app.py

import os
import re  # ← เพิ่ม
from fastapi import FastAPI, HTTPException, Body
from fastapi.middleware.cors import CORSMiddleware
from fastapi.responses import StreamingResponse, JSONResponse
import azure.cognitiveservices.speech as speechsdk
from typing import Optional

# --- Config from ENV ---
AZURE_SPEECH_KEY = os.getenv("AZURE_SPEECH_KEY")
AZURE_SPEECH_REGION = os.getenv("AZURE_SPEECH_REGION", "southeastasia")

if not AZURE_SPEECH_KEY:
    print("[WARN] AZURE_SPEECH_KEY is not set. Set it in HF Spaces (Settings → Repository secrets).")

app = FastAPI(title="Azure TTS API", version="1.0.0")

app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

DEFAULT_VOICE = "th-TH-PremwadeeNeural"

# --- NEW: ตัวช่วยลบอิโมจิออกจากข้อความก่อนส่งให้ TTS ---
# ครอบคลุม emoji blocks + ตัวเชื่อม/ตัวเลือกเวอร์ชัน (ZWJ/VS-16) + โทนสีผิว + ธง
EMOJI_RE = re.compile(
    "["                                 # main emoji blocks
    "\U0001F300-\U0001F5FF"
    "\U0001F600-\U0001F64F"
    "\U0001F680-\U0001F6FF"
    "\U0001F700-\U0001F77F"
    "\U0001F780-\U0001F7FF"
    "\U0001F800-\U0001F8FF"
    "\U0001F900-\U0001F9FF"
    "\U0001FA00-\U0001FA6F"
    "\U0001FA70-\U0001FAFF"
    "\U00002600-\U000026FF"             # misc symbols
    "\U000023F0-\U000023FA"             # clocks/media
    "\U00002700-\U000027BF"             # dingbats
    "\U0001F1E6-\U0001F1FF"             # regional indicator (ธง)
    "\U0001F3FB-\U0001F3FF"             # skin tones
    "]",
    flags=re.UNICODE
)

def strip_emoji(s: str) -> str:
    # ตัด emoji ออก + เอา ZWJ/VS-16 ที่หลงเหลือออก + จัดช่องว่างให้เรียบ
    s = EMOJI_RE.sub("", s)
    s = s.replace("\u200d", "").replace("\ufe0f", "")
    return re.sub(r"\s{2,}", " ", s).strip()

# Map format → Azure SDK OutputFormat
FORMAT_MAP = {
    "wav": speechsdk.SpeechSynthesisOutputFormat.Riff16Khz16BitMonoPcm,
    "mp3": speechsdk.SpeechSynthesisOutputFormat.Audio16Khz32KBitRateMonoMp3,
}

@app.get("/")
def root():
    return {"ok": True, "service": "Azure TTS API", "region": AZURE_SPEECH_REGION}

@app.get("/health")
def health():
    return {"status": "healthy"}

@app.post("/synthesize")
def synthesize(
    text: str = Body(..., embed=True, description="ข้อความที่จะสังเคราะห์เสียง"),
    voice: Optional[str] = Body(DEFAULT_VOICE, embed=True),
    audio_format: Optional[str] = Body("mp3", embed=True, description="mp3 หรือ wav"),
    # ถ้าต้องการสวิตช์ได้ ก็เปิดพารามิเตอร์นี้ (ค่าเริ่มต้น True = ไม่อ่านอิโมจิ)
    strip_emoji_before_tts: Optional[bool] = Body(True, embed=True)
):
    """
    คืน audio bytes เป็น StreamingResponse (Content-Type ตามฟอร์แมต)
    """
    if not AZURE_SPEECH_KEY:
        raise HTTPException(status_code=500, detail="AZURE_SPEECH_KEY not set")

    audio_format = audio_format.lower()
    if audio_format not in FORMAT_MAP:
        raise HTTPException(status_code=400, detail=f"Unsupported audio_format: {audio_format}. Use one of {list(FORMAT_MAP.keys())}")

    try:
        # 1) สร้าง SpeechConfig
        speech_config = speechsdk.SpeechConfig(subscription=AZURE_SPEECH_KEY, region=AZURE_SPEECH_REGION)
        speech_config.speech_synthesis_voice_name = voice or DEFAULT_VOICE
        speech_config.set_speech_synthesis_output_format(FORMAT_MAP[audio_format])

        # 2) เตรียมข้อความสำหรับ TTS (ลบอิโมจิออกเฉพาะตอนสังเคราะห์เสียง)
        text_for_tts = strip_emoji(text) if strip_emoji_before_tts else text
        if not text_for_tts:
            raise HTTPException(status_code=400, detail="Text contains only emojis after sanitization")

        # 3) ไม่ส่ง audio_config -> จะได้ audio bytes กลับมาใน result.audio_data
        synthesizer = speechsdk.SpeechSynthesizer(speech_config=speech_config, audio_config=None)

        # 4) สังเคราะห์
        result = synthesizer.speak_text_async(text_for_tts).get()

        if result.reason == speechsdk.ResultReason.SynthesizingAudioCompleted:
            audio_bytes = result.audio_data  # bytes
            if not audio_bytes:
                raise HTTPException(status_code=500, detail="No audio data produced")

            mime = "audio/mpeg" if audio_format == "mp3" else "audio/wav"
            filename = f"speech.{audio_format}"

            return StreamingResponse(
                iter([audio_bytes]),
                media_type=mime,
                headers={"Content-Disposition": f'inline; filename="{filename}"'}
            )

        elif result.reason == speechsdk.ResultReason.Canceled:
            details = result.cancellation_details
            msg = f"Synthesis canceled: {details.reason}"
            if details.reason == speechsdk.CancellationReason.Error:
                msg += f" | error: {details.error_details}"
            raise HTTPException(status_code=500, detail=msg)

        else:
            raise HTTPException(status_code=500, detail=f"Unknown result: {result.reason}")

    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

@app.get("/voices")
def list_voices():
    if not AZURE_SPEECH_KEY:
        raise HTTPException(status_code=500, detail="AZURE_SPEECH_KEY not set")

    try:
        speech_config = speechsdk.SpeechConfig(subscription=AZURE_SPEECH_KEY, region=AZURE_SPEECH_REGION)
        synthesizer = speechsdk.SpeechSynthesizer(speech_config=speech_config, audio_config=None)
        voices_result = synthesizer.get_voices_async().get()

        if voices_result.reason == speechsdk.ResultReason.VoicesListRetrieved:
            voices = []
            for v in voices_result.voices:
                voices.append({
                    "name": v.name,
                    "locale": v.locale,
                    "gender": v.gender.name if v.gender else None,
                    "shortName": getattr(v, "short_name", None)
                })
            return JSONResponse(content={"count": len(voices), "voices": voices})
        else:
            raise HTTPException(status_code=500, detail=f"Failed to list voices: {voices_result.reason}")

    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))