food_analyzer.py

from fastapi import FastAPI, UploadFile, File, HTTPException
from fastapi.middleware.cors import CORSMiddleware
from PIL import Image
import io
import base64
from transformers import pipeline
import requests
import json
from typing import Dict, Any
from pydantic import BaseModel
import uvicorn

app = FastAPI(title="Health Assistant AI - Food Recognition API")

# CORS設定
app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],  # 生產環境請設定具體的域名
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

# 初始化Hugging Face模型
try:
    # 使用nateraw/food專門的食物分類模型
    food_classifier = pipeline(
        "image-classification",
        model="nateraw/food",
        device=-1  # 使用CPU，如果有GPU可以設為0
    )
    print("nateraw/food 食物辨識模型載入成功")
except Exception as e:
    print(f"模型載入失敗: {e}")
    food_classifier = None

# 食物營養資料庫（擴展版，涵蓋nateraw/food模型常見的食物類型）
NUTRITION_DATABASE = {
    # 水果類
    "apple": {"name": "蘋果", "calories_per_100g": 52, "protein": 0.3, "carbs": 14, "fat": 0.2, "fiber": 2.4, "sugar": 10.4, "vitamin_c": 4.6},
    "banana": {"name": "香蕉", "calories_per_100g": 89, "protein": 1.1, "carbs": 23, "fat": 0.3, "fiber": 2.6, "sugar": 12.2, "potassium": 358},
    "orange": {"name": "橘子", "calories_per_100g": 47, "protein": 0.9, "carbs": 12, "fat": 0.1, "fiber": 2.4, "sugar": 9.4, "vitamin_c": 53.2},
    "strawberry": {"name": "草莓", "calories_per_100g": 32, "protein": 0.7, "carbs": 7.7, "fat": 0.3, "fiber": 2, "sugar": 4.9, "vitamin_c": 58.8},
    "grape": {"name": "葡萄", "calories_per_100g": 62, "protein": 0.6, "carbs": 16.8, "fat": 0.2, "fiber": 0.9, "sugar": 16.1},
    
    # 主食類
    "bread": {"name": "麵包", "calories_per_100g": 265, "protein": 9, "carbs": 49, "fat": 3.2, "fiber": 2.7, "sodium": 491},
    "rice": {"name": "米飯", "calories_per_100g": 130, "protein": 2.7, "carbs": 28, "fat": 0.3, "fiber": 0.4},
    "pasta": {"name": "義大利麵", "calories_per_100g": 131, "protein": 5, "carbs": 25, "fat": 1.1, "fiber": 1.8},
    "noodles": {"name": "麵條", "calories_per_100g": 138, "protein": 4.5, "carbs": 25, "fat": 2.2, "fiber": 1.2},
    "pizza": {"name": "披薩", "calories_per_100g": 266, "protein": 11, "carbs": 33, "fat": 10, "sodium": 598},
    
    # 肉類
    "chicken": {"name": "雞肉", "calories_per_100g": 165, "protein": 31, "carbs": 0, "fat": 3.6, "iron": 0.9},
    "beef": {"name": "牛肉", "calories_per_100g": 250, "protein": 26, "carbs": 0, "fat": 15, "iron": 2.6, "zinc": 4.8},
    "pork": {"name": "豬肉", "calories_per_100g": 242, "protein": 27, "carbs": 0, "fat": 14, "thiamine": 0.7},
    "fish": {"name": "魚肉", "calories_per_100g": 206, "protein": 22, "carbs": 0, "fat": 12, "omega_3": "豐富"},
    
    # 蔬菜類
    "broccoli": {"name": "花椰菜", "calories_per_100g": 34, "protein": 2.8, "carbs": 7, "fat": 0.4, "fiber": 2.6, "vitamin_c": 89.2},
    "carrot": {"name": "胡蘿蔔", "calories_per_100g": 41, "protein": 0.9, "carbs": 10, "fat": 0.2, "fiber": 2.8, "vitamin_a": 835},
    "tomato": {"name": "番茄", "calories_per_100g": 18, "protein": 0.9, "carbs": 3.9, "fat": 0.2, "fiber": 1.2, "vitamin_c": 13.7},
    "lettuce": {"name": "萵苣", "calories_per_100g": 15, "protein": 1.4, "carbs": 2.9, "fat": 0.2, "fiber": 1.3, "folate": 38},
    
