适配上传不同类型的文件，适配gpt-4-all,支持包含vision字段的模型，自动打开上传文件按钮

app/api/file/upload/route.ts

@@ -32,7 +32,8 @@ async function handle(req: NextRequest) {
 
     const buffer = Buffer.from(imageData);
     
-    var fileName = `${Date.now()}.png`;
+    // 使用获取到的文件后缀
+    var fileName = image.name;
     var filePath = "";
     const serverConfig = getServerSideConfig();
     if (serverConfig.isStoreFileToLocal) {

app/client/platforms/openai.ts

@@ -83,7 +83,7 @@ export class ChatGPTApi implements LLMApi {
       const base64 = Buffer.from(response.data, "binary").toString("base64");
       return base64;
     };
-    if (options.config.model === "gpt-4-vision-preview") {
+    if (options.config.model.includes("vision")) {
       for (const v of options.messages) {
         let message: {
           role: string;
@@ -102,10 +102,91 @@ export class ChatGPTApi implements LLMApi {
         });
         if (v.image_url) {
           var base64Data = await getImageBase64Data(v.image_url);
+          interface MIMEMap {
+            [key: string]: string;
+          }
+
+          const extensionToMIME: MIMEMap = {
+            'png': 'image/png',
+            'jpg': 'image/jpeg',
+            'webp': 'image/webp',
+            'gif': 'image/gif',
+            'bmp': 'image/bmp',
+            'svg': 'image/svg+xml',
+            'txt': 'text/plain',
+            'pdf': 'application/pdf',
+            'doc': 'application/msword',
+            'docx': 'application/vnd.openxmlformats-officedocument.wordprocessingml.document',
+            'xls': 'application/vnd.ms-excel',
+            'xlsx': 'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet',
+            'ppt': 'application/vnd.ms-powerpoint',
+            'pptx': 'application/vnd.openxmlformats-officedocument.presentationml.presentation',
+            'zip': 'application/zip',
+            'rar': 'application/x-rar-compressed',
+            'bin': 'application/octet-stream',
+
+            // Audio
+            'mp3': 'audio/mpeg',
+            'wav': 'audio/wav',
+            'ogg': 'audio/ogg',
+            'flac': 'audio/flac',
+            'aac': 'audio/aac',
+            'weba': 'audio/webm',
+            'midi': 'audio/midi',
+
+            // Video
+            'mp4': 'video/mp4',
+            'webm': 'video/webm',
+            'avi': 'video/x-msvideo',
+            'wmv': 'video/x-ms-wmv',
+            'flv': 'video/x-flv',
+            '3gp': 'video/3gpp',
+            'mkv': 'video/x-matroska',
+
+            //编程
+            'js': 'application/javascript',
+            'json': 'application/json',
+            'html': 'text/html',
+            'css': 'text/css',
+            'xml': 'application/xml',
+            'csv': 'text/csv',
+            'ts': 'text/typescript',
+            'java': 'text/x-java-source',
+            'py': 'text/x-python',
+            'c': 'text/x-csrc',
+            'cpp': 'text/x-c++src',
+            'h': 'text/x-chdr',
+            'hpp': 'text/x-c++hdr',
+            'php': 'application/x-httpd-php',
+            'rb': 'text/x-ruby',
+            'go': 'text/x-go',
+            'rs': 'text/rust',
+            'swift': 'text/x-swift',
+            'kt': 'text/x-kotlin',
+            'scala': 'text/x-scala',
+          };
+
+          let mimeType: string | undefined;
+          try {
+            // 使用正则表达式获取文件后缀
+            const match = v.image_url.match(/\.(\w+)$/);
+            if (match) {
+              const fileExtension = match[1].toLowerCase();
+              mimeType = extensionToMIME[fileExtension];
+              if (!mimeType) {
+                throw new Error('Unknown file extension: ' + fileExtension);
+              }
+            } else {
+              throw new Error('Unable to extract file extension from the URL');
+            }
+          } catch (error) {
+            // 使用通用的MIME类型
+            mimeType = 'text/plain';
+          }
           message.content.push({
             type: "image_url",
             image_url: {
-              url: `data:image/jpeg;base64,${base64Data}`,
+              url: `data:${mimeType};base64,${base64Data}`,
             },
           });
         }
@@ -136,7 +217,7 @@ export class ChatGPTApi implements LLMApi {
       frequency_penalty: modelConfig.frequency_penalty,
       top_p: modelConfig.top_p,
       max_tokens:
-        modelConfig.model == "gpt-4-vision-preview"
+        modelConfig.model.includes("vision")
           ? modelConfig.max_tokens
           : null,
       // max_tokens: Math.max(modelConfig.max_tokens, 1024),
@@ -226,7 +307,7 @@ export class ChatGPTApi implements LLMApi {
               try {
                 const resJson = await res.clone().json();
                 extraInfo = prettyObject(resJson);
-              } catch {}
+              } catch { }
 
               if (res.status === 401) {
                 responseTexts.push(Locale.Error.Unauthorized);

