Update app.py

app.py

@@ -3,138 +3,252 @@ import os
 import time
 import uuid
 import tempfile
-from PIL import Image, ImageDraw, ImageFont
+from PIL import Image
 import gradio as gr
-import base64
-import mimetypes
 from io import BytesIO
 from google import genai
-from google.genai import types
+import gspread
+from oauth2client.service_account import ServiceAccountCredentials
+import requests
 
-def generate(text, images, api_key, model="gemini-2.5-flash-image-preview"):
-    # Initialize client using provided api_key (or fallback to env variable)
-    client = genai.Client(api_key=(api_key.strip() if api_key and api_key.strip() != ""
-                                     else os.environ.get("GEMINI_API_KEY")))
+# 請將你的 Google Sheets 服務帳戶金鑰檔案命名為 "service_account.json" 並放在同一目錄
+SERVICE_ACCOUNT_FILE = "service_account.json"
 
-    # Prepare contents with images first, then text
-    contents = images + [text]
+# --- Google Sheets 相關函式 ---
+def get_sheet_data(spreadsheet_url):
+    """從 Google Sheet 讀取所有資料並返回 DataFrame 格式。"""
+    try:
+        scope = ['https://spreadsheets.google.com/feeds', 'https://www.googleapis.com/auth/drive']
+        creds = ServiceAccountCredentials.from_json_keyfile_name(SERVICE_ACCOUNT_FILE, scope)
+        client = gspread.authorize(creds)
+        
+        spreadsheet = client.open_by_url(spreadsheet_url)
+        worksheet = spreadsheet.worksheet("Sheet1")
+        
+        list_of_lists = worksheet.get_all_values()
+        
+        if not list_of_lists or len(list_of_lists) < 2:
+            return [], "試算表為空或沒有資料。"
+
+        headers = list_of_lists[0]
+        data = []
+        for i, row in enumerate(list_of_lists[1:], start=2):
+            if not any(row):  # 跳過空行
+                continue
+            row_dict = dict(zip(headers, row))
+            data.append({
+                "Index": i,
+                "白背圖URL": row_dict.get("White_Back_Image_URL", ""),
+                "參考圖URL": row_dict.get("Ref_Image_URL", ""),
+                "提示詞": row_dict.get("Prompt", "")
+            })
+            
+        dataframe_output = [[row['Index'], row['白背圖URL'], row['參考圖URL'], row['提示詞']] for row in data]
+        
+        log_message = f"成功讀取 {len(data)} 筆數據。"
+        return dataframe_output, log_message
+        
+    except FileNotFoundError:
+        raise gr.Error(f"錯誤：找不到金鑰檔案 '{SERVICE_ACCOUNT_FILE}'。請確認已正確放置。", duration=10)
+    except gspread.exceptions.SpreadsheetNotFound:
+        raise gr.Error(f"錯誤：Google Sheet not found at URL: {spreadsheet_url}", duration=10)
+    except Exception as e:
+        raise gr.Error(f"讀取試算表時發生錯誤: {e}", duration=10)
+
+def get_row_data(spreadsheet_url, row_number):
+    """從 Google Sheet 讀取指定列的資料。"""
