open-brainy/src/utils/paramValidation.ts

/**
 * Zero-Config Parameter Validation
 * 
 * Self-configuring validation that adapts to system capabilities
 * Only enforces universal truths, learns everything else
 */

import { FindParams, AddParams, UpdateParams, RelateParams, UpdateRelationParams } from '../types/brainy.types.js'
import { NounType, VerbType } from '../types/graphTypes.js'
import { prodLog } from './logger.js'
import { findCallerLocation } from './callerLocation.js'

// 8.0 is Node/Bun/Deno-only (no browser path), so `node:os` / `node:fs` are
// always present — static ESM imports instead of a top-level `await import()`.
// The TLA form poisoned the module graph with a top-level await, which `bun
// build --compile` refuses; the static form also drops the browser/edge
// fallback branches that no supported runtime can reach.
import * as os from 'node:os'
import * as fs from 'node:fs'
import { parseFieldAddress, UnsupportedFindOptionError } from '../db/fieldAddressing.js'

const getSystemMemory = (): number => {
  if (os) {
    return os.totalmem()
  }
  return 4 * 1024 * 1024 * 1024
}

/**
 * Best-effort "memory we can actually use right now", in bytes.
 *
 * Prefers `/proc/meminfo` **MemAvailable**, which the kernel computes as the
 * memory obtainable for a new workload *including reclaimable page cache*. This
 * matters on hosts that memory-map large index files: the cache holding those
 * mmap'd pages is instantly reclaimable, yet `os.freemem()` reports only
 * **MemFree** (the unused sliver), which collapses to near-zero on a warm box.
 * Driving the auto query-cap off MemFree made the cap crater to its floor on
 * perfectly healthy machines (BRAINY-QUERYCAP-MISREAD); MemAvailable fixes that.
 *
 * Falls back to `os.freemem()` where `/proc/meminfo` is absent (non-Linux), then
 * to a conservative 2 GB when no OS module is available at all (browser/edge).
 *
 * @returns Usable memory in bytes.
 */
const getAvailableMemory = (): number => {
  if (fs) {
    try {
      const meminfo = fs.readFileSync('/proc/meminfo', 'utf8') as string
      const match = meminfo.match(/^MemAvailable:\s+(\d+)\s*kB/m)
      if (match) {
        const bytes = parseInt(match[1], 10) * 1024
        if (bytes > 0) {
          return bytes
        }
      }
    } catch {
      // /proc/meminfo unreadable (non-Linux, sandboxed) — fall through.
    }
  }
  if (os) {
    return os.freemem()
  }
  return 2 * 1024 * 1024 * 1024
}

/**
 * Detect container memory limit (Docker/Kubernetes/Cloud Run)
 *
 * Production-grade detection for containerized environments.
 * Supports:
 * - cgroup v1 (legacy Docker/K8s)
 * - cgroup v2 (modern systems)
 * - Environment variables (Cloud Run, GCP, AWS, Azure)
 *
 * @returns Container memory limit in bytes, or null if not containerized
 */
const getContainerMemoryLimit = (): number | null => {
  // Not in Node.js environment
  if (!fs) {
    return null
  }

  try {
    // 1. Check environment variables first (fastest, most reliable for Cloud Run)
    // Google Cloud Run
    if (process.env.CLOUD_RUN_MEMORY) {
      // Format: "512Mi", "1Gi", "2Gi", "4Gi"
      const match = process.env.CLOUD_RUN_MEMORY.match(/^(\d+)(Mi|Gi)$/)
      if (match) {
        const value = parseInt(match[1])
        const unit = match[2]
        return unit === 'Gi' ? value * 1024 * 1024 * 1024 : value * 1024 * 1024
      }
    }

    // Generic MEMORY_LIMIT env var (bytes)
    if (process.env.MEMORY_LIMIT) {
      const limit = parseInt(process.env.MEMORY_LIMIT)
      if (!isNaN(limit) && limit > 0) {
        return limit
      }
    }

    // 2. Check cgroup v2 (modern Docker/K8s)
    try {
      const cgroupV2Path = '/sys/fs/cgroup/memory.max'
      const cgroupV2Content = fs.readFileSync(cgroupV2Path, 'utf8').trim()

