"""Multi-language i18n endpoint tests for RBS REGAL Business Suite.""" import os import re import requests import pytest BASE_URL = os.environ.get("REACT_APP_BACKEND_URL", "https://offline-billing-pro-2.preview.emergentagent.com").rstrip("/") EXPECTED_CODES = {"en", "hi", "gu", "mr", "pa", "bn", "ta", "te", "kn", "ml", "or", "ur", "kok"} # ----- /api/i18n/languages (public) ----- class TestLanguages: def test_list_languages(self): r = requests.get(f"{BASE_URL}/api/i18n/languages", timeout=30) assert r.status_code == 200, r.text data = r.json() assert isinstance(data, list) assert len(data) == 13, f"Expected 13 languages, got {len(data)}" codes = {row["code"] for row in data} assert codes == EXPECTED_CODES, f"Mismatch: {codes ^ EXPECTED_CODES}" for row in data: assert {"code", "name", "native", "dir", "flag"} <= set(row.keys()), row assert row["dir"] in ("ltr", "rtl") ur = next(row for row in data if row["code"] == "ur") assert ur["dir"] == "rtl" # ----- /api/i18n/translate ----- class TestTranslate: SAMPLE = ["Total Sales", "New Invoice", "Customers", "Stock", "Dashboard"] def test_translate_gujarati(self, admin_session): # Cache-bust: append a unique nonce so we definitely exercise the LLM path import uuid nonce = uuid.uuid4().hex[:6] texts = [f"{t} {nonce}" for t in self.SAMPLE] r = admin_session.post(f"{BASE_URL}/api/i18n/translate", json={"lang": "gu", "texts": texts, "use_cache": True}, timeout=120) assert r.status_code == 200, r.text data = r.json() assert data["lang"] == "gu" assert "translations" in data # All five sample strings should have a translation for t in texts: assert t in data["translations"] # At least some should come from LLM (since cache-busted) — or fallback to source if LLM fails assert data["from_llm"] + data["from_cache"] >= len(texts) - 1 # Verify Gujarati script presence in at least one translation when LLM succeeded gu_chars = re.compile(r"[\u0A80-\u0AFF]") translated_count = sum(1 for t in texts if gu_chars.search(data["translations"][t] or "")) # Soft-assert: warn but don't crash if LLM didn't respond if data["from_llm"] > 0: assert translated_count >= 1, f"No Gujarati script found in {data['translations']}" def test_translate_urdu_rtl_script(self, admin_session): texts = ["New Invoice", "Customers", "GST", "WhatsApp", "Save"] r = admin_session.post(f"{BASE_URL}/api/i18n/translate", json={"lang": "ur", "texts": texts, "use_cache": True}, timeout=120) assert r.status_code == 200, r.text data = r.json() assert data["lang"] == "ur" for t in texts: assert t in data["translations"] # Brand names should remain in Latin assert "GST" in data["translations"]["GST"] or data["translations"]["GST"] == "GST" # At least one Arabic-script translation should exist ar_chars = re.compile(r"[\u0600-\u06FF]") has_arabic = any(ar_chars.search(v or "") for v in data["translations"].values()) # Soft: only enforce if LLM produced fresh translations if data["from_llm"] > 0 or data["from_cache"] > 0: assert has_arabic, f"No Arabic script found in Urdu translations: {data['translations']}" def test_cache_hit_on_repeat(self, admin_session): texts = ["Dashboard", "Reports", "Settings"] # First call — warm the cache r1 = admin_session.post(f"{BASE_URL}/api/i18n/translate", json={"lang": "gu", "texts": texts, "use_cache": True}, timeout=120) assert r1.status_code == 200, r1.text # Second call — should be all cache r2 = admin_session.post(f"{BASE_URL}/api/i18n/translate", json={"lang": "gu", "texts": texts, "use_cache": True}, timeout=30) assert r2.status_code == 200, r2.text d2 = r2.json() assert d2["from_llm"] == 0, f"Expected 0 LLM calls on cache hit, got {d2['from_llm']}" assert d2["from_cache"] >= len(texts), f"Expected cache hits >= {len(texts)}, got {d2['from_cache']}" def test_unsupported_language(self, admin_session): r = admin_session.post(f"{BASE_URL}/api/i18n/translate", json={"lang": "xx", "texts": ["Hello"], "use_cache": True}, timeout=30) assert r.status_code == 400, r.text def test_english_identity(self, admin_session): r = admin_session.post(f"{BASE_URL}/api/i18n/translate", json={"lang": "en", "texts": ["Hello", "World"], "use_cache": True}, timeout=30) assert r.status_code == 200 d = r.json() assert d["translations"] == {"Hello": "Hello", "World": "World"} # ----- /api/i18n/set-user-lang + /me ----- class TestUserLang: def test_set_and_get_user_lang(self, admin_session): # Set Marathi r = admin_session.post(f"{BASE_URL}/api/i18n/set-user-lang", json={"lang": "mr"}, timeout=30) assert r.status_code == 200, r.text assert r.json()["lang"] == "mr" # Fetch — verify persistence r2 = admin_session.get(f"{BASE_URL}/api/i18n/me", timeout=30) assert r2.status_code == 200, r2.text assert r2.json()["lang"] == "mr" # Restore to English r3 = admin_session.post(f"{BASE_URL}/api/i18n/set-user-lang", json={"lang": "en"}, timeout=30) assert r3.status_code == 200 def test_set_invalid_lang(self, admin_session): r = admin_session.post(f"{BASE_URL}/api/i18n/set-user-lang", json={"lang": "zz"}, timeout=30) assert r.status_code == 400 # ----- /api/i18n/ai-system-prompt/{lang} ----- class TestAiSystemPrompt: def test_english_prompt(self): r = requests.get(f"{BASE_URL}/api/i18n/ai-system-prompt/en", timeout=30) assert r.status_code == 200 d = r.json() assert d["lang"] == "en" assert "English" in d["system_prompt"] def test_hindi_prompt(self): r = requests.get(f"{BASE_URL}/api/i18n/ai-system-prompt/hi", timeout=30) assert r.status_code == 200 d = r.json() assert d["lang"] == "hi" assert d["native_name"] == "हिन्दी" def test_gujarati_prompt(self): r = requests.get(f"{BASE_URL}/api/i18n/ai-system-prompt/gu", timeout=30) assert r.status_code == 200 d = r.json() assert d["lang"] == "gu" assert "Gujarati" in d["system_prompt"] assert d["native_name"] == "ગુજરાતી" def test_invalid_lang_falls_back_english(self): r = requests.get(f"{BASE_URL}/api/i18n/ai-system-prompt/xx", timeout=30) assert r.status_code == 200 assert r.json()["lang"] == "en"