fix(runner): normalize missing tool call IDs (#1958)

HAYDEN-OAI · web-flow · commit 6371623aadd2 · 2026-06-25T09:48:34.000-07:00
Normalize missing or empty tool call IDs before completion messages are stored and emitted, including streaming responses. Add coverage for parallel same-function calls and generated IDs.\n\nCo-authored-by: anish &lt;anishesg@users.noreply.github.com&gt;
diff --git a/src/lib/AbstractChatCompletionRunner.ts b/src/lib/AbstractChatCompletionRunner.ts
@@ -1,6 +1,7 @@
 import { OpenAIError } from '../error';
 import type OpenAI from '../index';
 import type { RequestOptions } from '../internal/request-options';
+import { uuid4 } from '../internal/utils/uuid';
 import { isAutoParsableTool, parseChatCompletion } from '../lib/parser';
 import type {
   ChatCompletion,
@@ -30,6 +31,19 @@ import {
 
 const DEFAULT_MAX_CHAT_COMPLETIONS = 10;
 
+function normalizeToolCallIds(chatCompletion: ChatCompletion): void {
+  for (const choice of chatCompletion.choices) {
+    for (const toolCall of choice.message.tool_calls ?? []) {
+      // Some OpenAI-compatible providers omit tool call IDs or return an empty string.
+      // Generate a unique ID before the completion is stored or emitted so the assistant
+      // tool call and its result message always reference the same value.
+      if (!toolCall.id) {
+        toolCall.id = `call_${uuid4()}`;
+      }
+    }
+  }
+}
+
 export interface ChatCompletionRunnerContext {
   messages: ChatCompletionMessageParam[];
   abort(): void;
@@ -58,6 +72,7 @@ export class AbstractChatCompletionRunner<
     this: AbstractChatCompletionRunner<AbstractChatCompletionRunnerEvents, ParsedT>,
     chatCompletion: ParsedChatCompletion<ParsedT>,
   ): ParsedChatCompletion<ParsedT> {
+    normalizeToolCallIds(chatCompletion);
     this._chatCompletions.push(chatCompletion);
     this._emit('chatCompletion', chatCompletion);
     const message = chatCompletion.choices[0]?.message;
diff --git a/src/lib/ChatCompletionStream.ts b/src/lib/ChatCompletionStream.ts
@@ -8,6 +8,7 @@ import {
 import OpenAI from '../index';
 import { RequestOptions } from '../internal/request-options';
 import { type ReadableStream } from '../internal/shim-types';
+import { uuid4 } from '../internal/utils/uuid';
 import {
   AutoParseableResponseFormat,
   hasAutoParseableInput,
@@ -657,9 +658,6 @@ function finalizeChatCompletion<ParsedT>(
               tool_calls: tool_calls.map((tool_call, i) => {
                 const { function: fn, type, id, ...toolRest } = tool_call;
                 const { arguments: args, name, ...fnRest } = fn || {};
-                if (id == null) {
-                  throw new OpenAIError(`missing choices[${index}].tool_calls[${i}].id\n${str(snapshot)}`);
-                }
                 if (type == null) {
                   throw new OpenAIError(`missing choices[${index}].tool_calls[${i}].type\n${str(snapshot)}`);
                 }
@@ -674,7 +672,12 @@ function finalizeChatCompletion<ParsedT>(
                   );
                 }
 
