/TikTokLive

Python library to receive live stream events (comments, gifts, etc.) in realtime from TikTok LIVE.

Primary LanguagePythonMIT LicenseMIT

TikTokLive

A Python library to connect to and read events from TikTok's LIVE service.

LinkedIn HitCount Downloads/Month Issues Forks Stars Support Server

A Python library to receive and decode livestream events such as comments and gifts in real-time from TikTok's LIVE service by connecting to TikTok's internal Webcast service. This library includes a wrapper that connects to the Webcast service using only a user's unique_id and allows you to join your livestream as well as that of other streamers. No credentials are required to use TikTokLive.

⚠️ This library is undergoing a major revamp.

This library was originally based off of the TikTok-Live-Connector by @zerodytrash, but has since taken on its own identity as it has added more features & changed much of its core functionality.

Join the support discord and visit the #py-support channel for questions, contributions and ideas. Feel free to make pull requests with missing/new features, fixes, etc.

UPDATE:
Due to a change on the part of TikTok, versions prior to v4.3.8 are no longer functional. If you are using an unsupported version, upgrade to the latest version using the pip install TikTokLive --upgrade command.

Other Languages

TikTok LIVE is available in several alternative languages with only slight differences between the versions:

Table of Contents

Primary Information

Resources & Guides

  1. David's Intro Tutorial
  2. Getting Started
  3. Params & Options
  4. Client Methods
  5. Client Attributes
  6. TikTok Events

TikTokLive Intro Tutorial

I cannot recommend this tutorial enough for people trying to get started. It is succinct, informative and easy to understand, created by David Teather, the creator of the Python TikTok-Api package. Click the thumbnail to warp.

David's Tutorial

Getting Started

  1. Install the module via pip
> pip install TikTokLive
  1. Create your first chat connection
from TikTokLive import TikTokLiveClient
from TikTokLive.types.events import CommentEvent, ConnectEvent

# Instantiate the client with the user's username
client: TikTokLiveClient = TikTokLiveClient(unique_id="@isaackogz")


# Define how you want to handle specific events via decorator
@client.on("connect")
async def on_connect(_: ConnectEvent):
    print("Connected to Room ID:", client.room_id)


# Notice no decorator?
async def on_comment(event: CommentEvent):
    print(f"{event.user.nickname} -> {event.comment}")


# Define handling an event via a "callback"
client.add_listener("comment", on_comment)

if __name__ == '__main__':
    # Run the client and block the main thread
    # await client.start() to run non-blocking
    client.run()

For more examples, see the examples folder provided in the tree.

Params & Options

To create a new TikTokLiveClient object the following parameter is required. You can optionally add configuration options to this via kwargs.

TikTokLiveClient(unique_id, **options)

Param Name Required Description
unique_id Yes The unique username of the broadcaster. You can find this name in the URL.
Example: https://www.tiktok.com/@isaackogz => isaackogz
**options No Here you can set the following optional connection properties. If you do not specify a value, the default value will be used.

process_initial_data (default: True)
Define if you want to process initial data upon connecting (e.g. first 1-10 messages from BEFORE you connected).

fetch_room_info_on_connect (default: True)
Define if you want to fetch all room information on start. If this option is enabled, connection to offline rooms will be prevented. If enabled, the connect result contains the room info via the room_info attribute. You can also manually retrieve the room info (even in an unconnected state) using the retrieve_room_info() method.

enable_detailed_gifts (default: False)
Define if you want to receive extended information about gifts like gift name, cost and images which you can retrieve via the available_gifts attribute. When enabled, the details attribute in a Gift object will be populated.

ws_ping_interval (default: 10.0)
The interval between keepalive pings on the websocket connection (in seconds).

ws_timeout (default: 10.0)
How long to wait before the websocket connection is considered timed out (in seconds).

http_timeout (default: 10.0)
How long to wait before considering an HTTP request in the http client timed out (in seconds).

http_headers (default: {})
Additional HTTP client headers to include when making requests to the Webcast API AND connecting to the websocket server.

http_params (default: {})
Additional HTTP client parameters to include when making requests to the Webcast API AND connecting to the websocket.

loop (default: None)
Optionally supply your own asyncio event loop for usage by the client. When set to None, the client pulls the current active loop or creates a new one. This option is mostly useful for people trying to nest asyncio.

trust_env (default: False)
Whether to trust environment variables that provide proxies in httpx requests

proxies (default: None)
Enable proxied requests by turning on forwarding for the HTTP "proxies" argument. Websocket connections will NOT be proxied

lang (default: en-US)
Change the language. Payloads will be in English, but front-end content will be in the desired language!

sign_api_key (default: None)
Parameter to increase the amount of connections allowed to be made per minute via a Sign Server API key. If you need this, contact the project maintainer.

ssl_context (default: None)
Providing an SSLContext object overrides the default context used in HTTP requests made to TikTok. This is useful if TikTok sends invalid SSL signatures, as you can provide a "no verify" context.

