Imhotep/packages/imhotep-core/src/geometry-cache.ts

/**
 * Geometry snapshot caching for Imhotep.
 *
 * Serializes and deserializes GeometryWorld instances so extracted
 * browser geometry can be cached to disk and reused across assertion
 * batches without re-extraction.
 */

import type { GeometryWorld, WorldEnvironment, WorldSource } from './world.js'
import type { ImhotepId } from './types.js'

// ---------------------------------------------------------------------------
// Serialization
// ---------------------------------------------------------------------------

interface SerializedTypedArray {
  type: 'Uint32Array' | 'Uint16Array' | 'Uint8Array' | 'Int32Array' | 'Int16Array' | 'Float64Array' | 'Float32Array'
  data: number[]
}

interface SerializedTable {
  [column: string]: SerializedTypedArray | string[] | number[]
}

interface SerializedWorld {
  version: number
  sceneId: ImhotepId
  snapshotId: ImhotepId
  env: WorldEnvironment
  source: WorldSource
  strings: { values: string[] }
  subjects: SerializedTable
  dom: SerializedTable
  frames: SerializedTable
  matrices: SerializedTable
  rects: SerializedTable
  boxes: SerializedTable
  visualBoxes: SerializedTable
  fragments: SerializedTable
  transforms: SerializedTable
  styles: SerializedTable
  text: SerializedTable
  topology: SerializedTable
  scroll: SerializedTable
  clipping: SerializedTable
  paint: SerializedTable
  visibility: SerializedTable
  provenance: SerializedTable
  confidence: SerializedTable
}

function serializeTypedArray(arr: TypedArray): SerializedTypedArray {
  const ctor = arr.constructor.name as SerializedTypedArray['type']
  return { type: ctor, data: Array.from(arr) }
}

function deserializeTypedArray(serialized: SerializedTypedArray): TypedArray {
  const ctor = globalThis[serialized.type] as new (data: number[]) => TypedArray
  return new ctor(serialized.data)
}

type TypedArray =
  | Uint32Array
  | Uint16Array
  | Uint8Array
  | Int32Array
  | Int16Array
  | Float64Array
  | Float32Array

function isTypedArray(value: unknown): value is TypedArray {
  return (
    value instanceof Uint32Array ||
    value instanceof Uint16Array ||
    value instanceof Uint8Array ||
    value instanceof Int32Array ||
    value instanceof Int16Array ||
    value instanceof Float64Array ||
    value instanceof Float32Array
  )
}

function serializeTable(table: Record<string, unknown>): SerializedTable {
  const result: SerializedTable = {}
  if (!table) return result
  for (const [key, value] of Object.entries(table)) {
    if (isTypedArray(value)) {
      result[key] = serializeTypedArray(value)
    } else if (Array.isArray(value)) {
      result[key] = value as number[] | string[]
    } else if (typeof value === 'number' || typeof value === 'string' || typeof value === 'boolean') {
      result[key] = [value] as number[] | string[]
    } else {
      result[key] = value as SerializedTable[string]
    }
  }
  return result
}

function deserializeTable(serialized: SerializedTable): Record<string, unknown> {
  const result: Record<string, unknown> = {}
  for (const [key, value] of Object.entries(serialized)) {
    if (value && typeof value === 'object' && 'type' in value && 'data' in value) {
      result[key] = deserializeTypedArray(value as SerializedTypedArray)
    } else if (Array.isArray(value)) {
      result[key] = value
    } else {
      result[key] = value
    }
  }
  return result
}

