fix(gauge): count Unicode prompt characters correctly

Author: waleedlatif1

apps/sim/tools/gauge/create_prompts.ts

@@ -17,6 +17,16 @@ import {
 import type { ToolConfig } from '@/tools/types'
 import { safeUrlPathSegment } from '@/tools/url-path'
 
+const MAX_PROMPT_TEXT_LENGTH = 500
+
+/** Matches Gauge's code-point limit without expanding oversized strings. */
+function isPromptTextWithinLimit(text: string): boolean {
+  return (
+    text.length <= MAX_PROMPT_TEXT_LENGTH ||
+    (text.length <= MAX_PROMPT_TEXT_LENGTH * 2 && Array.from(text).length <= MAX_PROMPT_TEXT_LENGTH)
+  )
+}
+
 /** Validates the batch before selecting only the text Gauge sends to AI engines. */
 function parseGaugePrompts(value: unknown): GaugePromptIn[] {
   const parsed = parseGaugeJson(value, 'prompts')
@@ -28,7 +38,7 @@ function parseGaugePrompts(value: unknown): GaugePromptIn[] {
       !isRecordLike(prompt) ||
       typeof prompt.text !== 'string' ||
       !prompt.text.trim() ||
-      prompt.text.length > 500 ||
+      !isPromptTextWithinLimit(prompt.text) ||
       typeof prompt.topicId !== 'string' ||
       !prompt.topicId.trim()
     ) {

apps/sim/tools/gauge/utils.test.ts

@@ -61,6 +61,25 @@ describe('Gauge tracked prompt model input', () => {
     { text: 'Compare analytics tools', topicId: 'top_other' },
   ]
 
+  it.each([
+    ['surrogate pairs', '😀'.repeat(500)],
+    ['mixed Unicode and ASCII', `${'😀'.repeat(250)}${'a'.repeat(250)}`],
+  ])('accepts 500 code points containing %s without changing the text', (_label, text) => {
+    expect(body({ ...requestParams, prompts: [{ text, topicId: 'top_example' }] })).toMatchObject({
+      prompts: [{ text, topicId: 'top_example' }],
+    })
+  })
+
+  it.each([
+    ['mixed Unicode and ASCII', `${'😀'.repeat(251)}${'a'.repeat(250)}`],
+    ['ASCII', 'a'.repeat(501)],
+    ['surrogate pairs', '😀'.repeat(501)],
+  ])('rejects 501 code points containing %s', (_label, text) => {
+    expect(() => body({ ...requestParams, prompts: [{ text, topicId: 'top_example' }] })).toThrow(
+      'Each prompt must have text (1 to 500 characters) and a topicId'
+    )
+  })
+
   it('selects prompt text and preserves control fields when rebuilding projected JSON', () => {
     const original = JSON.stringify(prompts)
     expect(select({ ...requestParams, prompts: original })).toEqual({