Update app.py

app.py

@@ -24,6 +24,9 @@ import httpx
 import aiohttp
 import uuid
 import tqdm
+import base64, pathlib
+from io import BytesIO
+from pdf2image import convert_from_bytes, convert_from_path     # pip install pdf2image
 
 import requests
 
@@ -275,16 +278,59 @@ def to_file(image_path):
     return image_path
 
 
-# async def process_file(file_path):
-#     if not file_path.endswith(".pdf"):
-#         tmp_path = Path(file_path).with_suffix(".pdf")
-#         images_to_pdf(file_path, tmp_path)
-#     else:
-#         tmp_path = Path(file_path)
 
-#     async with httpx.AsyncClient() as client:
-#         await send_pdf_to_parse_async(client, str(tmp_path), IP, PORT)
-#     return str(tmp_path)
+
+def files_to_b64(files, pdf_dpi: int = 200):
+    """
+    把上传文件统一转成 base64 图片列表。
+    • 普通图片：直接 base64
+    • PDF     ：按页转成 PNG 再 base64
+
+    返回
+    ----
+    List[str]  每个元素形如  "data:image/png;base64,xxxx..."
+    """
+    out: list[str] = []
+
+    for f in files:
+        # ---------- Gradio 4.x NamedString / FileData ----------
+        if hasattr(f, "data"):             # 新版 FileData / NamedString
+            raw_bytes = f.data
+            suffix    = pathlib.Path(f.name).suffix.lower()
+
+            # -- PDF --
+            if suffix == ".pdf":
+                pages = convert_from_bytes(raw_bytes, dpi=pdf_dpi)
+                for page in pages:
+                    buf = BytesIO()
+                    page.save(buf, format="PNG")
+                    b64 = base64.b64encode(buf.getvalue()).decode()
+                    out.append(f"data:image/png;base64,{b64}")
+                continue  # 进入下一文件
+
+            # -- 普通图片 --
+            b64 = base64.b64encode(raw_bytes).decode()
+            out.append(f"data:image/{suffix[1:]};base64,{b64}")
+
+        # ---------- 旧版：直接获得文件路径 ----------
+        else:
+            path   = pathlib.Path(f)
+            suffix = path.suffix.lower()
+
+            if suffix == ".pdf":
+                pages = convert_from_path(str(path), dpi=pdf_dpi)
+                for page in pages:
+                    buf = BytesIO()
+                    page.save(buf, format="PNG")
+                    b64 = base64.b64encode(buf.getvalue()).decode()
+                    out.append(f"data:image/png;base64,{b64}")
+                continue
+
+            raw_bytes = path.read_bytes()
+            b64 = base64.b64encode(raw_bytes).decode()
+            out.append(f"data:image/{suffix[1:]};base64,{b64}")
+
+    return out
 
 
 async def process_file(file_path):
@@ -307,35 +353,6 @@ async def process_file(file_path):
 import gradio as gr
 import base64, pathlib
 
-# def embed_file(file, scale):
-#     """根据后缀决定用 PDF 还是 <img>."""
-#     path = pathlib.Path(file.name)
-#     data = base64.b64encode(path.read_bytes()).decode()
-#     if path.suffix.lower() == ".pdf":
-#         b64 = base64.b64encode(pathlib.Path(file.name).read_bytes()).decode()
-#     # 注意：把 JS/HTML 写在一起返回给 gr.HTML
-#         return f"""
-#             <script>
-#             const pdf = atob("{b64}");
-#             const bytes = Uint8Array.from(pdf, c => c.charCodeAt(0));
-#             const blob = new Blob([bytes], {{type: "application/pdf"}});
-#             const url = URL.createObjectURL(blob);
-#             const iframe = document.getElementById("pdf_view") || document.createElement("iframe");
-#             iframe.id = "pdf_view";
-#             iframe.style = "width:100%;height:800px;border:none";
-#             iframe.src = url + "#zoom={int(scale*100)}";
-#             document.currentScript.parentElement.replaceChildren(iframe);
-#             </script>
-#         """
-#     else:
-#         # 图片：CSS transform 实现 zoom；支持拖拽滚轮缩放
-#         return (
-#             f'<div style="overflow:auto;border:1px solid #ccc;'
-#             f'width:100%;height:800px;text-align:center">'
-#             f'<img src="data:image/{path.suffix[1:]};base64,{data}" '
-#             f'style="transform:scale({scale});transform-origin:0 0;" />'
-#             f'</div>'
-#         )
 
 
 def embed_file(file, scale: float):
@@ -353,28 +370,8 @@ def embed_file(file, scale: float):
 
