Skip to content

Official Repository: Telegram bot which can download direct links, torrents, nzb, google drive, telegram document, any file/folder from rclone supported clouds, all yt-dlp supported sites and jdownloader supported sites, then upload them to google drive, telegram cloud or to one of rclone supported clouds

License

Notifications You must be signed in to change notification settings

anasty17/mirror-leech-telegram-bot

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

This Telegram Bot, based on python-aria-mirror-bot, has undergone substantial modifications and is designed for efficiently mirroring or leeching files from the Internet to various destinations, including Google Drive, Telegram, or any rclone-supported cloud. It is built using asynchronous programming in Python.

Features

qBittorrent

  • External access to webui, so you can remove files or edit settings. Then you can sync settings in database with sync button in bsetting
  • Select files from a Torrent before and during download using mltb file selector (Requires Base URL) (task option)
  • Seed torrents to a specific ratio and time (task option)
  • Edit Global Options while the bot is running from bot settings (global option)

Aria2c

  • Select files from a Torrent before and during download (Requires Base URL) (task option)
  • Seed torrents to a specific ratio and time (task option)
  • Netrc support (global option)
  • Direct link authentication for a specific link while using the bot (it will work even if only the username or password is provided) (task option)
  • Edit Global Options while the bot is running from bot settings (global option)

Sabnzbd

  • External access to web interface, so you can remove files or edit settings. Then you can sync settings in database with sync button in bsetting
  • Remove files from job before and during download using mltb file selector (Requires Base URL) (task option)
  • Edit Global Options while the bot is running from bot settings (global option)
  • Servers menu to edit/add/remove usenet servers

TG Upload/Download

  • Split size (global, user, and task option)
  • Thumbnail (user and task option)
  • Leech filename prefix (user option)
  • Set upload as a document or as media (global, user and task option)
  • Upload all files to a specific chat (superGroup/channel/private/topic) (global, user, and task option)
  • Equal split size settings (global and user option)
  • Ability to leech split file parts in a media group (global and user option)
  • Download restricted messages (document or link) by tg private/public/super links (task option)
  • Choose transfer by bot or user session in case you have a premium plan (global, user option and task option)
  • Mix upload between user and bot session with respect to file size (global, user option and task option)
  • Upload with custom layout multiple thubnmail (global, user option and task option)
  • Topics support

Google Drive

  • Download/Upload/Clone/Delete/Count from/to Google Drive
  • Count Google Drive files/folders
  • Search in multiple Drive folder/TeamDrive
  • Use Token.pickle if the file is not found with a Service Account, for all Gdrive functions
  • Random Service Account for each task
  • Recursive Search (only with root or TeamDrive ID, folder ids will be listed with a non-recursive method). Based on Sreeraj searchX-bot. (task option)
  • Stop Duplicates (global and user option)
  • Custom upload destination (global, user, and task option)
  • Index link support only for Bhadoo

Status

  • Download/Upload/Extract/Archive/Seed/Clone Status
  • Status Pages for an unlimited number of tasks, view a specific number of tasks in a message (global option)
  • Interval message update (global option)
  • Next/Previous buttons to get different pages (global and user option)
  • Status buttons to get specific tasks for the chosen status regarding transfer type if the number of tasks is more than 30 (global and user option)
  • Steps buttons for how much next/previous buttons should step backward/forward (global and user option)
  • Status for each user (no auto refresh)

Yt-dlp

  • Yt-dlp quality buttons (task option)
  • Ability to use a specific yt-dlp option (global, user, and task option)
  • Netrc support (global option)
  • Cookies support (global option)
  • Embed the original thumbnail and add it for leech
  • All supported audio formats

JDownloader

  • Synchronize Settings (global option)
  • Waiting to select (enable/disable files or change variants) before download start
  • DLC file support
  • All settings can be edited from the remote access to your JDownloader with Web Interface, Android App, iPhone App or Browser Extensions

Mongo Database

  • Store bot settings
  • Store user settings including thumbnails and all private files
  • Store RSS data
  • Store incompleted task messages
  • Store JDownloader settings
  • Store config.env file on first build and incase any change occured to it, then next build it will define variables from config.env instead of database

Torrents Search

  • Search on torrents with Torrent Search API
  • Search on torrents with variable plugins using qBittorrent search engine

Archives

  • Extract splits with or without password
  • Zip file/folder with or without password
  • Decompress zst files
  • Using 7-zip tool to extract with or without password all supported types:

