Skip to main content

Modern fork of opentele for Python 3.10–3.14 — no Qt runtime dependency (pure-Python QDataStream replacement). Convert Telegram Desktop tdata to Telethon sessions; supports current TDesktop 5.x–6.x tdata format.

Project description

opentele-ng

Modern fork of thedemons/opentele. Python 3.10 – 3.14 • pure-Python runtime, no Qt dependency • reads current Telegram Desktop 5.x – 6.x tdata format • drop-in import opentele compatibility.

PyPI version Python CI Lint License: MIT No Qt

Why this fork

Upstream thedemons/opentele was last touched in 2022. By mid-2026 it stopped working on modern Python (3.13+ broke the metaclass), missed several lskType keys added to tdata in 2024-2025 (silently dropping data on read), shipped stale device fingerprints, and required ~50 MB of PyQt5 wheels just to parse binary streams.

opentele-ng is a clean-room modernization done over 11 release phases with 3-AI code review (OpenAI Codex, Cursor, Google Gemini) at every milestone. None of the 132 community forks of upstream attempted the pure-Python rewrite — this one ships it.

Highlights

Phase What
Phase 5 Pure-Python QDataStream / QByteArray / QFile / QBuffer — byte-identical to PyQt6 (17 byte-for-byte equivalence tests + 247 integration tests). PyQt6 removed from runtime, install is now telethon + tgcrypto-pyrofork only.
Phase 1.5 Wire-format fixes for lskWebviewTokens (QByteArray, not uint64), lskBotStorages (Dict[PeerId, FileKey] map, not single key), lskPrefs 0x1E (missed by every upstream fork — verified against TDesktop C++ source).
Phase 2 2026 device fingerprints: iPhone 17 / Air, M5 / M5 Pro / M5 Max Macs, Galaxy S25 / S26 series, Pixel 10 / 10 Pro / 10 Pro XL, Android SDK 33-37 (Android 13 → 17 beta), macOS 26 Tahoe, iOS 26. Deterministic _generate_tdesktop_app_version(unique_id) for stable fingerprints across runs.
Phase 3 kMaxAccounts = 6 (was 3, matches TDesktop's kPremiumMaxAccounts). **kwargs forward in FromTelethonQRLoginToNewClient for proxy/connection/timeout. Nuitka-compatible sharemethod. Ruff lint replaces broken upstream pylint workflow.
Phase 4 168 → 247 tests: QDataStream golden bytes, hypothesis property-based fuzzing (~1000 cases/run), real TDesktop.SaveTData → load roundtrip through MapData.prepareToWrite().
Phase 1.0.3 Security: 6 DoS guards on attacker-controlled count fields in MapData.read / _setMtpAuthorization.readKeys / account-list. Pre-loop cap by bytesAvailable() // pair_size + hard regression tests (no fail-open xfail).

Install

pip install opentele-ng
from opentele.td import TDesktop
from opentele.tl import TelegramClient
from opentele.api import API, CreateNewSession

Runtime deps: telethon>=1.36,<2, tgcrypto-pyrofork>=1.2.7. No Qt.

System libraries (libgl1, libegl1, libxkbcommon-x11-0, etc.) are not required — you can deploy on Alpine, distroless, serverless, or any minimal Linux container.

CLI — one-shot workflows (v1.1.0+)

# Read-only inspection of a tdata folder
opentele-ng info /path/to/Telegram/tdata

# Convert tdata → Telethon .session file
opentele-ng convert /path/to/Telegram/tdata --output ./me.session

See docs/examples/cli-quick-start.md for flags and exit codes.

Quick start (Python API)

import asyncio
from opentele.api import API, CreateNewSession
from opentele.td import TDesktop
from opentele.tl import TelegramClient

async def main() -> None:
    # 1. Load tdata produced by Telegram Desktop.
    tdata_path = r"C:\Users\<user>\AppData\Roaming\Telegram Desktop\tdata"
    tdesk = TDesktop(tdata_path)

    # 2. Pick an official API (TelegramIOS / TelegramAndroid / TelegramDesktop / TelegramMacOS).
    #    .Generate() builds a deterministic-or-random device fingerprint.
    api = API.TelegramIOS.Generate(unique_id="my-host")

    # 3. Convert TDesktop session → Telethon. CreateNewSession links a new
    #    device via QR code on the existing TDesktop session (no phone OTP).
    client: TelegramClient = await tdesk.ToTelethon(
        "new_session.session", CreateNewSession, api
    )

    async with client:
        await client.PrintSessions()

asyncio.run(main())

Examples

Configuration

Env var Default Effect
OPENTELE_EXTEND_STRICT 1 @extend_class raises TypeError on attribute conflicts. Set to 0 to fall back to RuntimeWarning (legacy upstream behaviour).
OPENTELE_REAL_TDATA_PATH unset When set to an absolute path of a production tdata folder, enables the opt-in real-data smoke test in tests/integration/test_real_tdata_smoke.py. CI never sets it.

