webmail-client/backend/tests/test_translation_text.py

"""What is sent to the translation service (app/translation.py)."""

from __future__ import annotations

import pytest

from app import translation
from app.translation import detect, languages_of, names_to_keep, protect_addresses, translate_html


def test_addresses_in_text_are_kept_from_translation() -> None:
    html = (
        '<p>My reel: <a href="https://vimeo.com/1">https://vimeo.com/1</a>, '
        "write to anna@x.example or see www.example.com today.</p>"
    )
    out = protect_addresses(html)
    assert '<span translate="no">https://vimeo.com/1</span>' in out
    assert '<span translate="no">anna@x.example</span>' in out
    assert '<span translate="no">www.example.com</span>' in out
    assert '<a href="https://vimeo.com/1">' in out  # attributes untouched


def test_style_blocks_are_left_alone() -> None:
    html = "<style>.a{background:url(https://x.example/a.png)}</style><p>Hi</p>"
    assert protect_addresses(html) == html


async def test_only_names_the_model_changes_are_kept(monkeypatch: pytest.MonkeyPatch) -> None:
    sent: list[object] = []

    async def post(path: str, payload: dict[str, object]) -> object:
        sent.append(payload["q"])
        alone = {
            "Lukas Remis": "Lukas Remis.",
            "Milan": "Mediolan",
            "Imaginary Pixels": "Imaginanal",
        }
        return {"translatedText": [alone[q] for q in payload["q"]]}  # type: ignore[attr-defined]

    monkeypatch.setattr(translation, "_post", post)
    kept = await names_to_keep(
        ["Milan", "Lukas Remis", "Imaginary Pixels", "Milan", "x"], "en", "pl"
    )
    assert kept == {"Milan", "Imaginary Pixels"}  # "Lukas Remis" stays in its sentence
    assert len(sent) == 1  # one request for all


async def test_all_names_kept_when_the_check_fails(monkeypatch: pytest.MonkeyPatch) -> None:
    async def post(path: str, payload: dict[str, object]) -> object:
        raise translation.TranslationError("down")

    monkeypatch.setattr(translation, "_post", post)
    assert await names_to_keep(["Milan", "Lukas Remis"], "en", "pl") == {"Milan", "Lukas Remis"}


def _fake_detector(monkeypatch: pytest.MonkeyPatch) -> list[str]:
    """A detector that knows Chinese signs, Polish letters, else English."""
    asked: list[str] = []

    async def detect_one(text: str) -> str:
        asked.append(text)
        if any("\u4e00" <= c <= "\u9fff" for c in text):
            return "zh-Hans"
        return "pl" if any(c in "ąćęłńóśźż" for c in text.lower()) else "en"

    monkeypatch.setattr(translation, "_detect_one", detect_one)
    return asked


async def test_short_parts_take_the_next_language(monkeypatch: pytest.MonkeyPatch) -> None:
    asked = _fake_detector(monkeypatch)
    parts = [
        "您好！",
        "Szanowny Kliencie,",  # short: goes with the Polish that follows
        "Widzę, że złożyłeś wniosek o otwarcie sklepu na platformie.",
        "-----",
        "Hello, my name is Jacqueline, the manager for Europe.",
        "Jacqueline Lian",  # short, last: goes with the one before
    ]
    assert await languages_of(parts) == ["zh-Hans", "pl", "pl", "", "en", "en"]
    assert "Jacqueline Lian" not in asked  # too short to ask about


async def test_a_message_in_three_languages(monkeypatch: pytest.MonkeyPatch) -> None:
    _fake_detector(monkeypatch)
    text = (
        "尊敬的客户，我是速卖通的官方客户经理，我从后台看到您申请入驻速卖通平台开店。\n"
        "Szanowny Kliencie,\n"
        "Widzę, że złożyłeś wniosek o otwarcie sklepu na platformie AliExpress.\n"
        "Dears,\nI noticed from the backend that you have applied to join the platform.\n"
    )
    assert set((await detect(text)).split(",")) == {"zh-Hans", "pl", "en"}
    # One language: one code.
    assert await detect("I noticed from the backend that you applied, thanks a lot.") == "en"


async def test_each_paragraph_from_its_own_language(monkeypatch: pytest.MonkeyPatch) -> None:
    _fake_detector(monkeypatch)
    calls: list[tuple[str, list[str]]] = []

    async def post(path: str, payload: dict[str, object]) -> object:
        q = payload["q"]
        if path == "/translate" and isinstance(q, list) and payload.get("source") != "auto":
            calls.append((str(payload["source"]), list(q)))
            return {"translatedText": [f"PL({x})" for x in q]}
        return {"translatedText": list(q) if isinstance(q, list) else q}  # names: unchanged

    monkeypatch.setattr(translation, "_post", post)
    html = (
        "<div><span>我是速卖通的官方客户经理奎琳，</span>我从后台看到您申请入驻速卖通平台开店。</div>"
        "<div>Widzę, że złożyłeś wniosek o otwarcie sklepu &amp; więcej.</div>"
        "<div>I noticed that you applied to join the platform &amp; more.</div>"
        "<div>WhatsApp: +48 780 549 877</div>"
    )
    out = await translate_html(html, "zh-Hans,pl,en", "pl")
    assert [c[0] for c in calls] == ["zh-Hans", "en"]  # the Polish is left alone
    assert "<div>Widzę, że złożyłeś wniosek o otwarcie sklepu &amp; więcej.</div>" in out
    # Two pieces of one Chinese sentence get a space between them.
    assert "PL(我是速卖通的官方客户经理奎琳，) </span>PL(" in out
    assert "PL(I noticed that you applied to join the platform & more.)" in out.replace(
        "&amp;", "&"
    )
    assert '<span translate="no">+48 780 549 877</span>' in out