+    try:
+        scope = ['https://spreadsheets.google.com/feeds', 'https://www.googleapis.com/auth/drive']
+        creds = ServiceAccountCredentials.from_json_keyfile_name(SERVICE_ACCOUNT_FILE, scope)
+        client = gspread.authorize(creds)
+
+        spreadsheet = client.open_by_url(spreadsheet_url)
+        worksheet = spreadsheet.worksheet("Sheet1")
+        
+        headers = worksheet.row_values(1)
+        row_data = worksheet.row_values(row_number)
+        
+        row_dict = dict(zip(headers, row_data))
+        
+        white_back_image_url = row_dict.get("White_Back_Image_URL", "")
+        ref_image_url = row_dict.get("Ref_Image_URL", "")
+        prompt_text = row_dict.get("Prompt", "")
+        
+        return white_back_image_url, ref_image_url, prompt_text
+
+    except FileNotFoundError:
+        raise gr.Error(f"錯誤：找不到金鑰檔案 '{SERVICE_ACCOUNT_FILE}'。", duration=10)
+    except gspread.exceptions.SpreadsheetNotFound:
+        raise gr.Error(f"錯誤：Google Sheet not found at URL: {spreadsheet_url}", duration=10)
+    except IndexError:
+        raise gr.Error(f"錯誤：指定的行數 {row_number} 不存在。", duration=10)
+    except Exception as e:
+        raise gr.Error(f"讀取指定行時發生錯誤: {e}", duration=10)
+
+# --- 下載圖片函式 (已更新) ---
+def load_image_from_url(url: str):
+    """
+    從 URL 下載圖片並以 PIL Image 格式回傳，使用更穩健的請求標頭。
+    """
+    if not url:
+        return None
+    try:
+        # 使用更完整的標頭來模擬瀏覽器請求
+        headers = {
+            'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36',
+            'Referer': 'https://www.google.com/',  # 模擬從 Google 搜尋點擊過來的
+            'Accept-Encoding': 'gzip, deflate, br',
+            'Accept-Language': 'en-US,en;q=0.9',
+            'Connection': 'keep-alive',
+        }
+        
+        response = requests.get(url, timeout=20, headers=headers)
+        response.raise_for_status()  # 如果狀態碼是 4xx 或 5xx，將引發異常
+
+        image = Image.open(BytesIO(response.content)).convert("RGB")
+        print(f"Debug: Successfully loaded image from URL: {url}")
+        return image
+    except requests.exceptions.HTTPError as e:
+        print(f"Error downloading image from {url}: HTTP Error {e.response.status_code}")
+        gr.Warning(f"從 URL '{url}' 下載圖片失敗：HTTP 錯誤 {e.response.status_code}", duration=10)
+        return None
+    except Exception as e:
+        print(f"An unexpected error occurred: {e}")
+        gr.Warning(f"從 URL '{url}' 下載圖片時發生意外錯誤：{e}", duration=10)
+        return None
+
+# --- Gemini 核心函式 ---
+def generate_image(text, images, api_key, model="gemini-2.5-flash-image-preview"):
+    """使用 Gemini 模型生成圖片。"""
+    if not api_key or api_key.strip() == "":
+        raise gr.Error("錯誤：請輸入有效的 Gemini API 金鑰。", duration=10)
 
