justvitamin/scripts/validate-source-acquisition-budget.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 budgetRoot = join(root, 'data', 'sources', 'budgets')
mkdirSync(outRoot, { recursive: true })
mkdirSync(budgetRoot, { 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 budgetPath = join(budgetRoot, 'source-acquisition-budget.json')
const pilot = readJson(join(outRoot, 'dataforseo_pilot_manifest.json'), {})
const apify = readJson(join(outRoot, 'apify_gapfill_manifest.json'), {})
const defaultBudget = {
  currency: 'GBP',
  owner: '',
  updatedAt: '',
  notes: 'Fill this before any paid DataForSEO or Apify run. Do not store provider credentials here.',
  dataforseoStage1: {
    enabled: false,
    maxSpend: 0,
    maxTasks: pilot.counts?.stageOneTargets || 9,
    marketplace: 'amazon.co.uk',
    maxReviewsPerAsin: pilot.guardrails?.recommendedDepth || 10,
    hardStopAfterFirstRun: true,
    approvalRef: 'data/sources/approvals/dataforseo-stage1-approval.json'
  },
  apifyStage2: {
    enabled: false,
    maxSpend: 0,
    maxActorRuns: 1,
    maxTargets: apify.counts?.firstBatchTargets || 3,
    maxRequestsPerBrand: 3,
    hardStopAfterFirstDataset: true,
    approvalRef: 'data/sources/approvals/apify-stage2-approval.json'
  }
}
if (!existsSync(budgetPath)) writeFileSync(budgetPath, JSON.stringify(defaultBudget, null, 2) + '\n', 'utf8')
const budget = readJson(budgetPath, defaultBudget)
const currency = budget.currency || 'GBP'

const dataforseo = budget.dataforseoStage1 || {}
const apifyBudget = budget.apifyStage2 || {}
const stageOneTargets = Number(pilot.counts?.stageOneTargets || 0)
const firstBatchTargets = Number(apify.counts?.firstBatchTargets || 0)

const rows = [
  {
    Provider: 'DataForSEO',
    Stage: '1',
    Check: 'Budget lane enabled',
    Status: dataforseo.enabled === true ? 'pass' : 'blocked',
    Blocking: 'yes',
    Evidence: dataforseo.enabled === true ? 'enabled=true' : 'enabled=false in data/sources/budgets/source-acquisition-budget.json'
  },
  {
    Provider: 'DataForSEO',
    Stage: '1',
    Check: 'Max spend recorded',
    Status: Number(dataforseo.maxSpend || 0) > 0 ? 'pass' : 'blocked',
    Blocking: 'yes',
    Evidence: `${currency} ${Number(dataforseo.maxSpend || 0)}`
  },
  {
    Provider: 'DataForSEO',
    Stage: '1',
    Check: 'Task cap covers pilot only',
    Status: Number(dataforseo.maxTasks || 0) > 0 && Number(dataforseo.maxTasks || 0) <= 12 && Number(dataforseo.maxTasks || 0) >= stageOneTargets ? 'pass' : 'blocked',
    Blocking: 'yes',
    Evidence: `${Number(dataforseo.maxTasks || 0)} max tasks for ${stageOneTargets} pilot targets`
  },
  {
    Provider: 'DataForSEO',
    Stage: '1',
    Check: 'Review depth capped',
    Status: Number(dataforseo.maxReviewsPerAsin || 0) >= 100 && Number(dataforseo.maxReviewsPerAsin || 9999) <= 1000 ? 'pass' : 'blocked',
    Blocking: 'yes',
    Evidence: `${Number(dataforseo.maxReviewsPerAsin || 0)} reviews per ASIN (target range 100-1000)`
  },
  {
    Provider: 'DataForSEO',
    Stage: '1',
    Check: 'Hard stop after first run',
    Status: dataforseo.hardStopAfterFirstRun === true ? 'pass' : 'blocked',
    Blocking: 'yes',
    Evidence: `hardStopAfterFirstRun=${dataforseo.hardStopAfterFirstRun === true}`
  },
  {
    Provider: 'Apify',
    Stage: '2',
    Check: 'Budget lane enabled',
    Status: apifyBudget.enabled === true ? 'pass' : 'blocked',
    Blocking: 'yes',
    Evidence: apifyBudget.enabled === true ? 'enabled=true' : 'enabled=false in data/sources/budgets/source-acquisition-budget.json'
  },
  {
    Provider: 'Apify',
    Stage: '2',
    Check: 'Max spend recorded',
    Status: Number(apifyBudget.maxSpend || 0) > 0 ? 'pass' : 'blocked',
    Blocking: 'yes',
    Evidence: `${currency} ${Number(apifyBudget.maxSpend || 0)}`
  },
  {
    Provider: 'Apify',
    Stage: '2',
    Check: 'Actor and target caps preserve selective gap-fill',
    Status: Number(apifyBudget.maxActorRuns || 0) >= 1 && Number(apifyBudget.maxTargets || 0) >= firstBatchTargets && Number(apifyBudget.maxRequestsPerBrand || 0) >= 100 && Number(apifyBudget.maxRequestsPerBrand || 9999) <= 1000 ? 'pass' : 'blocked',
    Blocking: 'yes',
    Evidence: `${Number(apifyBudget.maxActorRuns || 0)} actor runs, ${Number(apifyBudget.maxTargets || 0)} targets, ${Number(apifyBudget.maxRequestsPerBrand || 0)} requests/brand for ${firstBatchTargets} first-batch targets (target range 100-1000)`
  },
  {
    Provider: 'Apify',
    Stage: '2',
    Check: 'Hard stop after first dataset',
    Status: apifyBudget.hardStopAfterFirstDataset === true ? 'pass' : 'blocked',
    Blocking: 'yes',
    Evidence: `hardStopAfterFirstDataset=${apifyBudget.hardStopAfterFirstDataset === true}`
  }
]