app/components/chat.tsx

@@ -373,17 +373,17 @@ function ChatAction(props: {
       style={
         props.icon && !props.loding
           ? ({
-              "--icon-width": `${width.icon}px`,
-              "--full-width": `${width.full}px`,
-              ...props.style,
-            } as React.CSSProperties)
+            "--icon-width": `${width.icon}px`,
+            "--full-width": `${width.full}px`,
+            ...props.style,
+          } as React.CSSProperties)
           : props.loding
-          ? ({
+            ? ({
               "--icon-width": `30px`,
               "--full-width": `30px`,
               ...props.style,
             } as React.CSSProperties)
-          : props.style
+            : props.style
       }
     >
       {props.icon ? (
@@ -538,7 +538,7 @@ export function ChatActions(props: {
         }
       }
     };
-    if (currentModel === "gpt-4-vision-preview") {
+    if (currentModel.includes("vision")) {
       window.addEventListener("paste", onPaste);
       return () => {
         window.removeEventListener("paste", onPaste);
@@ -609,7 +609,7 @@ export function ChatActions(props: {
 
         {config.pluginConfig.enable &&
           /^gpt(?!.*03\d{2}$).*$/.test(currentModel) &&
-          currentModel != "gpt-4-vision-preview" && (
+          !currentModel.includes("vision") && (
             <ChatAction
               onClick={switchUsePlugins}
               text={
@@ -620,16 +620,16 @@ export function ChatActions(props: {
               icon={usePlugins ? <EnablePluginIcon /> : <DisablePluginIcon />}
             />
           )}
-        {currentModel == "gpt-4-vision-preview" && (
+        {currentModel.includes("vision") && (
           <ChatAction
             onClick={selectImage}
-            text="选择图片"
+            text="选择文档"
             loding={uploadLoading}
             icon={<UploadIcon />}
             innerNode={
               <input
                 type="file"
-                accept=".png,.jpg,.webp,.jpeg"
+                accept="*/*"
                 id="chat-image-file-select-upload"
                 style={{ display: "none" }}
                 onChange={onImageSelected}
@@ -1031,28 +1031,28 @@ function _Chat() {
       .concat(
         isLoading
           ? [
-              {
-                ...createMessage({
-                  role: "assistant",
-                  content: "……",
-                }),
-                preview: true,
-              },
-            ]
+            {
+              ...createMessage({
+                role: "assistant",
+                content: "……",
+              }),
+              preview: true,
+            },
+          ]
           : [],
       )
       .concat(
         userInput.length > 0 && config.sendPreviewBubble
           ? [
-              {
-                ...createMessage({
-                  role: "user",
-                  content: userInput,
-                  image_url: userImage?.fileUrl,
-                }),
-                preview: true,
-              },
-            ]
+            {
+              ...createMessage({
+                role: "user",
+                content: userInput,
+                image_url: userImage?.fileUrl,
+              }),
+              preview: true,
+            },
+          ]
           : [],
       );
   }, [
@@ -1149,7 +1149,7 @@ function _Chat() {
         if (payload.key || payload.url) {
           showConfirm(
             Locale.URLCommand.Settings +
-              `\n${JSON.stringify(payload, null, 4)}`,
+            `\n${JSON.stringify(payload, null, 4)}`,
           ).then((res) => {
             if (!res) return;
             if (payload.key) {
@@ -1412,7 +1412,7 @@ function _Chat() {
                       defaultShow={i >= messages.length - 6}
                     />
                   </div>
-                  {!isUser && message.model == "gpt-4-vision-preview" && (
+                  {!isUser && message?.model?.includes("vision") && (
                     <div
                       className={[
                         styles["chat-message-actions"],

app/components/markdown.tsx

@@ -121,10 +121,25 @@ function _MarkDownContent(props: { content: string; imageBase64?: string }) {
     () => escapeDollarNumber(props.content),
     [props.content],
   );
+  // 判断文件路径
+  const isImage = (filename: any) => {
+    const imageExtensions = ['.png', '.jpg', '.jpeg', '.gif', '.bmp', '.webp', '.tiff'];
+    return imageExtensions.some(ext => filename.toLowerCase().endsWith(ext));
+  };
 
+  const show_filename = (base64: any) => {
+    let parts = base64.split("/");
+    return parts.pop();
+  };
   return (
     <div style={{ fontSize: "inherit" }}>
-      {props.imageBase64 && <img src={props.imageBase64} alt="" />}
+      {props.imageBase64 && isImage(props.imageBase64) && <img src={props.imageBase64} alt="" />}
+      {props.imageBase64 && !isImage(props.imageBase64) &&
+        <a href={props.imageBase64} style={{ fontWeight: 'bold' }}>
+          文件：{show_filename(props.imageBase64)}
+        </a>
+      }
+      
       <ReactMarkdown
         remarkPlugins={[RemarkMath, RemarkGfm, RemarkBreaks]}
         rehypePlugins={[

app/store/chat.ts

@@ -335,8 +335,7 @@ export const useChatStore = createPersistStore(
           config.pluginConfig.enable &&
           session.mask.usePlugins &&
           allPlugins.length > 0 &&
-          modelConfig.model.startsWith("gpt") &&
-          modelConfig.model != "gpt-4-vision-preview"
+          !modelConfig.model.includes("vision")
         ) {
           console.log("[ToolAgent] start");
           const pluginToolNames = allPlugins.map((m) => m.toolName);