life-echo/api/tests/test_prompt_language_branching.py

"""Prompt builders 在 language='en' 时返回纯 ASCII；language='zh' 时含中文（防回归）。"""

from __future__ import annotations

from app.agents.chat.output_rules import (
    chat_output_rules,
    chat_output_rules_en,
    chat_voice_style,
    chat_voice_style_en,
)
from app.agents.chat.personas import AGENT_NAME_EN, AGENT_NAME_ZH, agent_name
from app.agents.chat.prompts_conversation import (
    get_guided_conversation_prompt,
    get_opening_prompt,
)
from app.agents.chat.prompts_profile import (
    get_profile_followup_prompt,
    get_profile_greeting_prompt,
)
from app.agents.memoir.prompts import (
    get_creative_title_json_prompt,
    get_memoir_fidelity_facts_only_prompt,
    get_memoir_fidelity_system_prompt,
    get_narrative_json_prompt,
)
from app.agents.stage_constants import (
    chapter_category_display,
    stage_display_name,
)


def _has_cjk(s: str) -> bool:
    return any("\u4e00" <= ch <= "\u9fff" for ch in s)


# ── chat output rules ────────────────────────────────────────────────


def test_chat_output_rules_en_has_no_cjk() -> None:
    txt = chat_output_rules_en()
    assert txt.strip()
    assert not _has_cjk(txt)


def test_chat_voice_style_en_has_no_cjk() -> None:
    txt = chat_voice_style_en()
    assert txt.strip()
    assert not _has_cjk(txt)


def test_chat_output_rules_zh_unchanged_has_cjk() -> None:
    assert _has_cjk(chat_output_rules())


def test_chat_voice_style_zh_unchanged_has_cjk() -> None:
    assert _has_cjk(chat_voice_style())


# ── memoir narrative ────────────────────────────────────────────────


def test_memoir_fidelity_facts_only_en_branch_has_no_cjk() -> None:
    assert not _has_cjk(get_memoir_fidelity_facts_only_prompt(language="en"))


def test_memoir_fidelity_facts_only_zh_default_has_cjk() -> None:
    assert _has_cjk(get_memoir_fidelity_facts_only_prompt())


def test_memoir_fidelity_system_en_branch_has_no_cjk() -> None:
    assert not _has_cjk(get_memoir_fidelity_system_prompt(language="en"))


def test_get_narrative_json_prompt_en_has_no_cjk() -> None:
    out = get_narrative_json_prompt(
        stage="childhood",
        slots={"place": "the village"},
        new_content="[User's oral memory this turn]\nI grew up by the river.",
        language="en",
    )
    assert not _has_cjk(out)
    assert "paragraphs" in out


def test_get_narrative_json_prompt_zh_default_has_cjk() -> None:
    out = get_narrative_json_prompt(
        stage="childhood",
        slots={"place": "村里"},
        new_content="【本段用户口述】\n我小时候住在河边。",
    )
    assert _has_cjk(out)


# ── memoir title ────────────────────────────────────────────────────


def test_creative_title_json_prompt_en_has_no_cjk() -> None:
    out = get_creative_title_json_prompt(
        stage="childhood",
        emotion="warm",
        slots={"place": "the river"},
        language="en",
    )
    assert not _has_cjk(out)
    assert "title" in out


def test_creative_title_json_prompt_zh_default_has_cjk() -> None:
    out = get_creative_title_json_prompt(
        stage="childhood",
        emotion="warm",
        slots={"place": "河边"},
    )
    assert _has_cjk(out)


# ── stage / category display helpers ────────────────────────────────


def test_stage_display_name_branches() -> None:
    assert stage_display_name("childhood", language="en") == "Childhood"
    assert stage_display_name("childhood", language="zh") == "童年时光"
    # unknown stages pass through
    assert stage_display_name("unknown", language="en") == "unknown"


def test_chapter_category_display_branches() -> None:
    assert chapter_category_display("childhood", language="en") == "Childhood & Early Years"
    assert chapter_category_display("childhood", language="zh") == "童年与成长背景"


# ── agent brand name (interviewer identity) ─────────────────────────


def test_agent_name_constants_aligned_with_brand() -> None:
    """中英品牌名是单一来源，prompt / UI / 兜底标题统一引用此处。"""
    assert AGENT_NAME_ZH == "岁月知己"
    assert AGENT_NAME_EN == "Life Echo"


def test_agent_name_helper_handles_inputs() -> None:
    assert agent_name("zh") == "岁月知己"
    assert agent_name("en") == "Life Echo"
    assert agent_name("EN") == "Life Echo"
    assert agent_name(" en ") == "Life Echo"
    assert agent_name(None) == "岁月知己"  # type: ignore[arg-type]
    assert agent_name("ja") == "岁月知己"


def test_profile_greeting_prompt_introduces_life_echo_in_en() -> None:
    en = get_profile_greeting_prompt(
        ["birth_year", "occupation"], nickname="Sam", language="en"
    )
    assert "Life Echo" in en
    assert "岁月知己" not in en


def test_profile_greeting_prompt_keeps_chinese_brand_in_zh() -> None:
    zh = get_profile_greeting_prompt(["birth_year"], nickname="老王")
    assert "岁月知己" in zh
    assert "Life Echo" not in zh


def test_profile_followup_prompt_introduces_life_echo_in_en() -> None:
    en = get_profile_followup_prompt(
        missing_fields=["occupation"],
        filled_fields={"birth_year": "1990"},
        language="en",
    )
    assert "Life Echo" in en
    assert "岁月知己" not in en


def test_profile_followup_prompt_full_basics_branch_introduces_life_echo() -> None:
    en = get_profile_followup_prompt(
        missing_fields=[],
        filled_fields={
            "birth_year": "1990",
            "birth_place": "Boston",
            "grew_up_place": "Boston",
            "occupation": "engineer",
        },
        language="en",
    )
    assert "Life Echo" in en
    assert "岁月知己" not in en


def test_opening_prompt_introduces_life_echo_in_en() -> None:
    out = get_opening_prompt(
        current_stage="childhood",
        empty_slots_readable=["place", "people"],
        language="en",
    )
    assert "Life Echo" in out
    assert "岁月知己" not in out


def test_guided_conversation_prompt_introduces_life_echo_in_en() -> None:
    out = get_guided_conversation_prompt(
        current_stage="childhood",
        empty_slots=["place"],
        filled_slots={},
        language="en",
    )
    assert "Life Echo" in out
    assert "岁月知己" not in out