      // "max" means no limit, otherwise it's bytes
      if (cgroupV2Content !== 'max') {
        const limit = parseInt(cgroupV2Content)
        if (!isNaN(limit) && limit > 0) {
          return limit
        }
      }
    } catch (e) {
      // cgroup v2 not available, try v1
    }

    // 3. Check cgroup v1 (legacy Docker/K8s)
    try {
      const cgroupV1Path = '/sys/fs/cgroup/memory/memory.limit_in_bytes'
      const cgroupV1Content = fs.readFileSync(cgroupV1Path, 'utf8').trim()

      const limit = parseInt(cgroupV1Content)

      // Very large values (> 1 PB) indicate no limit
      const ONE_PETABYTE = 1024 * 1024 * 1024 * 1024 * 1024
      if (!isNaN(limit) && limit > 0 && limit < ONE_PETABYTE) {
        return limit
      }
    } catch (e) {
      // cgroup v1 not available
    }

    // Not containerized or no limit set
    return null

  } catch (e) {
    // Error reading cgroup files
    return null
  }
}

/**
 * Memory budget per query result, in KB. Used by the auto-configured
 * `maxLimit` formula across all three memory-derived priorities (reserved /
 * container / free).
 *
 * Calibration history:
 * - Pre-7.30.2: `100` (assumed 100 KB per result). Way too conservative —
 *   actual entity footprint in Brainy is 7-10 KB (384-dim float32 vector ≈
 *   1.5 KB + standard fields + metadata). On a 900 MB free-memory box this
 *   capped `limit` at 9_000, breaking common safety-cap call patterns like
 *   `find({ type, where, limit: 10_000 })` that typically return 10-500
 *   entities. Reported by a production consumer whose 10K safety-cap
 *   queries started failing after the cap landed.
 * - 7.30.2+: `25` (assumes 25 KB per result). Generous over typical (7-10 KB),
 *   comfortably under the worst case (~20 KB with large metadata blobs).
 *   Same 900 MB box now gives ~36_000 — typical 10_000 limits pass silently,
 *   the warning tier surfaces around 36-72 K (encouraging pagination), and
 *   the throw tier still fires before OOM territory (72 K+).
 */
const MAX_LIMIT_KB_PER_RESULT = 25

/**
 * Floor for *auto-detected* query caps (the container- and free-memory tiers).
 *
 * Memory probing is a heuristic; a transiently low reading must never silently
 * throttle legitimate queries to a near-useless ceiling. 10 000 is a generous
 * working limit that still sits far below any OOM danger (10 000 × 25 KB ≈
 * 250 MB worst case). It is a floor, not an override: consumer-supplied
 * `maxQueryLimit` / `reservedQueryMemory` bypass it entirely (Priorities 1–2),
 * because an explicit caller knows their box better than the probe does.
 */
const MIN_AUTO_QUERY_LIMIT = 10000

/**
 * One-time-per-call-site warning dedup. Keyed on the caller location returned
 * by `findCallerLocation()` plus the exceeding limit value so the warning fires
 * once per offending source line (not once per query). Survives the lifetime of
 * the process — that's intentional: the warning is a teaching signal, not a
 * recurring nag. Used by the two-tier limit enforcement in `validateFindParams`.
 */
const seenLimitWarnings = new Set<string>()

/**
 * Reset the limit-warning dedup. For tests that need a clean slate; production
 * code should never call this.
 */
export function resetLimitWarningCache(): void {
  seenLimitWarnings.clear()
}

/**
 * Configuration options for ValidationConfig
 */
export interface ValidationConfigOptions {
  /**
   * Explicit maximum query limit override
   * Bypasses all auto-detection
   */
  maxQueryLimit?: number

  /**
   * Memory reserved for query operations (in bytes)
   * Bypasses auto-detection but still applies safety limits
   */
  reservedQueryMemory?: number
}

/**
 * Auto-configured limits based on system resources.
 * Derived from memory (explicit overrides > reserved memory > container limit >
 * free memory). Query timing is recorded for diagnostics only — it never
 * changes the cap (see `recordQuery`).
 */
export class ValidationConfig {
  private static instance: ValidationConfig | null = null

  // Dynamic limits based on system
  public maxLimit: number
  public maxQueryLength: number
  public maxVectorDimensions: number

  // Tracking for diagnostics
  public limitBasis: 'override' | 'reservedMemory' | 'containerMemory' | 'freeMemory'
  public detectedContainerLimit: number | null

