Update pdf_to_jpg_gradio5_en.py

pdf_to_jpg_gradio5_en.py

@@ -1,10 +1,13 @@
 # pdf_to_jpg_gradio.py
+
 import gradio as gr
 import io
 import uuid
 import numpy as np
 from PIL import Image, ImageDraw
 from pdf2image import convert_from_bytes
+import zipfile
+import os
 
 def convert_pdf_to_combined_image(pdf_bytes, output_size="A4", dpi=96, line_color=(200, 200, 200)):
     """
@@ -47,38 +50,84 @@ def convert_pdf_to_combined_image(pdf_bytes, output_size="A4", dpi=96, line_colo
     
     return combined_image
 
+def convert_pdf_to_multiple_images(pdf_bytes, output_size="A4", dpi=96):
+    """
+    将 PDF 文件的字节数据(pdf_bytes)转换为多个单独的图片（每一页一个）。
+    参数:
+        pdf_bytes:  PDF 文件的字节内容
+        output_size: 输出尺寸，支持 Original / A4 / A3
+        dpi:         像素密度，默认96
+    返回:
+        images_list: PIL Image 对象的列表，每个元素对应一页 PDF。
+    """
+    images = convert_from_bytes(pdf_bytes)
 
-def pdf_to_jpg(pdf_file, output_size):
+    sizes = {
+        "Original": images[0].size,
+        "A4": (int(210 * dpi / 25.4), int(297 * dpi / 25.4)),
+        "A3": (int(297 * dpi / 25.4), int(420 * dpi / 25.4))
+    }
+    output_width, output_height = sizes[output_size]
+
+    images_list = []
+    for img in images:
+        # 调整每页大小
+        img_resized = img.resize((output_width, output_height))
+        images_list.append(img_resized)
+    
+    return images_list
+
+def pdf_to_jpg_or_zip(pdf_file, output_size, convert_mode):
     """
-    将上传的 PDF 文件转换为长图，并返回 (用于显示的图像[numpy], 用于下载的唯一文件路径)
+    根据用户选择的 convert_mode，决定输出单张长图或多个图片(压缩包)。
     参数:
-        pdf_file:    Gradio 上传的 PDF 文件对象
-        output_size: 用户选择的输出尺寸（Original / A4 / A3）
+        pdf_file:     Gradio 上传的 PDF 文件对象 (binary)
+        output_size:  'Original' / 'A4' / 'A3'
+        convert_mode: 'Single Image' 或 'Multiple Images'
     返回:
         (display_image_array, file_path)
-        display_image_array: numpy 数组，用于在 gr.Image(type="numpy") 中显示
-        file_path: string，唯一文件路径，用于 gr.File(type="filepath") 下载
+        如果是单图:
+            display_image_array: numpy 数组（用于预览）
+            file_path:           合并后的单张图片的唯一文件路径
+        如果是多图:
+            display_image_array: None（不预览）
+            file_path:           打包好的 zip 文件的唯一文件路径
     """
     if pdf_file is None:
         return None, None
     
-    # 读取 PDF 文件的字节内容
+    # 将上传的pdf_file（字节流）取出来
     pdf_bytes = pdf_file
-    
-    # 调用拼接函数
-    combined_image = convert_pdf_to_combined_image(pdf_bytes, output_size=output_size)
-
-    # 转为 numpy 数组，供 gr.Image(type="numpy") 显示
-    display_image_array = np.array(combined_image)
-
-    # 使用 uuid 生成唯一文件名
-    unique_filename = f"{uuid.uuid4()}.jpg"
-    combined_image.save(unique_filename, format="JPEG")
-
-    return display_image_array, unique_filename
 
