Bugfix: removing partial unicode when streaming

packages/ai-jsx/package.json

@@ -4,7 +4,7 @@
   "repository": "fixie-ai/ai-jsx",
   "bugs": "https://github.com/fixie-ai/ai-jsx/issues",
   "homepage": "https://ai-jsx.com",
-  "version": "0.28.1",
+  "version": "0.28.2",
   "volta": {
     "extends": "../../package.json"
   },

packages/ai-jsx/src/lib/openai.tsx

@@ -648,9 +648,19 @@ export async function* OpenAIChatModel(
               argsJson += toolCall.function.arguments;
             }
 
-            yield (
-              <FunctionCall id={id} partial name={name} args={JSON.parse(patchedUntruncateJson(argsJson || '{}'))} />
-            );
+            let partialArgs: Record<string, string | number | boolean | null> | undefined = undefined;
+            try {
+              partialArgs = JSON.parse(patchedUntruncateJson(argsJson || '{}'));
+            } catch (e: any) {
+              // If the JSON is incomplete and we get an error, we can ignore it.
+              const acceptedErrorPattern = /Unexpected .* JSON/;
+              if (!acceptedErrorPattern.test(e.message)) {
+                throw e;
+              }
+            }
+            if (partialArgs !== undefined) {
+              yield <FunctionCall id={id} partial name={name} args={partialArgs} />;
+            }
 
             delta = await advance();
           }

packages/ai-jsx/src/lib/util.ts

@@ -36,4 +36,9 @@ export function getEnvVar(name: string, shouldThrow: boolean = true) {
  * There's an ESM issue with untruncate-json, so we need to do this to support running on both client & server.
  */
 /** @hidden */
-export const patchedUntruncateJson = 'default' in untruncateJson ? untruncateJson.default : untruncateJson;
+const _patchedUntruncateJson = 'default' in untruncateJson ? untruncateJson.default : untruncateJson;
+
+export function patchedUntruncateJson(str: string) {
+  // Remove partial unicode characters: e.g. "\\u5728\\u5fA" -> "\\u5728"
+  return _patchedUntruncateJson(str.replace(/\\u[\dA-F]{0,3}$/gi, ''));
+}

packages/ai-jsx/test/lib/util.test.ts

@@ -1,4 +1,4 @@
-import { getEnvVar } from '../../dist/cjs/lib/util.cjs';
+import { getEnvVar, patchedUntruncateJson } from '../../dist/cjs/lib/util.cjs';
 
 process.env.EXISTS = 'exists';
 process.env.REACT_APP_ONLY = 'react-value';
@@ -42,3 +42,20 @@ test('env is not defined', () => {
 
   globalThis.process.env = originalEnv;
 });
+
+test('Basic untrucation of JSON', () => {
+  expect(patchedUntruncateJson('{"a":')).toEqual('{}');
+  expect(patchedUntruncateJson('{"a":"b')).toEqual('{"a":"b"}');
+  expect(patchedUntruncateJson('{"a":"b"')).toEqual('{"a":"b"}');
+});
+
+test('Partial unicode characters are removed', () => {
+  expect(patchedUntruncateJson('{"a":"\\u5728\\u5fA')).toEqual('{"a":"\\u5728"}');
+  expect(patchedUntruncateJson('"\\u5728\\u')).toEqual('"\\u5728"');
+  expect(patchedUntruncateJson('"\\u5728\\u0')).toEqual('"\\u5728"');
+  expect(patchedUntruncateJson('"\\u5728\\u5fA')).toEqual('"\\u5728"');
+});
+
+test('Unicode characters are allowed', () => {
+  expect(patchedUntruncateJson('{"a":"\\u5728什么是"}')).toEqual('{"a":"\\u5728什么是"}');
+});

packages/docs/docs/changelog.md

@@ -1,6 +1,10 @@
 # Changelog
 
-## 0.28.1
+## 0.28.2
+
+- Fix bug where partially streamed unicode characters (e.g. Chinese) would cause an error in OpenAI function calls.
+
+## [0.28.1](https://github.com/fixie-ai/ai-jsx/tree/4c67d845f48585dc3f26e90a9a656471f40c82ed)
 
 - Add `openai.finish_reason` span attribute for `OpenAIChatModel`