import gradio as gr from openai import OpenAI from huggingface_hub import InferenceClient import os import time import requests import urllib.parse import pandas as pd from langchain_community.document_loaders import DataFrameLoader from langchain_text_splitters import RecursiveCharacterTextSplitter from langchain_huggingface import HuggingFaceEmbeddings from langchain_community.vectorstores import FAISS from PIL import Image, ImageFont, ImageDraw import warnings import cv2 import numpy as np from deepface import DeepFace # 忽略 pandas 的一些警告 warnings.filterwarnings("ignore") # ========================================== # 0. 環境變數與 RAG 初始化 # ========================================== GROQ_API_KEY = os.getenv("GROQ_API_KEY") HF_TOKEN = os.getenv("HF_TOKEN") # --- 1. 資料讀取 (主要資料庫 - 使用 restaurants.csv) --- df = pd.DataFrame() try: # 確保你有 restaurants.csv 在專案根目錄 df = pd.read_csv('restaurants.csv', engine='python') df.columns = df.columns.str.strip() # 清理欄位空白 def classify_mood(row): name_str = str(row.get('Name', '')) rag_str = str(row.get('RAG_Content', '')) text = (name_str + " " + rag_str).lower() tags = [] rules = { "開心/慶祝": ["牛排", "steak", "pizza", "炸", "雞排", "甜點", "蛋糕", "cake", "冰", "waffle", "吃到飽", "buffet", "burger", "bistro", "餐酒館"], "傷心/疲憊": ["粥", "湯", "warm", "congee", "麵", "noodle", "小吃", "comfort food", "豆花", "關東煮", "soup"], "生氣/發洩": ["辣", "spicy", "麻辣", "鍋", "curry", "咖哩", "燒肉", "bbq", "臭豆腐", "fry"], "平靜/放鬆": ["cafe", "coffee", "tea", "茶", "素食", "vegetable", "早午餐", "brunch", "壽司", "sushi", "居酒屋"] } for mood, keywords in rules.items(): for kw in keywords: if kw in text: tags.append(mood) if not tags: tags.append("隨意/探索") return ", ".join(list(set(tags))) if 'Mood_Tags' not in df.columns: df['Mood_Tags'] = df.apply(classify_mood, axis=1) print("✅ 成功讀取 restaurants.csv 並建立 Mood_Tags") except Exception as e: # 如果找不到檔案或讀取失敗，這裡會提醒 print(f"⚠️ 讀取 restaurants.csv 失敗: {e}") # --- 2. 資料讀取 (Prompt 資料庫) --- df_prompts = pd.DataFrame() PROMPT_COL_NAME = 'Visual_prompt' try: if not df.empty: df_prompts = df.copy() if 'Name' in df_prompts.columns: df_prompts.set_index('Name', inplace=True) except Exception as e: print(f"⚠️ 處理 Prompt 資料發生錯誤: {e}") # --- RAG 建置 --- retriever = None if not df.empty and 'RAG_Content' in df.columns: try: loader = DataFrameLoader(df, page_content_column="RAG_Content") documents = loader.load() text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50) docs = text_splitter.split_documents(documents) embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2") vectorstore = FAISS.from_documents(docs, embeddings) retriever = vectorstore.as_retriever(search_kwargs={"k": 3}) except Exception as e: print(f"⚠️ RAG 初始化失敗: {e}") # ========================================== # 1. 