Mofilo Content Filter #

Multi-layer content filtering for Flutter apps. Blocks offensive language, hate speech, and harmful content with 12+ detection layers.

Quick Start #

import 'package:mofilo_content_filter/mofilo_content_filter.dart';

// Check any text input
final result = ContentFilter.check(userInput);

if (result.isBlocked) {
  showError('This content is not allowed');
} else {
  submitContent(userInput);
}

That's it! The filter handles the rest.

Features #

• 🛡️ 12+ Detection Layers - Catches sophisticated bypass attempts
• 🌍 Multi-Language - English, Spanish, French, German, Polish, Russian, Portuguese, Italian, Dutch
• ⚡ Fast & Offline - <5ms latency, no external API calls
• 🔒 Privacy-Friendly - All processing happens locally
• 🎯 Comprehensive - Profanity, slurs, hate speech, violence, CSAM, terrorism
• 🧪 Well-Tested - 870+ comprehensive tests

Installation #

Add to your pubspec.yaml:

dependencies:
  mofilo_content_filter: ^1.0.3

Then run:

flutter pub get

Usage Examples #

Basic Usage (Recommended) #

import 'package:mofilo_content_filter/mofilo_content_filter.dart';

final result = ContentFilter.check("chocolate cake");

if (result.isBlocked) {
  print('Blocked: ${result.reason}');
  print('Severity: ${result.severity}');
} else {
  print('Content is safe!');
}

Validate Food Names #

final result = ContentFilter.validateFoodName("apple pie");

if (result.isAllowed) {
  saveFoodName("apple pie");
}

Validate Multiple Fields #

final results = ContentFilter.validateAll(
  foodName: "pizza",
  brandName: "Pizza Hut",
);

if (results.isEmpty) {
  // All fields are safe
  saveToDatabase();
} else {
  // Show which fields failed
  results.forEach((field, result) {
    print('$field: ${result.reason}');
  });
}

What Gets Blocked #

Language-Based Content #

• Profanity and vulgar language
• Racial, ethnic, and religious slurs
• Homophobic and transphobic slurs
• Sexist and misogynistic language
• Ableist slurs

Harmful Content #

• Hate speech and Nazi/white supremacy symbols
• Terrorism and extremism references
• Self-harm and suicide encouragement
• Child abuse (CSAM) references
• Sexual harassment and violence
• Drug references

Technical Attacks #

• SQL injection attempts
• XSS (Cross-Site Scripting)
• Spam and gibberish patterns
• Encoded malicious content

Advanced Bypass Detection #

Defends against sophisticated bypass techniques:

1. Character Substitution - l33tspeak (f0ck, sh1t), symbols (@ss, f*ck)
2. Spacing/Separators - f.u.c.k, f-u-c-k, f u c k
3. Unicode Tricks - Cyrillic (фuck), fullwidth (ｆｕｃｋ), circled (ⓕⓤⓒⓚ)
4. Zero-Width Characters - Invisible Unicode characters
5. Emojis - f😀u😀c😀k
6. Encoding - Base64, ROT13, Hex, reversed text
7. Mixed Case - FuCk, FUCK, fUcK
8. Asterisks - fk, ner
9. Homoglyphs - Visually similar characters
10. Fuzzy Spelling - Levenshtein distance matching

Best Practices & Limitations #

✅ Best Used For: #

• Food names and brand names
• Product titles and descriptions
• Usernames and display names
• Form inputs with limited scope
• First-pass content filtering

⚠️ Important to Know: #

This is a first-line defense tool. Like all automated filters, it's not perfect:

• Not 100% accurate - May produce false positives or miss some content
• Context matters - Best for names/titles, not ideal for full conversations
• Human review recommended - Especially for chat, comments, and user-generated content
• Best practices - Use as your first filter, add human moderation as backup

For comprehensive content moderation, combine this filter with:

• User reporting system
• Human review queue for flagged content
• Regular monitoring and updates

See DISCLAIMER.md for complete details.

Result Object #

class ContentFilterResult {
  final bool isAllowed;          // Whether content passes filter
  final bool isBlocked;           // Opposite of isAllowed
  final String? reason;           // Reason for blocking (if blocked)
  final List<String> violations;  // Specific violations found
  final ContentFilterSeverity severity;
}

enum ContentFilterSeverity {
  none,      // Content is safe
  low,       // Suspicious but might be legitimate
  medium,    // Likely violation
  high,      // Definite violation
  critical,  // Severe violation (slurs, hate speech, CSAM)
}

Performance #

• ⚡ Latency: <5ms on average
• 🔒 Privacy: All processing happens locally, no external APIs
• 📱 Offline: Works without internet connection
• 🧪 Tested: 870+ comprehensive tests across bypass techniques

Use Cases #

Companies Using This #

Great for apps that need content filtering for:

• Food & recipe apps (filter inappropriate food names)
• E-commerce platforms (filter product names)
• Social platforms (filter usernames, bio text)
• Gaming platforms (filter clan names, character names)
• Business directories (filter business names)

License & Legal #

MIT License - Copyright (c) 2025 Mofilo

This software is provided "AS IS" without warranty of any kind. See LICENSE for full terms.

Important: By using this package, you accept full responsibility for content moderation in your application. See DISCLAIMER.md for complete legal terms and limitations.