Detox ✓

Web-Based Automata Intelligence for Toxic Content Detection
Bersihkan toksikmu

---

📋 Deskripsi

Detox adalah sistem moderasi konten berbasis rule engine yang menggunakan tiga layer pemrosesan:

1. NFA — Normalisasi slang dan variasi ejaan (gblk → goblok)
2. Regex — Pendefinisian pattern toxic secara formal
3. DFA — Eksekusi matching deterministik O(n)

Dikembangkan sebagai project mata kuliah Formal Language and Automata (FLA)
Group Greenflag · President University Pekanbaru · 2026

---

🗂️ Struktur Project

detox/
├── frontend/                 # SvelteKit
│   ├── src/
│   │   ├── routes/
│   │   │   ├── +page.svelte          # Home / Analyzer
│   │   │   ├── bulk/
│   │   │   │   └── +page.svelte      # Bulk Checker
│   │   │   └── riwayat/
│   │   │       └── +page.svelte      # Riwayat
│   │   ├── lib/
│   │   │   ├── components/
│   │   │   │   ├── Navbar.svelte
│   │   │   │   ├── Footer.svelte
│   │   │   │   ├── ResultPanel.svelte
│   │   │   │   └── StatBar.svelte
│   │   │   └── api.js               # Fetch helper ke backend
│   │   └── app.css                  # Global styles + CSS vars
│   ├── package.json
│   ├── svelte.config.js
│   └── vite.config.js
│
├── backend/                  # FastAPI
│   ├── main.py               # Entry point + routes
│   ├── engine/
│   │   ├── nfa.py            # NFA normalization engine
│   │   ├── regex_patterns.py # Regex pattern definitions
│   │   └── dfa.py            # DFA matching engine
│   ├── lexicon/
│   │   ├── slang.json        # Slang → canonical mapping
│   │   └── toxic.json        # Toxic word patterns
│   └── requirements.txt
│
├── docs/
│   ├── diagrams.md           # Mermaid diagrams
│   └── design.md             # Design system
│
└── README.md

---

🚀 Cara Setup

Prerequisites

Pastikan sudah terinstall:

• Node.js v18+
• Python 3.10+
• npm (sudah include dengan Node.js)

---

1. Clone Repository

git clone https://github.com/username/detox.git
cd detox

---

2. Setup Backend (FastAPI)

# Masuk ke folder backend
cd backend

# Buat virtual environment
python -m venv venv

# Aktifkan virtual environment
# Windows:
venv\Scripts\activate
# Mac/Linux:
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Jalankan backend
uvicorn main:app --reload --port 8000

Backend berjalan di: http://localhost:8000
API docs: http://localhost:8000/docs

---

3. Setup Frontend (SvelteKit)

Buka terminal baru:

# Masuk ke folder frontend
cd frontend

# Install dependencies
npm install

# Jalankan frontend
npm run dev

Frontend berjalan di: http://localhost:5173

---

4. Environment Variables

Buat file .env di folder frontend/:

VITE_API_URL=http://localhost:8000

---

📡 API Endpoints

POST /analyze

Analisis satu teks.

Request:

{
  "text": "lo gblk bngst sih"
}

Response:

{
  "original": "lo gblk bngst sih",
  "normalized": "lo goblok bangsat sih",
  "result": "TOXIC",
  "detected_words": ["goblok", "bangsat"],
  "trace": {
    "nfa_steps": ["gblk → goblok", "bngst → bangsat"],
    "dfa_path": "q0 → q1 → q6 → q11 → q12 → q13 → qTOXIC"
  }
}

---

POST /analyze/bulk

Analisis banyak teks sekaligus.

Request:

{
  "texts": [
    "lo gblk bngst sih",
    "makasih udh bantu gw"
  ]
}

Response:

{
  "total": 2,
  "toxic_count": 1,
  "safe_count": 1,
  "results": [
    {
      "text": "lo gblk bngst sih",
      "normalized": "lo goblok bangsat sih",
      "result": "TOXIC",
      "detected_words": ["goblok", "bangsat"]
    },
    {
      "text": "makasih udh bantu gw",
      "normalized": "makasih sudah bantu aku",
      "result": "SAFE",
      "detected_words": []
    }
  ]
}

---

📦 Dependencies

Frontend

{
  "@sveltejs/kit": "^2.0.0",
  "svelte": "^4.0.0",
  "vite": "^5.0.0"
}

Backend

fastapi==0.110.0
uvicorn==0.29.0
python-multipart==0.0.9
pydantic==2.6.0

---

👥 Tim Pengembang

Nama	NIM	Role
Cut Sarah Alisa	030202400012	NFA Engine + Slang Lexicon
M. Rizal Basri	030202400001	Backend API + DFA Engine
Wesnita Ruth Angie	030202400019	Frontend SvelteKit + UI

---

📄 Lisensi

Project ini dibuat untuk keperluan akademik.
President University Pekanbaru · Informasi Teknologi · 2026

---

Lecturer: Shella Eldwina Fitri S.T., M.Eng