import json
import os
from collections import Counter, defaultdict

MEMORY_FILE = "data/corpora/memory_cards.jsonl"

def analyze_memory():
    if not os.path.exists(MEMORY_FILE):
        print(f"Error: {MEMORY_FILE} not found.")
        return

    print(f"Analyzing {MEMORY_FILE}...")
    
    total_cards = 0
    user_counts = Counter()
    content_hashes = defaultdict(int)
    user_content_hashes = defaultdict(int)
    
    with open(MEMORY_FILE, "r", encoding="utf-8") as f:
        for line in f:
            if not line.strip(): continue
            try:
                card = json.loads(line)
                total_cards += 1
                uid = card.get("user_id", "unknown")
                text = card.get("note_text", "").strip()
                
                user_counts[uid] += 1
                content_hashes[text] += 1
                user_content_hashes[(uid, text)] += 1
            except:
                pass

    print("\n" + "="*40)
    print("MEMORY STORE ANALYSIS")
    print("="*40)
    print(f"Total Cards: {total_cards}")
    print(f"Unique Users: {len(user_counts)}")
    print("-" * 40)
    
    print("\nTop 10 Users by Card Count:")
    for uid, count in user_counts.most_common(10):
        print(f"  {uid}: {count}")
        
    print("\nTop 10 Most Frequent Contents (Global):")
    sorted_content = sorted(content_hashes.items(), key=lambda x: x[1], reverse=True)[:10]
    for text, count in sorted_content:
        display_text = (text[:50] + '...') if len(text) > 50 else text
        print(f"  [{count}] {display_text}")

    print("\nTop 10 Most Frequent (User, Content) Duplicates:")
    sorted_user_content = sorted(user_content_hashes.items(), key=lambda x: x[1], reverse=True)[:10]
    for (uid, text), count in sorted_user_content:
        display_text = (text[:50] + '...') if len(text) > 50 else text
        print(f"  [{count}] User: {uid} | {display_text}")

    print("="*40)

if __name__ == "__main__":
    analyze_memory()