const dataforseoAllowed = rows.filter(row => row.Provider === 'DataForSEO').every(row => row.Status === 'pass')
const apifyAllowed = rows.filter(row => row.Provider === 'Apify').every(row => row.Status === 'pass')
const manifest = {
  generatedAt,
  ready: dataforseoAllowed && apifyAllowed,
  budgetAllowed: {
    dataforseoStage1: dataforseoAllowed,
    apifyStage2: apifyAllowed
  },
  budgetFile: 'data/sources/budgets/source-acquisition-budget.json',
  currency,
  counts: {
    checks: rows.length,
    passing: rows.filter(row => row.Status === 'pass').length,
    blocked: rows.filter(row => row.Status !== 'pass').length
  },
  checks: rows,
  nextActions: rows.filter(row => row.Status !== 'pass').map(row => `${row.Provider} Stage ${row.Stage}: ${row.Check} - ${row.Evidence}`),
  budget
}

writeFileSync(join(outRoot, 'source_acquisition_budget_gate.json'), JSON.stringify(manifest, null, 2) + '\n', 'utf8')
writeFileSync(join(outRoot, 'source_acquisition_budget_gate.csv'), csv(rows, ['Provider', 'Stage', 'Check', 'Status', 'Blocking', 'Evidence']), 'utf8')
writeFileSync(join(outRoot, 'source_acquisition_budget_gate.md'), [
  '# Source acquisition budget gate',
  '',
  `Generated: ${generatedAt}`,
  '',
  `DataForSEO Stage 1 budget allowed: ${dataforseoAllowed ? 'YES' : 'NO'}`,
  `Apify Stage 2 budget allowed: ${apifyAllowed ? 'YES' : 'NO'}`,
  '',
  'This gate prevents provider subscriptions or paid calls from starting without an explicit local budget cap. It does not store credentials or pricing assumptions.',
  '',
  `Budget file: \`${manifest.budgetFile}\``,
  '',
  '## Checks',
  ...rows.map(row => `- ${row.Status === 'pass' ? '[pass]' : '[blocked]'} ${row.Provider} Stage ${row.Stage}: ${row.Check} — ${row.Evidence}`),
  '',
  '## Rule',
  'A human approval file is not enough by itself. The matching budget lane must also be enabled with a max spend and hard-stop caps before `source:spend-gate` can allow paid provider usage.',
  ''
].join('\n'), 'utf8')

console.log(`Source acquisition budget gate ready: ${manifest.ready}`)
console.log(`DataForSEO budget allowed: ${dataforseoAllowed}`)
console.log(`Apify budget allowed: ${apifyAllowed}`)
console.log(`Checks passing: ${manifest.counts.passing}/${manifest.counts.checks}`)