    # 飲品類
    "coffee": {"name": "咖啡", "calories_per_100g": 2, "protein": 0.3, "carbs": 0, "fat": 0, "caffeine": 95},
    "tea": {"name": "茶", "calories_per_100g": 1, "protein": 0, "carbs": 0.3, "fat": 0, "antioxidants": "豐富"},
    "milk": {"name": "牛奶", "calories_per_100g": 42, "protein": 3.4, "carbs": 5, "fat": 1, "calcium": 113},
    "juice": {"name": "果汁", "calories_per_100g": 45, "protein": 0.7, "carbs": 11, "fat": 0.2, "vitamin_c": "因果汁種類而異"},
    
    # 甜點類
    "cake": {"name": "蛋糕", "calories_per_100g": 257, "protein": 4, "carbs": 46, "fat": 6, "sugar": 35},
    "cookie": {"name": "餅乾", "calories_per_100g": 502, "protein": 5.9, "carbs": 64, "fat": 25, "sugar": 39},
    "ice_cream": {"name": "冰淇淋", "calories_per_100g": 207, "protein": 3.5, "carbs": 24, "fat": 11, "sugar": 21},
    "chocolate": {"name": "巧克力", "calories_per_100g": 546, "protein": 4.9, "carbs": 61, "fat": 31, "sugar": 48},
    
    # 其他常見食物
    "egg": {"name": "雞蛋", "calories_per_100g": 155, "protein": 13, "carbs": 1.1, "fat": 11, "choline": 294},
    "cheese": {"name": "起司", "calories_per_100g": 113, "protein": 7, "carbs": 1, "fat": 9, "calcium": 200},
    "yogurt": {"name": "優格", "calories_per_100g": 59, "protein": 10, "carbs": 3.6, "fat": 0.4, "probiotics": "豐富"},
    "nuts": {"name": "堅果", "calories_per_100g": 607, "protein": 15, "carbs": 7, "fat": 54, "vitamin_e": 26},
    "salad": {"name": "沙拉", "calories_per_100g": 20, "protein": 1.5, "carbs": 4, "fat": 0.2, "fiber": 2, "vitamins": "多種維生素"}
}

# 回應模型
class FoodAnalysisResponse(BaseModel):
    success: bool
    food_name: str
    confidence: float
    nutrition_info: Dict[str, Any]
    ai_suggestions: list
    message: str

class HealthResponse(BaseModel):
    status: str
    message: str

def get_nutrition_info(food_name: str) -> Dict[str, Any]:
    """根據食物名稱獲取營養資訊"""
    # 將食物名稱轉為小寫並清理
    food_key = food_name.lower().strip()
    
    # 移除常見的修飾詞和格式化字符
    food_key = food_key.replace("_", " ").replace("-", " ")
    
    # 直接匹配
    if food_key in NUTRITION_DATABASE:
        return NUTRITION_DATABASE[food_key]
    
    # 模糊匹配 - 檢查是否包含關鍵字
    for key, value in NUTRITION_DATABASE.items():
        if key in food_key or food_key in key:
            return value
        # 也檢查中文名稱
        if value["name"] in food_name:
            return value
    
    # 更智能的匹配 - 處理複合詞
    food_words = food_key.split()
    for word in food_words:
        for key, value in NUTRITION_DATABASE.items():
            if word == key or word in key:
                return value
    
