justvitamin/scripts/validate-source-acquisition-approvals.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 approvalRoot = join(root, 'data', 'sources', 'approvals')
mkdirSync(outRoot, { recursive: true })
mkdirSync(approvalRoot, { 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 budgetGate = readJson(join(outRoot, 'source_acquisition_budget_gate.json'), {})
const costPlan = readJson(join(outRoot, 'source_acquisition_cost_plan.json'), {})
const dataforseoPilot = readJson(join(outRoot, 'dataforseo_pilot_manifest.json'), {})
const apifyGapfill = readJson(join(outRoot, 'apify_gapfill_manifest.json'), {})
const dataforseoReview = readJson(join(outRoot, 'dataforseo_pilot_review.json'), {})
const budget = budgetGate.budget || {}
const currency = budget.currency || 'GBP'

const lanes = [
  {
    key: 'dataforseoStage1',
    provider: 'DataForSEO',
    stage: 1,
    approvalFile: 'dataforseo-stage1-approval.json',
    budgetLane: budget.dataforseoStage1 || {},
    expectedScope: 'DataForSEO Stage 1 Amazon Reviews pilot only',
    expectedMaxTargets: Number(dataforseoPilot.counts?.stageOneTargets || budget.dataforseoStage1?.maxTasks || 9),
    expectedRunLimit: 1,
    expectedDepthOrRequests: Number(budget.dataforseoStage1?.maxReviewsPerAsin || 10),
    decisionRef: 'content_population_exports/dataforseo_task_preflight.json'
  },
  {
    key: 'apifyStage2',
    provider: 'Apify',
    stage: 2,
    approvalFile: 'apify-stage2-approval.json',
    budgetLane: budget.apifyStage2 || {},
    expectedScope: 'Apify Stage 2 named gap-fill only after DataForSEO pilot review',
    expectedMaxTargets: Number(apifyGapfill.counts?.firstBatchTargets || budget.apifyStage2?.maxTargets || 3),
    expectedRunLimit: Number(budget.apifyStage2?.maxActorRuns || 1),
    expectedDepthOrRequests: Number(budget.apifyStage2?.maxRequestsPerBrand || 3),
    decisionRef: 'data/sources/reviews/dataforseo-stage1-review.json'
  }
]

function approvalTemplate(lane: any, existing: any = {}) {
  const maxSpend = Number(lane.budgetLane?.maxSpend || 0)
  return {
    provider: lane.provider,
    stage: lane.stage,
    approved: existing.approved === true,
    approvedBy: existing.approvedBy || '',
    approvedAt: existing.approvedAt || '',
    approvalScope: existing.approvalScope || existing.scope || lane.expectedScope,
    budgetRef: existing.budgetRef || 'data/sources/budgets/source-acquisition-budget.json',
    decisionRef: existing.decisionRef || lane.decisionRef,
    currency,
    maxSpendApproved: Number(existing.maxSpendApproved || 0),
    expectedMaxSpend: maxSpend,
    targetCapApproved: Number(existing.targetCapApproved || 0),
    expectedTargetCap: lane.expectedMaxTargets,
    runLimitApproved: Number(existing.runLimitApproved || 0),
    expectedRunLimit: lane.expectedRunLimit,
    depthOrRequestCapApproved: Number(existing.depthOrRequestCapApproved || 0),
    expectedDepthOrRequestCap: lane.expectedDepthOrRequests,
    hardStopAcknowledged: existing.hardStopAcknowledged === true,
    noCredentialStorageAcknowledged: existing.noCredentialStorageAcknowledged === true,
    postRunReviewRequired: existing.postRunReviewRequired !== false,
    namedGapForApify: existing.namedGapForApify || dataforseoReview.review?.namedGapForApify || '',
    notes: existing.notes || ''
  }
}

