import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import timesfm
from datetime import datetime, timedelta

# 导入必要的库
# pandas用于数据处理
# numpy用于数值计算
# matplotlib用于数据可视化
# timesfm用于时间序列预测
# datetime用于日期时间处理

# 读取CSV文件
# 从iMac的下载文件夹读取
file_path = '/Users/ningli/Downloads/rwa-asset-timeseries-export-1744949365800.csv'  # 替换"你的用户名"为您的Mac用户名
stablecoin_data = pd.read_csv(file_path)

# 将时间戳转换为datetime类型，便于处理
stablecoin_data['Date'] = pd.to_datetime(stablecoin_data['Date'])

# 打印数据集的基本信息
print(f"Dataset shape: {stablecoin_data.shape}")  # 显示行数和列数
print(f"Date range: {stablecoin_data['Date'].min()} to {stablecoin_data['Date'].max()}")  # 显示数据的时间跨度

# 获取所有稳定币列的列表（排除非稳定币列）
stablecoin_columns = [col for col in stablecoin_data.columns 
                      if col not in ['Timestamp', 'Date', 'Measure']]

print(f"Number of stablecoin types: {len(stablecoin_columns)}")
print(f"Stablecoin types: {stablecoin_columns}")

# 计算每日稳定币市场总规模
# 按日期分组并对所有稳定币值求和
daily_totals = stablecoin_data.groupby('Date')[stablecoin_columns].sum().sum(axis=1).reset_index()
daily_totals.columns = ['Date', 'TotalStablecoinSize']  # 重命名列

# 按日期排序
daily_totals = daily_totals.sort_values('Date')

# 显示基本统计数据
print("\nDaily Total Stablecoin Market Size Statistics:")
print(daily_totals['TotalStablecoinSize'].describe())  # 计算均值、标准差、最小值、最大值等统计量

# 显示前几行和后几行记录
print("\nFirst few days of data:")
print(daily_totals.head())
print("\nLast few days of data:")
print(daily_totals.tail())

# 创建TimesFM模型实例，遵循示例代码
tfm = timesfm.TimesFm(
    hparams=timesfm.TimesFmHparams(
        context_len=32,  # 模型的最大上下文长度
        horizon_len=1825,  # 预测5年（365天 × 5）
        input_patch_len=32,  # 固定值
        output_patch_len=128,  # 固定值
        num_layers=20,  # 固定值
        model_dims=1280,  # 固定值
        backend="cpu"  # 可以是 "cpu"、"gpu" 或 "tpu"
    ),
    checkpoint=timesfm.TimesFmCheckpoint(
        huggingface_repo_id="google/timesfm-1.0-200m-pytorch"),  # 加载预训练检查点
)

# 为模型准备输入数据
forecast_input = daily_totals['TotalStablecoinSize'].values.reshape(1, -1)  # 转换为二维数组
frequency_input = [0]  # 使用高频率设置

# 生成预测
point_forecast, experimental_quantile_forecast = tfm.forecast(
    forecast_input, 
    freq=frequency_input,
)

# 获取历史数据的最后一个日期
last_date = daily_totals['Date'].iloc[-1]

# 为预测创建日期序列（5年）
forecast_dates = pd.date_range(start=last_date, periods=len(point_forecast[0]) + 1, inclusive='right')

# 创建用于绘图的DataFrame
forecast_df = pd.DataFrame({
    'Date': forecast_dates, 
    'Forecast': point_forecast[0]
})

# 绘制历史数据和预测数据
plt.figure(figsize=(16, 8))  # 设置图形大小
plt.plot(daily_totals['Date'], daily_totals['TotalStablecoinSize'], label='Historical Stablecoin Market Size')
plt.plot(forecast_df['Date'], forecast_df['Forecast'], label='Forecast (5 Years)')
plt.xlabel('Date')  # X轴标签
plt.ylabel('Total Stablecoin Market Size (USD)')  # Y轴标签
plt.title('Stablecoin Market Size Forecast (5 Years)')  # 图表标题
plt.grid(True, alpha=0.3)  # 添加网格线
plt.legend()  # 显示图例

# 添加关键统计信息的注释
current_size = daily_totals['TotalStablecoinSize'].iloc[-1]  # 当前市场规模
future_size = forecast_df['Forecast'].iloc[-1]  # 5年后的市场规模
growth_pct = ((future_size - current_size) / current_size) * 100  # 增长百分比

# 为当前市场规模添加注释
plt.annotate(f'Current Size: ${current_size/1e9:.2f}B', 
             xy=(last_date, current_size),
             xytext=(10, 30), textcoords='offset points',
             arrowprops=dict(arrowstyle='->'))

# 为5年后市场规模添加注释
plt.annotate(f'Projected Size in 5 Years: ${future_size/1e9:.2f}B\nGrowth: {growth_pct:.2f}%', 
             xy=(forecast_dates[-1], future_size),
             xytext=(-100, 30), textcoords='offset points',
             arrowprops=dict(arrowstyle='->'))

# 优化X轴日期格式，提高可读性
plt.gcf().autofmt_xdate()

# 为主要稳定币单独创建一个图表
# 首先检查哪些稳定币确实存在于我们的数据中
available_stablecoins = [coin for coin in stablecoin_columns if coin in ['USDT', 'USDC', 'BUSD', 'FRAX', 'DAI']]
print(f"Available major stablecoins: {available_stablecoins}")

if available_stablecoins:  # 如果有可用的稳定币
    # 将数据透视为按日期索引的形式
    stablecoin_data_pivoted = stablecoin_data.pivot_table(
        index='Date', 
        values=available_stablecoins,  # 只使用实际存在的稳定币
        aggfunc='sum'
    ).reset_index()

    # 创建新的图形
    plt.figure(figsize=(16, 8))

else:
    print("No major stablecoins found in data. Skipping individual stablecoin chart.")

# 显示所有图表
plt.show()

# 打印预测摘要
print("\nForecast Summary:")
print(f"Current total stablecoin market size: ${current_size/1e9:.2f} billion")
print(f"Projected market size in 5 years: ${future_size/1e9:.2f} billion")
print(f"Projected growth: {growth_pct:.2f}%")
print(f"Compound Annual Growth Rate (CAGR): {((future_size/current_size)**(1/5) - 1) * 100:.2f}%")