Methods

A TikTokLiveClient object contains the following methods:

Method Name Description
run Starts a connection to the live chat while blocking the main thread
start (async) Connects to the live chat without blocking the main thread
stop Turns off the connection to the live chat.
retrieve_room_info (async) Gets the current room info from TikTok API
retrieve_available_gifts (async) Retrieves a list of the available gifts for the room and adds it to the extended_gift attribute of the Gift object on the gift event, when enabled.
add_listener Adds an asynchronous listener function (or, you can decorate a function with @client.on("<event>")) and takes two parameters, an event name and the payload, an AbstractEvent
download Start downloading the livestream video for a given duration or until stopped via the stop_download method. Supports the ability to add different flags, like -c copy which may reduce CPU usage by disabling transcoding.
stop_download Stop downloading the livestream video if currently downloading, otherwise throws an error

Attributes

A TikTokLiveClient object contains the following attributes:

Attribute Name Description
room_id The Room ID of the livestream room the client is currently connected to
room_info Information about the given livestream room
unique_id The TikTok username of the person whose livestream the client is currently connected to (e.g. @charlidamelio)
connected Whether the client is currently connected to a livestream
connecting Whether the client is currently connecting to a livestream
available_gifts A dictionary containing K:V pairs of Dict[int, GiftDetailed], where the int is the internal TikTok gift id
proxies Get the current proxies being used for HTTP requests.

Note: To set the active proxies, set the value of this attribute
loop The asyncio event loop the client is running off of
http The HTTP client TikTokLive uses to make all HTTP-based requests
websocket The WebcastWebsocketConnection websocket client TikTokLive uses to manage its websocket connection
ffmpeg The ffmpeg wrapper TikTokLive uses to manage ffmpeg-based stream downloads
viewer_count The number of people currently watching the livestream broadcast. Updated automatically on a viewer_update event
top_viewers The top N (usually ~1-20) users, ranked by coins gifted to the streamer, for the broadcast. Updated automatically on a viewer_update event

Events

A TikTokLiveClient object has the following events. You can add events either by doing client.add_listener("event_name", callable) or by decorating a function with @client.on("event_name") that includes an event payload parameter.

connect

Triggered when the websocket connection is successfully established.

@client.on("connect")
async def on_connect(event: ConnectEvent):
    print("Connected")

disconnect

Triggered when the connection is terminated. You can call start() to reconnect . Note that you should wait a little bit before attempting a reconnect to to avoid being rate-limited.

@client.on("disconnect")
async def on_disconnect(event: DisconnectEvent):
    print("Disconnected")

like

Triggered every time someone likes the stream.

@client.on("like")
async def on_like(event: LikeEvent):
    print(f"@{event.user.unique_id} liked the stream!")

join

Triggered every time a new person joins the stream.

@client.on("join")
async def on_join(event: JoinEvent):
    print(f"@{event.user.unique_id} joined the stream!")

gift

Triggered every time a gift arrives. Extra information can be gleamed off the available_gifts client attribute.

NOTE: Users have the capability to send gifts in a streak. This increases the event.gift.count value until the user terminates the streak. During this time new gift events are triggered again and again with an increased event.gift.count value. It should be noted that after the end of the streak, another gift event is triggered, which signals the end of the streak via event.gift.is_repeating:1. This applies only to gifts with event.gift.info.type:1. This means that even if the user sends an event.gift.info.type:1 gift only once, you may receive the event twice. Once with event.gift.is_repeating:0 and once with event.gift.is_repeating:1. Therefore, the event should be handled as follows in one of TWO ways. These are the same, except the second is a 'higher level' implementation using TikTokLive API features:

@client.on("gift")
async def on_gift(event: GiftEvent):
    # If it's type 1 and the streak is over
    if event.gift.info.type == 1:
        if event.gift.is_repeating == 1:
            print(f"{event.user.unique_id} sent {event.gift.count}x \"{event.gift.info.name}\"")

    # It's not type 1, which means it can't have a streak & is automatically over
    elif event.gift.info.type != 1:
        print(f"{event.user.unique_id} sent \"{event.gift.info.name}\"")
@client.on("gift")
async def on_gift(event: GiftEvent):
    # Streakable gift & streak is over
    if event.gift.streakable and not event.gift.streaking:
        print(f"{event.user.unique_id} sent {event.gift.count}x \"{event.gift.info.name}\"")

    # Non-streakable gift
    elif not event.gift.streakable:
        print(f"{event.user.unique_id} sent \"{event.gift.info.name}\"")

follow

Triggered every time someone follows the streamer.

@client.on("follow")
async def on_follow(event: FollowEvent):
    print(f"@{event.user.unique_id} followed the streamer!")

share

Triggered every time someone shares the stream.

