"""Every user-visible string has German text (AD-15); the informal catalogue only differs."""

import re
from pathlib import Path

import polib
import pytest

PORTAL = Path(__file__).resolve().parent.parent
LOCALE = PORTAL / "locale"

TRANS_TAG = re.compile(r"""\{%\s*trans\s+(?:"([^"]+)"|'([^']+)')""")
BLOCKTRANS = re.compile(r"\{%\s*blocktrans[^%]*%\}(.*?)\{%\s*endblocktrans\s*%\}", re.S)
VARIABLE = re.compile(r"\{\{\s*(\w+)\s*\}\}")
PY_CALL = re.compile(r"""(?<![\w.])(?:_|gettext|gettext_lazy)\(\s*(?:"([^"]+)"|'([^']+)')""")


def msgids_in_code() -> set[str]:
    found = set()
    for template in (PORTAL / "templates").rglob("*.html"):
        text = template.read_text(encoding="utf-8")
        found.update(a or b for a, b in TRANS_TAG.findall(text))
        found.update(VARIABLE.sub(r"%(\1)s", body).strip() for body in BLOCKTRANS.findall(text))
    for source in PORTAL.rglob("*.py"):
        if "tests" in source.parts or "migrations" in source.parts:
            continue
        found.update(a or b for a, b in PY_CALL.findall(source.read_text(encoding="utf-8")))
    return found


def catalogue(name: str) -> polib.POFile:
    return polib.pofile(str(LOCALE / name / "LC_MESSAGES" / "django.po"))


def test_the_scan_finds_the_strings():
    assert {"Sign out", "Home", "Hello, %(name)s", "Audit log"} <= msgids_in_code()


def test_every_string_in_the_code_has_german_text():
    german = {e.msgid: e for e in catalogue("de")}
    missing = sorted(msgids_in_code() - german.keys())
    assert not missing, f"no German entry for: {missing}"
    for entry in german.values():
        assert entry.msgstr.strip(), f"empty German text for {entry.msgid!r}"
        assert "fuzzy" not in entry.flags, f"fuzzy entry {entry.msgid!r}"


def test_the_german_catalogue_has_no_stale_entries():
    stale = sorted({e.msgid for e in catalogue("de")} - msgids_in_code())
    assert not stale, f"entries no longer used: {stale}"


def test_the_informal_catalogue_only_holds_differences_to_known_strings():
    german = {e.msgid: e.msgstr for e in catalogue("de")}
    for entry in catalogue("de_DU"):
        assert entry.msgid in german, f"unknown string {entry.msgid!r}"
        assert entry.msgstr.strip() and "fuzzy" not in entry.flags
        assert entry.msgstr != german[entry.msgid], f"{entry.msgid!r} does not differ"


@pytest.mark.parametrize("name", ["de", "de_DU"])
def test_placeholders_survive_translation(name):
    for entry in catalogue(name):
        assert set(re.findall(r"%\(\w+\)s", entry.msgid)) == set(
            re.findall(r"%\(\w+\)s", entry.msgstr)
        ), entry.msgid


def test_the_formal_text_does_not_use_du_and_the_informal_does_not_use_sie():
    formal = " ".join(e.msgstr for e in catalogue("de"))
    assert not re.search(r"\b(du|dir|dich|dein|deine)\b", formal, re.I)
    informal = " ".join(e.msgstr for e in catalogue("de_DU"))
    assert not re.search(r"\b(Sie|Ihnen|Ihr|Ihre)\b", informal)