@ai-sdk/openai 4.0.11 → 4.0.14

package/dist/internal/index.js

@@ -2320,16 +2320,16 @@ import {
   convertBase64ToUint8Array as convertBase64ToUint8Array2,
   convertToBase64 as convertToBase642,
   createJsonResponseHandler as createJsonResponseHandler5,
-  getWebSocketConstructor,
+  connectToWebSocket,
   mediaTypeToExtension,
   parseProviderOptions as parseProviderOptions5,
   postFormDataToApi as postFormDataToApi2,
-  readWebSocketMessageText,
   safeParseJSON,
   serializeModelOptions as serializeModelOptions5,
   toWebSocketUrl,
   WORKFLOW_DESERIALIZE as WORKFLOW_DESERIALIZE5,
-  WORKFLOW_SERIALIZE as WORKFLOW_SERIALIZE5
+  WORKFLOW_SERIALIZE as WORKFLOW_SERIALIZE5,
+  waitForWebSocketBufferDrain
 } from "@ai-sdk/provider-utils";
 
 // src/transcription/openai-transcription-api.ts
@@ -2678,21 +2678,18 @@ function createOpenAIRealtimeTranscriptionStream({
   };
   return new ReadableStream({
     start: (controller) => {
-      const WebSocketConstructor = getWebSocketConstructor(webSocket);
-      const ws = new WebSocketConstructor(
-        url,
-        getOpenAIRealtimeProtocols(headers),
-        { headers }
-      );
+      const realtimeConnection = getOpenAIRealtimeConnection(headers);
       let audioReader;
+      let connection;
       cleanup = (closeCode) => {
-        abortSignal == null ? void 0 : abortSignal.removeEventListener("abort", abort);
-        void (audioReader == null ? void 0 : audioReader.cancel().catch(() => {
-        }));
-        try {
-          ws.close(closeCode);
-        } catch (e) {
+        if (audioReader != null) {
+          void audioReader.cancel().catch(() => {
+          });
+        } else {
+          void audio.cancel().catch(() => {
+          });
         }
+        connection == null ? void 0 : connection.close(closeCode);
       };
       const finishWithError = (error) => {
         if (finished) return;
@@ -2715,42 +2712,42 @@ function createOpenAIRealtimeTranscriptionStream({
         controller.close();
         cleanup(1e3);
       };
-      const abort = () => {
-        var _a;
-        finishWithError((_a = abortSignal == null ? void 0 : abortSignal.reason) != null ? _a : new Error("Aborted"));
-      };
-      if (abortSignal == null ? void 0 : abortSignal.aborted) {
-        abort();
-        return;
-      }
-      abortSignal == null ? void 0 : abortSignal.addEventListener("abort", abort, { once: true });
-      const sendAudio = async () => {
+      const sendAudio = async (socket) => {
         audioReader = audio.getReader();
         try {
           while (true) {
             const { done, value } = await audioReader.read();
             if (done || finished) break;
-            ws.send(
+            socket.send(
               JSON.stringify({
                 type: "input_audio_buffer.append",
                 audio: convertToBase642(value)
               })
             );
+            await waitForWebSocketBufferDrain(socket);
           }
         } finally {
           audioReader.releaseLock();
+          audioReader = void 0;
         }
         if (!finished) {
-          ws.send(JSON.stringify({ type: "input_audio_buffer.commit" }));
+          socket.send(JSON.stringify({ type: "input_audio_buffer.commit" }));
         }
       };
-      ws.onopen = () => {
-        controller.enqueue({ type: "stream-start", warnings });
-        ws.send(JSON.stringify(sessionUpdate));
-        void sendAudio().catch(finishWithError);
-      };
-      ws.onmessage = (event) => {
-        void readWebSocketMessageText(event.data).then(async (text) => {
+      connection = connectToWebSocket({
+        url,
+        protocols: realtimeConnection.protocols,
+        headers: realtimeConnection.headers,
+        webSocket,
+        abortSignal,
+        onAbort: finishWithError,
+        onProcessingError: finishWithError,
+        onOpen: (socket) => {
+          controller.enqueue({ type: "stream-start", warnings });
+          socket.send(JSON.stringify(sessionUpdate));
+          void sendAudio(socket).catch(finishWithError);
+        },
+        onMessageText: async (text) => {
           var _a, _b, _c, _d;
           const parsed = await safeParseJSON({ text });
           if (!parsed.success) return;
@@ -2778,17 +2775,17 @@ function createOpenAIRealtimeTranscriptionStream({
               break;
             }
           }
-        }).catch(finishWithError);
-      };
-      ws.onerror = () => {
-        finishWithError(new Error("OpenAI realtime transcription error"));
-      };
-      ws.onclose = () => {
-        if (finished) return;
-        finished = true;
-        cleanup();
-        controller.close();
-      };
+        },
+        onSocketError: () => {
+          finishWithError(new Error("OpenAI realtime transcription error"));
+        },
+        onClose: () => {
+          if (finished) return;
+          finished = true;
+          cleanup();
+          controller.close();
+        }
+      });
     },
     cancel: () => {
       if (finished) return;
@@ -2825,11 +2822,26 @@ function buildOpenAIRealtimeTranscriptionSession({
     }
   };
 }
-function getOpenAIRealtimeProtocols(headers) {
+function getOpenAIRealtimeConnection(headers) {
   var _a;
-  const authorization = (_a = headers.Authorization) != null ? _a : headers.authorization;
-  const token = (authorization == null ? void 0 : authorization.startsWith("Bearer ")) ? authorization.slice("Bearer ".length) : void 0;
-  return token == null ? ["realtime"] : ["realtime", `openai-insecure-api-key.${token}`];
+  let authorization;
+  for (const [key, value] of Object.entries(headers)) {
+    if (key.toLowerCase() === "authorization" && value != null) {
+      authorization = value;
+    }
+  }
+  const token = (_a = authorization == null ? void 0 : authorization.match(/^bearer\s+(.+)$/i)) == null ? void 0 : _a[1];
+  if (token == null) {
+    return { protocols: ["realtime"], headers };
+  }
+  return {
+    protocols: ["realtime", `openai-insecure-api-key.${token}`],
+    headers: Object.fromEntries(
+      Object.entries(headers).filter(
+        ([key]) => key.toLowerCase() !== "authorization"
+      )
+    )
+  };
 }
 
 // src/speech/openai-speech-model.ts