@ai-sdk/openai 2.0.128 → 2.0.129

package/dist/internal/index.js

@@ -419,6 +419,9 @@ var openaiChatResponseSchema = (0, import_provider_utils3.lazyValidator)(
           message: import_v42.z.object({
             role: import_v42.z.literal("assistant").nullish(),
             content: import_v42.z.string().nullish(),
+            audio: import_v42.z.object({
+              transcript: import_v42.z.string().nullish()
+            }).nullish(),
             tool_calls: import_v42.z.array(
               import_v42.z.object({
                 id: import_v42.z.string().nullish(),
@@ -954,7 +957,7 @@ var OpenAIChatLanguageModel = class {
     };
   }
   async doGenerate(options) {
-    var _a2, _b, _c, _d, _e, _f, _g, _h, _i, _j, _k, _l, _m, _n;
+    var _a2, _b, _c, _d, _e, _f, _g, _h, _i, _j, _k, _l, _m, _n, _o;
     const { args: body, warnings } = await this.getArgs(options);
     const {
       responseHeaders,
@@ -982,19 +985,19 @@ var OpenAIChatLanguageModel = class {
       });
     }
     const content = [];
-    const text = choice.message.content;
+    const text = choice.message.content != null && choice.message.content.length > 0 ? choice.message.content : (_a2 = choice.message.audio) == null ? void 0 : _a2.transcript;
     if (text != null && text.length > 0) {
       content.push({ type: "text", text });
     }
-    for (const toolCall of (_a2 = choice.message.tool_calls) != null ? _a2 : []) {
+    for (const toolCall of (_b = choice.message.tool_calls) != null ? _b : []) {
       content.push({
         type: "tool-call",
-        toolCallId: (_b = toolCall.id) != null ? _b : (0, import_provider_utils5.generateId)(),
+        toolCallId: (_c = toolCall.id) != null ? _c : (0, import_provider_utils5.generateId)(),
         toolName: toolCall.function.name,
         input: toolCall.function.arguments
       });
     }
-    for (const annotation of (_c = choice.message.annotations) != null ? _c : []) {
+    for (const annotation of (_d = choice.message.annotations) != null ? _d : []) {
       content.push({
         type: "source",
         sourceType: "url",
@@ -1003,8 +1006,8 @@ var OpenAIChatLanguageModel = class {
         title: annotation.url_citation.title
       });
     }
-    const completionTokenDetails = (_d = response.usage) == null ? void 0 : _d.completion_tokens_details;
-    const promptTokenDetails = (_e = response.usage) == null ? void 0 : _e.prompt_tokens_details;
+    const completionTokenDetails = (_e = response.usage) == null ? void 0 : _e.completion_tokens_details;
+    const promptTokenDetails = (_f = response.usage) == null ? void 0 : _f.prompt_tokens_details;
     const providerMetadata = { openai: {} };
     if ((completionTokenDetails == null ? void 0 : completionTokenDetails.accepted_prediction_tokens) != null) {
       providerMetadata.openai.acceptedPredictionTokens = completionTokenDetails == null ? void 0 : completionTokenDetails.accepted_prediction_tokens;
@@ -1017,18 +1020,18 @@ var OpenAIChatLanguageModel = class {
         cacheWriteTokens: promptTokenDetails.cache_write_tokens
       };
     }
-    if (((_f = choice.logprobs) == null ? void 0 : _f.content) != null) {
+    if (((_g = choice.logprobs) == null ? void 0 : _g.content) != null) {
       providerMetadata.openai.logprobs = choice.logprobs.content;
     }
     return {
       content,
       finishReason: mapOpenAIFinishReason(choice.finish_reason),
       usage: {
-        inputTokens: (_h = (_g = response.usage) == null ? void 0 : _g.prompt_tokens) != null ? _h : void 0,
-        outputTokens: (_j = (_i = response.usage) == null ? void 0 : _i.completion_tokens) != null ? _j : void 0,
-        totalTokens: (_l = (_k = response.usage) == null ? void 0 : _k.total_tokens) != null ? _l : void 0,
-        reasoningTokens: (_m = completionTokenDetails == null ? void 0 : completionTokenDetails.reasoning_tokens) != null ? _m : void 0,
-        cachedInputTokens: (_n = promptTokenDetails == null ? void 0 : promptTokenDetails.cached_tokens) != null ? _n : void 0
+        inputTokens: (_i = (_h = response.usage) == null ? void 0 : _h.prompt_tokens) != null ? _i : void 0,
+        outputTokens: (_k = (_j = response.usage) == null ? void 0 : _j.completion_tokens) != null ? _k : void 0,
+        totalTokens: (_m = (_l = response.usage) == null ? void 0 : _l.total_tokens) != null ? _m : void 0,
+        reasoningTokens: (_n = completionTokenDetails == null ? void 0 : completionTokenDetails.reasoning_tokens) != null ? _n : void 0,
+        cachedInputTokens: (_o = promptTokenDetails == null ? void 0 : promptTokenDetails.cached_tokens) != null ? _o : void 0
       },
       request: { body },
       response: {
@@ -4654,7 +4657,7 @@ var OpenAIResponsesLanguageModel = class {
                   providerExecuted: true
                 });
               } else if (value.item.type === "message") {
-                ongoingAnnotations.splice(0, ongoingAnnotations.length);
+                ongoingAnnotations.splice(0);
                 activeMessagePhase = (_a2 = value.item.phase) != null ? _a2 : void 0;
                 controller.enqueue({
                   type: "text-start",