Create utils.py

utils.py

@@ -0,0 +1,67 @@
+"""
+Utility functions for SEA Prep Tutor
+"""
+import re
+import json
+from typing import List, Dict, Optional
+
+def clean_text(text: str) -> str:
+    """Clean and normalize text"""
+    # Remove extra whitespace
+    text = re.sub(r'\s+', ' ', text)
+    # Remove special characters but keep basic punctuation
+    text = re.sub(r'[^\w\s\.\?\!,;:\-\(\)]', '', text)
+    return text.strip()
+
+def parse_mcq_options(text: str) -> Optional[Dict]:
+    """Parse MCQ options from text"""
+    options = {}
+    
+    # Pattern for A), B), etc.
+    pattern = r'([A-D])[\)\.]\s*([^A-D]+?)(?=\s+[A-D][\)\.]|\s*$)'
+    matches = re.findall(pattern, text, re.IGNORECASE | re.DOTALL)
+    
+    for letter, content in matches:
+        content = content.strip()
+        if content and len(content) > 1:
+            options[letter.upper()] = content
+    
+    return options if options else None
+
+def calculate_difficulty(text: str, subject: str) -> int:
+    """Calculate question difficulty (1-5)"""
+    word_count = len(text.split())
+    sentence_count = len(re.split(r'[.!?]', text))
+    
+    base_score = min(5, word_count // 20 + sentence_count // 2)
+    
+    if subject == "Math":
+        # Additional factors for math
+        numbers = len(re.findall(r'\d+', text))
+        operations = len(re.findall(r'[+\-×÷=]', text))
+        base_score += min(2, (numbers + operations) // 3)
+    
+    return min(5, max(1, base_score))
+
+def format_test_for_display(test: Dict) -> str:
+    """Format test as HTML for display"""
+    html = f"""
+    <div style="padding: 20px; background: #f8f9fa; border-radius: 10px;">
+        <h3>{test.get('title', 'Practice Test')}</h3>
+        <p><strong>Date:</strong> {test.get('date', '')}</p>
+        <p><strong>Questions:</strong> {len(test.get('questions', []))}</p>
+        <p><strong>Time:</strong> {test.get('total_time', '60 minutes')}</p>
+        <hr>
+    """
+    
+    for i, q in enumerate(test.get('questions', []), 1):
+        html += f"""
+        <div style="margin: 15px 0; padding: 15px; background: white; border-radius: 5px;">
+            <h4>Question {i}</h4>
+            <p>{q.get('text', '')}</p>
+            <p><small>Type: {q.get('type', '')} | Difficulty: {q.get('difficulty_label', 'Medium')}</small></p>
+        </div>
+        """
+    
+    html += "</div>"
+    return html