  // Performance observations
  private avgQueryTime: number = 0
  private queryCount: number = 0

  private constructor(options?: ValidationConfigOptions) {
    // Vector dimensions (standard for all-MiniLM-L6-v2)
    this.maxVectorDimensions = 384

    // Detect container memory limit
    this.detectedContainerLimit = getContainerMemoryLimit()

    // Priority 1: Explicit override (highest priority)
    if (options?.maxQueryLimit !== undefined) {
      this.maxLimit = Math.min(options.maxQueryLimit, 100000)  // Still cap at 100k for safety
      this.limitBasis = 'override'

      // Scale query length with limit
      this.maxQueryLength = Math.min(50000, this.maxLimit * 5)
      return
    }

    // Priority 2: Reserved memory specified
    if (options?.reservedQueryMemory !== undefined) {
      this.maxLimit = Math.min(
        100000,
        Math.floor(options.reservedQueryMemory / (1024 * 1024 * MAX_LIMIT_KB_PER_RESULT)) * 1000
      )
      this.limitBasis = 'reservedMemory'

      this.maxQueryLength = Math.min(
        50000,
        Math.floor(options.reservedQueryMemory / (1024 * 1024 * 10)) * 1000
      )
      return
    }

    // Priority 3: Container detected (smart containerized behavior)
    if (this.detectedContainerLimit) {
      // In containers, assume 75% used by graph data (EXPECTED)
      // Reserve 25% for query operations
      const queryMemory = this.detectedContainerLimit * 0.25

      this.maxLimit = Math.max(
        MIN_AUTO_QUERY_LIMIT,
        Math.min(
          100000,
          Math.floor(queryMemory / (1024 * 1024 * MAX_LIMIT_KB_PER_RESULT)) * 1000
        )
      )
      this.limitBasis = 'containerMemory'

      this.maxQueryLength = Math.min(
        50000,
        Math.floor(queryMemory / (1024 * 1024 * 10)) * 1000
      )
      return
    }

    // Priority 4: Free memory (fallback, current behavior)
    const availableMemory = getAvailableMemory()

    this.maxLimit = Math.max(
      MIN_AUTO_QUERY_LIMIT,
      Math.min(
        100000,
        Math.floor(availableMemory / (1024 * 1024 * MAX_LIMIT_KB_PER_RESULT)) * 1000
      )
    )
    this.limitBasis = 'freeMemory'

    this.maxQueryLength = Math.min(
      50000,
      Math.floor(availableMemory / (1024 * 1024 * 10)) * 1000
    )
  }

  static getInstance(options?: ValidationConfigOptions): ValidationConfig {
    if (!ValidationConfig.instance) {
      ValidationConfig.instance = new ValidationConfig(options)
    }
    return ValidationConfig.instance
  }

  /**
   * Reset singleton (for testing or reconfiguration)
   */
  static reset(): void {
    ValidationConfig.instance = null
  }

  /**
   * Reconfigure with new options
   */
  static reconfigure(options: ValidationConfigOptions): ValidationConfig {
    ValidationConfig.instance = new ValidationConfig(options)
    return ValidationConfig.instance
  }

  /**
   * Record query timing for diagnostics. Telemetry ONLY — never mutates the cap.
   *
   * `maxLimit` is a MEMORY-protection bound; query duration says nothing about
   * memory-per-result, so duration must never drive it. An earlier version
   * "learned" here: while the lifetime-average query time exceeded 1s it shrank
   * `maxLimit` by 20% per recorded query down to a floor of 1000 — below the
   * documented `MIN_AUTO_QUERY_LIMIT` (10 000) and with no recovery once slow
   * samples poisoned the cumulative average. On a production host a burst of
   * slow aggregate queries silently strangled every consumer's `find()` to
   * 1000 while the error message blamed "available free memory" — exactly the
   * silent throttling this module's own contract forbids. The cap now comes
   * from its construction-time basis (or explicit overrides) alone.
   */
  recordQuery(duration: number, resultCount: number) {
    void resultCount
    this.queryCount++
    this.avgQueryTime = (this.avgQueryTime * (this.queryCount - 1) + duration) / this.queryCount
  }
}