-                return { ...toolRest, id, type, function: { ...fnRest, name, arguments: args } };
+                return {
+                  ...toolRest,
+                  id: id || `call_${uuid4()}`,
+                  type,
+                  function: { ...fnRest, name, arguments: args },
+                };
               }),
             },
           };
diff --git a/tests/lib/ChatCompletionRunFunctions.test.ts b/tests/lib/ChatCompletionRunFunctions.test.ts
@@ -662,6 +662,93 @@ describe('resource completions', () => {
       expect(listener.functionCallResults).toEqual([`it's raining`]);
       await listener.sanityCheck({ ignoredMessages: new Set([injectedMessage]) });
     });
+    test('generates unique IDs for parallel tool calls with empty IDs', async () => {
+      const { fetch, handleRequest } = mockChatCompletionFetch();
+
+      const openai = new OpenAI({ apiKey: 'something1234', baseURL: 'http://127.0.0.1:4010', fetch });
+      const runner = openai.chat.completions.runTools({
+        messages: [{ role: 'user', content: 'echo both values' }],
+        model: 'gpt-3.5-turbo',
+        tools: [
+          {
+            type: 'function',
+            function: {
+              function: function echo(value: string) {
+                return value;
+              },
+              parameters: {},
+              description: 'echoes a value',
+            },
+          },
+        ],
+      });
+
+      await handleRequest(async () => ({
+        id: '1',
+        choices: [
+          {
+            index: 0,
+            finish_reason: 'tool_calls',
+            logprobs: null,
+            message: {
+              role: 'assistant',
+              content: null,
+              refusal: null,
+              tool_calls: [
+                {
+                  type: 'function',
+                  id: '',
+                  function: { arguments: 'first', name: 'echo' },
+                },
+                {
+                  type: 'function',
+                  id: '',
+                  function: { arguments: 'second', name: 'echo' },
+                },
+              ],
+            },
+          },
+        ],
+        created: Math.floor(Date.now() / 1000),
+        model: 'gpt-3.5-turbo',
+        object: 'chat.completion',
+      }));
+
+      await handleRequest(async (request) => {
+        const assistantMessage = request.messages[1];
+        if (assistantMessage?.role !== 'assistant' || !assistantMessage.tool_calls) {
+          throw new Error('expected an assistant message with tool calls');
+        }
+
+        const generatedIDs = assistantMessage.tool_calls.map((toolCall) => toolCall.id);
+        expect(generatedIDs).toHaveLength(2);
+        expect(generatedIDs.every((id) => id.startsWith('call_'))).toBe(true);
+        expect(new Set(generatedIDs).size).toBe(2);
+
+        const toolMessages = request.messages.slice(2);
+        expect(
+          toolMessages.map((message) => (message.role === 'tool' ? message.tool_call_id : undefined)),
+        ).toEqual(generatedIDs);
+
+        return {
+          id: '2',
+          choices: [
+            {
+              index: 0,
+              finish_reason: 'stop',
+              logprobs: null,
+              message: { role: 'assistant', content: 'done', refusal: null },
+            },
+          ],
+          created: Math.floor(Date.now() / 1000),
+          model: 'gpt-3.5-turbo',
+          object: 'chat.completion',
+        };
+      });
+
+      await runner.done();
+      await expect(runner.finalFunctionToolCallResult()).resolves.toBe('second');
+    });
     test('runs tool calls concurrently and preserves their result order', async () => {
       const { fetch, handleRequest } = mockChatCompletionFetch();
 
@@ -1827,6 +1914,69 @@ describe('resource completions', () => {
       expect(listener.eventFunctionCallResults).toEqual([`it's raining`]);
       await listener.sanityCheck({ ignoredMessages: new Set([injectedMessage]) });
     });
+    test('generates an ID for streamed tool calls with an empty ID', async () => {
+      const { fetch, handleRequest } = mockStreamingChatCompletionFetch();
+
+      const openai = new OpenAI({ apiKey: 'something1234', baseURL: 'http://127.0.0.1:4010', fetch });
+      const runner = openai.chat.completions.runTools({
+        stream: true,
+        messages: [{ role: 'user', content: 'tell me what the weather is like' }],
+        model: 'gpt-3.5-turbo',
+        tools: [
+          {
+            type: 'function',
+            function: {
+              function: function getWeather() {
+                return `it's raining`;
+              },
+              parameters: {},
+              description: 'gets the weather',
+            },
+          },
+        ],
+      });
+
+      await Promise.all([
+        handleRequest(async function* (): AsyncIterable<OpenAI.Chat.ChatCompletionChunk> {
+          for (const choice of functionCallDeltas('', { id: '', name: 'getWeather' })) {
+            yield {
+              id: '1',
+              choices: [choice],
+              created: Math.floor(Date.now() / 1000),
+              model: 'gpt-3.5-turbo',
+              object: 'chat.completion.chunk',
+            };
+          }
+        }),
+        handleRequest(async function* (request): AsyncIterable<OpenAI.Chat.ChatCompletionChunk> {
+          const assistantMessage = request.messages[1];
+          const toolMessage = request.messages[2];
+          if (assistantMessage?.role !== 'assistant' || !assistantMessage.tool_calls?.[0]) {
+            throw new Error('expected an assistant message with a tool call');
+          }
+          if (toolMessage?.role !== 'tool') {
+            throw new Error('expected a tool result message');
+          }
+
+          const generatedID = assistantMessage.tool_calls[0].id;
+          expect(generatedID).toMatch(/^call_/);
+          expect(toolMessage.tool_call_id).toBe(generatedID);
+
+          for (const choice of contentChoiceDeltas(`it's raining`)) {
+            yield {
+              id: '2',
+              choices: [choice],
+              created: Math.floor(Date.now() / 1000),
+              model: 'gpt-3.5-turbo',
+              object: 'chat.completion.chunk',
+            };
+          }
+        }),
+        runner.done(),
+      ]);
+
+      await expect(runner.finalFunctionToolCallResult()).resolves.toBe(`it's raining`);
+    });
     test('flow with abort', async () => {
       const { fetch, handleRequest } = mockStreamingChatCompletionFetch();
 
