Operations: Dump Compliance

This page is the practical runbook for validating SQL dumps against plaintext PII leakage.

Validate SQL database dumps before sharing, restoring, or promoting them to production. This runbook helps ensure no plaintext PII leaked into a dump that should only contain Veilio tokens (tok_...).

What this checks

The compliance scanner scans dump text for patterns that indicate plaintext sensitive data:

Rule ID	Detects
`email-plaintext`	Email addresses
`phone-e164-fr-like`	Phone numbers (E.164 / FR formats)
`iban-like`	IBAN-like values
`credit-card-like`	Card numbers (13–19 digits)
`ssn-like`	US SSN format (`###-##-####`)

Excluded by default: Veilio tokens matching tok_[A-Za-z0-9_-]{10,}.

Any match above the allowed threshold → FAIL.

Quick start (standalone script)

No Veilio repository access required. You need Node.js 18+.

1) Save the policy file

Create dump-policy.json:

json

{
  "description": "Dump compliance policy. Any match above maxAllowed fails the check.",
  "maxFindingsPreview": 20,
  "rules": [
    {
      "id": "email-plaintext",
      "description": "Detect plaintext email addresses",
      "regex": "\\b[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}\\b",
      "maxAllowed": 0
    },
    {
      "id": "phone-e164-fr-like",
      "description": "Detect phone numbers likely in E.164/FR formats",
      "regex": "(\\+\\d{6,15}|\\b0[1-9](?:[ .-]?\\d{2}){4}\\b)",
      "maxAllowed": 0
    },
    {
      "id": "iban-like",
      "description": "Detect IBAN-like values",
      "regex": "\\b[A-Z]{2}\\d{2}[A-Z0-9]{11,30}\\b",
      "maxAllowed": 0
    },
    {
      "id": "credit-card-like",
      "description": "Detect possible card numbers",
      "regex": "\\b(?:\\d[ -]*?){13,19}\\b",
      "maxAllowed": 0
    },
    {
      "id": "ssn-like",
      "description": "Detect SSN-like US format",
      "regex": "\\b\\d{3}-\\d{2}-\\d{4}\\b",
      "maxAllowed": 0
    }
  ],
  "exclusions": [
    {
      "id": "token-prefix",
      "description": "Ignore Veilio tokenized values",
      "regex": "\\btok_[A-Za-z0-9_-]{10,}\\b"
    }
  ]
}

2) Save the checker script

Create check-dump-pii.mjs:

javascript

#!/usr/bin/env node
import fs from "fs";

const [dumpPath, policyPath = "dump-policy.json"] = process.argv.slice(2);
if (!dumpPath) {
  console.error("Usage: node check-dump-pii.mjs <dump.sql> [dump-policy.json]");
  process.exit(1);
}

const policy = JSON.parse(fs.readFileSync(policyPath, "utf8"));
const text = fs.readFileSync(dumpPath, "utf8");

function stripExclusions(input, exclusions) {
  let out = input;
  for (const ex of exclusions ?? []) {
    out = out.replace(new RegExp(ex.regex, "g"), "");
  }
  return out;
}

const scrubbed = stripExclusions(text, policy.exclusions);
const findings = [];

for (const rule of policy.rules ?? []) {
  const re = new RegExp(rule.regex, "g");
  const matches = [...scrubbed.matchAll(re)];
  const count = matches.length;
  const maxAllowed = rule.maxAllowed ?? 0;
  if (count > maxAllowed) {
    findings.push({
      id: rule.id,
      description: rule.description,
      count,
      maxAllowed,
      samples: matches.slice(0, 5).map((m) => m[0]),
    });
  }
}

if (findings.length > 0) {
  console.error("FAIL — plaintext PII detected:");
  console.error(JSON.stringify(findings, null, 2));
  process.exit(2);
}

console.log("PASS — no disallowed plaintext findings");
process.exit(0);

3) Run the check

bash

node check-dump-pii.mjs /path/to/dump.sql dump-policy.json

Exit codes

Code	Meaning
`0`	PASS — no disallowed plaintext findings
`2`	FAIL — policy violation (plaintext PII found)
`1`	Script / config / runtime error

Suggested CI gate

Generate dump artifact (e.g. pg_dump).
Run node check-dump-pii.mjs dump.sql.
Block pipeline on non-zero exit code.

GitHub Actions example

yaml

- name: Check dump for plaintext PII
  run: node check-dump-pii.mjs artifacts/db-dump.sql dump-policy.json

Incident handling

If the checker fails:

Stop dump promotion or import immediately.
Identify the table / column source of plaintext values.
Verify tokenization is enforced at the ingestion path (write path must call /tokenize before DB insert).
Re-run dump generation and checker after fix.

Customizing the policy

Add rules for national ID formats, passport numbers, or internal identifiers.
Adjust maxAllowed for known false positives (e.g. test data patterns).
Add exclusions regexes for hashed values or known-safe placeholders.

Contact support@veilio.fr if you need help tuning policies for your schema.