fix: skip create_repo (causes 403) — upload_file directly; retry with create only on 404

fix: restore TELEGRAM_API_ID/HASH defaults (Pyrogram was failing to start); keep env override

fix: Pyrogram proper reconnect lifecycle + remove hardcoded credentials

fix: channel_id int cast, archive_id top-level read, dataset per-account, pyrogram reconnect, better error handling v4

Add timeout and is_connected check to download_from_channel

Make _receive async: return 202 immediately, process in bg thread

Use HF_TOKEN_USER as primary, remove HF_TOKEN secret collision

trigger rebuild

fix: correct Dockerfile pip install continuation

v3: Pyrogram download from channel + huggingface_hub upload

support dataset_pushed pull mode, detailed progress

v2.1: progress reporting, download_url, quality

fix: accept download_url from upload-bot relay (Telegram blocked on user HF Spaces)

fix: Pyrogram MTProto download; copy file to dataset dir before push

Fix: add build-essential, git, pyrogram for direct Telegram downloads

Fix: Python 3.12, env var fallback

Fix: runtime dataset init (no submodule), read from env

Fix: add proper YAML frontmatter with sdk: docker

Archiver v2

Archiver service with Telegram receive and dataset push

Link archive dataset as submodule

Initial commit

initial commit