/**
 * Serialize a GeometryWorld to a JSON string.
 */
export function serializeGeometryWorld(world: GeometryWorld): string {
  const serialized: SerializedWorld = {
    version: 1,
    sceneId: world.sceneId,
    snapshotId: world.snapshotId,
    env: world.env,
    source: world.source,
    strings: world.strings,
    subjects: serializeTable(world.subjects as unknown as Record<string, unknown>),
    dom: serializeTable(world.dom as unknown as Record<string, unknown>),
    frames: serializeTable(world.frames as unknown as Record<string, unknown>),
    matrices: serializeTable(world.matrices as unknown as Record<string, unknown>),
    rects: serializeTable(world.rects as unknown as Record<string, unknown>),
    boxes: serializeTable(world.boxes as unknown as Record<string, unknown>),
    visualBoxes: serializeTable(world.visualBoxes as unknown as Record<string, unknown>),
    fragments: serializeTable(world.fragments as unknown as Record<string, unknown>),
    transforms: serializeTable(world.transforms as unknown as Record<string, unknown>),
    styles: serializeTable(world.styles as unknown as Record<string, unknown>),
    text: serializeTable(world.text as unknown as Record<string, unknown>),
    topology: serializeTable(world.topology as unknown as Record<string, unknown>),
    scroll: serializeTable(world.scroll as unknown as Record<string, unknown>),
    clipping: serializeTable(world.clipping as unknown as Record<string, unknown>),
    paint: serializeTable(world.paint as unknown as Record<string, unknown>),
    visibility: serializeTable(world.visibility as unknown as Record<string, unknown>),
    provenance: serializeTable(world.provenance as unknown as Record<string, unknown>),
    confidence: serializeTable(world.confidence as unknown as Record<string, unknown>),
  }
  return JSON.stringify(serialized)
}

/**
 * Deserialize a JSON string back into a GeometryWorld.
 */
export function deserializeGeometryWorld(json: string): GeometryWorld {
  const serialized = JSON.parse(json) as SerializedWorld

  if (serialized.version !== 1) {
    throw new Error(`Unsupported geometry cache version: ${serialized.version}`)
  }

  return {
    sceneId: serialized.sceneId,
    snapshotId: serialized.snapshotId,
    env: serialized.env,
    source: serialized.source,
    strings: serialized.strings,
    subjects: deserializeTable(serialized.subjects) as unknown as GeometryWorld['subjects'],
    dom: deserializeTable(serialized.dom) as unknown as GeometryWorld['dom'],
    frames: deserializeTable(serialized.frames) as unknown as GeometryWorld['frames'],
    matrices: deserializeTable(serialized.matrices) as unknown as GeometryWorld['matrices'],
    rects: deserializeTable(serialized.rects) as unknown as GeometryWorld['rects'],
    boxes: deserializeTable(serialized.boxes) as unknown as GeometryWorld['boxes'],
    visualBoxes: deserializeTable(serialized.visualBoxes) as unknown as GeometryWorld['visualBoxes'],
    fragments: deserializeTable(serialized.fragments) as unknown as GeometryWorld['fragments'],
    transforms: deserializeTable(serialized.transforms) as unknown as GeometryWorld['transforms'],
    styles: deserializeTable(serialized.styles) as unknown as GeometryWorld['styles'],
    text: deserializeTable(serialized.text) as unknown as GeometryWorld['text'],
    topology: deserializeTable(serialized.topology) as unknown as GeometryWorld['topology'],
    scroll: deserializeTable(serialized.scroll) as unknown as GeometryWorld['scroll'],
    clipping: deserializeTable(serialized.clipping) as unknown as GeometryWorld['clipping'],
    paint: deserializeTable(serialized.paint) as unknown as GeometryWorld['paint'],
    visibility: deserializeTable(serialized.visibility) as unknown as GeometryWorld['visibility'],
    provenance: deserializeTable(serialized.provenance) as unknown as GeometryWorld['provenance'],
    confidence: deserializeTable(serialized.confidence) as unknown as GeometryWorld['confidence'],
  }
}

// ---------------------------------------------------------------------------
// Cache key computation
// ---------------------------------------------------------------------------

function djb2Hash(str: string): string {
  let hash = 5381
  for (let i = 0; i < str.length; i++) {
    hash = ((hash << 5) + hash + str.charCodeAt(i)) & 0xffffffff
  }
  return hash.toString(16).padStart(8, '0')
}