-
-# ========== Gradio 应用界面部分 (英文界面) ==========
+    if convert_mode == "Single Image":
+        # 调用原有拼接逻辑
+        combined_image = convert_pdf_to_combined_image(pdf_bytes, output_size=output_size)
+        display_image_array = np.array(combined_image)
+        
+        # 使用 uuid 生成唯一文件名，保存单张 JPG
+        unique_filename = f"{uuid.uuid4()}.jpg"
+        combined_image.save(unique_filename, format="JPEG")
+        return display_image_array, unique_filename
+    
+    else:  # convert_mode == "Multiple Images"
+        # 将 PDF 转换为多个图片
+        images_list = convert_pdf_to_multiple_images(pdf_bytes, output_size=output_size)
+        
+        # 创建唯一 zip 文件
+        zip_filename = f"{uuid.uuid4()}.zip"
+        with zipfile.ZipFile(zip_filename, 'w') as zipf:
+            # 依次将每张图片写入压缩包
+            for idx, img in enumerate(images_list):
+                page_jpg_filename = f"page_{idx+1}.jpg"
+                img.save(page_jpg_filename, "JPEG")
+                zipf.write(page_jpg_filename)
+                # 写入后删除临时文件
+                os.remove(page_jpg_filename)
+        
+        # 返回 None 不显示预览
+        return None, zip_filename
+
+# ========== Gradio 应用界面部分 ==========
 with gr.Blocks(
     title="PDF to JPG Converter - High Quality PDF Merging",
     css=".gradio-container {max-width: 800px; margin: 0 auto;}"
@@ -87,8 +136,9 @@ with gr.Blocks(
     
     gr.Markdown("""
     <p style='text-align: center; font-size: 16px;'>
-    Welcome to our PDF to JPG Online Converter! This tool supports merging all PDF pages into one single long image for easy viewing and sharing.<br/>
-    <b>SEO Summary:</b> Supports Original, A4, and A3 page sizes for stitching, fast conversion, high-quality output, and free to use.
+    Welcome to our PDF to JPG Online Converter! This tool supports merging all PDF pages into one single long image 
+    or converting each page into separate images for easy viewing and sharing.<br/>
+    Supports Original, A4, and A3 page sizes for stitching, fast conversion, high-quality output, and free to use.
     </p>
     """)
     
@@ -98,8 +148,9 @@ with gr.Blocks(
     <ol>
         <li>Upload your PDF file in the "Choose PDF File" section below.</li>
         <li>Select your desired output size (Original, A4, or A3) from the dropdown menu.</li>
-        <li>Click the "Convert" button and wait a few seconds to preview the result.</li>
-        <li>Click the "Download File" button to save the JPG image to your local device.</li>
+        <li>Select whether you want "Multiple Images" or "Single Image".</li>
+        <li>Click the "Convert" button and wait a few seconds to preview the result (only if single image).</li>
+        <li>Click the "Download File" button to save the result to your local device.</li>
     </ol>
     <hr/>
     """)
@@ -112,18 +163,33 @@ with gr.Blocks(
             value="A4", 
             label="Choose Output Size"
         )
+        # 默认选项为 Multiple Images
+        conversion_mode = gr.Radio(
+            choices=["Multiple Images", "Single Image"],
+            value="Multiple Images",  
+            label="Conversion Mode"
+        )
+
+    # 在 Single Image 模式下增加英文提示说明
+    gr.Markdown("""
+    <p style='color: red; font-size: 14px;'>
+    Notice: If you choose <b>Single Image</b> mode but your PDF has many pages, you might encounter 
+    maximum height limitations that can cause errors or improper display. 
+    In that case, please consider choosing <b>Multiple Images</b> mode.
+    </p>
+    """)
 
     convert_button = gr.Button("Convert")
 
     with gr.Row():
-        # type="numpy"，便于直接显示 numpy 数组
         image_output = gr.Image(label="Preview", type="numpy")
-        # type="filepath"，返回本地文件路径，避免串名
         download_output = gr.File(label="Download File", type="filepath")
 
+    # 输入: pdf_input, size_dropdown, conversion_mode
+    # 输出: image_output, download_output
     convert_button.click(
-        fn=pdf_to_jpg, 
-        inputs=[pdf_input, size_dropdown],
+        fn=pdf_to_jpg_or_zip,
+        inputs=[pdf_input, size_dropdown, conversion_mode],
         outputs=[image_output, download_output]
     )