ZIP, RAR, TAR, 7z, ISO, WIM, CAB, GZIP, BZIP2, APM, ARJ, CHM, CPIO, CramFS, DEB, DMG, FAT, HFS, LZH, LZMA, LZMA2,MBR, MSI, MSLZ, NSIS, NTFS, RPM, SquashFS, UDF, VHD, XAR, Z, TAR.XZ

RSS

  • Based on this repository rss-chan
  • Rss feed (user option)
  • Title Filters (feed option)
  • Edit any feed while running: pause, resume, edit command and edit filters (feed option)
  • Sudo settings to control users feeds
  • All functions have been improved using buttons from one command.

Rclone

  • Transfer (download/upload/clone-server-side) without or with random service accounts (global and user option)
  • Ability to choose config, remote and path from list with or without buttons (global, user and task option)
  • Ability to set flags for each task or globally from config (global, user and task option)
  • Abitity to select specific files or folders to download/copy using buttons (task option)
  • Rclone.conf (global and user option)
  • Rclone serve for combine remote to use it as index from all remotes (global option)
  • Upload destination (global, user and task option)

Overall

  • Docker image support for linux amd64, arm64/v8, arm/v7
  • Edit variables and overwrite the private files while bot running (bot, user settings)
  • Update bot at startup and with restart command using UPSTREAM_REPO
  • Telegraph. Based on Sreeraj loaderX-bot
  • Mirror/Leech/Watch/Clone/Count/Del by reply
  • Mirror/Leech/Clone multi links/files with one command
  • Custom name for all links except torrents. For files you should add extension except yt-dlp links (global and user option)
  • Extensions Filter for the files to be uploaded/cloned (global and user option)
  • View Link button. Extra button to open index link in broswer instead of direct download for file
  • Queueing System for all tasks (global option)
  • Ability to zip/unzip multi links in same directory. Mostly helpful in unziping tg file parts (task option)
  • Bulk download from telegram txt file or text message contains links seperated by new line (task option)
  • Join splitted files that have splitted before by split(linux pkg) (task option)
  • Sample video Generator (task option)
  • Screenshots Generator (task option)
  • Ability to cancel upload/clone/archive/extract/split/queue (task option)
  • Cancel all buttons for choosing specific tasks status to cancel (global option)
  • Convert videos and audios to specific format with filter (task option)
  • Force start to upload or download or both from queue using cmds or args once you add the download (task option)
  • Shell and Executor
  • Add sudo users
  • Ability to save upload Paths
  • Name Substitution to rename the files before upload
  • Supported Direct links Generators:

mediafire (file/folders), hxfile.co (need cookies txt with name) [hxfile.txt], streamtape.com, streamsb.net, streamhub.ink, streamvid.net, doodstream.com, feurl.com, upload.ee, pixeldrain.com, racaty.net, 1fichier.com, 1drv.ms (Only works for file not folder or business account), filelions.com, streamwish.com, send.cm (file/folders), solidfiles.com, linkbox.to (file/folders), shrdsk.me ( sharedisk.io), akmfiles.com, wetransfer.com, pcloud.link, gofile.io (file/folders), easyupload.io, mdisk.me (with ytdl), tmpsend.com, qiwi.gg, berkasdrive.com, mp4upload.com, terabox.com (videos only file/folders).

How to deploy?

Prerequisites

  • Tutorial Video from A to Z:

1. Installing requirements

  • Clone this repo:
git clone https://github.com/anasty17/mirror-leech-telegram-bot mirrorbot/ && cd mirrorbot
  • For Debian based distros
sudo apt install python3 python3-pip

Install Docker by following the official Docker docs

  • For Arch and it's derivatives:
sudo pacman -S docker python
  • Install dependencies for running setup scripts:
pip3 install -r requirements-cli.txt

2. Setting up config file

cp config_sample.env config.env
  • Remove the first line saying:
_____REMOVE_THIS_LINE_____=True

Fill up rest of the fields. Meaning of each field is discussed below. NOTE: All values must be filled between quotes, even if it's Int, Bool or List.