核心功能函式 (餐廳推薦邏輯) # ========================================== def get_restaurant_data(mood_score_str, food_choice): if df.empty: return None, True # 定義分數與 Mood_Tags 的對應關係 (與舊版維持一致) score_map = { "1 (心情差)": ["傷心/疲憊", "生氣/發洩"], "2 (不太好)": ["傷心/疲憊", "生氣/發洩"], "3 (普通)": ["平靜/放鬆", "隨意/探索"], "4 (不錯)": ["開心/慶祝", "平靜/放鬆"], "5 (超棒)": ["開心/慶祝"] } target_moods = score_map.get(mood_score_str, []) food_keyword = "" if food_choice == "吃飯": food_keyword = "飯" elif food_choice == "吃麵": food_keyword = "麵" candidates = df.copy() if 'Mood_Tags' not in candidates.columns: return df.sample(1).iloc[0], True if target_moods: pattern = "|".join(target_moods) candidates = candidates[candidates['Mood_Tags'].str.contains(pattern, regex=True, na=False)] if food_keyword: candidates = candidates[ candidates['Name'].str.contains(food_keyword, case=False, na=False) | candidates['RAG_Content'].str.contains(food_keyword, case=False, na=False) ] if candidates.empty: return df.sample(1).iloc[0], True return candidates.sample(1).iloc[0], False # 注意：因為移除了測驗，這裡我們用 '無測驗' 作為預設結果 def generate_content_with_groq(restaurant_name, restaurant_detail, user_diary, mood_score): if not GROQ_API_KEY: return "⚠️ 請設定 GROQ_API_KEY (系統變數)" client = OpenAI(api_key=GROQ_API_KEY, base_url="https://api.groq.com/openai/v1") system_prompt = "你是一個幽默、懂吃且善解人意的 AI 朋友。請根據使用者的日記和心情來推薦餐廳。" user_msg = f""" 【使用者狀態】 心情分數：{mood_score} 心情日記：{user_diary} 【推薦餐廳】 名稱：{restaurant_name} 資料：{restaurant_detail} 請完成任務： 【回應內容】：用繁體中文寫一段溫暖有趣的回覆。 - 結合「日記」與「心情」給予回應。 - 推薦上述餐廳，說明為什麼這家店適合他。 """ try: response = client.chat.completions.create( model="llama-3.3-70b-versatile", messages=[{"role": "system", "content": system_prompt}, {"role": "user", "content": user_msg}] ) return response.choices[0].message.content except Exception as e: return f"Groq Error: {str(e)}" def generate_image_huggingface(prompt): if not HF_TOKEN: return None try: hf_client = InferenceClient(token=HF_TOKEN) return hf_client.text_to_image(prompt=prompt, model="stabilityai/stable-diffusion-xl-base-1.0") except: return None # --- 主邏輯 Agent --- # 移除了 quiz_state 參數 def mood_agent_logic(score_input, food_input, diary_input, debug_mode): # 確保 score_input 有值 if not score_input: yield "⚠️ 請先選擇心情分數，或使用上方相機偵測！", None, "" return restaurant, is_random = get_restaurant_data(score_input, food_input) if restaurant is None: yield "資料庫讀取錯誤或為空", None, "" return name = str(restaurant['Name']) url = str(restaurant.get('URL', f'https://www.google.com/maps/search/?api=1&query={urllib.parse.quote(name)}')) note = "(隨機推薦)" if is_random else "" # 1. 處理 RAG rag_info = str(restaurant.get('RAG_Content', '')) if retriever: try: docs = retriever.invoke(name) if docs: rag_info = "\n".join([d.page_content for d in docs]) except: pass # 2. 準備圖片 Prompt img_prompt = f"Delicious food from {name}, cinematic lighting, 8k, photorealistic" prompt_source = "⚠️ 預設生成" if not df_prompts.empty and PROMPT_COL_NAME in df_prompts.columns: if name in df_prompts.index: try: csv_prompt = df_prompts.loc[name, PROMPT_COL_NAME] if isinstance(csv_prompt, pd.Series): csv_prompt = csv_prompt.iloc[0] if pd.notna(csv_prompt) and str(csv_prompt).strip() != "": img_prompt = str(csv_prompt) prompt_source = "✅ CSV 檔案" except: pass # 3. 呼叫 LLM (移除了 quiz_state) ai_text = generate_content_with_groq(name, rag_info, diary_input, score_input) # 4. 組合回應 debug_text = "" if debug_mode: debug_text = f"\n\n---\n**🛠️ Prompt 來源**: {prompt_source}\n**Prompt**: `{img_prompt}`" final_response = f"### 🍽️ 推薦：{name} {note}\n\n{ai_text}{debug_text}" map_html = f'
🗺️ Google Map 導航
