webmail-client/backend/tests/test_translation_prep.py

"""Getting a message ready for translation (app/translation_prep.py)."""

from __future__ import annotations

from app.translation_prep import (
    find_names,
    find_names_in_signature,
    join_wrapped_lines,
    keep_symbols,
    keep_terms,
    split_pieces,
    tidy,
    translate_phrases,
)

KEPT = '<span translate="no">'


def test_wrapped_lines_are_joined_but_paragraphs_kept() -> None:
    assert join_wrapped_lines("<p>ask you about<br>the schedule.</p>") == (
        "<p>ask you about the schedule.</p>"
    )
    assert join_wrapped_lines("<div>ask you about</div><div>the schedule.</div>") == (
        "<div>ask you about the schedule.</div>"
    )
    text = '<div class="wm-plaintext">ask you about\nthe schedule.\n\nThanks,\nAnna</div>'
    assert join_wrapped_lines(text) == (
        '<div class="wm-plaintext">ask you about the schedule.\n\nThanks,\nAnna</div>'
    )
    # A new sentence, or a capital after the break: left alone.
    for html in ("<p>Done.<br>then more</p>", "<p>Hi Lukas,<br>Natalia</p>"):
        assert join_wrapped_lines(html) == html


def test_sign_offs_come_from_the_phrase_table() -> None:
    assert translate_phrases("<p>Best,<br>Natalia</p>", "en", "pl") == (
        f"<p>{KEPT}Pozdrawiam,</span><br>Natalia</p>"
    )
    assert translate_phrases("<div>Mit freundlichen Grüßen</div>", "de", "en") == (
        f"<div>{KEPT}Kind regards</span></div>"
    )
    assert "Pozdrawiam" not in translate_phrases("<p>Best results so far</p>", "en", "pl")
    assert translate_phrases("<p>Best,</p>", "fr", "pl") == "<p>Best,</p>"  # no table


def test_names_mid_sentence_but_not_headlines_or_german() -> None:
    html = "<p>I came across Imaginary Pixels and Trusted Partner Network, in Spain.</p>"
    assert find_names(html, "en") == {"Imaginary Pixels", "Trusted Partner Network"}
    assert find_names("<p>Why Matchmoving Matters, Editing Digger And More</p>", "en") == set()
    assert find_names("<p>Wir haben das Neue Angebot gesehen.</p>", "de") == set()
    # Not across a sentence's end.
    assert find_names("<p>working mostly in NukeX. My work includes paint.</p>", "en") == set()
    assert find_names_in_signature("Imaginary Pixels – visual effects studio") == [
        "Imaginary Pixels"
    ]


def test_kept_terms_and_spacing() -> None:
    out = keep_terms(
        "<p>Hi Natalia Raz, see Imaginary Pixels.</p>", ["Natalia Raz", "Imaginary Pixels"]
    )
    assert out == f"<p>Hi {KEPT}Natalia Raz</span>, see {KEPT}Imaginary Pixels</span>.</p>"
    assert keep_terms("<style>.Natalia{}</style>", ["Natalia"]) == "<style>.Natalia{}</style>"
    assert tidy(f"nad {KEPT}Bruton</span>i dalej") == f"nad {KEPT}Bruton</span> i dalej"
    assert tidy("<span>a</span>b") == "<span>a</span>b"  # the sender's own spans


def test_emoji_kept() -> None:
    html = (
        '<p>nice to e-meet you <span style="font-family:Segoe UI Emoji">\n😊</span> &nbsp;</p>'
        "<p>Great 👍🏽 work ❤️</p>"
    )
    out = keep_symbols(html)
    assert f"{KEPT}😊</span>" in out
    assert f"{KEPT}👍🏽</span>" in out
    assert f"{KEPT}❤️</span>" in out
    assert "nice to e-meet you" in out  # the words still go to translation


def test_file_names_kept() -> None:
    html = (
        "<li>A_0023C017_260521_130157_p1CAC.mov</li>"
        "<p>See final_v2 and plate-03.exr, e.g. the U.S cut.</p>"
    )
    out = keep_symbols(html)
    assert f"{KEPT}A_0023C017_260521_130157_p1CAC.mov</span>" in out
    assert f"{KEPT}final_v2</span>" in out
    assert f"{KEPT}plate-03.exr</span>" in out
    assert "e.g. the U.S cut" in out  # no file names there


def test_symbols_not_marked_twice_or_in_attributes() -> None:
    html = (
        f'<a href="https://x.io/a_b.mov">{KEPT}https://x.io/a_b.mov</span></a>'
        "<style>.a_b{}</style>"
    )
    assert keep_symbols(html) == html


def test_text_without_letters_is_kept() -> None:
    # A byte-order mark alone between tags came back as "Uzasadnienie".
    out = keep_symbols(
        '<div dir="ltr">\ufeff<div>Dear Lukas,</div><div> &nbsp; </div><p> – </p></div>'
    )
    assert f"{KEPT}\ufeff</span>" in out
    assert "<div> &nbsp; </div>" in out  # only spaces: nothing to keep
    assert f"<p> {KEPT}–</span> </p>" in out
    assert "<div>Dear Lukas,</div>" in out


def test_handles_and_phone_numbers_kept() -> None:
    out = keep_symbols(
        "<p>WeChat: Jacquelineeeee_ WhatsApp: +48 780 549 877, for 1300 - 2500 Euro</p>"
    )
    assert f"{KEPT}Jacquelineeeee_</span>" in out
    assert f"{KEPT}+48 780 549 877</span>" in out
    assert "1300 - 2500 Euro" in out  # plain numbers stay in their sentence


def test_paragraphs_split_at_block_tags() -> None:
    html = (
        f"<style>p{{}}</style><div>Dear <b>Anna</b>,<br>hello {KEPT}x.mov</span> there</div>"
        "<p> &nbsp; </p><li>Item</li>"
    )
    pieces, paragraphs = split_pieces(html)
    assert [p.text for p in paragraphs] == ["Dear Anna,", "hello x.mov there", "Item"]
    # The kept file name counts for the language but is not translated.
    assert [pieces[i] for i in paragraphs[1].pieces] == ["hello ", " there"]