app.py

import os
import json
import pandas as pd
import numpy as np
from flask import Flask, render_template, request, jsonify
from werkzeug.utils import secure_filename

app = Flask(__name__)
app.config['MAX_CONTENT_LENGTH'] = 16 * 1024 * 1024  # 16MB limit
app.config['UPLOAD_FOLDER'] = 'uploads'

if not os.path.exists(app.config['UPLOAD_FOLDER']):
    os.makedirs(app.config['UPLOAD_FOLDER'])

# --- Helper Functions ---
def analyze_dataframe(df):
    """Generates a summary of the dataframe."""
    summary = {
        "columns": list(df.columns),
        "row_count": len(df),
        "dtypes": {k: str(v) for k, v in df.dtypes.items()},
        "numeric_columns": list(df.select_dtypes(include=[np.number]).columns),
        "categorical_columns": list(df.select_dtypes(include=['object', 'category']).columns),
        "sample": df.head(5).to_dict(orient='records'),
        "missing_values": df.isnull().sum().to_dict()
    }
    return summary

def generate_demo_data():
    """Generates a robust demo dataset for retail sales."""
    dates = pd.date_range(start='2024-01-01', periods=100)
    categories = ['电子产品', '服装', '家居', '图书', '美妆']
    data = {
        '日期': dates,
        '类别': np.random.choice(categories, 100),
        '销售额': np.random.randint(100, 5000, 100),
        '利润': np.random.randint(10, 1000, 100),
        '客户满意度': np.random.uniform(1.0, 5.0, 100).round(1)
    }
    return pd.DataFrame(data)

# Global store for demo (in production use session or db)
current_df = None

@app.route('/')
def index():
    return render_template('index.html')

@app.route('/api/load_demo', methods=['POST'])
def load_demo():
    global current_df
    current_df = generate_demo_data()
    summary = analyze_dataframe(current_df)
    return jsonify({"status": "success", "message": "Demo data loaded", "summary": summary})

@app.route('/api/upload', methods=['POST'])
def upload_file():
    global current_df
    if 'file' not in request.files:
        return jsonify({"error": "No file part"}), 400
    file = request.files['file']
    if file.filename == '':
        return jsonify({"error": "No selected file"}), 400
    
    if file:
        filename = secure_filename(file.filename)
        filepath = os.path.join(app.config['UPLOAD_FOLDER'], filename)
        file.save(filepath)
        
        try:
            if filename.endswith('.csv'):
                current_df = pd.read_csv(filepath)
            elif filename.endswith('.xlsx'):
                current_df = pd.read_excel(filepath)
            else:
                return jsonify({"error": "Unsupported file format"}), 400
                
            summary = analyze_dataframe(current_df)
            return jsonify({"status": "success", "summary": summary})
        except Exception as e:
            return jsonify({"error": str(e)}), 500

@app.route('/api/chat', methods=['POST'])
def chat():
    global current_df
    if current_df is None:
        return jsonify({"error": "No data loaded"}), 400
        
    data = request.json
    query = data.get('message', '').lower()
    
    response = {
        "text": "",
        "chart": None
    }
    
    # Simple "Agent" Heuristics
    numeric_cols = list(current_df.select_dtypes(include=[np.number]).columns)
    cat_cols = list(current_df.select_dtypes(include=['object', 'category']).columns)
    
    # Keyword matching for Chinese and English
    is_trend = any(k in query for k in ["trend", "time", "趋势", "时间", "变化"])
    is_dist = any(k in query for k in ["distribution", "bar", "breakdown", "分布", "柱状", "分类"])
    is_corr = any(k in query for k in ["correlation", "相关性", "关系"])
    
    if is_trend:
        # Time series analysis
        date_cols = [c for c in current_df.columns if any(k in c.lower() for k in ['date', 'time', '日期', '时间'])]
        if date_cols and numeric_cols:
            d_col = date_cols[0]
            v_col = numeric_cols[0]
            # Aggregate
            try:
                df_agg = current_df.groupby(d_col)[v_col].sum().reset_index()
                # Convert to JSON for frontend
                response["text"] = f"这是 {v_col} 随 {d_col} 变化的趋势图。数据已按时间聚合。"
                response["chart"] = {
                    "type": "line",
                    "x": df_agg[d_col].astype(str).tolist(),
                    "y": df_agg[v_col].tolist(),
                    "label": v_col,
                    "title": f"{v_col} 趋势分析"
                }
            except Exception as e:
                response["text"] = f"生成趋势图时出错: {str(e)}"
        else:
            response["text"] = "我没找到日期列来展示趋势。请尝试询问数据分布情况。"

    elif is_dist:
        if cat_cols and numeric_cols:
            c_col = cat_cols[0]
            v_col = numeric_cols[0]
            try:
                df_agg = current_df.groupby(c_col)[v_col].sum().reset_index()
                response["text"] = f"这是 {v_col} 按 {c_col} 分组的分布情况。"
                response["chart"] = {
                    "type": "bar",
                    "x": df_agg[c_col].tolist(),
                    "y": df_agg[v_col].tolist(),
                    "label": v_col,
                    "title": f"{c_col} - {v_col} 分布"
                }
            except Exception as e:
                response["text"] = f"生成分布图时出错: {str(e)}"
        else:
            response["text"] = "我需要类别数据来展示分布。"
            
    elif is_corr:
        if len(numeric_cols) >= 2:
            try:
                corr = current_df[numeric_cols].corr()
                response["text"] = "这是数值变量之间的相关性矩阵。"
                # Simplified heatmap data structure
                response["chart"] = {
                    "type": "heatmap",
                    "x": numeric_cols,
                    "y": numeric_cols,
                    "z": corr.values.tolist(),
                    "title": "相关性矩阵"
                }
            except Exception as e:
                 response["text"] = f"计算相关性时出错: {str(e)}"
        else:
            response["text"] = "数值列不足，无法进行相关性分析。"
            
    else:
        # General summary
        stats = current_df.describe().to_html(classes="table table-sm")
        response["text"] = f"我已分析数据集。以下是基本统计信息：<br>{stats}<br>您可以问我 '展示趋势'、'分析分布' 或 '显示相关性'。"
        
    return jsonify(response)

if __name__ == '__main__':
    app.run(host='0.0.0.0', port=7860, debug=True)