import tkinter as tk
from tkinter import ttk, messagebox
import pandas as pd
import numpy as np
from sklearn.decomposition import PCA
from sklearn.preprocessing import StandardScaler
import json
import os

# --- 設定 ---
DATA_FILE = "sample.cvd"

class MatchingApp:
    def __init__(self, root):
        self.root = root
        self.root.title("AI Compatibility Matcher")
        self.root.geometry("600x700")
        
        # 1. データの読み込みと前処理
        self.load_and_process_data()
        
        # 2. UIの構築
        self.create_widgets()
        
        # 3. 初期表示
        self.update_student_list()

    def load_and_process_data(self):
        # --- データ読み込み (main.pyと同様のロジック) ---
        if os.path.exists(DATA_FILE):
            try:
                with open(DATA_FILE, 'r', encoding='utf-8') as f:
                    raw_data = json.load(f)
            except Exception:
                raw_data = self.get_default_data()
        else:
            raw_data = self.get_default_data()
            
        self.df = pd.DataFrame(raw_data)
        
        # --- PCA計算 (6次元) ---
        features = [
            'Score_Math_Logic', 'Score_English', 'Score_Science', 'Score_Humanities',
            'Extraversion', 'Sensitivity', 'Independence', 'Gender_Code'
        ]
        
        # データが少なすぎる場合のガード
        if len(self.df) < 6:
            messagebox.showerror("Error", "データが少なすぎて分析できません（最低6人必要です）")
            self.root.destroy()
            return

        # 標準化
        scaler = StandardScaler()
        df_scaled = scaler.fit_transform(self.df[features])
        
        # PCA (6成分)
        # 1-3: 位置 (x, y, z), 4-6: ベクトル (u, v, w)
        pca = PCA(n_components=6)
        components = pca.fit_transform(df_scaled)
        
        self.df[['x', 'y', 'z', 'u', 'v', 'w']] = components
        
        # 生徒と講師に分離して保持
        self.students = self.df[self.df['Type'] == 'Student'].copy()
        self.teachers = self.df[self.df['Type'] == 'Teacher'].copy()

    def get_default_data(self):
        # デフォルトデータ（ファイルがない場合用）
        return {
            'Name': ['講師A', '生徒1'], 
            'Type': ['Teacher', 'Student'],
            'Score_Math_Logic': [0.9, 0.1], 'Score_English': [0.9, 0.1],
            'Score_Science': [0.9, 0.1], 'Score_Humanities': [0.9, 0.1],
            'Extraversion': [0.5, 0.5], 'Sensitivity': [0.5, 0.5],
            'Independence': [0.5, 0.5], 'Gender_Code': [1.0, 1.0]
        }

    def create_widgets(self):
        # --- 上部: コントロールエリア ---
        control_frame = tk.LabelFrame(self.root, text="検索条件", padx=10, pady=10)
        control_frame.pack(fill=tk.X, padx=10, pady=10)
        
        # 生徒選択
        tk.Label(control_frame, text="対象の生徒を選択:").grid(row=0, column=0, sticky="w")
        self.student_var = tk.StringVar()
        self.student_combo = ttk.Combobox(control_frame, textvariable=self.student_var, state="readonly", width=30)
        self.student_combo.grid(row=0, column=1, padx=10, sticky="w")
        self.student_combo.bind("<<ComboboxSelected>>", self.calculate_match)
        
        # アルファ値 (隠れた特性の重み)
        tk.Label(control_frame, text="隠れた特性(ベクトル)の重視度 α:").grid(row=1, column=0, sticky="w", pady=10)
        self.alpha_var = tk.DoubleVar(value=1.0)
        self.alpha_scale = tk.Scale(control_frame, variable=self.alpha_var, from_=0.0, to=5.0, resolution=0.1, orient=tk.HORIZONTAL, length=200, command=lambda x: self.calculate_match())
        self.alpha_scale.grid(row=1, column=1, padx=10, sticky="w")
        
        # --- 下部: 結果エリア ---
        result_frame = tk.LabelFrame(self.root, text="AI推奨講師ベスト3", padx=10, pady=10)
        result_frame.pack(fill=tk.BOTH, expand=True, padx=10, pady=10)
        