     # ---------- 1. PDF ----------
     if ext == ".pdf":
-        # 输出 JS：把 Base64 转 Blob，然后注入/更新 iframe
-        pdf_b64 = base64.b64encode(pathlib.Path(file.name).read_bytes()).decode()
-        return f"""
-            <script>
-            // -------- 把 base64 转为二进制 Blob ----------
-            const bytes = Uint8Array.from(
-                atob("{pdf_b64}"), ch => ch.charCodeAt(0)
-            );
-            const blobUrl = URL.createObjectURL(
-                new Blob([bytes], {{type: "application/pdf"}})
-            );
-            // -------- 创建或更新 iframe ----------
-            let frame = document.getElementById("pdf_view");
-            if (!frame) {{
-              frame = document.createElement("iframe");
-              frame.id = "pdf_view";
-              frame.style = "width:100%;height:800px;border:none;";
-              document.currentScript.parentElement.appendChild(frame);
-            }}
-            frame.src = blobUrl + "#zoom={int(scale*100)}";
-            </script>
-            """
+
+
     else:
         return (
             f'<div style="overflow:auto;border:1px solid #ccc;'
@@ -406,14 +403,58 @@ if __name__ == '__main__':
                     clear_bu = gr.ClearButton(value='Clear')
                 pdf_show = PDF(label='Preview', interactive=False, visible=True, height=800)
 
-                slider = gr.Slider(0.5, 3.0, value=1.0, step=0.1, label="缩放倍数")
-                out = gr.HTML(sanitize_html=False)
-            
-                file.change(embed_file, [file, slider], out)
-                slider.change(embed_file, [file, slider], out)
-    
 
 
+                with gr.Row():
+                    prev_btn = gr.Button("⬅️ 上一张")
+                    next_btn = gr.Button("下一张 ➡️")
+                    zoom = gr.Slider(0.5, 3, value=1, step=0.1, label="缩放倍数")
+            
+                # HTML 容器
+                viewer = gr.HTML()
+
+                # ---- 状态变量 ----
+                img_list_state = gr.State([])   # base64 列表
+                idx_state = gr.State(0)         # 当前索引
+            
+                # ---- 事件逻辑 ----
+                def upload_handler(files):
+                    b64s = files_to_b64(files)
+                    return b64s, 0, render_img(b64s, 0, 1)
+            
+                file.change(
+                    upload_handler,
+                    inputs=file,
+                    outputs=[img_list_state, idx_state, viewer],
+                )
+            
+                def show_prev(b64s, idx, scale):
+                    idx -= 1
+                    return idx, render_img(b64s, idx, scale)
+            
+                prev_btn.click(
+                    show_prev,
+                    inputs=[img_list_state, idx_state, zoom],
+                    outputs=[idx_state, viewer],
+                )
+            
+                def show_next(b64s, idx, scale):
+                    idx += 1
+                    return idx, render_img(b64s, idx, scale)
+            
+                next_btn.click(
+                    show_next,
+                    inputs=[img_list_state, idx_state, zoom],
+                    outputs=[idx_state, viewer],
+                )
+            
+                # 缩放滑块变化
+                zoom.change(
+                    lambda b64s, idx, scale: render_img(b64s, idx, scale),
+                    inputs=[img_list_state, idx_state, zoom],
+                    outputs=viewer,
+                )
+    
          
                 example_root = os.path.join(os.path.dirname(__file__), 'examples')
                 images = [
@@ -464,7 +505,7 @@ if __name__ == '__main__':
                             
                         
                
-        file.change(fn=process_file, inputs=file, outputs=pdf_show)
+        # file.change(fn=process_file, inputs=file, outputs=pdf_show)
 
 
         change_bu.click(