1. Required Fields

  • BOT_TOKEN: The Telegram Bot Token that you got from @BotFather. Str
  • OWNER_ID: The Telegram User ID (not username) of the Owner of the bot. Int
  • TELEGRAM_API: This is to authenticate your Telegram account for downloading Telegram files. You can get this from https://my.telegram.org. Int
  • TELEGRAM_HASH: This is to authenticate your Telegram account for downloading Telegram files. You can get this from https://my.telegram.org. Str

2. Optional Fields

  • USER_SESSION_STRING: To download/upload from your telegram account if user is PREMIUM and to send rss. To generate session string use this command python3 generate_string_session.py after mounting repo folder for sure. Str. * NOTE*: You can't use bot with private message. Use it with superGroup.
  • DATABASE_URL: Your Mongo Database URL (Connection string). Follow this Generate Database to generate database. Data will be saved in Database: bot settings, users settings, rss data and incomplete tasks. NOTE: You can always edit all settings that saved in database from the official site -> (Browse collections). Str
  • DOWNLOAD_DIR: The path to the vps local folder where the downloads should be downloaded to. Str
  • CMD_SUFFIX: Commands index number. This number will added at the end all commands. Str|Int
  • AUTHORIZED_CHATS: Fill user_id and chat_id of groups/users you want to authorize. To auth only specific topic(s) write it in this format chat_id|thread_id Ex:-100XXXXXXXXXXX|10 or Ex:-100XXXXXXXXXXX|10|12. Separate them by space. Int
  • SUDO_USERS: Fill user_id of users whom you want to give sudo permission. Separate them by space. Int
  • DEFAULT_UPLOAD: Whether rc to upload to RCLONE_PATH or gd to upload to GDRIVE_ID. Default is gd. Read More HERE.Str
  • STATUS_UPDATE_INTERVAL: Time in seconds after which the progress/status message will be updated. Recommended 10 seconds at least. Int
  • STATUS_LIMIT: Limit the no. of tasks shown in status message with buttons. Default is 10. NOTE: Recommended limit is 4 tasks. Int
  • EXTENSION_FILTER: File extensions that won't upload/clone. Separate them by space. Str
  • INCOMPLETE_TASK_NOTIFIER: Get incomplete task messages after restart. Require database and superGroup. Default is False. Bool
  • FILELION_API: Filelion api key to mirror Filelion links. Get it from Filelion. str
  • STREAMWISH_API: Streamwish api key to mirror Streamwish links. Get it from Streamwish. str
  • YT_DLP_OPTIONS: Default yt-dlp options. Check all possible options HERE or use this script to convert cli arguments to api options. Format: key:value|key: value|key:value. Add ^ before integer or float, some numbers must be numeric and some string. str
    • Example: "format:bv*+mergeall[vcodec=none]|nocheckcertificate:True"
  • USE_SERVICE_ACCOUNTS: Whether to use Service Accounts or not, with google-api-python-client. For this to work see Using Service Accounts section below. Default is False. Bool
  • NAME_SUBSTITUTE: Add word/letter/character/sentense/pattern to remove or replace with other words with sensitive case or without. Notes:
    1. Seed will get disbaled while using this option
    2. Before any character you must add \BACKSLASH, those are the characters: \^$.|?*+()[]{}-
    • Example: script/code/s | mirror/leech | tea/ /s | clone | cpu/ | [mltb]/mltb | \text\/text/s
    • script will get replaced by code with sensitive case
    • mirror will get replaced by leech
    • tea will get replaced by space with sensitive case
    • clone will get removed
    • cpu will get replaced by space
    • [mltb] will get replaced by mltb
    • \text\ will get replaced by text with sensitive case

3. GDrive Tools

  • GDRIVE_ID: This is the Folder/TeamDrive ID of the Google Drive OR root to which you want to upload all the mirrors using google-api-python-client. Str
  • IS_TEAM_DRIVE: Set True if uploading to TeamDrive using google-api-python-client. Default is False. Bool
  • INDEX_URL: Refer to https://gitlab.com/ParveenBhadooOfficial/Google-Drive-Index. Str
  • STOP_DUPLICATE: Bot will check file/folder name in Drive incase uploading to GDRIVE_ID. If it's present in Drive then downloading or cloning will be stopped. (NOTE: Item will be checked using name and not hash, so this feature is not perfect yet). Default is False. Bool