function validateApproval(lane: any, approval: any) {
  const checks = []
  const maxSpend = Number(lane.budgetLane?.maxSpend || 0)
  checks.push(['Approved flag true', approval.approved === true, approval.approved ? 'approved=true' : 'approved=false'])
  checks.push(['Approver recorded', Boolean(approval.approvedBy), approval.approvedBy || 'missing approver'])
  checks.push(['Approval timestamp recorded', Boolean(approval.approvedAt), approval.approvedAt || 'missing timestamp'])
  checks.push(['Budget lane enabled', lane.budgetLane?.enabled === true, `enabled=${lane.budgetLane?.enabled === true}`])
  checks.push(['Cost plan ready', costPlan.costReady?.[lane.key] === true, costPlan.costReady?.[lane.key] === true ? 'cost plan ready' : `source_acquisition_cost_plan.json not ready for ${lane.provider}`])
  checks.push(['Max spend matches budget cap', maxSpend > 0 && Number(approval.maxSpendApproved || 0) === maxSpend, `${approval.currency || currency} ${Number(approval.maxSpendApproved || 0)} approved / ${maxSpend} budget`])
  checks.push(['Target cap matches staged lane', Number(approval.targetCapApproved || 0) === lane.expectedMaxTargets, `${Number(approval.targetCapApproved || 0)} approved / ${lane.expectedMaxTargets} expected`])
  checks.push(['Run limit matches staged lane', Number(approval.runLimitApproved || 0) === lane.expectedRunLimit, `${Number(approval.runLimitApproved || 0)} approved / ${lane.expectedRunLimit} expected`])
  checks.push(['Depth/request cap matches staged lane', Number(approval.depthOrRequestCapApproved || 0) === lane.expectedDepthOrRequests, `${Number(approval.depthOrRequestCapApproved || 0)} approved / ${lane.expectedDepthOrRequests} expected`])
  checks.push(['Hard stop acknowledged', approval.hardStopAcknowledged === true, `hardStopAcknowledged=${approval.hardStopAcknowledged === true}`])
  checks.push(['No credential storage acknowledged', approval.noCredentialStorageAcknowledged === true, `noCredentialStorageAcknowledged=${approval.noCredentialStorageAcknowledged === true}`])
  if (lane.key === 'apifyStage2') {
    checks.push(['Named Apify gap recorded', Boolean(String(approval.namedGapForApify || '').trim()), approval.namedGapForApify || 'missing named Apify gap'])
  }
  return checks.map(([Check, pass, Evidence]) => ({
    Provider: lane.provider,
    Stage: String(lane.stage),
    Check,
    Status: pass ? 'pass' : 'blocked',
    Blocking: 'yes',
    Evidence
  }))
}

const approvals: any[] = []
const rows: any[] = []
for (const lane of lanes) {
  const path = join(approvalRoot, lane.approvalFile)
  const existing = readJson(path, {})
  const approval = approvalTemplate(lane, existing)
  writeFileSync(path, JSON.stringify(approval, null, 2) + '\n', 'utf8')
  const checks = validateApproval(lane, approval)
  approvals.push({
    key: lane.key,
    provider: lane.provider,
    stage: lane.stage,
    file: `data/sources/approvals/${lane.approvalFile}`,
    approved: approval.approved,
    valid: checks.every(row => row.Status === 'pass'),
    approval
  })
  rows.push(...checks)
}

const manifest = {
  generatedAt,
  ready: approvals.every(row => row.valid),
  counts: {
    providers: approvals.length,
    approvalsValid: approvals.filter(row => row.valid).length,
    checks: rows.length,
    passing: rows.filter(row => row.Status === 'pass').length,
    blocked: rows.filter(row => row.Status !== 'pass').length
  },
  files: {
    dataforseoApproval: 'data/sources/approvals/dataforseo-stage1-approval.json',
    apifyApproval: 'data/sources/approvals/apify-stage2-approval.json',
    reportMd: 'content_population_exports/source_acquisition_approval_gate.md',
    reportCsv: 'content_population_exports/source_acquisition_approval_gate.csv',
    reportJson: 'content_population_exports/source_acquisition_approval_gate.json'
  },
  guardrails: [
    'Approval is local and explicit; it does not store provider credentials.',
    'Approval must match the budget cap and staged run limits exactly.',
    'A human approval cannot bypass source, budget, provider preflight, or post-run review gates.',
    'Apify approval also requires a named gap from the DataForSEO pilot review.'
  ],
  approvals,
  checks: rows,
  nextActions: rows.filter(row => row.Status !== 'pass').map(row => `${row.Provider} Stage ${row.Stage}: ${row.Check} - ${row.Evidence}`)
}

writeFileSync(join(outRoot, 'source_acquisition_approval_gate.json'), JSON.stringify(manifest, null, 2) + '\n', 'utf8')
writeFileSync(join(outRoot, 'source_acquisition_approval_gate.csv'), csv(rows, ['Provider', 'Stage', 'Check', 'Status', 'Blocking', 'Evidence']), 'utf8')
writeFileSync(join(outRoot, 'source_acquisition_approval_gate.md'), [
  '# Source acquisition approval gate',
  '',
  `Generated: ${generatedAt}`,
  '',
  `Approvals valid: ${manifest.counts.approvalsValid}/${manifest.counts.providers}`,
  `Checks passing: ${manifest.counts.passing}/${manifest.counts.checks}`,
  '',
  'This gate validates the local approval files that must be filled before any paid provider run. It does not call DataForSEO, Apify, or any paid API.',
  '',
  '## Guardrails',
  ...manifest.guardrails.map(item => `- ${item}`),
  '',
  '## Checks',
  ...rows.map(row => `- ${row.Status === 'pass' ? '[pass]' : '[blocked]'} ${row.Provider} Stage ${row.Stage}: ${row.Check} — ${row.Evidence}`),
  '',
  '## Approval files',
  ...approvals.map(row => `- ${row.provider}: \`${row.file}\` (${row.valid ? 'valid' : 'blocked'})`),
  ''
].join('\n'), 'utf8')

console.log(`Source acquisition approval gate ready: ${manifest.ready}`)
console.log(`Approvals valid: ${manifest.counts.approvalsValid}/${manifest.counts.providers}`)
console.log(`Checks passing: ${manifest.counts.passing}/${manifest.counts.checks}`)