' yield final_response, None, map_html image_output = generate_image_huggingface(img_prompt) yield final_response, image_output, map_html # ========================================== # 2. 情緒辨識整合模組 (維持不變) # ========================================== # 定義中文字典 (繪圖用) emotion_text_obj = { 'angry': '生氣', 'disgust': '噁心', 'fear': '害怕', 'happy': '開心', 'sad': '難過', 'surprise': '驚訝', 'neutral': '正常' } # 繪圖函式 (維持不變) def putText(img, x, y, text, size=50, color=(255, 255, 255)): try: fontpath = 'NotoSansTC-VariableFont_wght.ttf' if not os.path.exists(fontpath): return img font = ImageFont.truetype(fontpath, size) imgPil = Image.fromarray(img) draw = ImageDraw.Draw(imgPil) displayText = emotion_text_obj.get(text, text) draw.text((x, y), displayText, fill=color, font=font) return np.array(imgPil) except: return img # 關鍵功能：偵測情緒並回傳「分數選項」 (維持不變) def detect_emotion_and_map(frame): if frame is None: return frame, None detected_emotion = "neutral" mapped_score = "3 (普通)" try: # 1. 辨識情緒 analyze = DeepFace.analyze(frame, actions=['emotion'], enforce_detection=False) if isinstance(analyze, list): analyze = analyze[0] detected_emotion = analyze['dominant_emotion'] # 2. 畫在圖片上 frame = putText(frame, 20, 40, detected_emotion) # 3. 橋樑：將情緒轉換為餐廳系統的分數 if detected_emotion == 'happy': mapped_score = "5 (超棒)" elif detected_emotion == 'surprise': mapped_score = "4 (不錯)" elif detected_emotion == 'neutral': mapped_score = "3 (普通)" elif detected_emotion in ['sad', 'fear']: mapped_score = "2 (不太好)" elif detected_emotion in ['angry', 'disgust']: mapped_score = "1 (心情差)" except Exception as e: print(f"DeepFace Error: {e}") pass return frame, mapped_score # ========================================== # 5. Gradio 介面建構 (修改版：鏡頭與結果分離) # ========================================== with gr.Blocks(title="AI 心情食堂") as demo: with gr.Column(visible=True) as main_app_col: gr.Markdown(f"## 🍱 AI 心情食堂導航") gr.Markdown("請看著鏡頭，讓 AI 幫你判斷今天的心情分數！") with gr.Row(): # 左側：輸入區 with gr.Column(scale=1): # ★ 修改重點：將鏡頭與結果分開顯示 ★ gr.Markdown("### 📸 步驟 1：心情偵測 (選用)") # 這裡用 Row 把兩個影像並排 (左邊鏡頭，右邊截圖) with gr.Row(): # 左邊：永遠是即時鏡頭 (不設為 output) webcam_input = gr.Image(sources=["webcam"], label="即時鏡頭 (請看這裡)", streaming=True) # 右邊：顯示偵測後的「靜態截圖」 captured_image = gr.Image(label="偵測結果截圖", interactive=False) detect_btn = gr.Button("📸 截圖並偵測心情 👇", variant="secondary") gr.Markdown("### 📝 步驟 2：確認與補充") score_input = gr.Radio( ["1 (心情差)", "2 (不太好)", "3 (普通)", "4 (不錯)", "5 (超棒)"], label="1. 心情分數 (AI 會自動填入，也可手動改)", value="3 (普通)" ) food_input = gr.Radio(["吃飯", "吃麵", "隨便"], label="2. 想吃什麼", value="隨便") diary_input = gr.Textbox(lines=3, label="3. 心情日記 (選填)", placeholder="例如：今天被老闆罵了，想吃點好料的...") debug_mode_btn = gr.Checkbox(label="🔧 顯示 Prompt 除錯資訊", value=False) submit_btn = gr.Button("🚀 送出給 Agent", variant="primary") # 右側：Agent 輸出區 (維持不變) with gr.Column(scale=1): agent_output = gr.Markdown(label="AI 回應") image_output = gr.Image(label="AI 推薦美食圖", type="pil", width=400) map_output = gr.HTML(label="地圖導航") # Events # ★ 修改重點：按鈕點擊後，輸出目標改為 captured_image，不再覆蓋 webcam_input detect_btn.click( fn=detect_emotion_and_map, inputs=[webcam_input], outputs=[captured_image, score_input] # 輸出到「右邊的截圖」和「分數選項」 ) # 提交按鈕 submit_btn.click( fn=mood_agent_logic, inputs=[score_input, food_input, diary_input, debug_mode_btn], outputs=[agent_output, image_output, map_output] ) if __name__ == "__main__": demo.launch()