4. Rclone

  • RCLONE_PATH: Default rclone path to which you want to upload all the files/folders using rclone. Str
  • RCLONE_FLAGS: key:value|key|key|key:value . Check here all RcloneFlags. Str
  • RCLONE_SERVE_URL: Valid URL where the bot is deployed to use rclone serve. Format of URL should be http://myip, where myip is the IP/Domain(public) of your bot or if you have chosen port other than 80 so write it in this format http://myip:port (http and not https). Str
  • RCLONE_SERVE_PORT: Which is the RCLONE_SERVE_URL Port. Default is 8080. Int
  • RCLONE_SERVE_USER: Username for rclone serve authentication. Str
  • RCLONE_SERVE_PASS: Password for rclone serve authentication. Str

5. Update

  • UPSTREAM_REPO: Your github repository link, if your repo is private add https://username:{githubtoken}@github.com/{username}/{reponame} format. Get token from Github settings. So you can update your bot from filled repository on each restart. Str.
    • NOTE: Any change in docker or requirements you need to deploy/build again with updated repo to take effect. DON'T delete .gitignore file. For more information read THIS.
  • UPSTREAM_BRANCH: Upstream branch for update. Default is master. Str

6. Leech

  • LEECH_SPLIT_SIZE: Size of split in bytes. Default is 2GB. Default is 4GB if your account is premium. Int
  • AS_DOCUMENT: Default type of Telegram file upload. Default is False mean as media. Bool
  • EQUAL_SPLITS: Split files larger than LEECH_SPLIT_SIZE into equal parts size (Not working with zip cmd). Default is False. Bool
  • MEDIA_GROUP: View Uploaded splitted file parts in media group. Default is False. Bool.
  • USER_TRANSMISSION: Upload/Download by user session. Only in superChat. Default is False. Bool
  • MIXED_LEECH: Upload by user and bot session with respect to file size. Only in superChat. Default is False. Bool
  • LEECH_FILENAME_PREFIX: Add custom word to leeched file name. Str
  • LEECH_DUMP_CHAT: ID or USERNAME or PM(private message) to where files would be uploaded. Int|Str. Add -100 before channel/superGroup id.
  • THUMBNAIL_LAYOUT: Thumbnail layout (widthxheight, 2x2, 3x3, 2x4, 4x4, ...) of how many photo arranged for the thumbnail.Str

7. qBittorrent/Aria2c/Sabnzbd

  • TORRENT_TIMEOUT: Timeout of dead torrents downloading with qBittorrent and Aria2c in seconds. Int
  • BASE_URL: Valid BASE URL where the bot is deployed to use torrent/nzb web files selection. Format of URL should be http://myip, where myip is the IP/Domain(public) of your bot or if you have chosen port other than 80 so write it in this format http://myip:port (http and not https). Str
  • BASE_URL_PORT: Which is the BASE_URL Port. Default is 80. Int
  • WEB_PINCODE: Whether to ask for pincode before selecting files from torrent in web or not. Default is False. Bool.
    • Qbittorrent NOTE: If your facing ram issues then set limit for MaxConnections, decrease AsyncIOThreadsCount, set limit of DiskWriteCacheSize to 32 and decrease MemoryWorkingSetLimit from qbittorrent.conf or bsetting command.
    • Open port 8090 in your vps to access webui from any device. username: mltb, password: mltbmltb

8. JDownloader

  • JD_EMAIL: jdownlaoder email sign up on JDownloader
  • JD_PASS: jdownlaoder password
    • JDownloader Config: You can use your config from local device to bot by zipping cfg folder (cfg.zip) and add it in repo folder but before zip you must change the downloads directory to /root/Downloads.

9. Sabnzbd

  • USENET_SERVERS: list of dictionaries, you can add as much as you want and there is a button for servers in sabnzbd settings to edit current servers and add new servers.

    [{'name': 'main', 'host': '', 'port': 563, 'timeout': 60, 'username': '', 'password': '', 'connections': 8, 'ssl': 1, 'ssl_verify': 2, 'ssl_ciphers': '', 'enable': 1, 'required': 0, 'optional': 0, 'retention': 0, 'send_group': 0, 'priority': 0}]

10. RSS

  • RSS_DELAY: Time in seconds for rss refresh interval. Recommended 600 second at least. Default is 600 in sec. Int
  • RSS_CHAT: Chat ID or USERNAME or ID|TOPIC_ID or USERNAME|TOPIC_ID where rss links will be sent. If you want message to be sent to the channel then add channel id. Add -100 before channel id. Int|Str
    • RSS NOTES: RSS_CHAT is required, otherwise monitor will not work. You must use USER_STRING_SESSION --OR-- CHANNEL. If using channel then bot should be added in both channel and group(linked to channel) and RSS_CHAT is the channel id, so messages sent by the bot to channel will be forwarded to group. Otherwise with USER_STRING_SESSION add group id for RSS_CHAT. If DATABASE_URL not added you will miss the feeds while bot offline.