Status

  • Latest: v1.1.0 (2026-05-20). PyPI: opentele-ng 1.1.0. Production-ready. 1.1.0 adds the opentele-ng CLI, PEP 621 pyproject.toml, Trusted Publishing on tag push, py.typed marker, and a GitHub Pages docs site.
  • 262 tests pass on Python 3.10 / 3.11 / 3.12 / 3.13 / 3.14 (Docker matrix + GitHub Actions matrix × Ubuntu / macOS / Windows).
  • opentele.td package coverage: 94.83% (CI gate 90%).
  • See CHANGELOG.md for the full per-release breakdown.

Security

opentele-ng accepts attacker-influenced binary blobs (tdata files from the filesystem), and version 1.0.3 added bounded-count guards on every loop that reads a count field from a decrypted payload. If you find a malformed tdata input that bypasses these guards or causes the library to read unbounded memory or CPU, please report it privately: see SECURITY.md.

Differences from upstream

  • Import name unchanged (import opentele) — drop-in for code that already uses thedemons/opentele.
  • PyPI dist name is opentele-ng to avoid collision with the original package.
  • kMaxAccounts is 6 (Telegram's premium limit), not 3.
  • @extend_class raises on attribute conflicts by default; set OPENTELE_EXTEND_STRICT=0 for the old warning-only behaviour.
  • _settingsKey is FileKey(0) by default; the upstream FileKey(1851671142505648812) magic was removed in 1.0.1 (proper AES block padding added to Storage.PrepareEncrypted instead).
  • PyQt6 is NOT a runtime dependencyopentele.td.qdatastream provides byte-compatible pure-Python replacements for QDataStream, QByteArray, QBuffer, QFile, QDir, QSysInfo.

Authorization

opentele-ng keeps the upstream's ability to use official APIs through the API class (API.TelegramDesktop, API.TelegramAndroid, API.TelegramAndroidX, API.TelegramIOS, API.TelegramMacOS, API.TelegramWeb_K, API.TelegramWeb_Z).

Per Telegram Terms of Service:

All accounts that sign up or log in using unofficial Telegram API clients are automatically put under observation to avoid violations of the Terms of Service.

Using an official API id/hash + lang_pack="tdesktop" (or ios, android, etc.) makes the session indistinguishable from the corresponding official client, which reduces spam-detection risk.

Credits

See ACKNOWLEDGMENTS.md.

License

MIT (same as upstream). See LICENSE.

logo

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

opentele_ng-1.1.0.tar.gz (121.5 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

opentele_ng-1.1.0-py3-none-any.whl (101.7 kB view details)

Uploaded Python 3

File details

Details for the file opentele_ng-1.1.0.tar.gz.

File metadata

  • Download URL: opentele_ng-1.1.0.tar.gz
  • Upload date:
  • Size: 121.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.13

File hashes

Hashes for opentele_ng-1.1.0.tar.gz
Algorithm Hash digest
SHA256 f02a3dcfead5de94446720f84f685cd0b44693522999e03d89c9749af179e6b3
MD5 05b59a1f8c298091ceca99df98820a07
BLAKE2b-256 83b7ef7afa49d99157af679171e901913bb444dcbd150a4b0cefb7e4b57e2323

See more details on using hashes here.

File details

Details for the file opentele_ng-1.1.0-py3-none-any.whl.

File metadata

  • Download URL: opentele_ng-1.1.0-py3-none-any.whl
  • Upload date:
  • Size: 101.7 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.13

File hashes

Hashes for opentele_ng-1.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 892e6896c6b059d87a64848306057a2c0371c7317fee5e44828cf803783a7301
MD5 c9a0d35a0482bb1b20b7fd9bdf55ab07
BLAKE2b-256 66065a685ea70e1e33159b65c84c7b60da0ccfddd6a55ebdfe13c04ca9b106b8

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page