aiturk-hermes-ide/apps/desktop/codex-link-json.ts

// Strict P1 JSON domain/parser, shared algorithm with router codexLinkValidation.
// Browser-safe; never apply to the opaque native/provider JSON body.
type Data = Record<string, any>

export class CodexLinkValidationError extends Error {
  readonly code: string
  constructor(code: string) {
    super(code)
    this.code = code
    this.name = 'CodexLinkValidationError'
  }
}

function requireValue(condition: unknown, code: string): asserts condition {
  if (!condition) {
    throw new CodexLinkValidationError(code)
  }
}

function object(value: unknown): Data {
  requireValue(value !== null && typeof value === 'object' && !Array.isArray(value), 'schema_rejected')
  const prototype = Object.getPrototypeOf(value)
  requireValue(prototype === Object.prototype || prototype === null, 'schema_rejected')

  return value as Data
}

function integer(value: unknown, positive = false): asserts value is number {
  requireValue(Number.isSafeInteger(value) && (value as number) >= (positive ? 1 : 0), 'schema_rejected')
}

function id(value: unknown): asserts value is string {
  requireValue(typeof value === 'string' && /^[A-Za-z0-9][A-Za-z0-9._:/-]{0,255}$/.test(value), 'schema_rejected')
}

function exactKeys(value: Data, keys: string[]) {
  requireValue(
    Object.keys(value).length === keys.length && keys.every(key => Object.hasOwn(value, key)),
    'schema_rejected'
  )
}

function unicode(value: string) {
  for (let i = 0; i < value.length; i++) {
    const code = value.charCodeAt(i)

    if (code >= 0xd800 && code <= 0xdbff) {
      const low = value.charCodeAt(++i)
      requireValue(low >= 0xdc00 && low <= 0xdfff, 'invalid_unicode')
    } else {
      requireValue(code < 0xdc00 || code > 0xdfff, 'invalid_unicode')
    }
  }
}

/** P1 canonical domain: ASCII object keys, safe integers, exact Unicode text. */
export function canonicalJson(value: unknown): string {
  const ancestors = new Set<object>()

  function encode(item: unknown, depth: number): string {
    requireValue(depth <= 256, 'invalid_json')

    if (item === null || typeof item === 'boolean') {
      return JSON.stringify(item)
    }

    if (typeof item === 'string') {
      unicode(item)

      return JSON.stringify(item)
    }

    if (typeof item === 'number') {
      requireValue(Number.isInteger(item), 'floats_or_non_json_type')
      requireValue(Number.isSafeInteger(item), 'unsafe_integer')

      return JSON.stringify(item)
    }

    requireValue(typeof item === 'object' && item !== null, 'floats_or_non_json_type')
    requireValue(!ancestors.has(item), 'invalid_json')
    ancestors.add(item)
    let result: string

    if (Array.isArray(item)) {
      result = '[' + item.map(child => encode(child, depth + 1)).join(',') + ']'
    } else {
      const data = object(item)
      const keys = Object.keys(data).sort()
      requireValue(
        keys.every(key => [...key].every(char => char.charCodeAt(0) <= 127)),
        'non_ascii_key'
      )
      result = '{' + keys.map(key => JSON.stringify(key) + ':' + encode(data[key], depth + 1)).join(',') + '}'
    }

    ancestors.delete(item)

    return result
  }

  return encode(value, 0)
}

/** A bounded parser that detects duplicates before JSON.parse can erase them. */
export function parseStrictJson(raw: string | Uint8Array, maxBytes: number): unknown {
  integer(maxBytes, true)
  const rawBytes = typeof raw === 'string' ? new TextEncoder().encode(raw).byteLength : raw.byteLength
  requireValue(rawBytes <= maxBytes, 'transport_bytes_exceeded')
  let text: string

  try {
    text = typeof raw === 'string' ? raw : new TextDecoder('utf-8', { fatal: true }).decode(raw)
  } catch {
    throw new CodexLinkValidationError('invalid_json')
  }

  unicode(text)
  let position = 0

  function whitespace() {
    while (/[ \t\r\n]/.test(text[position] ?? '') && position < text.length) {
      position++
    }
  }

  function quoted(): string {
    const start = position++

    while (position < text.length) {
      const next = text[position++]

      if (next === '\\') {
        position++

        continue
      }

      if (next === '"') {
        try {
          const value = JSON.parse(text.slice(start, position))
          unicode(value)

          return value
        } catch (error) {
          if (error instanceof CodexLinkValidationError) {
            throw error
          }

          throw new CodexLinkValidationError('invalid_json')
        }
      }
    }

    throw new CodexLinkValidationError('invalid_json')
  }

  function read(depth: number): unknown {
    requireValue(depth <= 256, 'invalid_json')
    whitespace()
    const token = text[position]

    if (token === '"') {
      return quoted()
    }

    if (token === '{') {
      position++
      whitespace()
      const value: Data = Object.create(null)

      if (text[position] === '}') {
        position++

        return value
      }

      while (position < text.length) {
        requireValue(text[position] === '"', 'invalid_json')
        const key = quoted()
        requireValue(
          [...key].every(char => char.charCodeAt(0) <= 127),
          'non_ascii_key'
        )
        requireValue(!Object.hasOwn(value, key), 'duplicate_json_key')
        whitespace()
        requireValue(text[position++] === ':', 'invalid_json')
        value[key] = read(depth + 1)
        whitespace()
        const separator = text[position++]

        if (separator === '}') {
          return value
        }

        requireValue(separator === ',', 'invalid_json')
        whitespace()
      }
    } else if (token === '[') {
      position++
      whitespace()
      const value: unknown[] = []

      if (text[position] === ']') {
        position++

        return value
      }

      while (position < text.length) {
        value.push(read(depth + 1))
        whitespace()
        const separator = text[position++]

        if (separator === ']') {
          return value
        }

        requireValue(separator === ',', 'invalid_json')
      }
    } else {
      for (const [word, value] of [
        ['true', true],
        ['false', false],
        ['null', null]
      ] as const) {
        if (text.startsWith(word, position)) {
          position += word.length

          return value
        }
      }

      if (/^(?:NaN|[-+]?Infinity)/.test(text.slice(position))) {
        throw new CodexLinkValidationError('floats_or_non_json_type')
      }

      const match = /^-?(?:0|[1-9]\d*)/.exec(text.slice(position))

      if (match) {
        position += match[0].length
        requireValue(!/[.eE]/.test(text[position] ?? ''), 'floats_or_non_json_type')
        const number = Number(match[0])
        requireValue(Number.isSafeInteger(number), 'unsafe_integer')

        return number
      }
    }

    throw new CodexLinkValidationError('invalid_json')
  }

  const value = read(0)
  whitespace()
  requireValue(position === text.length, 'invalid_json')
  canonicalJson(value)

  return value
}