/**
 * Two-tier limit enforcement for `find({ limit })`. Compares the requested
 * limit against the auto-configured (or consumer-overridden) `maxLimit` and
 * picks one of three outcomes:
 *
 * 1. **Pass** — `limit <= maxLimit`. The configured cap was set with this
 *    consumer's use case in mind; no signal, no friction.
 *
 * 2. **Warn** — `maxLimit < limit <= 2 * maxLimit`. The consumer is asking
 *    for more than the cap, but not enough to be in OOM territory. Log a
 *    one-time warning per call site (dedup keyed on stack location + limit
 *    value) explaining the cap, the three escape valves, and the docs link.
 *    The query proceeds. Pre-7.30.2 code that relied on the cap silently
 *    allowing typical safety-cap limits (`10_000`) keeps working; the
 *    warning teaches the migration.
 *
 * 3. **Throw** — `limit > 2 * maxLimit`. Real OOM danger territory. Throw
 *    with the same message format the warning uses so the consumer gets
 *    consistent guidance whether they hit the soft or hard threshold.
 *
 * The 2× soft margin is chosen to absorb existing safety-cap patterns
 * (`limit: 10_000` against a `maxLimit: 9_000` box, common case at 7.30 GA)
 * without disabling OOM protection. Real OOM territory on a JS in-memory
 * brain is hundreds of thousands of results, not 10× the safety cap.
 *
 * Both warning and throw paths use the same formatter so the rendered message
 * is identical — consumers see the same recipe regardless of which threshold
 * they crossed.
 *
 * @param limit - The requested `limit` value (already validated non-negative)
 * @param config - The active ValidationConfig (carries `maxLimit` + basis)
 * @throws When `limit > 2 * config.maxLimit`
 *
 * @since 7.30.2 (replaced the unconditional throw added in 7.30.0)
 */
function enforceLimitCap(limit: number, config: ValidationConfig): void {
  if (limit <= config.maxLimit) return

  const message = formatLimitMessage(limit, config)
  const hardCeiling = config.maxLimit * 2

  if (limit > hardCeiling) {
    // OOM danger zone — throw to prevent runaway memory allocation
    throw new Error(message)
  }

  // Soft margin: warn once per call site + limit value, then let the query proceed.
  // Survives the lifetime of the process; the goal is to teach the migration recipe,
  // not to spam the log every query.
  const caller = findCallerLocation(['validateFindParams', 'enforceLimitCap', 'formatLimitMessage']) ?? '<unknown caller>'
  const dedupKey = `${caller}|${limit}`
  if (!seenLimitWarnings.has(dedupKey)) {
    seenLimitWarnings.add(dedupKey)
    prodLog.warn('[Brainy] ' + message)
  }
}

/**
 * Render the user-facing message used by both the warning tier and the throw
 * tier of `enforceLimitCap`. Same body shape as the 7.30.1 enforcement-error
 * messages: state the problem, name the recipe, point at the call site, link
 * the docs.
 */
function formatLimitMessage(limit: number, config: ValidationConfig): string {
  const cap = config.maxLimit
  const basis = config.limitBasis
  const basisLabel: Record<string, string> = {
    override: 'consumer-supplied `maxQueryLimit`',
    reservedMemory: 'consumer-supplied `reservedQueryMemory`',
    containerMemory: 'detected container memory limit',
    freeMemory: 'available free memory'
  }
  const basisText = basisLabel[basis] ?? 'available memory'
  const caller = findCallerLocation(['enforceLimitCap', 'formatLimitMessage'])

  const lines = [
    `find({ limit: ${limit} }) exceeds the auto-configured query limit of ${cap} (basis: ${basisText}). Choose one:`,
    `  • Increase the cap: new Brainy({ maxQueryLimit: ${Math.min(limit, 100000)} })`,
    `  • Reserve more memory: new Brainy({ reservedQueryMemory: ${limit * MAX_LIMIT_KB_PER_RESULT * 1024} })`,
    '  • Paginate: split the query with { limit, offset } pages'
  ]
  if (caller) {
    lines.push(`  at ${caller}`)
  }
  lines.push('Docs: https://soulcraft.com/docs/guides/find-limits')

  return lines.join('\n')
}

/**
 * Universal validations - things that are always invalid
 * These are mathematical/logical truths, not configuration
 */
export function validateFindParams(params: FindParams): void {
  const config = ValidationConfig.getInstance()
  
