justvitamin/scripts/validate-dataforseo-pilot-review.ts

#!/usr/bin/env bun
import { existsSync, mkdirSync, readFileSync, writeFileSync } from 'fs'
import { join } from 'path'

const root = process.cwd()
const outRoot = join(root, 'content_population_exports')
const reviewRoot = join(root, 'data', 'sources', 'reviews')
mkdirSync(outRoot, { recursive: true })
mkdirSync(reviewRoot, { recursive: true })

function readJson(path: string, fallback: any = null) {
  if (!existsSync(path)) return fallback
  return JSON.parse(readFileSync(path, 'utf8').replace(/^\uFEFF/, ''))
}
function csv(rows: any[], headers: string[]) {
  const esc = (value: any) => {
    if (value === undefined || value === null) value = ''
    if (Array.isArray(value)) value = value.join(' | ')
    value = String(value)
    return /[",\n\r]/.test(value) ? `"${value.replace(/"/g, '""')}"` : value
  }
  return [headers.join(','), ...rows.map(row => headers.map(header => esc(row[header])).join(','))].join('\n') + '\n'
}

const generatedAt = new Date().toISOString()
const ingest = readJson(join(outRoot, 'dataforseo_ingest_validation.json'), {})
const spendGate = readJson(join(outRoot, 'source_acquisition_spend_gate.json'), {})
const apifyGapfill = readJson(join(outRoot, 'apify_gapfill_manifest.json'), {})
const reviewPath = join(reviewRoot, 'dataforseo-stage1-review.json')
const reviewTemplate = {
  reviewed: false,
  reviewedBy: '',
  reviewedAt: '',
  provider: 'DataForSEO',
  stage: 1,
  rawOutputInspected: false,
  costReviewed: false,
  outputHasReviewText: false,
  outputHasRating: false,
  outputHasDate: false,
  outputHasSourceUrl: false,
  costAcceptable: false,
  decision: 'pending',
  allowedDecisions: ['scale_dataforseo', 'run_apify_gapfill', 'stop_adjust_source_map'],
  namedGapForApify: '',
  notes: ''
}
if (!existsSync(reviewPath)) writeFileSync(reviewPath, JSON.stringify(reviewTemplate, null, 2) + '\n', 'utf8')
const review = readJson(reviewPath, reviewTemplate)

const convertedReviews = Number(ingest.counts?.convertedReviews || 0)
const convertedFiles = Number(ingest.counts?.convertedFiles || 0)
const issues = Number(ingest.counts?.issues || 0)
const apifyNamedGap = String(review.namedGapForApify || '').trim()
const checks = [
  {
    Check: 'Raw DataForSEO result ingested',
    Status: convertedFiles > 0 && convertedReviews > 0 ? 'pass' : 'blocked',
    Evidence: `${convertedFiles} converted files / ${convertedReviews} converted reviews`
  },
  {
    Check: 'Ingest has no structural issues',
    Status: convertedFiles > 0 && issues === 0 ? 'pass' : 'blocked',
    Evidence: `${issues} ingest issues`
  },
  {
    Check: 'Reviewer inspected raw output',
    Status: review.reviewed === true && review.rawOutputInspected === true ? 'pass' : 'blocked',
    Evidence: review.reviewed ? `${review.reviewedBy || 'unknown'} at ${review.reviewedAt || 'unknown'}` : 'review not recorded'
  },
  {
    Check: 'Required evidence fields present',
    Status: review.outputHasReviewText && review.outputHasRating && review.outputHasDate && review.outputHasSourceUrl ? 'pass' : 'blocked',
    Evidence: `text=${!!review.outputHasReviewText}, rating=${!!review.outputHasRating}, date=${!!review.outputHasDate}, url=${!!review.outputHasSourceUrl}`
  },
  {
    Check: 'Cost reviewed and acceptable',
    Status: review.costReviewed === true && review.costAcceptable === true ? 'pass' : 'blocked',
    Evidence: `costReviewed=${!!review.costReviewed}, costAcceptable=${!!review.costAcceptable}`
  },
  {
    Check: 'Decision recorded',
    Status: review.reviewed === true && reviewTemplate.allowedDecisions.includes(review.decision) ? 'pass' : 'blocked',
    Evidence: `decision=${review.decision || 'pending'}`
  },
  {
    Check: 'Apify gap named if selected',
    Status: review.decision === 'run_apify_gapfill' ? (apifyNamedGap ? 'pass' : 'blocked') : 'pass',
    Evidence: review.decision === 'run_apify_gapfill' ? (apifyNamedGap || 'missing named gap') : 'not selecting Apify gap-fill'
  }
].map(row => ({ ...row, Blocking: row.Status === 'pass' ? 'no' : 'yes' }))

const ready = checks.every(row => row.Status === 'pass')
const decision = ready ? review.decision : 'pending'
const recommendations = ready
  ? decision === 'scale_dataforseo'
    ? ['Scale DataForSEO only to the next controlled priority batch; keep Apify off.']
    : decision === 'run_apify_gapfill'
      ? [`Run only the Apify first batch (${apifyGapfill.counts?.firstBatchTargets || 0} targets) for named gap: ${apifyNamedGap}.`]
      : ['Stop and adjust the source map before any further provider spend.']
  : ['Do not scale DataForSEO or run Apify yet.', 'Complete raw ingest, evidence-field review, cost review, and explicit decision first.']

const manifest = {
  generatedAt,
  ready,
  decision,
  reviewFile: 'data/sources/reviews/dataforseo-stage1-review.json',
  counts: {
    checks: checks.length,
    passing: checks.filter(row => row.Status === 'pass').length,
    blocked: checks.filter(row => row.Status !== 'pass').length,
    convertedFiles,
    convertedReviews,
    ingestIssues: issues
  },
  checks,
  recommendations,
  review
}
writeFileSync(join(outRoot, 'dataforseo_pilot_review.json'), JSON.stringify(manifest, null, 2) + '\n', 'utf8')
writeFileSync(join(outRoot, 'dataforseo_pilot_review.csv'), csv(checks, ['Check', 'Status', 'Blocking', 'Evidence']), 'utf8')
writeFileSync(join(outRoot, 'dataforseo_pilot_review.md'), [
  '# DataForSEO Stage 1 pilot review gate',
  '',
  `Generated: ${generatedAt}`,
  '',
  `Ready: ${ready ? 'YES' : 'NO'}`,
  `Decision: ${decision}`,
  '',
  'This gate is the stop/review step between paid Stage 1 output and any scale-up or Apify gap-fill spend.',
  '',
  '## Checks',
  ...checks.map(row => `- ${row.Status === 'pass' ? '[pass]' : '[blocked]'} ${row.Check} — ${row.Evidence}`),
  '',
  '## Recommendations',
  ...recommendations.map(item => `- ${item}`),
  '',
  '## Review file',
  '- `data/sources/reviews/dataforseo-stage1-review.json`',
  ''
].join('\n'), 'utf8')

console.log(`DataForSEO pilot review ready: ${ready}`)
console.log(`Decision: ${decision}`)
console.log(`Checks passing: ${manifest.counts.passing}/${manifest.counts.checks}`)