Merge remote-tracking branch 'upstream/main' into klaas20240303

app/client/platforms/openai.ts

@@ -1,3 +1,4 @@
+"use client";
 import {
   ApiPath,
   DEFAULT_API_HOST,
@@ -8,7 +9,14 @@ import {
 } from "@/app/constant";
 import { useAccessStore, useAppConfig, useChatStore } from "@/app/store";
 
-import { ChatOptions, getHeaders, LLMApi, LLMModel, LLMUsage } from "../api";
+import {
+  ChatOptions,
+  getHeaders,
+  LLMApi,
+  LLMModel,
+  LLMUsage,
+  MultimodalContent,
+} from "../api";
 import Locale from "../../locales";
 import {
   EventStreamContentType,
@@ -17,6 +25,11 @@ import {
 import { prettyObject } from "@/app/utils/format";
 import { getClientConfig } from "@/app/config/client";
 import { makeAzurePath } from "@/app/azure";
+import {
+  getMessageTextContent,
+  getMessageImages,
+  isVisionModel,
+} from "@/app/utils";
 
 //# for AdEx custom usage tracking (calls per model per user per datekey)
 // import { NextApiRequest, NextApiResponse } from 'next';
@@ -63,7 +76,9 @@ export class ChatGPTApi implements LLMApi {
 
     if (baseUrl.length === 0) {
       const isApp = !!getClientConfig()?.isApp;
-      baseUrl = isApp ? DEFAULT_API_HOST : ApiPath.OpenAI;
+      baseUrl = isApp
+        ? DEFAULT_API_HOST + "/proxy" + ApiPath.OpenAI
+        : ApiPath.OpenAI;
     }
 
     if (baseUrl.endsWith("/")) {
@@ -77,6 +92,8 @@ export class ChatGPTApi implements LLMApi {
       path = makeAzurePath(path, accessStore.azureApiVersion);
     }
 
+    console.log("[Proxy Endpoint] ", baseUrl, path);
+
     return [baseUrl, path].join("/");
   }
 
@@ -85,9 +102,10 @@ export class ChatGPTApi implements LLMApi {
   }
 
   async chat(options: ChatOptions) {
+    const visionModel = isVisionModel(options.config.model);
     const messages = options.messages.map((v) => ({
       role: v.role,
-      content: v.content,
+      content: visionModel ? v.content : getMessageTextContent(v),
     }));
 
     const modelConfig = {
@@ -123,6 +141,16 @@ export class ChatGPTApi implements LLMApi {
       // Please do not ask me why not send max_tokens, no reason, this param is just shit, I dont want to explain anymore.
     };
 
+    // add max_tokens to vision model
+    if (visionModel) {
+      Object.defineProperty(requestPayload, "max_tokens", {
+        enumerable: true,
+        configurable: true,
+        writable: true,
+        value: modelConfig.max_tokens,
+      });
+    }
+
     console.log("[Request] openai payload: ", requestPayload);
 
     const modelIdentifier = modelConfig.model; 
@@ -177,6 +205,9 @@ export class ChatGPTApi implements LLMApi {
           if (finished || controller.signal.aborted) {
             responseText += remainText;
             console.log("[Response Animation] finished");
+            if (responseText?.length === 0) {
+              options.onError?.(new Error("empty response from server"));
+            }
             return;
           }
 
@@ -377,6 +408,11 @@ export class ChatGPTApi implements LLMApi {
     return chatModels.map((m) => ({
       name: m.id,
       available: true,
+      provider: {
+        id: "openai",
+        providerName: "OpenAI",
+        providerType: "openai",
+      },
     }));
   }
 }