  // Universal truth: negative pagination never makes sense
  if (params.limit !== undefined) {
    if (params.limit < 0) {
      throw new Error('limit must be non-negative')
    }
    enforceLimitCap(params.limit, config)
  }
  
  if (params.offset !== undefined && params.offset < 0) {
    throw new Error('offset must be non-negative')
  }
  
  // Universal truth: probability/similarity must be 0-1
  if (params.near?.threshold !== undefined) {
    const t = params.near.threshold
    if (t < 0 || t > 1) {
      throw new Error('threshold must be between 0 and 1')
    }
  }
  
  // Universal truth: can't specify both query and vector (they're alternatives)
  if (params.query !== undefined && params.vector !== undefined) {
    throw new Error('cannot specify both query and vector - they are mutually exclusive')
  }
  
  // ACCEPTED-AND-IGNORED DIED AS A CLASS (sealed 2026-08-03): options the
  // engine does not implement REFUSE with a typed error instead of silently
  // doing nothing — a production consumer discovered a no-op by measurement
  // once; never again.
  if (params.cursor !== undefined) {
    throw new UnsupportedFindOptionError('cursor')
  }
  if ((params as Record<string, unknown>).includeRelations !== undefined) {
    throw new UnsupportedFindOptionError('includeRelations')
  }
  if ((params as Record<string, unknown>).writeOnly !== undefined) {
    throw new UnsupportedFindOptionError('writeOnly')
  }

  // THE ONE ADDRESSING LAW: the orderBy address must PARSE (bare/metadata. =
  // user field, system.<field> = the ruled map, anything else refuses typed
  // with the valid map in the message) and order must be a real direction.
  if (params.orderBy !== undefined) {
    if (typeof params.orderBy !== 'string') {
      throw new Error('orderBy must be a string field address')
    }
    parseFieldAddress(params.orderBy, 'entity') // throws InvalidFieldAddressError on a bad address
  }
  if (params.order !== undefined && params.order !== 'asc' && params.order !== 'desc') {
    throw new Error(`order must be 'asc' or 'desc', got '${String(params.order)}'`)
  }
  
  // Auto-limit query length based on memory
  if (params.query && params.query.length > config.maxQueryLength) {
    throw new Error(`query exceeds auto-configured maximum length of ${config.maxQueryLength} characters`)
  }
  
  // Validate vector dimensions if provided
  if (params.vector && params.vector.length !== config.maxVectorDimensions) {
    throw new Error(`vector must have exactly ${config.maxVectorDimensions} dimensions`)
  }
  
  // Validate enum types if specified
  if (params.type) {
    const types = Array.isArray(params.type) ? params.type : [params.type]
    for (const type of types) {
      if (!Object.values(NounType).includes(type)) {
        throw new Error(`invalid NounType: ${type}`)
      }
    }
  }
}

/**
 * Validate add parameters
 */

/**
 * The namespace cannot be forged: a USER metadata key literally spelled
 * 'system.<anything>' would collide with the engine's explicit address
 * namespace at read time — refuse it at the write door, loudly, with the
 * fix in the message (sealed 2026-08-03).
 */
function rejectForgedSystemKeys(metadata: Record<string, unknown> | undefined, site: string): void {
  if (!metadata) return
  for (const key of Object.keys(metadata)) {
    if (key.startsWith('system.')) {
      throw new Error(
        `${site}: metadata key '${key}' is not allowed — the 'system.' prefix is the ` +
          `engine's explicit address namespace and cannot be used as a user field name. ` +
          `Rename the field (e.g. '${key.slice('system.'.length)}').`
      )
    }
  }
}

