ChatPack

01 // SectionWhat it is

ChatPack is a Python library of interactive UI components for Telegram bots built on Pyrogram. Each component — a confirmation, a menu, a form step — is awaited like a function call, so a multi-step conversation is written as sequential code instead of a set of callback handlers and stored state.

It is published on PyPI as chatpack-ui and requires Python 3.8+.

pip install chatpack-ui

02 // SectionThe flow it replaces

Without ChatPack
  state (DB) ──> handler ──> edit UI ──> next state (DB)

With ChatPack
  data = await Form([Step1, Step2]).run()

03 // SectionArchitecture

Every component extends a shared BaseField class and can be called in two ways:

• ask(client, chat_id) returns the validated value together with the ids of the messages it sent, so a Form can clean the conversation up afterwards.
• ask_value(client, chat_id) returns only the value, for one-off prompts inside existing handlers.

Inline buttons carry namespaced callback data — cp:<field key>:<action>[:<value>] — and each field ignores payloads that belong to another key, so stale buttons from an earlier step cannot interfere.

04 // SectionComponents

• ConfirmDialog — yes/no confirmation before destructive actions
• JoinChecker — blocks access until the user joins required channels
• RatingStars — star rating on a single message
• NestedMenu — multi-level menus built from nested dictionaries
• BroadcastSender — copies a message to a list of users
• ChunkSender — splits text that exceeds Telegram's 4,096-character limit
• BranchingForm — forms whose next step depends on earlier answers
• MultiSelectMenu — checkbox menu with minimum and maximum limits
• PhotoCollector — collects a batch of photos within set limits

from chatpack import ConfirmDialog

dialog = ConfirmDialog("Are you sure you want to permanently delete your data?")
confirmed = await dialog.ask_value(client, message.chat.id, timeout=30)

05 // SectionTesting

The test suite runs offline: real Pyrogram update objects drive the components while a fake client records every outgoing API call. It uses pytest, with flake8 for linting.

06 // SectionStatus

Released as v1.0.3 in June 2026; the September 2026 release (V1.0.4) added more components. Distributed under the MIT License.