Merge files

merge.py

import pandas as pd
from tqdm import tqdm

# List of file names to be processed
file_names = [
    "2e522b3e-c97c-472f-9c75-83dec7217079.xlsx",
    "2f6e37ac-9b52-4f86-a5dc-1b1cffdea581.xlsx",
    "4b0ece5b-9800-4544-8d0b-d1cffa069ad0.xlsx",
    "5bc0f302-d03e-444e-916a-61bb2bcde488.xlsx",
    "99dceb6d-2ba2-4944-a4e7-dbd7c2994fc4.xlsx",
    "8785f01a-193a-4bfc-928e-27bf9c674fcb.xlsx",
    "71386e43-0d98-4c22-b03c-ecbc41006d1d.xlsx",
    "b30b7d9c-95bb-4d85-8e04-1694bb60f442.xlsx",
    "b956fd31-d034-4a74-9f8e-16a31ba650e3.xlsx",
    "d2ddebe8-331a-4be5-a42a-cfca3372b090.xlsx",
    "dfcc9dea-9f0e-425e-9c36-eac2a6bfed5f.xlsx",
    "f7851b00-d7b4-442f-af11-547b1dea3bdd.xlsx",
    "ff12987c-c9c0-4ea8-82fc-cf5235dc385f.xlsx",
]

file_names.reverse()

# Initialize empty DataFrames for profiles and posts
df_profiles = pd.DataFrame()
df_posts = pd.DataFrame()

# Process each file and append data to the main DataFrames
for file_name in tqdm(file_names, desc="Processing files", unit="file"):
    df_temp_profile = pd.read_excel(f"./data/{file_name}", engine="openpyxl", sheet_name="Profiles")
    df_temp_post = pd.read_excel(f"./data/{file_name}", engine="openpyxl", sheet_name="Posts")

    df_profiles = df_profiles.append(df_temp_profile)
    df_posts = df_posts.append(df_temp_post)

with pd.ExcelWriter('./output/merged.xlsx') as writer:
    df_profiles.to_excel(writer, sheet_name='Profiles', index=False)
    df_posts.to_excel(writer, sheet_name='Posts', index=False)

print("Data processing complete. Profiles and Posts data have been combined and stored in DataFrames.")

requirements.txt

pandas
tqdm
openpyxl