database.py

"""
Comprehensive Homeopathic Medicine Database with Common Misspellings
"""

import json
from typing import Dict, List, Tuple

# Main remedy database with common misspellings
REMEDY_DATABASE = {
    "Arnica Montana": {
        "correct_name": "Arnica Montana",
        "common_names": ["Arnica", "Mountain Tobacco", "Leopard's Bane"],
        "common_misspellings": ["Arnika", "Arnicha", "Arnoca", "Arnicum"],
        "description": "First remedy for trauma, injuries, and shock",
        "indications": ["Physical trauma", "Bruising", "Muscle soreness", "Post-surgical", "Shock"],
        "potencies": ["6C", "30C", "200C", "1M"],
        "dosage": "3-5 pellets every 2-4 hours for acute"
    },
    "Belladonna": {
        "correct_name": "Belladonna",
        "common_names": ["Deadly Nightshade"],
        "common_misspellings": ["Belladona", "Beladona", "Belladone"],
        "description": "Sudden violent onset with intense symptoms",
        "indications": ["Sudden fever", "Throbbing headache", "Acute inflammation"],
        "potencies": ["30C", "200C", "1M"],
        "dosage": "Every 1-2 hours in acute fever"
    },
    "Argentum Nitricum": {
        "correct_name": "Argentum Nitricum",
        "common_names": ["Silver Nitrate"],
        "common_misspellings": [
            "Argentic Nitricum", "Argentum Nitric", "Argent Nitricum",
            "Argentum Nitrate", "Argentic Nitrate", "Argent Nitric"
        ],
        "description": "Anxiety, anticipation, digestive issues",
        "indications": ["Anxiety", "Stage fright", "Diarrhea from anxiety", "Vertigo"],
        "potencies": ["30C", "200C", "1M"],
        "dosage": "3 times daily for anxiety"
    },
    "Nux Vomica": {
        "correct_name": "Nux Vomica",
        "common_names": ["Poison Nut", "Quaker Buttons"],
        "common_misspellings": ["Nuxvomica", "Nux Vomic", "Nuxvomic", "Nux Vom"],
        "description": "Irritable, impatient, digestive complaints",
        "indications": ["Digestive disorders", "Overindulgence", "Insomnia", "Irritability"],
        "potencies": ["6C", "30C", "200C"],
        "dosage": "3 times daily for digestive issues"
    },
    "Pulsatilla": {
        "correct_name": "Pulsatilla",
        "common_names": ["Wind Flower", "Pasque Flower"],
        "common_misspellings": ["Pulsatila", "Pulsitilla", "Pulsitila", "Pulsatella"],
        "description": "Mild, yielding, changeable symptoms",
        "indications": ["Changeable symptoms", "Digestive issues", "Colds", "Weepy"],
        "potencies": ["6C", "30C", "200C"],
        "dosage": "Every 4 hours for acute"
    },
    # Additional 15 common remedies
    "Bryonia Alba": {
        "correct_name": "Bryonia Alba",
        "common_names": ["White Bryony", "Wild Hops"],
        "common_misspellings": ["Briyonia", "Bryonia", "Bironia", "Bryonia Alba"],
        "description": "Irritable, thirsty, worse from motion",
        "indications": ["Dry mucous", "Joint pains", "Constipation", "Pleurisy"],
        "potencies": ["6C", "30C", "200C"],
        "dosage": "Every 4 hours"
    },
    "Aconitum Napellus": {
        "correct_name": "Aconitum Napellus",
        "common_names": ["Monkshood", "Aconite"],
        "common_misspellings": ["Aconite", "Aconitum", "Aconite Napellus", "Aconiteum"],
        "description": "Sudden onset after fright, intense anxiety",
        "indications": ["Sudden fever", "Panic attacks", "Croup", "Exposure"],
        "potencies": ["30C", "200C", "1M"],
        "dosage": "Every 30 minutes in acute panic"
    },
    "Chamomilla": {
        "correct_name": "Chamomilla",
        "common_names": ["German Chamomile"],
        "common_misspellings": ["Camomilla", "Chamomile", "Chamomila", "Camomile"],
        "description": "Extreme irritability, oversensitivity to pain",
        "indications": ["Teething", "Colic", "Irritability", "Childbirth pain"],
        "potencies": ["6C", "30C", "200C"],
        "dosage": "Every 15-30 minutes for acute pain"
    },
    "Gelsemium": {
        "correct_name": "Gelsemium",
        "common_names": ["Yellow Jasmine"],
        "common_misspellings": ["Gelsem", "Gelsimium", "Gelseminum", "Gelsemia"],
        "description": "Weakness, trembling, anticipatory anxiety",
        "indications": ["Flu", "Stage fright", "Headache", "Weakness"],
        "potencies": ["30C", "200C", "1M"],
        "dosage": "Every 2-4 hours"
    },
    "Ignatia": {
        "correct_name": "Ignatia",
        "common_names": ["St. Ignatius Bean"],
        "common_misspellings": ["Ignatia Amara", "Ignat", "Ignatia", "Ignatiam"],
        "description": "Acute grief, hysteria, contradictory symptoms",
        "indications": ["Grief", "Hysteria", "Headache from grief", "Sighing"],
        "potencies": ["30C", "200C", "1M"],
        "dosage": "As needed for emotional distress"
    },
    "Phosphorus": {
        "correct_name": "Phosphorus",
        "common_names": [],
        "common_misspellings": ["Phosporus", "Phos", "Phosphorous", "Phosph"],
        "description": "Bleeding, fear, thirst for cold drinks",
        "indications": ["Bleeding", "Anxiety", "Pneumonia", "Burning pains"],
        "potencies": ["30C", "200C", "1M"],
        "dosage": "3 times daily"
    },
    "Sepia": {
        "correct_name": "Sepia",
        "common_names": ["Cuttlefish Ink"],
        "common_misspellings": ["Sep", "Sepiah", "Sapia", "Seppia"],
        "description": "Indifference, bearing-down sensation",
        "indications": ["Menopausal symptoms", "Indifference", "Prolapse", "Headache"],
        "potencies": ["30C", "200C", "1M"],
        "dosage": "Once daily or weekly"
    },
    "Silicea": {
        "correct_name": "Silicea",
        "common_names": ["Silica", "Flint"],
        "common_misspellings": ["Silica", "Silice", "Silicaea", "Sil"],
        "description": "Lack of confidence, suppurations, sensitive to cold",
        "indications": ["Abscesses", "Lack of confidence", "Sweaty feet", "Headache"],
        "potencies": ["30C", "200C", "1M"],
        "dosage": "Once weekly"
    },
    "Sulphur": {
        "correct_name": "Sulphur",
        "common_names": ["Brimstone"],
        "common_misspellings": ["Sulfur", "Sulpur", "Sulfer", "Sulph"],
        "description": "Philosophical, untidy, skin complaints",
        "indications": ["Skin disorders", "Itching", "Philosophical", "Hungry at 11am"],
        "potencies": ["30C", "200C", "1M"],
        "dosage": "Once weekly"
    },
    "Thuja": {
        "correct_name": "Thuja",
        "common_names": ["Arbor Vitae"],
        "common_misspellings": ["Thuga", "Thuja Occidentalis", "Thuje", "Thuya"],
        "description": "Fixed ideas, warts, vaccination effects",
        "indications": ["Warts", "Vaccination effects", "Fixed ideas", "Skin growths"],
        "potencies": ["30C", "200C", "1M"],
        "dosage": "Once weekly"
    }
}

