import fs from 'fs'
import path from 'path'
import { BenchmarkData } from './types'
import { BenchmarkDataSchema } from './schema'

let cached: BenchmarkData | null = null

export async function loadBenchmarks(): Promise<BenchmarkData> {
  if (cached) return cached
  const filePath = path.join(process.cwd(), 'data/benchmarks.json')
  const raw = JSON.parse(fs.readFileSync(filePath, 'utf-8'))

  // Sanitize patterns: Ollama sometimes returns arrays instead of strings, nulls, or missing fields
  if (Array.isArray(raw.patterns)) {
    raw.patterns = raw.patterns
      .map((p: Record<string, unknown>) => ({
        ...p,
        signature: Array.isArray(p.signature) ? p.signature.join(', ') : p.signature,
        interpretation: Array.isArray(p.interpretation) ? p.interpretation.join(' ') : p.interpretation,
        improvements: Array.isArray(p.improvements)
          ? p.improvements.map((i: unknown) => (Array.isArray(i) ? i.join(', ') : String(i)))
          : [],
      }))
      // Drop patterns with missing or null required string fields
      .filter((p: Record<string, unknown>) =>
        typeof p.id === 'string' && p.id &&
        typeof p.signature === 'string' && p.signature &&
        typeof p.interpretation === 'string' && p.interpretation
      )
  }

  cached = BenchmarkDataSchema.parse(raw)
  return cached
}