Hey there! Let's break down how BitTorrent clients track those upload/download stats, exactly how they crunch the numbers, and how you can pull those metrics without ever opening a GUI window.

At their core, BitTorrent clients rely on granular monitoring of peer-to-peer communication to track transfer data:

• When your client establishes a connection (TCP or uTP) with a peer, it distinguishes between control messages (like handshake requests, piece requests, or cancellation signals) and actual payload data (the real file chunks you're downloading or uploading). Only the payload counts toward your stats.
• Clients maintain two persistent counters: one for total downloaded bytes, one for total uploaded bytes. These are usually saved to local state files (like .torrent.resume or .fastresume) so your stats don't reset when you restart the client.
• A critical step here is piece validation: any chunk you download first goes into a temporary cache. Only after the client verifies the chunk's hash matches what's listed in the torrent file does it get added to your download count. If the hash check fails, those bytes are discarded and don't count.

Let's get specific about how these numbers are tallied:

Download Calculation #

• Only hash-validated piece data counts. Each torrent defines fixed piece sizes (except the final piece, which is often smaller), and the client adds the size of every successfully validated piece to your download counter.
• If you download the same piece fragment from multiple peers (a common trick to speed up transfers), you only count the full piece once—duplicate fragments don't inflate your download stats.
• Some clients differentiate between "raw network download" and "disk-write size," but the standard download stat you see is the former: the total valid data pulled from the network.

Upload Calculation #

• Stats track successfully delivered payload data to peers. When your client sends a piece fragment, it waits for confirmation (either via TCP ACKs or peer-specific acknowledgment messages) before adding those bytes to your upload counter.
• Like with downloads, control messages don't count toward upload stats—only the actual file chunks you share with peers.
• A few clients track "upload cache" data (bytes ready to send but not yet delivered), but the public upload stat always reflects data that's been successfully sent.

You don't need to click around a window to get these numbers—here are your best options:

Command-Line Native Clients #

Many BitTorrent tools are built for CLI use, or have CLI companions:

• Transmission: Use transmission-remote to query stats. For example, transmission-remote -t <torrent-ID> -i pulls detailed stats for a specific torrent, while transmission-remote -l lists all active torrents with their upload/download totals.
• qBittorrent: Run the headless qbittorrent-nox version, then use either the qbittorrent-cli tool or its Web API. You can even use curl to fetch data: curl -s http://localhost:8080/api/v2/torrents/info (just enable the Web UI in settings first).
• rTorrent: This is a CLI-only client—you can view stats directly in its terminal interface, or use rtxmlrpc to pull specific values, like rtxmlrpc get_upload_total for your overall upload count.

Use GUI Client APIs #

Even if you normally use a GUI client, most have built-in APIs to pull stats programmatically:

• Deluge: Use the deluge-console tool (e.g., deluge-console "info <torrent-hash>") or the JSON-RPC interface with Python's deluge-client library to script stats retrieval.
• uTorrent: Enable the Web UI in settings, then send HTTP requests to fetch data—for example, http://localhost:8080/gui/?list=1 (you'll need to authenticate first) returns a JSON object with all torrent stats.

Parse Local State Files #

If all else fails, you can directly read the client's local state files:

• Transmission stores resume data in ~/.config/transmission/resume (binary files, but readable via transmission-remote).
• qBittorrent uses .fastresume files in ~/.config/qBittorrent/BT_backup, which are in Bencode format. You can parse these with Python's bencode library to extract uploaded and downloaded fields directly.

内容的提问来源于stack exchange，提问作者ShatteredSaint