MDOF/knet-250/FolderScripts/save_h5.py

"""
Author: Jason Jiang
Date: 2024.01.30
用于处理单个文件夹内的加速度数据，然后保存为HDF5文件，后续再合并所有文件夹的HDF5文件

"""
import h5py
import numpy as np
import os
import re  # 引入正则表达式库
import sys

def delete_file(file_path:str) -> None:
    """
    先删除本来的 h5 文件
    """
    if os.path.exists(file_path):
        os.remove(file_path)
        print(f"Delete file: {file_path}")

def extract_floor_response(folder_path:str, 
                           start_gms_number:int = 0,
                           number_of_gm:int=250) -> np.ndarray:
    """
    读取所有 results 文件中包含 Acc_Roof 的文件，并存为一个大数组。
    
    """
    # 初始化一个大型的 NumPy 数组，假设每个地震动数据有3000个时间点
    Acc_Floor_Response = np.empty((57, number_of_gm, 3000))

    for file_name in os.listdir(folder_path):
        if all(keyword in file_name for keyword in ['Blg', 'RoofAcc']) and file_name.endswith('.txt'):
            # 使用正则表达式从文件名中提取数字
            numbers = re.findall(r'\d+', file_name)
            # print(f"number: {numbers}")
            blg_number = int(numbers[0]) -1 # 第一个数字是楼宇号
            gms_number = int(numbers[1]) - start_gms_number -1 # 第二个数字是地震动号

            # 读取文件
            file_path = os.path.join(folder_path, file_name)
            data = np.loadtxt(file_path, skiprows=1, usecols=1)  # 只读取第二列

            # 如果数据长度小于3000，补充0到长度为3000
            if data.shape[0] < 3000:
                data = np.pad(data, (0, 3000 - data.shape[0]), 'constant')
            
            Acc_Floor_Response[blg_number, gms_number, :] = data
    
    Acc_Floor_Response = np.transpose(Acc_Floor_Response, (1, 0, 2))
    Acc_Floor_Response = Acc_Floor_Response.reshape(57*number_of_gm, 3000)
    print(f"Acc_Blg_GMs.shape = {Acc_Floor_Response.shape}")

    return Acc_Floor_Response       # (14250, 3000)

def extract_damage_state(folder_path:str,
                         start_gms_number:int = 0,
                         number_of_gm:int=250) -> np.ndarray:
    """
    读取所有 results 文件中损伤等级的数据，并存为一个大数组。
    
    """
    # 初始化numpy数组
    damage_states = np.zeros((57, number_of_gm, 1))
    file_path = os.path.join(folder_path, 'DamageState.txt')
    # 读取txt文件并处理数据
    with open(file_path, 'r') as file:
        next(file)  # 跳过标题行
        for line in file:
            data = line.strip().split()
            building_index = int(data[0]) - 1
            earthquake_index = int(data[1].split('_')[1]) - start_gms_number - 1
            damage_state = max(map(int, data[2:8]))
            damage_states[building_index, earthquake_index, 0] = damage_state

    damage_states = np.transpose(damage_states, (1, 0, 2))
    damage_states = damage_states.reshape(57*number_of_gm, 1)
    print(f"damage_states.shape = {damage_states.shape}")

    return damage_states       # (14250, 1)

def save_2_hdf5(file_name:str,
                dataset_name:str,
                array:np.ndarray) -> None:
    """存放数据到 HDF5 文件中。

    Args:
        file_name: 文件名
        dataset_name: 数据集名称
        array: 数据集
    
    """
    # 创建一个新的 HDF5 文件
    with h5py.File(file_name, 'a') as f:
        # 将数据存储为数据集
        f.create_dataset(dataset_name, data=array)

def save_h5():
    # 确定当前工作目录是脚本所在的目录
    current_folder = os.path.dirname(os.path.abspath(__file__))
    # current_folder = os.getcwd()
    os.chdir(current_folder)
    print(f"current_folder = {current_folder}")
    # 处理当前文件夹中的数据
    folder_name = os.path.basename(current_folder)
    print(f"folder_name = {folder_name}")
    # 首先删除本来的 h5 文件
    delete_file(os.path.join(f"{folder_name}.h5"))

    match = re.match(r'^(\d+)-(\d+)$', folder_name)
    if not match:
        sys.exit(f"Folder name '{folder_name}' does not match the expected format.")

    start_num = int(match.group(1))
    end_num = int(match.group(2))
    start_gms_number = start_num - 1
    number_of_gm = end_num - start_num + 1

    results_folder_path = os.path.join(current_folder, 'Results')
    if not os.path.exists(results_folder_path):
        sys.exit(f"The results folder does not exist: {results_folder_path}")

    # 提取数据
    print(f"Processing folder: {folder_name}")
    Acc_Floor_Response = extract_floor_response(results_folder_path, start_gms_number, number_of_gm)
    DS_Blg = extract_damage_state(results_folder_path, start_gms_number, number_of_gm)

    # 存储数据到 HDF5 文件中
    hdf5_filename = os.path.join(current_folder, f"{folder_name}.h5")
    save_2_hdf5(hdf5_filename, 'Acc_Floor_Response', Acc_Floor_Response)
    save_2_hdf5(hdf5_filename, 'DS_Blg', DS_Blg)
    print(f"Data saved to {hdf5_filename}")

if __name__ == "__main__":
    save_h5()