def search_remedy_by_name(query: str, use_spell_check: bool = True) -> List[Tuple[str, Dict, float]]:
    """
    Search for remedies by name with spell correction
    Returns: [(remedy_name, remedy_data, confidence_score)]
    """
    query = query.strip().lower()
    if not query or len(query) < 2:
        return []
    
    results = []
    
    for remedy_name, remedy_data in REMEDY_DATABASE.items():
        confidence = 0
        
        # 1. Exact match (highest confidence)
        if query == remedy_name.lower():
            confidence = 100
        
        # 2. Partial match in correct name
        elif query in remedy_name.lower():
            confidence = 80
        
        # 3. Match in common names
        elif any(query in name.lower() for name in remedy_data.get("common_names", [])):
            confidence = 70
        
        # 4. Spell correction matching
        elif use_spell_check and _spell_correction_match(query, remedy_name, remedy_data):
            confidence = _calculate_spell_confidence(query, remedy_name, remedy_data)
        
        if confidence > 0:
            results.append((remedy_name, remedy_data, confidence))
    
    # Sort by confidence
    results.sort(key=lambda x: x[2], reverse=True)
    return results[:10]

def _spell_correction_match(query: str, remedy_name: str, remedy_data: Dict) -> bool:
    """Check if query matches through spell correction"""
    # Check common misspellings
    if query in [m.lower() for m in remedy_data.get("common_misspellings", [])]:
        return True
    
    # Simple Levenshtein-like check for short remedies
    if len(query) > 3 and len(remedy_name) > 3:
        # Check if query is close to remedy name
        if _string_similarity(query, remedy_name.lower()) > 0.7:
            return True
    
    return False

def _calculate_spell_confidence(query: str, remedy_name: str, remedy_data: Dict) -> float:
    """Calculate confidence score for spell-corrected matches"""
    # Exact misspelling match
    if query in [m.lower() for m in remedy_data.get("common_misspellings", [])]:
        return 90
    
    # Calculate string similarity
    similarity = _string_similarity(query, remedy_name.lower())
    
    if similarity > 0.9:
        return 85
    elif similarity > 0.8:
        return 75
    elif similarity > 0.7:
        return 65
    else:
        return 50

def _string_similarity(s1: str, s2: str) -> float:
    """Calculate simple string similarity"""
    if not s1 or not s2:
        return 0
    
    # Common prefix length
    prefix_len = 0
    for i in range(min(len(s1), len(s2))):
        if s1[i] == s2[i]:
            prefix_len += 1
        else:
            break
    
    return prefix_len / max(len(s1), len(s2))

def get_remedy_details(name: str) -> Dict:
    """Get complete remedy details"""
    return REMEDY_DATABASE.get(name)

def get_all_remedies() -> List[str]:
    """Get list of all remedy names"""
    return list(REMEDY_DATABASE.keys())