Environment

• Operating System: Darwin
• Node Version: v20.19.0
• Nuxt Version: 3.16.2
• Nuxt Content Version: 3.6.0
• Nuxt Sitemap Version: 7.4.1
• Better-SQLite3 Version: 11.10.0
• Package Manager: yarn@4.2.2
• Builder: -
• User Config: -
• Runtime Modules: -
• Build Modules: -

Version

v3

Reproduction

1. Create a sample.json file in the content directory
2. Ensure that a UTF-8 multibyte character (e.g., Japanese character "心" = 0xE5 0xBF 0x83) starts at byte position 349,999

json structure

{
  "data": [
    {
      "id": "1",
      "content": "... [text padding to reach exactly 349,999 bytes] ... 中心 ..."
    }
  ]
}

nuxt.config.ts

import { defineContentConfig, defineCollection, z } from '@nuxt/content'

export default defineContentConfig({
  collections: {
    content: defineCollection({
      type: 'data',
      source: '**/*.json',
      schema: z.object({}),
    }),
  },
})

Description

When a UTF-8 multibyte character boundary falls exactly at byte position 350,000 in a JSON file, Nuxt Content's SQLite implementation fails with:

SQLite3Error: SQLITE_ERROR: unrecognized token: "[token_hash]"

And on the client side:
Unterminated string in JSON at position 349999 (line 1 column 350000)

Additional context

Investigation Results

Proof:

• Character "心" in UTF-8: 0xE5 0xBF 0x83 (3 bytes)
• At position 349,999: 0xE5 (first byte of "心")
• At position 350,000: 0xBF (second byte of "心") <- Buffer cuts here
• At position 350,001: 0x83 (third byte of "心")

Verification

Tested with multiple files:

• File A: 685,647 bytes total, multibyte character at position 350,000 → ❌ Error
• File B: 3,150,308 bytes total, ASCII character at position 350,000 → ✅ Success
• File C: 3,751,769 bytes total, ASCII character at position 350,000 → ✅ Success

The error only occurs when position 350,000 falls within a multibyte UTF-8 character.

Logs