/**
 * Compute a stable cache key for a geometry extraction.
 *
 * The key incorporates:
 * - page URL
 * - sorted selectors (so order doesn't matter)
 * - environment (viewport dimensions, color scheme, etc.)
 */
export function computeGeometryCacheKey(
  pageUrl: string,
  selectors: string[],
  env: Partial<WorldEnvironment>,
): string {
  const selectorsHash = djb2Hash([...selectors].sort().join('\x00'))
  const envHash = djb2Hash(
    JSON.stringify({
      viewportWidth: env.viewportWidth ?? 0,
      viewportHeight: env.viewportHeight ?? 0,
      deviceScaleFactor: env.deviceScaleFactor ?? 1,
      colorScheme: env.colorScheme ?? 'light',
      pointer: env.pointer ?? 'fine',
      hover: env.hover ?? false,
      reducedMotion: env.reducedMotion ?? 'no-preference',
      locale: env.locale ?? 'en',
      writingMode: env.writingMode ?? 'horizontal-tb',
    }),
  )
  const urlHash = djb2Hash(pageUrl)
  return `${urlHash}-${selectorsHash}-${envHash}`
}

// ---------------------------------------------------------------------------
// File-based cache storage
// ---------------------------------------------------------------------------

import { existsSync, mkdirSync, rmSync, readdirSync, statSync } from 'node:fs'
import { readFile, writeFile, unlink } from 'node:fs/promises'
import { join } from 'node:path'

const DEFAULT_CACHE_DIR = join(process.cwd(), '.imhotep-cache')
const DEFAULT_MAX_CACHE_ENTRIES = 100

function cacheFilePath(cacheDir: string, cacheKey: string): string {
  return join(cacheDir, `${cacheKey}.json`)
}

/**
 * Evict oldest cache entries when exceeding maxEntries.
 */
function evictOldestEntries(cacheDir: string, maxEntries: number): void {
  if (!existsSync(cacheDir)) return
  const files = readdirSync(cacheDir)
    .filter((f) => f.endsWith('.json'))
    .map((f) => {
      const fullPath = join(cacheDir, f)
      try {
        return { name: f, path: fullPath, mtime: statSync(fullPath).mtimeMs }
      } catch (err) {
        console.warn(`[imhotep-core] cache eviction: stat failed for ${f}: ${err instanceof Error ? err.message : err}`)
        return null
      }
    })
    .filter((f): f is NonNullable<typeof f> => f !== null)
    .sort((a, b) => a.mtime - b.mtime)

  if (files.length > maxEntries) {
    const toRemove = files.length - maxEntries
    for (let i = 0; i < toRemove; i++) {
      try {
        rmSync(files[i].path)
      } catch (err) {
        console.warn(`[imhotep-core] cache eviction: failed to remove ${files[i].path}: ${err instanceof Error ? err.message : err}`)
      }
    }
  }
}

/**
 * Read a cached GeometryWorld from disk (async, non-blocking).
 */
export async function readCachedWorld(cacheDir: string, cacheKey: string): Promise<GeometryWorld | null> {
  const path = cacheFilePath(cacheDir, cacheKey)
  try {
    const json = await readFile(path, 'utf-8')
    return deserializeGeometryWorld(json)
  } catch (err) {
    console.warn(`[imhotep-core] readCachedWorld failed for ${cacheKey}: ${err instanceof Error ? err.message : err}. Cache miss.`)
    return null
  }
}

/**
 * Write a GeometryWorld to the cache on disk (async, non-blocking).
 */
export async function writeCachedWorld(cacheDir: string, cacheKey: string, world: GeometryWorld): Promise<void> {
  mkdirSync(cacheDir, { recursive: true })
  evictOldestEntries(cacheDir, DEFAULT_MAX_CACHE_ENTRIES)
  const path = cacheFilePath(cacheDir, cacheKey)
  await writeFile(path, serializeGeometryWorld(world))
}

