{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Module 1: Language Detection\n",
    "\n",
    "This notebook documents the language detection module for the RAG-based mental health support chatbot. The goal is to classify the user question language so later modules can route retrieval and response generation correctly."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Approach\n",
    "\n",
    "- Use traditional NLP as required by the project brief.\n",
    "- Extract character-level TF-IDF features because language identity is strongly reflected in scripts, accents, and short character patterns.\n",
    "- Train a Multinomial Naive Bayes classifier because it is fast, explainable, and strong for sparse text features."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import sys\n",
    "from pathlib import Path\n",
    "\n",
    "PROJECT_ROOT = Path.cwd().resolve().parent if Path.cwd().name == 'notebooks' else Path.cwd().resolve()\n",
    "sys.path.append(str(PROJECT_ROOT / 'src' / 'models'))\n",
    "\n",
    "from language_classifier import LanguageDetector, DATA_DIR"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training character n-gram TF-IDF language detector...\n",
      "Validation accuracy: 99.35%\n",
      "Test accuracy: 99.40%\n",
      "Saved model to <PROJECT_ROOT>\\src\\models\\saved_lang_model.pkl\n",
      "Saved evaluation reports to <PROJECT_ROOT>\\reports\\module_1_language_detection\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.994"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "detector = LanguageDetector()\n",
    "results = detector.train(\n",
    "    train_path=DATA_DIR / 'lang_train.csv',\n",
    "    validation_path=DATA_DIR / 'lang_val.csv',\n",
    "    test_path=DATA_DIR / 'lang_test.csv',\n",
    ")\n",
    "results['test']['accuracy']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "I feel anxious and need help. -> {'language_code': 'en', 'language_name': 'English', 'confidence': 0.9971600701429593, 'is_confident': True, 'message': None}\n",
      "انا اشعر بالقلق واحتاج الى المساعدة. -> {'language_code': 'ar', 'language_name': 'Arabic', 'confidence': 0.9999992535818422, 'is_confident': True, 'message': None}\n",
      "Je me sens stresse aujourd'hui. -> {'language_code': 'fr', 'language_name': 'French', 'confidence': 0.9966388195110697, 'is_confident': True, 'message': None}\n"
     ]
    }
   ],
   "source": [
    "samples = [\n",
    "    'I feel anxious and need help.',\n",
    "    'انا اشعر بالقلق واحتاج الى المساعدة.',\n",
    "    \"Je me sens stresse aujourd'hui.\",\n",
    "]\n",
    "\n",
    "for sample in samples:\n",
    "    print(sample, '->', detector.predict_with_confidence(sample))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "base",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}