export function validateAddParams(params: AddParams): void {
  rejectForgedSystemKeys(params.metadata as Record<string, unknown> | undefined, 'add()')
  // 'data' is ABSENT only when null/undefined — an empty string ('') is real
  // content (a legitimate empty file's first write) and must not be treated
  // as missing. Falsy-but-present values (0, false, '') all count as present;
  // only the true "nothing was given" case is absent.
  const hasData = params.data !== undefined && params.data !== null
  // MT5 deferred embedding: an explicit vector has nothing to defer, and a
  // deferral without data has nothing to embed — both are caller bugs that
  // must refuse with the fix, never be silently reinterpreted.
  if ((params as AddParams & { deferEmbedding?: boolean }).deferEmbedding === true) {
    if (params.vector) {
      throw new Error(
        `add(): deferEmbedding cannot be combined with an explicit 'vector' — ` +
          `the vector is already computed; drop one of the two.`
      )
    }
    if (!hasData) {
      throw new Error(
        `add(): deferEmbedding requires 'data' (the content the background worker will embed).`
      )
    }
  }
  // Universal truth: must have data or vector
  if (!hasData && !params.vector) {
    throw new Error(
      `Invalid add() parameters: Missing required field 'data'\n` +
      `\nReceived: ${JSON.stringify({
        type: params.type,
        hasMetadata: !!params.metadata,
        hasId: !!params.id
      }, null, 2)}\n` +
      `\nExpected one of:\n` +
      `  { data: 'text to store', type?: 'note', metadata?: {...} }\n` +
      `  { vector: [0.1, 0.2, ...], type?: 'embedding', metadata?: {...} }\n` +
      `\nExamples:\n` +
      `  await brain.add({ data: 'Machine learning is AI', type: 'concept' })\n` +
      `  await brain.add({ data: { title: 'Doc', content: '...' }, type: 'document' })`
    )
  }
  
  // Validate noun type
  if (!Object.values(NounType).includes(params.type)) {
    throw new Error(
      `Invalid NounType: '${params.type}'\n` +
      `\nValid types: ${Object.values(NounType).join(', ')}\n` +
      `\nExample: await brain.add({ data: 'text', type: NounType.Document })`
    )
  }
  
  // Validate vector dimensions if provided. A length-0 vector is the
  // "unvectored" shape — an explicit `vector: []` (e.g. the VFS root's
  // permanently-vectorless creation, see
  // VirtualFileSystem.doInitializeRoot()'s zero-norm fix) carries no
  // dimension information, exactly like an absent vector or a deferred
  // embed's internal stub, so it is exempt from the dimension check rather
  // than refused as a "0-dimensional vector".
  if (params.vector && params.vector.length > 0) {
    const config = ValidationConfig.getInstance()
    if (params.vector.length !== config.maxVectorDimensions) {
      throw new Error(`vector must have exactly ${config.maxVectorDimensions} dimensions`)
    }
  }
}

/**
 * Validate update parameters
 */
export function validateUpdateParams(params: UpdateParams): void {
  rejectForgedSystemKeys(params.metadata as Record<string, unknown> | undefined, 'update()')
  // Same absent-vs-empty distinction as validateAddParams: '' is a real new
  // value (e.g. truncating a file to empty content via overwrite), only
  // null/undefined means "no new data was given".
  const hasData = params.data !== undefined && params.data !== null
  if ((params as UpdateParams & { deferEmbedding?: boolean }).deferEmbedding === true) {
    if (params.vector && params.vector.length === 0) {
      // The nonsensical combination Leg D of the zero-norm/unvector-door law
      // refuses: `vector: []` is the SANCTIONED UNVECTOR DOOR — an explicit
      // instruction to remove the vector NOW, never "please embed" — so it
      // cannot be paired with a request to defer an embed.
      throw new Error(
        `update(): 'vector: []' (the unvector door) cannot be combined with ` +
          `'deferEmbedding: true' — an unvector is an explicit instruction to remove ` +
          `the vector now, not a request to defer an embed. Drop one of the two.`
      )
    }
    if (params.vector) {
      throw new Error(
        `update(): deferEmbedding cannot be combined with an explicit 'vector' — ` +
          `the vector is already computed; drop one of the two.`
      )
    }
    if (!hasData) {
      throw new Error(
        `update(): deferEmbedding requires new 'data' — without a data change there is nothing to re-embed.`
      )
    }
  }
  // Universal truth: must have an ID
  if (!params.id) {
    throw new Error('id is required for update')
  }

  // Universal truth: must update something
  if (
    !hasData &&
    !params.metadata &&
    !params.type &&
    !params.vector &&
    params.subtype === undefined &&
    params.visibility === undefined &&
    params.confidence === undefined &&
    params.weight === undefined
  ) {
    throw new Error('must specify at least one field to update')
  }
  
  // Validate type if changing
  if (params.type && !Object.values(NounType).includes(params.type)) {
    throw new Error(`invalid NounType: ${params.type}`)
  }
  