@client.on("share")
async def on_share(event: ShareEvent):
    print(f"@{event.user.unique_id} shared the stream!")

more_share

Triggered when 5 or 10 users join from a viewer's share link.

@client.on("more_share")
async def on_connect(event: MoreShareEvent):
    print(f"More than {event.amount} users have joined from {event.user.unique_id}'s share link!")

viewer_update

Triggered every time the viewer count is updated. This event also updates the cached viewer count by default.

@client.on("viewer_update")
async def on_connect(event: ViewerUpdateEvent):

    # Viewer Count
    print("Received a new viewer count:", event.viewer_count)
    print("The client automatically sets the count as an attribute too:", client.viewer_count)
    
    # Top VIewers
    print("You can even get the top viewers (by coins gifted)!:", event.top_viewers)
    print("The client automatically sets the top viewers as an attribute too:", client.top_viewers)

comment

Triggered every time someone comments on the live.

NOTE: Some comments will be missing. Certain "low quality" comments will ONLY show up when a session_id is passed to the client.

@client.on("comment")
async def on_connect(event: CommentEvent):
    print(f"{event.user.nickname} -> {event.comment}")

emote

Triggered when someone sends a subscription emote comment to the live chat.

@client.on("emote")
async def on_connect(event: EmoteEvent):
    print(f"{event.user.nickname} -> {event.emote.image.url}")

envelope

Triggered when someone sends an envelope (treasure box) to the TikTok streamer.

@client.on("envelope")
async def on_connect(event: EnvelopeEvent):
    print(f"{event.treasure_box_user.unique_id} -> {event.treasure_box_data}")

ranking_update

Triggered when a stream rank update is sent out. Can be Weekly Ranking or Rising Star!

@client.on("ranking_update")
async def on_connect(event: RankingUpdateEvent):
    print(f"{event.user.unique_id} has the rank #{event.rank} for the {event.type} leaderboard.")

user_ranking_update

Triggered when a user rank update is sent out. Can be Top Viewer status.

@client.on("user_ranking_update")
async def on_connect(event: UserRankingUpdateEvent):
    print(f"{event.user.unique_id} just became a #{event.rank} top viewer!")

mic_battle_start

Triggered when a Mic Battle starts!

@client.on("mic_battle_start")
async def on_connect(event: MicBattleStartEvent):
    print(f"A Mic battle has started!")

mic_battle_update

Triggered when information is received about a mic battle's progress.

@client.on("mic_battle_update")
async def on_connect(event: MicBattleUpdateEvent):
    print(f"An army in the mic battle has received points, or the status of the battle has changed!")

live_end

Triggered when the live stream gets terminated by the host.

@client.on("live_end")
async def on_connect(event: LiveEndEvent):
    print(f"Livestream ended :(")

intro_message

Triggered when an intro message is sent to the live room. An intro message is basically a pinned message at the top of chat when you join a room.

This event only fires if "process_initial_data" is enabled and the streamer has an intro message configured.

@client.on("intro_message")
async def on_connect(event: IntroMessageEvent):
    print(f"Message: {event.message}")

unknown

Triggered when ANY unknown event is received that is not yet handled by this client.

This includes both events where the protobuf has NOT been decoded, as well as events where it has been decoded, but no event object has been created (e.g. it's useless data).

Use this event to debug and find new events to add to TikTokLive. Mention them here when you do.

This event is very advanced and handles both types of cases, an API to help you decode including offering the binary as base64. You can plug base64 into https://protobuf-decoder.netlify.app/ to reverse-engineer the protobuf schema.

@client.on("unknown")
async def on_connect(event: UnknownEvent):
    print(f"Event Type: {event.type}")
    print(f"Event Base64: {event.base64}")

error

Triggered when there is an error in the client or error handlers.

If this handler is not present in the code, an internal default handler will log errors in the console. If a handler is added, all error handling (including logging) is up to the individual.

Warning: If you listen for the error event and do not log errors, you will not see when an error occurs. This expected behaviour, listening to the error event overrides & disables the built-in one.

@client.on("error")
async def on_connect(error: Exception):
    # Handle the error
    if isinstance(error, SomeRandomError):
        print("Handle some error!")
        return

    # Otherwise, log the error
    # You can use the internal method, but ideally your own
    client._log_error(error)

Contributors

  • Isaac Kogan - Creator, Primary Maintainer, and Reverse-Engineering - isaackogan
  • Zerody - Initial Reverse-Engineering Protobuf & Support - Zerody
  • Davincible - Reverse-Engineering Stream Downloads - davincible
  • David Teather - TikTokLive Introduction Tutorial - davidteather

See also the full list of contributors who have participated in this project.

License

This project is licensed under the MIT License - see the LICENSE file for details.