@ogcio/o11y-sdk-react/lib/internals/pii-detection.ts

Opentelemetry standard instrumentation SDK for React based project

import { faro, TransportItemType } from "@grafana/faro-web-sdk";

const EMAIL_REGEX = /[a-zA-Z0-9._%+-]+@([a-zA-Z0-9.-]+\.[a-z]{2,})/gi;

/**
 * Redacts all email addresses in the input string and collects metadata.
 *
 * @param {string} value The input string potentially containing email addresses.
 * @returns {{
 *   redacted: string,
 *   count: number,
 *   domains: Record<string, number>
 * }}
 *
 * An object containing:
 *   - `redacted`: the string with email addresses replaced by `[REDACTED EMAIL]`
 *   - `count`: total number of email addresses redacted
 *   - `domains`: a map of domain names to the number of times they were redacted
 */
function _redactEmails(value: string): {
  redacted: string;
  count: number;
  domains: Record<string, number>;
} {
  let count = 0;
  const domains: Record<string, number> = {};

  const redacted = value.replace(EMAIL_REGEX, (_, domain) => {
    count++;
    domains[domain] = (domains[domain] || 0) + 1;
    return "[REDACTED EMAIL]";
  });

  return { redacted, count, domains };
}

/**
 * Checks whether a string contains URI-encoded components.
 *
 * @param {string} value - The string to inspect.
 * @returns {boolean} `true` if the string is encoded, `false` otherwise.
 */
function _containsEncodedComponents(value: string) {
  try {
    return decodeURI(value) !== decodeURIComponent(value);
  } catch {
    return false;
  }
}

/**
 * Cleans a string by redacting email addresses and emitting metrics for PII.
 *
 * If the string is URL-encoded, it will be decoded before redaction.
 * Metrics are emitted for each domain found in redacted email addresses.
 *
 * @param {string} value - The input string to sanitize.
 * @param {TransportItemType} source - The source context of the input.
 * @returns {string} The cleaned string with any email addresses replaced by `[REDACTED EMAIL]`.
 */
export function _cleanStringPII(
  value: string,
  source: TransportItemType,
): string {
  let kind: "string" | "url" = "string";
  let decodedValue = value;

  if (_containsEncodedComponents(value)) {
    decodedValue = decodeURIComponent(value);
    kind = "url";
  }

  const { redacted, count, domains } = _redactEmails(decodedValue);

  if (count > 0) {
    for (const [domain, domainCount] of Object.entries(domains)) {
      faro.api.pushMeasurement(
        {
          type: "faro_o11y_pii_redaction",
          values: {
            redacted: domainCount,
          },
        },
        {
          context: {
            pii_type: "email",
            redaction_source: source,
            pii_email_domain: domain,
            pii_format: kind,
          },
        },
      );
    }
  }
  return redacted;
}