/**
 * Clear all cached geometry snapshots.
 */
export function clearGeometryCache(cacheDir: string = DEFAULT_CACHE_DIR): void {
  if (!existsSync(cacheDir)) {
    return
  }
  for (const file of readdirSync(cacheDir)) {
    if (file.endsWith('.json')) {
      try {
        rmSync(join(cacheDir, file))
      } catch (err) {
        console.warn(`[imhotep-core] cache clear: failed to remove ${file}: ${err instanceof Error ? err.message : err}`)
      }
    }
  }
}

// ---------------------------------------------------------------------------
// Full extraction result caching (world + selector mapping + errors)
// ---------------------------------------------------------------------------

interface CachedExtractionResult {
  version: number
  world: string
  selectorToIds: [string, number[]][]
  errors: Array<{
    code: string
    severity: string
    category: string
    message: string
    source: string
    related: string[]
    fixHints: string[]
    metrics: Record<string, unknown>
    sourceRef: Record<string, unknown>
    clauseLabel?: string
  }>
}

/**
 * Serialize a full extraction result (world + selector mappings + errors).
 */
export function serializeExtractionResult(
  world: GeometryWorld,
  selectorToIds: Map<string, number[]>,
  errors: Array<{
    code: string
    severity: string
    category: string
    message: string
    source: string
    related: string[]
    fixHints: string[]
    metrics: Record<string, unknown>
    sourceRef: Record<string, unknown>
    clauseLabel?: string
  }>,
): string {
  const cached: CachedExtractionResult = {
    version: 1,
    world: serializeGeometryWorld(world),
    selectorToIds: Array.from(selectorToIds.entries()),
    errors,
  }
  return JSON.stringify(cached)
}

/**
 * Deserialize a full extraction result from JSON.
 */
export function deserializeExtractionResult(json: string): {
  world: GeometryWorld
  selectorToIds: Map<string, number[]>
  errors: CachedExtractionResult['errors']
} {
  const cached = JSON.parse(json) as CachedExtractionResult
  if (cached.version !== 1) {
    throw new Error(`Unsupported extraction cache version: ${cached.version}`)
  }
  return {
    world: deserializeGeometryWorld(cached.world),
    selectorToIds: new Map(cached.selectorToIds),
    errors: cached.errors,
  }
}

/**
 * Read a cached extraction result from disk (async, non-blocking).
 */
export async function readCachedExtractionResult(
  cacheDir: string,
  cacheKey: string,
): Promise<{ world: GeometryWorld; selectorToIds: Map<string, number[]>; errors: CachedExtractionResult['errors'] } | null> {
  const path = cacheFilePath(cacheDir, cacheKey)
  try {
    const json = await readFile(path, 'utf-8')
    return deserializeExtractionResult(json)
  } catch (err) {
    console.warn(`[imhotep-core] readCachedExtractionResult failed for cacheKey=${cacheKey}: ${err instanceof Error ? err.message : err}. Cache miss.`)
    return null
  }
}

/**
 * Write an extraction result to the cache on disk (async, non-blocking).
 */
export async function writeCachedExtractionResult(
  cacheDir: string,
  cacheKey: string,
  world: GeometryWorld,
  selectorToIds: Map<string, number[]>,
  errors: CachedExtractionResult['errors'],
): Promise<void> {
  mkdirSync(cacheDir, { recursive: true })
  evictOldestEntries(cacheDir, DEFAULT_MAX_CACHE_ENTRIES)
  const path = cacheFilePath(cacheDir, cacheKey)
  await writeFile(path, serializeExtractionResult(world, selectorToIds, errors))
}

/**
 * Return the default cache directory path.
 */
export function getDefaultCacheDir(): string {
  return DEFAULT_CACHE_DIR
}