11. Queue System

  • QUEUE_ALL: Number of parallel tasks of downloads and uploads. For example if 20 task added and QUEUE_ALL is 8, then the summation of uploading and downloading tasks are 8 and the rest in queue. Int. NOTE: if you want to fill QUEUE_DOWNLOAD or QUEUE_UPLOAD, then QUEUE_ALL value must be greater than or equal to the greatest one and less than or equal to summation of QUEUE_UPLOAD and QUEUE_DOWNLOAD.
  • QUEUE_DOWNLOAD: Number of all parallel downloading tasks. Int
  • QUEUE_UPLOAD: Number of all parallel uploading tasks. Int

12. Torrent Search

  • SEARCH_API_LINK: Search api app link. Get your api from deploying this repository. Str
    • Supported Sites:

    1337x, Piratebay, Nyaasi, Torlock, Torrent Galaxy, Zooqle, Kickass, Bitsearch, MagnetDL, Libgen, YTS, Limetorrent, TorrentFunk, Glodls, TorrentProject and YourBittorrent

  • SEARCH_LIMIT: Search limit for search api, limit for each site and not overall result limit. Default is zero ( Default api limit for each site). Int
  • SEARCH_PLUGINS: List of qBittorrent search plugins (github raw links). I have added some plugins, you can remove/add plugins as you want. Main Source: qBittorrent Search Plugins (Official/Unofficial). List

3. Build And Run the Docker Image

Make sure you still mount the app folder and installed the docker from official documentation.

  • There are two methods to build and run the docker:
    1. Using official docker commands.
    2. Using docker-compose. (Recommended)

Build And Run The Docker Image Using Official Docker Commands

  • Start Docker daemon (SKIP if already running, mostly you don't need to do this):
sudo dockerd
  • Build Docker image:
sudo docker build . -t mltb
  • Run the image:
sudo docker run -p 80:80 -p 8080:8080 mltb
  • To stop the running image:
sudo docker ps
sudo docker stop id

Build And Run The Docker Image Using docker-compose

NOTE: If you want to use ports other than 80 and 8080 for torrent file selection and rclone serve respectively, change it in docker-compose.yml also.

  • Install docker compose plugin
sudo apt install docker-compose-plugin
  • Build and run Docker image or to view current running image:
sudo docker compose up
  • After editing files with nano for example (nano start.sh) or git pull:
sudo docker compose up --build
  • To stop the running image:
sudo docker compose stop
  • To run the image:
sudo docker compose start
  • To get log from already running image (after mounting the folder):
sudo docker compose logs --follow

IMPORTANT NOTES:

  1. Set BASE_URL_PORT and RCLONE_SERVE_PORT variables to any port you want to use. Default is 80 and 8080 respectively.

  2. Check the number of processing units of your machine with nproc cmd and times it by 4, then edit AsyncIOThreadsCount in qBittorrent.conf.


Extras

Bot commands to be set in @BotFather

mirror - or /m Mirror
qbmirror - or /qm Mirror torrent using qBittorrent
jdmirror - or /jm Mirror using jdownloader
nzbmirror - or /nm Mirror using sabnzbd
ytdl - or /y Mirror yt-dlp supported links
leech - or /l Upload to telegram
qbleech - or /ql Leech torrent using qBittorrent
jdleech - or /jl Leech using jdownloader
nzbleech - or /nl Leech using sabnzbd
ytdlleech - or /yl Leech yt-dlp supported links
clone - Copy file/folder to Drive
count - Count file/folder from GDrive
usetting - or /us User settings
bsetting - or /bs Bot settings
status - Get Mirror Status message
sel - Select files from torrent
rss - Rss menu
list - Search files in Drive
search - Search for torrents with API
cancel - or /c Cancel a task
cancelall - Cancel all tasks
forcestart - or /fs to start task from queue
del - Delete file/folder from GDrive
log - Get the Bot Log
shell - Run commands in Shell
aexec - Execute async function
exec - Execute sync function
restart - Restart the Bot
stats - Bot Usage Stats
ping - Ping the Bot
help - All cmds with description

