HTML Document Redaction for Web Content Privacy

DOM-Aware Redaction Processing

Our HTML processor builds a complete Document Object Model representation before making any modifications. This ensures structural integrity throughout the redaction process—tag pairs remain matched, attributes stay properly quoted, and character entities are correctly encoded for their context.

Context-Aware Encoding

Different HTML contexts require different character encoding. Our processor automatically applies the correct encoding based on where redacted content appears:

<a href="mailto:[email protected]">Contact John Smith</a>
<img src="/photos/john-smith.jpg" alt="John Smith, CEO">
<script>
  var userData = { name: "John Smith", email: "[email protected]" };
</script>


<a href="mailto:[EMAIL_REDACTED]">Contact [NAME_REDACTED]</a>
<img src="/photos/[REDACTED].jpg" alt="[NAME_REDACTED], CEO">
<script>
  var userData = { name: "[NAME_REDACTED]", email: "[EMAIL_REDACTED]" };
</script>

// Encoding applied:
// - Text content: HTML entities (&, <, >, etc.)
// - URLs: Percent encoding (%40 for @, %20 for space)
// - JavaScript: Unicode escapes (\u0040) or JSON escaping
// - Attribute values: HTML attribute encoding

Malformed HTML Recovery

Real-world HTML is often malformed—unclosed tags, improper nesting, missing quotes. Our parser implements HTML5's error recovery algorithms to build a valid DOM even from broken input, then produces well-formed output. The redacted document is often structurally cleaner than the original while maintaining visual fidelity.

Embedded Script and Structured Data Processing

Modern HTML documents embed significant data in script blocks—JavaScript configuration objects, JSON-LD for SEO, analytics initialization, and application state. This embedded data frequently contains PII that text-based redaction misses. RedactionAPI parses and processes these embedded formats while maintaining valid syntax.

// Original JSON-LD with author PII
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "Understanding Data Privacy",
  "author": {
    "@type": "Person",
    "name": "Dr. Jane Smith",
    "email": "[email protected]",
    "telephone": "+1-555-123-4567"
  },
  "publisher": {
    "@type": "Organization",
    "name": "Privacy Institute",
    "contactPoint": {
      "telephone": "+1-555-987-6543",
      "email": "[email protected]"
    }
  }
}
</script>

// Redacted JSON-LD preserving schema structure
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "Understanding Data Privacy",
  "author": {
    "@type": "Person",
    "name": "[NAME_REDACTED]",
    "email": "[EMAIL_REDACTED]",
    "telephone": "[PHONE_REDACTED]"
  },
  "publisher": {
    "@type": "Organization",
    "name": "Privacy Institute",
    "contactPoint": {
      "telephone": "[PHONE_REDACTED]",
      "email": "[EMAIL_REDACTED]"
    }
  }
}
</script>

Inline JavaScript Processing

JavaScript blocks may contain user data in variable assignments, configuration objects, or template literals:

CSS Content Processing

CSS content properties and generated content can contain PII, particularly in before/after pseudo-elements used for decorative text or accessibility labels. Our processor scans inline styles and embedded stylesheets for content declarations containing sensitive data.

Security and XSS Prevention

Processing untrusted HTML requires careful attention to security. Our redaction process not only protects PII but also ensures the output is safe to render—properly escaped to prevent cross-site scripting attacks. Whether the original content was benign or contained malicious payloads, the redacted output is safe for browser display.

Content Security Policy Compatibility

Our HTML output is compatible with strict Content Security Policy headers. Redaction markers use plain text that doesn't require unsafe-inline for scripts or styles. Organizations can serve redacted documents with restrictive CSP headers while maintaining full functionality.

// Security-focused HTML redaction configuration
{
  "security": {
    "sanitize_output": true,
    "strip_scripts": false,           // Keep scripts, sanitize content
    "strip_event_handlers": true,     // Remove onclick, onerror, etc.
    "strip_javascript_urls": true,    // Remove javascript: hrefs
    "strip_data_urls": false,         // Keep data: for images (optional)
    "encode_entities": true,          // Always encode special chars
    "normalize_unicode": true,        // Prevent homograph attacks
    "validate_urls": true,            // Check URL format validity
    "csp_compatible": true            // Ensure CSP-friendly output
  }
}