  // Validate vector dimensions if provided. A length-0 vector is the
  // SANCTIONED UNVECTOR DOOR (see brainy.ts update()'s matching comment): an
  // explicit `vector: []` — or a real all-zero vector, normalized to `[]`
  // upstream by the zero-norm law — carries no dimension information,
  // exactly like validateAddParams's identical exemption, so it is exempt
  // from the dimension check rather than refused as a "0-dimensional
  // vector". (The `deferEmbedding` combination above already refuses
  // `vector: []` paired with `deferEmbedding: true` — an empty array is
  // truthy, so that guard fires unconditionally on any explicit `vector`.)
  if (params.vector && params.vector.length > 0) {
    const config = ValidationConfig.getInstance()
    if (params.vector.length !== config.maxVectorDimensions) {
      throw new Error(`vector must have exactly ${config.maxVectorDimensions} dimensions`)
    }
  }
}

/**
 * Validate relate parameters
 */
export function validateRelateParams(params: RelateParams): void {
  rejectForgedSystemKeys(params.metadata as Record<string, unknown> | undefined, 'relate()')
  // 8.0 verb-id contract (L.7): verb ids are UUIDs, generated by brainy.
  // RelateParams has no `id` field — an untyped caller passing one would
  // previously have it silently ignored (a generated UUID was used instead).
  // Teach instead of surprise: explain the contract and how ids actually work.
  const suppliedId = (params as RelateParams & { id?: unknown }).id
  if (suppliedId !== undefined) {
    throw new Error(
      `relate() does not accept a custom id (got: ${JSON.stringify(suppliedId)}). ` +
      'Verb ids are UUIDs by contract in 8.0 — brainy generates one for every ' +
      'relationship and returns it from relate(). Graph-index providers key ' +
      'verb-int interning on the raw UUID bytes, so custom verb ids are not ' +
      'supported. Remove the id field and use the returned id instead.'
    )
  }

  // Universal truths
  if (!params.from) {
    throw new Error('from entity ID is required')
  }

  if (!params.to) {
    throw new Error('to entity ID is required')
  }

  // Allow self-referential relationships - they're valid in graph systems
  // (e.g., a person can be related to themselves, a file can reference itself, etc.)

  // Validate verb type - default to RelatedTo if not specified
  if (params.type === undefined) {
    params.type = VerbType.RelatedTo
  } else if (!Object.values(VerbType).includes(params.type)) {
    throw new Error(`invalid VerbType: ${params.type}`)
  }

  // Universal truth: weight must be 0-1
  if (params.weight !== undefined) {
    if (params.weight < 0 || params.weight > 1) {
      throw new Error('weight must be between 0 and 1')
    }
  }
}

/**
 * Validate UpdateRelationParams. Mirror of validateUpdateParams for verbs —
 * requires id + at least one field to change; bounds-checks weight/confidence;
 * accepts type/subtype/weight/confidence/data/metadata changes.
 */
export function validateUpdateRelationParams(params: UpdateRelationParams): void {
  rejectForgedSystemKeys(params.metadata as Record<string, unknown> | undefined, 'updateRelation()')
  if (!params.id) {
    throw new Error('id is required for updateRelation')
  }

  if (
    !params.data &&
    !params.metadata &&
    !params.type &&
    params.subtype === undefined &&
    params.weight === undefined &&
    params.confidence === undefined
  ) {
    throw new Error('updateRelation: must specify at least one field to update')
  }

  if (params.type !== undefined && !Object.values(VerbType).includes(params.type)) {
    throw new Error(`invalid VerbType: ${params.type}`)
  }

  if (params.weight !== undefined && (params.weight < 0 || params.weight > 1)) {
    throw new Error('weight must be between 0 and 1')
  }

  if (params.confidence !== undefined && (params.confidence < 0 || params.confidence > 1)) {
    throw new Error('confidence must be between 0 and 1')
  }
}

/**
 * Get current validation configuration
 * Useful for debugging and monitoring
 */
export function getValidationConfig() {
  const config = ValidationConfig.getInstance()
  return {
    maxLimit: config.maxLimit,
    maxQueryLength: config.maxQueryLength,
    maxVectorDimensions: config.maxVectorDimensions,
    systemMemory: getSystemMemory(),
    availableMemory: getAvailableMemory()
  }
}

/**
 * Record query performance for auto-tuning
 */
export function recordQueryPerformance(duration: number, resultCount: number) {
  ValidationConfig.getInstance().recordQuery(duration, resultCount)
}