notes-qpid-pro/app/routers/dictionary.py

import asyncio
import re
from typing import Optional
from fastapi import APIRouter, Depends, HTTPException, Query, Response
from fastapi.responses import FileResponse
from pydantic import BaseModel, Field

from app.auth import require_user
from app.database import get_db
from app.dictionary import lookup_term
from app.phrasal import detect_phrasal_verb
from app.ai import translate_sentence_ai, get_ollama_status
from app.tts import (
    synthesize_speech,
    prewarm_card_audio,
    AUDIO_CACHE_DIR,
    get_available_voices,
)

router = APIRouter(tags=["Dictionary & TTS"])


class SentenceTranslateRequest(BaseModel):
    sentence: str = Field(..., min_length=1)
    target_word: Optional[str] = ""
    context: Optional[str] = ""


@router.get("/api/lookup")
async def api_lookup(
    term: Optional[str] = None,
    q: Optional[str] = None,
    context: Optional[str] = None,
    user: dict = Depends(require_user),
):
    query_str = (term or q or "").strip()
    if not query_str:
        raise HTTPException(status_code=400, detail="Query term cannot be empty")
    data = await lookup_term(query_str)

    # Ensure term is always present in response
    target_term = data.get("term") or data.get("english_term") or query_str
    data["term"] = target_term

    # Check if this term already exists in user's dictionary
    with get_db() as conn:
        existing = conn.execute(
            "SELECT id, term, translation, context_sentence, tags, collection, srs_state, srs_reps FROM vocabulary_cards WHERE user_id = ? AND (lower(trim(term)) = lower(trim(?)) OR lower(trim(term)) = lower(trim(?)))",
            (user["id"], target_term, query_str),
        ).fetchone()
        if existing:
            data["in_library"] = True
            data["library_card"] = dict(existing)
        else:
            data["in_library"] = False
            data["library_card"] = None

        # Multi-Word & Phrasal Verb Detection
        data["phrasal_verb"] = None
        if context:
            pv = detect_phrasal_verb(
                sentence=context,
                target_word=query_str,
                user_id=user["id"],
                conn=conn,
            )
            if pv and pv["term"].lower() != query_str.lower():
                data["phrasal_verb"] = pv

    return data


@router.get("/api/detect-phrasal-verb")
async def api_detect_phrasal_verb(
    sentence: str = Query(..., min_length=1),
    word: str = Query(..., min_length=1),
    user: dict = Depends(require_user),
):
    """
    Analyzes sentence context around word to detect phrasal verbs, idioms, and collocations.
    """
    with get_db() as conn:
        pv = detect_phrasal_verb(sentence=sentence, target_word=word, user_id=user["id"], conn=conn)
        return {"detected": bool(pv), "phrasal_verb": pv}


@router.api_route("/api/audio", methods=["GET", "HEAD"])
async def api_audio_proxy():
    # Audio feature disabled
    return Response(status_code=204)


@router.post("/api/translate-sentence")
async def api_translate_sentence(
    req: SentenceTranslateRequest,
    user: dict = Depends(require_user),
):
    """
    Translates a sentence using local AI with in-context meaning of the target word.
    Uses SQLite sentence_translations cache for sub-millisecond retrieval on repeat queries.
    """
    clean_s = req.sentence.strip()
    if not clean_s:
        raise HTTPException(status_code=400, detail="Sentence cannot be empty")

    # Pre-warm neural audio for sentence and target word in background
    asyncio.create_task(prewarm_card_audio(req.target_word or "", clean_s))

    import sys
    main_mod = sys.modules.get("app.main")
    translator = getattr(main_mod, "translate_sentence_ai", translate_sentence_ai) if main_mod else translate_sentence_ai
    return await translator(clean_s, req.target_word or "", req.context or "")


@router.get("/api/ai/status")
async def api_ai_status(user: dict = Depends(require_user)):
    """
    Returns local AI status and available models.
    """
    return await get_ollama_status()


@router.get("/api/audio/{hash_val}.wav")
async def api_get_cached_audio(hash_val: str):
    """
    Streams pre-warmed / cached audio file directly with immutable cache headers.
    """
    clean_hash = hash_val.lower().strip()
    if not re.match(r"^[a-f0-9]{32}$", clean_hash):
        raise HTTPException(status_code=400, detail="Invalid audio hash")
    audio_path = AUDIO_CACHE_DIR / f"{clean_hash}.wav"
    if not audio_path.is_file():
        raise HTTPException(status_code=404, detail="Audio file not found")
    return FileResponse(
        path=str(audio_path),
        media_type="audio/wav",
        headers={"Cache-Control": "public, max-age=31536000, immutable"},
    )


@router.get("/api/tts")
async def api_tts(
    text: str = Query(..., min_length=1, max_length=1000),
    lang: str = Query("en"),
    speed: float = Query(1.0, ge=0.5, le=2.0),
    voice: Optional[str] = Query(None),
    user: dict = Depends(require_user),
):
    """
    Synthesizes speech on-demand using local Piper TTS.
    Instantly returns cached audio if previously synthesized for this voice, speed, and text.
    """
    clean_text = text.strip()
    if not clean_text:
        raise HTTPException(status_code=400, detail="Text cannot be empty")
    audio_path = await synthesize_speech(clean_text, lang=lang, speed=speed, voice=voice)
    if not audio_path or not audio_path.is_file():
        raise HTTPException(status_code=500, detail="TTS synthesis failed or unavailable")
    return FileResponse(
        path=str(audio_path),
        media_type="audio/wav",
        headers={"Cache-Control": "public, max-age=604800, immutable"},
    )


@router.get("/api/tts/status")
async def api_tts_status(user: dict = Depends(require_user)):
    """Returns status of local Piper TTS and all available voice models."""
    return get_available_voices()