    # 特殊情況處理
    special_mappings = {
        "french fries": "potato",
        "hamburger": "beef",
        "sandwich": "bread", 
        "soda": "juice",
        "water": {"name": "水", "calories_per_100g": 0, "protein": 0, "carbs": 0, "fat": 0},
        "soup": {"name": "湯", "calories_per_100g": 50, "protein": 2, "carbs": 8, "fat": 1, "sodium": 400}
    }
    
    for special_key, mapping in special_mappings.items():
        if special_key in food_key:
            if isinstance(mapping, str):
                return NUTRITION_DATABASE.get(mapping, {"name": food_name, "message": "營養資料不完整"})
            else:
                return mapping
    
    # 如果沒有找到，返回預設值
    return {
        "name": food_name,
        "calories_per_100g": "未知",
        "protein": "未知",
        "carbs": "未知", 
        "fat": "未知",
        "message": f"抱歉，暫時沒有「{food_name}」的詳細營養資料，建議查詢專業營養資料庫"
    }

def generate_ai_suggestions(food_name: str, nutrition_info: Dict) -> list:
    """根據食物和營養資訊生成AI建議"""
    suggestions = []
    food_name_lower = food_name.lower()
    
    # 檢查是否有完整的營養資訊
    if isinstance(nutrition_info.get("calories_per_100g"), (int, float)):
        calories = nutrition_info["calories_per_100g"]
        
        # 熱量相關建議
        if calories > 400:
            suggestions.append("⚠️ 這是高熱量食物，建議控制份量，搭配運動")
        elif calories > 200:
            suggestions.append("🍽️ 中等熱量食物，適量食用，建議搭配蔬菜")
        elif calories < 50:
            suggestions.append("✅ 低熱量食物，適合減重期間食用")
        
        # 營養素相關建議
        protein = nutrition_info.get("protein", 0)
        if isinstance(protein, (int, float)) and protein > 20:
            suggestions.append("💪 高蛋白食物，有助於肌肉發展和修復")
        
        fiber = nutrition_info.get("fiber", 0)
        if isinstance(fiber, (int, float)) and fiber > 3:
            suggestions.append("🌿 富含纖維，有助於消化健康和增加飽足感")
        
        sugar = nutrition_info.get("sugar", 0)
        if isinstance(sugar, (int, float)) and sugar > 20:
            suggestions.append("🍯 含糖量較高，建議適量食用，避免血糖快速上升")
        
        # 特殊營養素
        if nutrition_info.get("vitamin_c", 0) > 30:
            suggestions.append("🍊 富含維生素C，有助於增強免疫力和抗氧化")
        
        if nutrition_info.get("calcium", 0) > 100:
            suggestions.append("🦴 富含鈣質，有助於骨骼和牙齒健康")
        
        if nutrition_info.get("omega_3"):
            suggestions.append("🐟 含有Omega-3脂肪酸，對心血管健康有益")
    
    # 根據食物類型給出特定建議
    if any(fruit in food_name_lower for fruit in ["apple", "banana", "orange", "strawberry", "grape"]):
        suggestions.append("🍎 建議在餐前或運動前食用，提供天然糖分和維生素")
    
    elif any(meat in food_name_lower for meat in ["chicken", "beef", "pork", "fish"]):
        suggestions.append("🥩 建議搭配蔬菜食用，選擇健康的烹調方式（烤、蒸、煮）")
    
    elif any(sweet in food_name_lower for sweet in ["cake", "cookie", "ice_cream", "chocolate"]):
        suggestions.append("🍰 甜點建議偶爾享用，可在運動後適量食用")
        suggestions.append("💡 可以考慮與朋友分享，減少單次攝取量")
    
    elif any(drink in food_name_lower for drink in ["coffee", "tea"]):
        suggestions.append("☕ 建議控制咖啡因攝取量，避免影響睡眠")
    
    elif "salad" in food_name_lower:
        suggestions.append("🥗 很棒的選擇！可以添加堅果或橄欖油增加健康脂肪")
    
