_union"}
    assert set_union_paths == {
        "investment.product_preferences",
        "investment.excluded_products",
    }


def test_get_slot():
    assert get_slot("basic.city")["name"] == "所在城市"
    assert get_slot("not.a.slot") is None


def test_extraction_prompt_covers_all_slots():
    prompt = build_extraction_prompt()
    for s in SLOTS:
        assert s["path"] in prompt
        assert s["name"] in prompt
    # 高敏保护规则必须出现在 prompt 中
    assert "高敏感" in prompt
    assert "禁止" in prompt


@pytest.mark.parametrize(
    "raw,expected",
    [
        ("我今年28", "25-30"),
        ("30岁", "25-30"),       # 年龄档含上界
        ("我35了", "31-35"),
        ("今年40", "36-40"),
        ("我爸72", "61+"),
        ("刚工作22", "18-24"),
    ],
)
def test_normalize_age_band(raw, expected):
    assert normalize_value("basic.age_band", raw) == expected


@pytest.mark.parametrize("raw", ["很年轻", "年龄不大", ""])
def test_normalize_age_invalid(raw):
    assert normalize_value("basic.age_band", raw) is None


@pytest.mark.parametrize(
    "raw,expected",
    [
        ("年薪30万", "20-50万"),
        ("月入2万", "20-50万"),      # 2万*12=24万
        ("一年到手十几万", "10-20万"),  # 中文数字"十几万"取下界 10 万
        ("月薪8000", "10万以下"),     # 8000元*12=9.6万
        ("月薪一万", "10-20万"),     # 中文数字 1万*12=12万
        ("年收入80万", "50-100万"),
    ],
)
def test_normalize_income_band(raw, expected):
    assert normalize_value("financial.income_band", raw) == expected


@pytest.mark.parametrize(
    "raw,expected",
    [
        ("每月能拿3000定投", "3000-5000"),
        ("月可投4999", "3000-5000"),
        ("月可投5000左右", "5000-10000"),
        ("每月1万", "10000以上"),
        ("每个月2000块", "1000-3000"),
        ("月投800", "1000以下"),
        ("每月三千定投", "3000-5000"),
    ],
)
def test_normalize_monthly_investable(raw, expected):
    assert normalize_value("financial.monthly_investable", raw) == expected


def test_normalize_monthly_no_number():
    assert normalize_value("financial.monthly_investable", "一个月剩不下钱投") is None


@pytest.mark.parametrize(
    "raw,expected",
    [
        ("我本科毕业", "本科"),
        ("研究生学历", "硕士及以上"),
        ("大专", "大专"),
        ("博士在读", "硕士及以上"),
        ("高中毕业", "高中及以下"),
    ],
)
def test_normalize_education(raw, expected):
    assert normalize_value("basic.education", raw) == expected


def test_normalize_education_unknown():
    assert normalize_value("basic.education", "我学历还行") is None


@pytest.mark.parametrize(
    "raw,expected",
    [
        ("半年后要用钱", "1年以内"),
        ("这笔钱三年不用", "1-3年"),
        ("两三年吧", "1-3年"),
        ("计划持有三年以上", "3-5年"),
        ("长期放着养老", "5年以上"),
    ],
)
def test_normalize_horizon(raw, expected):
    assert normalize_value("investment.horizon", raw) == expected


@pytest.mark.parametrize(
    "raw,expected",
    [
        ("刚工作单身", "单身奋斗期"),
        ("刚结婚还没孩子", "家庭形成期"),
        ("孩子上小学", "家庭成长期"),
        ("孩子都上大学了", "家庭成熟期"),
        ("退休了", "退休养老期"),
    ],
)
def test_normalize_stage(raw, expected):
    assert normalize_value("lifecycle.stage", raw) == expected


def test_normalize_product_preferences():
    assert normalize_value("investment.product_preferences", "我喜欢指数基金和货币基金") == [
        "指数基金",
        "货币基金",
    ]


def test_normalize_excluded_products():
    assert normalize_value("investment.excluded_products", "私募不碰，期货也不做") == [
        "私募基金",
        "期货",
    ]


def test_normalize_list_alias():
    assert normalize_value("investment.product_preferences", "想买点债基") == ["债券基金"]


def test_normalize_list_non_vocab_dropped():
    """非词表自由值（如个股名）不进画像。"""
    assert normalize_value("investment.product_preferences", "我看好白酒股") is None


def test_normalize_text_slots():
    # basic.city Wave 3 起归一为干净城市名（剥前后缀）
    assert normalize_value("basic.city", "我在上海") == "上海"
    assert normalize_value("basic.city", "老家成都的") == "成都"
    assert normalize_value("basic.city", "我搬去杭州了") == "杭州"
    assert normalize_value("basic.city", "在深圳上班") == "深圳"
    assert normalize_value("basic.city", "浙江省杭州市") == "杭州"
    assert normalize_value("basic.city", "我在上海出差顺便开会") is None  # 清洗后超 7 字，非城市名
    assert normalize_value("investment.goal", "存钱买房") == "存钱买房"
    assert normalize_value("threshold_pref_summary", "亏10%就提醒我") == "亏10%就提醒我"
    assert normalize_value("basic.city", "") is None


def test_normalize_unknown_slot():
    assert normalize_value("not.exists", "随便") is None