        # 結果表示用ツリービュー
        columns = ('Rank', 'Name', 'Total_Dist', 'Pos_Dist', 'Vec_Dist', 'Major_Subj')
        self.tree = ttk.Treeview(result_frame, columns=columns, show='headings', height=10)
        
        self.tree.heading('Rank', text='順位')
        self.tree.heading('Name', text='講師名')
        self.tree.heading('Total_Dist', text='総合距離(相性)')
        self.tree.heading('Pos_Dist', text='位置差')
        self.tree.heading('Vec_Dist', text='ベクトル差')
        self.tree.heading('Major_Subj', text='得意科目')
        
        self.tree.column('Rank', width=50, anchor='center')
        self.tree.column('Name', width=150)
        self.tree.column('Total_Dist', width=100, anchor='center')
        self.tree.column('Pos_Dist', width=80, anchor='center')
        self.tree.column('Vec_Dist', width=80, anchor='center')
        self.tree.column('Major_Subj', width=100)
        
        self.tree.pack(fill=tk.BOTH, expand=True)

        # 解説ラベル
        note = "※ 総合距離が 0 に近いほど「AI判定による相性」が良いことを示します。\n" \
               "※ 位置差: 大まかなキャラの一致度\n" \
               "※ ベクトル差: こだわりや細部の感性の一致度"
        tk.Label(result_frame, text=note, justify="left", fg="gray").pack(anchor="w", pady=5)

    def update_student_list(self):
        student_names = self.students['Name'].tolist()
        self.student_combo['values'] = student_names
        if student_names:
            self.student_combo.current(0)
            self.calculate_match()

    def calculate_match(self, event=None):
        student_name = self.student_var.get()
        alpha = self.alpha_var.get()
        
        if not student_name:
            return

        # 選択された生徒のデータ取得
        target_s = self.students[self.students['Name'] == student_name].iloc[0]
        
        results = []
        
        # 全講師との距離を計算
        for _, teacher in self.teachers.iterrows():
            # 1. 空間距離 (位置の差) - Position Distance
            # (x, y, z) のユークリッド距離
            pos_dist = np.sqrt(
                (target_s['x'] - teacher['x'])**2 +
                (target_s['y'] - teacher['y'])**2 +
                (target_s['z'] - teacher['z'])**2
            )
            
            # 2. ベクトル距離 (向きと強さの差) - Vector Difference
            # (u, v, w) のユークリッド距離
            vec_dist = np.sqrt(
                (target_s['u'] - teacher['u'])**2 +
                (target_s['v'] - teacher['v'])**2 +
                (target_s['w'] - teacher['w'])**2
            )
            
            # 3. 総合距離 (Total Distance)
            # あなたの提案ロジック: Distance + alpha * VectorDiff
            total_dist = pos_dist + (alpha * vec_dist)
            
            # 講師の得意科目（スコア最大の科目）を取得（表示用）
            scores = {
                '数': teacher['Score_Math_Logic'],
                '英': teacher['Score_English'],
                '理': teacher['Score_Science'],
                '社国': teacher['Score_Humanities']
            }
            best_subj = max(scores, key=scores.get)
            
            results.append({
                'Name': teacher['Name'],
                'Total_Dist': total_dist,
                'Pos_Dist': pos_dist,
                'Vec_Dist': vec_dist,
                'Major_Subj': best_subj
            })
            
        # 距離が小さい順（相性が良い順）にソート
        results.sort(key=lambda x: x['Total_Dist'])
        
        # ツリービュー更新
        for item in self.tree.get_children():
            self.tree.delete(item)
            
        for i, res in enumerate(results[:5]): # トップ5まで表示
            self.tree.insert('', 'end', values=(
                f"{i+1}位",
                res['Name'],
                f"{res['Total_Dist']:.4f}",
                f"{res['Pos_Dist']:.4f}",
                f"{res['Vec_Dist']:.4f}",
                res['Major_Subj']
            ))

if __name__ == "__main__":
    root = tk.Tk()
    app = MatchingApp(root)
    root.mainloop()