Getting Google OAuth API credential file and token.pickle

NOTES

  • Old authentication changed, now we can't use bot or replit to generate token.pickle. You need OS with a local browser. For example Termux.
  • Windows users should install python3 and pip. You can find how to install and use them from google or from this telegraph from Wiszky tutorial.
  • You can ONLY open the generated link from generate_drive_token.py in local browser.
  1. Visit the Google Cloud Console
  2. Go to the OAuth Consent tab, fill it, and save.
  3. Go to the Credentials tab and click Create Credentials -> OAuth Client ID
  4. Choose Desktop and Create.
  5. Publish your OAuth consent screen App to prevent token.pickle from expire
  6. Use the download button to download your credentials.
  7. Move that file to the root of mirrorbot, and rename it to credentials.json
  8. Visit Google API page
  9. Search for Google Drive Api and enable it
  10. Finally, run the script to generate token.pickle file for Google Drive:
pip3 install google-api-python-client google-auth-httplib2 google-auth-oauthlib
python3 generate_drive_token.py

Getting rclone.conf

  1. Install rclone from Official Site
  2. Create new remote(s) using rclone config command.
  3. Copy rclone.conf from .config/rclone/rclone.conf to repo folder

Upload

  • RCLONE_PATH is like GDRIVE_ID a default path for mirror. In additional to those variables DEFAULT_UPLOAD to choose the default tool whether it's rclone or google-api-python-client.
  • If DEFAULT_UPLOAD = 'rc' then you must fill RCLONE_PATH with path as default one or with rcl to select destination path on each new task.
  • If DEFAULT_UPLOAD = 'gd' then you must fill GDRIVE_ID with folder/TD id.
  • rclone.conf can be added before deploy like token.pickle to repo folder root or use bsetting to upload it as private file.
  • If rclone.conf uploaded from usetting or added in rclone/{user_id}.conf then RCLONE_PATH must start with mrcc:.
  • Whenever you want to write path manually to use user rclone.conf that added from usetting then you must add the mrcc: at the beginning.
  • So in short, up: has 4 possible values which is: gd(Upload to GDRIVE_ID), rc(Upload to RCLONE_PATH), rcl(Select Rclone Path) and rclone_path(remote:path(owner rclone.conf) or mrcc:remote:path(user rclone.conf))

UPSTREAM REPO (Recommended)

  • UPSTREAM_REPO variable can be used for edit/add any file in repository.
  • You can add private/public repository link to grab/overwrite all files from it.
  • You can skip adding the privates files like token.pickle or accounts folder before deploying, simply fill UPSTREAM_REPO private one in case you want to grab all files including private files.
  • If you added private files while deploying and you have added private UPSTREAM_REPO and your private files in this private repository, so your private files will be overwritten from this repository. Also if you are using database for private files, then all files from database will override the private files that added before deploying or from private UPSTREAM_REPO.
  • If you filled UPSTREAM_REPO with the official repository link, then be carefull incase any change in requirements.txt your bot will not start after restart. In this case you need to deploy again with updated code to install the new requirements or simply by changing the UPSTREAM_REPO to you fork link with that old updates.
  • In case you you filled UPSTREAM_REPO with your fork link be carefull also if you fetched the commits from the official repository.
  • The changes in your UPSTREAM_REPO will take affect only after restart.

Bittorrent Seed

  • Using -d argument alone will lead to use global options for aria2c or qbittorrent.

Qbittorrent

  • Global options: GlobalMaxRatio and GlobalMaxSeedingMinutes in qbittorrent.conf, -1 means no limit, but you can cancel manually.
    • NOTE: Don't change MaxRatioAction.

Aria2c

  • Global options: --seed-ratio (0 means no limit) and --seed-time (0 means no seed) in aria.sh.

Using Service Accounts for uploading to avoid user rate limit

For Service Account to work, you must set USE_SERVICE_ACCOUNTS = "True" in config file or environment variables. NOTE: Using Service Accounts is only recommended while uploading to a Team Drive.

1. Generate Service Accounts. What is Service Account?

Let us create only the Service Accounts that we need.

Warning: Abuse of this feature is not the aim of this project and we do NOT recommend that you make a lot of projects, just one project and 100 SAs allow you plenty of use, its also possible that over abuse might get your projects banned by Google.

NOTE: If you have created SAs in past from this script, you can also just re download the keys by running:

python3 gen_sa_accounts.py --download-keys $PROJECTID

NOTE: 1 Service Account can upload/copy around 750 GB a day, 1 project can make 100 Service Accounts so you can upload 75 TB a day.

