Altpins-Instagram/dump_instagram.py

from BunnyCDN.Storage import Storage
from datetime import datetime
import os, config, funcs
from PIL import Image


def UploadMedia(filepath, username, media_id, timestamp = None, user_id = None):
    filename = os.path.basename(filepath)
    file_extension = os.path.splitext(filename)[1].lower()

    media_type = funcs.get_media_type(filename)

    post_type = funcs.determine_post_type(filepath, media_type)

    file_hash = funcs.calculate_file_hash(filepath)

    duration = funcs.get_video_duration(filepath) if media_type == 'video' else 0

    post_date = datetime.fromtimestamp(int(timestamp)) if timestamp else datetime.now()

    width, height = funcs.get_video_dimensions(filepath) if media_type == 'video' else Image.open(filepath).size

    server_path = f'media/{post_type}/{username}/{media_id}{file_extension}'

    file_url = f"https://storysave.b-cdn.net/{server_path}"

    if media_id and int(media_id) in existing_files:
        print('Duplicate file detected. Removing...')
        os.remove(filepath)
        return True

    obj_storage.PutFile(filepath, server_path)

    query = "INSERT IGNORE INTO media (username, media_type, media_url, width, height, media_id, post_type, date, user_id, hash, filename, duration) VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s)"
    values = (username, media_type, file_url, width, height, media_id, post_type, post_date, user_id, file_hash, filename, duration)

    newCursor.execute(query, values)
    newDB.commit()
    print(f'[{newCursor.rowcount}] records updated. File {filename} uploaded to {file_url}')

    os.remove(filepath)

    return True


def dump_instagram(folder_path):
    for filename in os.listdir(folder_path):
        parts = filename.split('~')
        if len(parts) < 4:
            continue 

        username = parts[0]
        timestamp = parts[1]
        media_id = parts[2]
        user_id = parts[3].split('_')[-1].split('.')[0]
        
        filepath = os.path.join(folder_path, filename)
        UploadMedia(username=username, filepath=filepath, media_id=media_id, timestamp=timestamp, user_id=user_id)
 

if __name__ == '__main__':
    print('Starting processing...')

    newDB, newCursor = config.gen_connection()

    obj_storage = Storage('345697f9-d9aa-4a6b-a5ec8bffc16d-ceaf-453e', 'storysave')

    newCursor.execute("SELECT media_id FROM media WHERE platform='instagram' AND media_id IS NOT NULL")
    existing_files = [image[0] for image in newCursor.fetchall()]

    dump_instagram('storysaver/')

    print("Processing completed.")