    # 通用健康建議
    if not suggestions:
        suggestions.extend([
            "🍽️ 建議均衡飲食，搭配多樣化的食物",
            "💧 記得多喝水，保持身體水分充足",
            "🏃‍♂️ 搭配適量運動，維持健康生活型態"
        ])
    else:
        # 添加一些通用的健康提醒
        suggestions.append("💧 記得多喝水，幫助營養吸收")
        if len(suggestions) < 4:
            suggestions.append("⚖️ 注意食物份量，適量攝取是健康飲食的關鍵")
    
    return suggestions[:5]  # 限制建議數量，避免過多

@app.get("/", response_model=HealthResponse)
async def root():
    """API根路径"""
    return HealthResponse(
        status="success",
        message="Health Assistant AI - Food Recognition API is running!"
    )

@app.get("/health", response_model=HealthResponse)
async def health_check():
    """健康檢查端點"""
    model_status = "正常" if food_classifier else "模型載入失敗"
    return HealthResponse(
        status="success",
        message=f"API運行正常，模型狀態: {model_status}"
    )

@app.post("/analyze-food", response_model=FoodAnalysisResponse)
async def analyze_food(file: UploadFile = File(...)):
    """分析上傳的食物圖片"""
    try:
        # 檢查模型是否載入成功
        if not food_classifier:
            raise HTTPException(status_code=500, detail="AI模型尚未載入，請稍後再試")
        
        # 檢查文件類型
        if not file.content_type.startswith("image/"):
            raise HTTPException(status_code=400, detail="請上傳圖片文件")
        
        # 讀取圖片
        image_data = await file.read()
        image = Image.open(io.BytesIO(image_data))
        
        # 確保圖片是RGB格式
        if image.mode != "RGB":
            image = image.convert("RGB")
        
        # 使用AI模型進行食物辨識
        results = food_classifier(image)
        
        # 獲取最高信心度的結果
        top_result = results[0]
        food_name = top_result["label"]
        confidence = top_result["score"]
        
        # 獲取營養資訊
        nutrition_info = get_nutrition_info(food_name)
        
        # 生成AI建議
        ai_suggestions = generate_ai_suggestions(food_name, nutrition_info)
        
        return FoodAnalysisResponse(
            success=True,
            food_name=food_name,
            confidence=round(confidence * 100, 2),
            nutrition_info=nutrition_info,
            ai_suggestions=ai_suggestions,
            message="食物分析完成"
        )
        
    except Exception as e:
        raise HTTPException(status_code=500, detail=f"分析失敗: {str(e)}")

@app.post("/analyze-food-base64", response_model=FoodAnalysisResponse)
async def analyze_food_base64(image_data: dict):
    """分析base64編碼的食物圖片"""
    try:
        # 檢查模型是否載入成功
        if not food_classifier:
            raise HTTPException(status_code=500, detail="AI模型尚未載入，請稍後再試")
        
        # 解碼base64圖片
        base64_string = image_data.get("image", "")
        if not base64_string:
            raise HTTPException(status_code=400, detail="缺少圖片資料")
        
        # 移除base64前綴（如果有的話）
        if "," in base64_string:
            base64_string = base64_string.split(",")[1]
        
        # 解碼圖片
        image_bytes = base64.b64decode(base64_string)
        image = Image.open(io.BytesIO(image_bytes))
        
        # 確保圖片是RGB格式
        if image.mode != "RGB":
            image = image.convert("RGB")
        
        # 使用AI模型進行食物辨識
        results = food_classifier(image)
        
        # 獲取最高信心度的結果
        top_result = results[0]
        food_name = top_result["label"]
        confidence = top_result["score"]
        
        # 獲取營養資訊
        nutrition_info = get_nutrition_info(food_name)
        
        # 生成AI建議
        ai_suggestions = generate_ai_suggestions(food_name, nutrition_info)
        
        return FoodAnalysisResponse(
            success=True,
            food_name=food_name,
            confidence=round(confidence * 100, 2),
            nutrition_info=nutrition_info,
            ai_suggestions=ai_suggestions,
            message="食物分析完成"
        )
        
    except Exception as e:
        raise HTTPException(status_code=500, detail=f"分析失敗: {str(e)}")

if __name__ == "__main__":
    uvicorn.run(app, host="0.0.0.0", port=8000)