NOTE: All people can copy 2TB/DAY from each file creator (uploader account), so if you got error userRateLimitExceeded that doesn't mean your limit exceeded but file creator limit have been exceeded which is 2TB/DAY.

Two methods to create service accounts

Choose one of these methods

1. Create Service Accounts in existed Project (Recommended Method)
  • List your projects ids
python3 gen_sa_accounts.py --list-projects
  • Enable services automatically by this command
python3 gen_sa_accounts.py --enable-services $PROJECTID
  • Create Sevice Accounts to current project
python3 gen_sa_accounts.py --create-sas $PROJECTID
  • Download Sevice Accounts as accounts folder
python3 gen_sa_accounts.py --download-keys $PROJECTID
2. Create Service Accounts in New Project
python3 gen_sa_accounts.py --quick-setup 1 --new-only

A folder named accounts will be created which will contain keys for the Service Accounts.

2. Add Service Accounts

Two methods to add service accounts

Choose one of these methods

1. Add Them To Google Group then to Team Drive (Recommended)
  • Mount accounts folder
cd accounts
  • Grab emails form all accounts to emails.txt file that would be created in accounts folder
  • For Windows using PowerShell
$emails = Get-ChildItem .\**.json |Get-Content -Raw |ConvertFrom-Json |Select -ExpandProperty client_email >>emails.txt
  • For Linux
grep -oPh '"client_email": "\K[^"]+' *.json > emails.txt
  • Unmount acounts folder
cd ..

Then add emails from emails.txt to Google Group, after that add this Google Group to your Shared Drive and promote it to manager and delete email.txt file from accounts folder

2. Add Them To Team Drive Directly
  • Run:
python3 add_to_team_drive.py -d SharedTeamDriveSrcID

Generate Database

  1. Go to https://mongodb.com/ and sign-up.
  2. Create Shared Cluster.
  3. Press on Database under Deployment Header, your created cluster will be there.
  4. Press on connect, choose Allow Access From Anywhere and press on Add IP Address without editing the ip, then create user.
  5. After creating user press on Choose a connection, then press on Connect your application. Choose Driver * python* and version 3.12 or later.
  6. Copy your connection string and replace <password> with the password of your user, then press close.

Multi Drive List

To use list from multi TD/folder. Run driveid.py in your terminal and follow it. It will generate list_drives.txt file or u can simply create list_drives.txt file in working directory and fill it, check below format:

DriveName folderID/tdID or `root` IndexLink(if available)
DriveName folderID/tdID or `root` IndexLink(if available)

Example:

TD1 root https://example.dev
TD2 0AO1JDB1t3i5jUk9PVA https://example.dev

Yt-dlp and Aria2c Authentication Using .netrc File

For using your premium accounts in yt-dlp or for protected Index Links, create .netrc file according to following format:

Note: Create .netrc and not netrc, this file will be hidden, so view hidden files to edit it after creation.

Format:

machine host login username password my_password

Example:

machine instagram login anas.tayyar password mypassword

Instagram Note: You must login even if you want to download public posts and after first try you must confirm that this was you logged in from different ip(you can confirm from phone app).

Youtube Note: For youtube authentication use cookies.txt file.

Using Aria2c you can also use built in feature from bot with or without username. Here example for index link without username.

machine example.workers.dev password index_password

Where host is the name of extractor (eg. instagram, Twitch). Multiple accounts of different hosts can be added each separated by a new line.


All Thanks To Our Contributors

Donations

If you feel like showing your appreciation for this project, then how about buying me a coffee.

"Buy Me A Coffee"

Binance ID:

52187862

USDT Address:

TEzjjfkxLKQqndpsdpkA7jgiX7QQCL5p4f

Network:

TRC20

TRX Address:

TEzjjfkxLKQqndpsdpkA7jgiX7QQCL5p4f

Network:

TRC20

BTC Address:

17dkvxjqdc3yiaTs6dpjUB1TjV3tD7ScWe

ETH Address:

0xf798a8a1c72d593e16d8f3bb619ebd1a093c7309

About

Official Repository: Telegram bot which can download direct links, torrents, nzb, google drive, telegram document, any file/folder from rclone supported clouds, all yt-dlp supported sites and jdownloader supported sites, then upload them to google drive, telegram cloud or to one of rclone supported clouds

Topics

Resources

License

Stars

Watchers

Forks

Sponsor this project

Languages