import os
from openpyxl import load_workbook
from tqdm import tqdm
# Define the base path for the dataset
audio_base_path = "/data5/Temp_ProjectData/QiJianProject/2022ZD0116304-202405/Videos/VFGR/"

# Load the Excel workbook and select the active sheet
workbook_path = "/data1/Hatcher/model/project/MPKFNet/大规模数据库-视频.xlsx"  # Update this with the actual path to the Excel file
workbook = load_workbook(workbook_path)
sheet = workbook.active  # Assuming the active sheet is the one we need

# Function to update the Excel with new audio data
def update_excel_with_audio_data(base_path, sheet):
    # Find the last row with data to continue from there
    last_row = sheet.max_row + 1
    a = 0
    # Iterate through all subfolders in the base directory
    for folder_name in tqdm(os.listdir(base_path)):
        folder_path = os.path.join(base_path, folder_name)
        # a=a+1
        # if a>2:
        #     break
        if os.path.isdir(folder_path):  # Ensure it's a directory
            # Iterate through all audio files in the folder

            for audio_name in os.listdir(folder_path):
                if audio_name.lower().endswith(('.mp4')):  # Check audio extensions
                    audio_path = os.path.join(folder_path, audio_name)

                    # Fill the Excel sheet
                    sheet.cell(row=last_row, column=1, value=last_row - 1)  # Sequence number
                    sheet.cell(row=last_row, column=2, value=audio_path)     # Audio data
                    sheet.cell(row=last_row, column=3, value=folder_name)    # Label (folder name)
                    sheet.cell(row=last_row, column=4, value=0)              # Data function type is always 1

                    # Increment the row index for the next entry
                    last_row += 1

    # Save the updated workbook
    workbook.save(workbook_path)  # Save the changes to the same file

# Uncomment the following line to run the function with actual file paths in a real environment
update_excel_with_audio_data(audio_base_path, sheet)