-    response = client.models.generate_content(
-        model=model,
-        contents=contents,
-    )
-    text_response = ""
-    image_path = None
-
-    for part in response.candidates[0].content.parts:
-        if part.text is not None:
-            text_response += part.text + "\n"
-        elif part.inline_data is not None:
-            # Create a temporary file to store the generated image
-            with tempfile.NamedTemporaryFile(suffix=".png", delete=False) as tmp:
-                temp_path = tmp.name
-                generated_image = Image.open(BytesIO(part.inline_data.data))
-                generated_image.save(temp_path)
-                image_path = temp_path
-                print(f"Generated image saved to: {temp_path} with prompt: {text}")
-    return image_path, text_response
-
-def load_uploaded_images(uploaded_files):
-    """Load and display uploaded images immediately"""
-    uploaded_images = []
-    if uploaded_files:
-        for file in uploaded_files:
-            if file.name.lower().endswith(('.png', '.jpg', '.jpeg', '.webp')):
-                img = Image.open(file.name)
-                if img.mode == "RGBA":
-                    img = img.convert("RGBA")
-                uploaded_images.append(img)
-    return uploaded_images
-
-def process_image_and_prompt(uploaded_files, prompt, gemini_api_key):
     try:
-        input_text = prompt
-        model = "gemini-2.5-flash-image-preview"
-        # Load images from uploaded files
-        images = []
-        uploaded_images = []
-        if uploaded_files:
-            for file in uploaded_files:
-                if file.name.lower().endswith(('.png', '.jpg', '.jpeg', '.webp')):
-                    img = Image.open(file.name)
-                    if img.mode == "RGBA":
-                        img = img.convert("RGBA")
-                    images.append(img)
-                    uploaded_images.append(img)
-        if not images:
-            raise gr.Error("Please upload at least one image", duration=5)
-        # Format: [dress_image, model_image, text_input] or [image1, image2, ..., text_input]
-        image_path, text_response = generate(text=input_text, images=images, api_key=gemini_api_key, model=model)
+        client = genai.Client(api_key=api_key.strip())
+        contents = images + [text]
+        response = client.models.generate_content(
+            model=model,
+            contents=contents,
+        )
+        text_response = ""
+        image_path = None
+        for part in response.candidates[0].content.parts:
+            if part.text is not None:
+                text_response += part.text + "\n"
+            elif part.inline_data is not None:
+                with tempfile.NamedTemporaryFile(suffix=".png", delete=False) as tmp:
+                    temp_path = tmp.name
+                    generated_image = Image.open(BytesIO(part.inline_data.data))
+                    generated_image.save(temp_path)
+                    image_path = temp_path
+        return image_path, text_response
+    except Exception as e:
+        raise gr.Error(f"Gemini API 呼叫失敗: {e}", duration=10)
+
+# --- Gradio 互動函式 ---
+def process_sheet_data(sheet_url):
+    """處理試算表，讀取所有數據並更新 DataFrame。"""
+    if not sheet_url:
+        raise gr.Error("請輸入 Google Sheet URL。", duration=5)
+    
+    data, log_message = get_sheet_data(sheet_url)
+    return data, log_message
+
+def generate_image_from_row(sheet_url, row_number, gemini_api_key):
+    """根據指定的行數，生成圖片。"""
+    if not sheet_url:
+        raise gr.Error("請先輸入 Google Sheet URL。", duration=5)
+    if not row_number or row_number <= 1:
+        raise gr.Error("請輸入有效的行數 (大於 1)。", duration=5)
+        
+    try:
+        row_num = int(row_number)
+        white_back_url, ref_image_url, prompt_text = get_row_data(sheet_url, row_num)
+        
+        log_message = f"開始處理第 {row_num} 行...\n"
+        log_message += f"白背圖URL: {white_back_url}\n"
+        log_message += f"參考圖URL: {ref_image_url}\n"
+        log_message += f"提示詞: {prompt_text}\n"
+
+        images_for_gemini = []
+        
+        # 使用新的函式
+        if white_back_url:
+            wb_img = load_image_from_url(white_back_url)
+            if wb_img:
+                images_for_gemini.append(wb_img)
+        if ref_image_url:
+            ref_img = load_image_from_url(ref_image_url)
+            if ref_img:
+                images_for_gemini.append(ref_img)
+
+        if not images_for_gemini:
+            return None, "警告：無效的圖片 URL，無法生成圖片。"
+        
+        log_message += "圖片已下載，開始呼叫 Gemini 模型...\n"
+        
+        image_path, text_response = generate_image(text=prompt_text, images=images_for_gemini, api_key=gemini_api_key)
 
