notes-qpid-pro/app/ai.py

import json
import logging
from typing import Optional, Dict, Any
import httpx
from app import config
from app.database import get_cached_sentence_translation, save_sentence_translation

logger = logging.getLogger(__name__)


def build_translation_prompt(sentence: str, target_word: str = "") -> str:
    clean_sentence = sentence.strip()
    clean_word = target_word.strip()

    if clean_word:
        return (
            "You are a professional English-to-Russian translator.\n"
            "Translate the following English sentence into clear, natural, and idiomatic Russian.\n"
            f"Also provide the precise in-context meaning and translation of the target word/phrase '{clean_word}'.\n\n"
            f'Sentence: "{clean_sentence}"\n'
            f'Target word: "{clean_word}"\n\n'
            "Respond ONLY with a valid JSON object in this format:\n"
            "{\n"
            '  "sentence_ru": "перевод предложения на русский язык",\n'
            '  "word_meaning": "значение слова в этом контексте"\n'
            "}"
        )
    else:
        return (
            "You are a professional English-to-Russian translator.\n"
            "Translate the following English sentence into clear, natural, and idiomatic Russian.\n\n"
            f'Sentence: "{clean_sentence}"\n\n'
            "Respond ONLY with a valid JSON object in this format:\n"
            "{\n"
            '  "sentence_ru": "перевод предложения на русский язык",\n'
            '  "word_meaning": ""\n'
            "}"
        )


async def translate_sentence_ai(sentence: str, target_word: str = "") -> Dict[str, Any]:
    """
    Translates a sentence using local AI (Ollama), with SQLite instant-cache lookup.
    Returns:
    {
        "original_sentence": "...",
        "sentence_ru": "...",
        "target_word": "...",
        "word_meaning": "...",
        "source": "ollama:model" | "cache" | "fallback",
        "cached": bool
    }
    """
    clean_sentence = sentence.strip()
    clean_word = target_word.strip()

    if not clean_sentence:
        return {
            "original_sentence": "",
            "sentence_ru": "",
            "target_word": clean_word,
            "word_meaning": "",
            "source": "none",
            "cached": False,
        }

    # 1. Fast SQLite Cache Lookup (0.01 ms)
    cached = get_cached_sentence_translation(clean_sentence, clean_word)
    if cached:
        return {
            "original_sentence": clean_sentence,
            "sentence_ru": cached["translated_text"],
            "target_word": clean_word,
            "word_meaning": cached.get("word_meaning") or "",
            "source": cached.get("source") or "cache",
            "cached": True,
        }

    # 2. Local AI Generation (Ollama)
    prompt = build_translation_prompt(clean_sentence, clean_word)
    payload = {
        "model": config.OLLAMA_MODEL,
        "prompt": prompt,
        "stream": False,
        "format": "json",
        "options": {
            "temperature": 0.2,
            "num_predict": 256,
        },
    }

    sentence_ru = ""
    word_meaning = ""
    source = f"ollama:{config.OLLAMA_MODEL}"

    try:
        async with httpx.AsyncClient(timeout=30.0) as client:
            resp = await client.post(f"{config.OLLAMA_BASE_URL}/api/generate", json=payload)
            if resp.status_code == 200:
                data = resp.json()
                raw_response = data.get("response", "{}")
                try:
                    parsed = json.loads(raw_response)
                    sentence_ru = parsed.get("sentence_ru", "").strip()
                    word_meaning = parsed.get("word_meaning", "").strip()
                except Exception:
                    # In case LLM emitted unescaped markdown or wrapper
                    clean_raw = raw_response.strip()
                    if clean_raw.startswith("```json"):
                        clean_raw = clean_raw[7:]
                    if clean_raw.startswith("```"):
                        clean_raw = clean_raw[3:]
                    if clean_raw.endswith("```"):
                        clean_raw = clean_raw[:-3]
                    parsed = json.loads(clean_raw.strip())
                    sentence_ru = parsed.get("sentence_ru", "").strip()
                    word_meaning = parsed.get("word_meaning", "").strip()
            else:
                logger.warning(f"Ollama returned HTTP {resp.status_code}: {resp.text}")
    except Exception as e:
        logger.warning(f"Ollama connection error: {e}")

    # 3. Graceful Fallback if Ollama is unreachable
    if not sentence_ru:
        sentence_ru = "(Локальный ИИ временно недоступен)"
        source = "fallback"

    # Save to SQLite Cache if valid translation obtained
    if sentence_ru and not sentence_ru.startswith("("):
        save_sentence_translation(
            sentence=clean_sentence,
            translated_text=sentence_ru,
            target_word=clean_word,
            word_meaning=word_meaning,
            source=source,
        )

    return {
        "original_sentence": clean_sentence,
        "sentence_ru": sentence_ru,
        "target_word": clean_word,
        "word_meaning": word_meaning,
        "source": source,
        "cached": False,
    }


async def get_ollama_status() -> Dict[str, Any]:
    """Checks Ollama server connectivity and lists installed models."""
    try:
        async with httpx.AsyncClient(timeout=3.0) as client:
            resp = await client.get(f"{config.OLLAMA_BASE_URL}/api/tags")
            if resp.status_code == 200:
                data = resp.json()
                models = [m.get("name") for m in data.get("models", [])]
                return {
                    "online": True,
                    "models": models,
                    "active_model": config.OLLAMA_MODEL,
                }
    except Exception:
        pass
    return {
        "online": False,
        "models": [],
        "active_model": config.OLLAMA_MODEL,
    }