         if image_path:
-            # Load and convert the image if needed.
-            result_img = Image.open(image_path)
-            if result_img.mode == "RGBA":
-                result_img = result_img.convert("RGBA")
-            return uploaded_images, [result_img], ""  # Return uploaded images, generated image, and empty text output.
+            log_message += "圖片生成成功！"
+            return image_path, log_message
         else:
-            # Return uploaded images, no generated image, and the text response.
-            return uploaded_images, None, text_response
+            log_message += "圖片生成失敗。\n"
+            log_message += f"Gemini 文字回應: {text_response}"
+            return None, log_message
+            
+    except ValueError:
+        raise gr.Error("行數必須為數字。", duration=5)
     except Exception as e:
-        raise gr.Error(f"Error Getting {e}", duration=5)
+        return None, f"生成圖片時發生錯誤: {e}"
 
-# Build a Blocks-based interface with a custom HTML header and CSS
+# --- Gradio 介面設定 ---
 with gr.Blocks() as demo:
-    # Custom HTML header with proper class for styling
-    
-
+    gr.Markdown("## Google Sheets 圖片生成器")
     
     with gr.Row(elem_classes="main-content"):
         with gr.Column(elem_classes="input-column"):
-            image_input = gr.File(
-                file_types=["image"],
-                file_count="multiple",
-                label="Upload Images ",
-                elem_id="image-input",
-                elem_classes="upload-box"
-            )
             gemini_api_key = gr.Textbox(
                 lines=1,
-                placeholder="Enter Gemini API Key (optional)",
-                label="Gemini API Key (optional)",
+                placeholder="請在此輸入你的 Gemini API 金鑰",
+                label="Gemini API 金鑰",
                 elem_classes="api-key-input"
             )
-            prompt_input = gr.Textbox(
-                lines=2,
-                placeholder="Enter prompt here...",
-                label="Prompt",
-                elem_classes="prompt-input"
+            sheet_url_input = gr.Textbox(
+                label="Google Sheet URL", 
+                value="https://docs.google.com/spreadsheets/d/1G3olHxydDIbnyXdh5nnw5TG0akZFeMeYm-25JmCGDLg/edit?gid=0#gid=0"
             )
-            submit_btn = gr.Button("Generate", elem_classes="generate-btn")
-
+            process_button = gr.Button("處理試算表", elem_classes="generate-btn")
+            
+            with gr.Row():
+                row_index_input = gr.Number(label="要生成的行數", precision=0, value=2)
+                generate_selected_button = gr.Button("生成所選行的圖片", elem_classes="generate-btn")
+        
         with gr.Column(elem_classes="output-column"):
-            uploaded_gallery = gr.Gallery(label="Uploaded Images", elem_classes="uploaded-gallery")
-            output_gallery = gr.Gallery(label="Generated Outputs", elem_classes="output-gallery")
-            output_text = gr.Textbox(
-                label="Gemini Output",
-                placeholder="Text response will appear here if no image is generated.",
-                elem_classes="output-text"
+            output_dataframe = gr.DataFrame(
+                headers=["Index", "白背圖URL", "參考圖URL", "提示詞"],
+                col_count=(4, "fixed"),
+                interactive=False,
+                label="已處理的試算表數據"
+            )
+            generated_image_output = gr.Image(label="生成的圖片", elem_classes="output-gallery")
+            operation_log_output = gr.Textbox(
+                label="操作日誌", 
+                lines=10,
+                placeholder="文字回應和日誌會顯示在這裡。"
             )
 
-    # Set up the interaction with three outputs.
-    submit_btn.click(
-        fn=process_image_and_prompt,
-        inputs=[image_input, prompt_input, gemini_api_key],
-        outputs=[uploaded_gallery, output_gallery, output_text],
+    # 按鈕的事件綁定
+    process_button.click(
+        fn=process_sheet_data,
+        inputs=[sheet_url_input],
+        outputs=[output_dataframe, operation_log_output]
     )
 
-    # Update uploaded gallery immediately when files are uploaded
-    image_input.upload(
-        fn=load_uploaded_images,
-        inputs=[image_input],
-        outputs=[uploaded_gallery],
+    generate_selected_button.click(
+        fn=generate_image_from_row,
+        inputs=[sheet_url_input, row_index_input, gemini_api_key],
+        outputs=[generated_image_output, operation_log_output]
     )
-
     
-
-demo.queue(max_size=50).launch(mcp_server=True, share=True)
+    demo.queue().launch(mcp_server=True, share=True)