AI Assistance

This project was developed with assistance from AI coding tools including Claude Sonnet 4 and Gemini Pro 2.5

DL

DL is a command-line tool written in Go for downloading multiple files concurrently from a list of URLs or a Hugging Face repository. It features a dynamic progress bar display for each download, showing speed, percentage, and downloaded/total size. The tool supports advanced Hugging Face repository handling, including interactive selection of specific .gguf files or series. Auto-update is available with -update.

Quick Start

1. Build the tool for all major platforms:

   ./build.sh

   Binaries will be placed in the build/ directory for macOS (Intel/ARM), Windows (x64/ARM), and Linux (x64/ARM).

2. Download from a URL list:

   ./dl -f ../download_links.txt -c 4

3. Download from a Hugging Face repo:

   ./dl -hf "Qwen/Qwen3-30B-A3B"

4. Select a GGUF file/series from a Hugging Face repo:

   ./dl -hf "unsloth/DeepSeek-R1-0528-GGUF" -select

5. Download a pre-defined model by alias:

   ./dl -m qwen3-0.6b

6. Search for models on Hugging Face:

   ./dl model search llama 7b gguf

7. Install, update, or remove llama.cpp binaries:

   ./dl install llama-mac-arm
   ./dl update llama
   ./dl remove llama-win-cuda

8. Show system hardware info:

   ./dl -t

9. Self-update the tool:

   ./dl --update

Features

• Concurrent Downloads: Download multiple files at once, with concurrency caps for file lists and Hugging Face downloads.
• Multiple Input Sources: Download from a URL list (-f), Hugging Face repo (-hf), or direct URLs.
• Model Registry: Use -m <alias> to download popular models by shortcut (see below).
• Model Search: Search Hugging Face models from the command line.
• Resume: Resume supported
• Llama.cpp App Management: Install, update, or remove pre-built llama.cpp binaries for your platform.
• Hugging Face GGUF Selection: Use -select to interactively choose .gguf files or series from Hugging Face repos.
• Dynamic Progress Bars: Per-download progress bars with speed, ETA, and more.
• Pre-scanning: HEAD requests to determine file size before download.
• Organized Output: Downloads go to downloads/, with subfolders for Hugging Face repos and models.
• Error Handling: Clear error messages and robust handling of download issues.
• Filename Derivation: Smart filename handling for URLs and Hugging Face files.
• Clean UI: ANSI escape codes for a tidy terminal interface.
• Debug Logging: Enable with -debug (logs to log.log).
• System Info: Show hardware info with -t.
• Self-Update: Update the tool with -update.
• Cross-Platform: Windows, macOS, and Linux supported.

Command-Line Arguments

Note: You must provide only one of the following: -f, -hf, -m, or direct URLs.

• -c <concurrency_level>: (Optional) Number of concurrent downloads. Defaults to 3. Capped at 4 for Hugging Face, 100 for file lists.
• -f <path_to_urls_file>: Download from a text file of URLs (one per line).
• -hf <repo_input>: Download all files from a Hugging Face repo (owner/repo_name or full URL).
• -m <model_alias>: Download a pre-defined model by alias (see Model Registry below).
• --token: Use the HF_TOKEN environment variable for Hugging Face API requests and downloads. Necessary for gated or private repositories. The HF_TOKEN variable must be set in your environment.
• -select: (Hugging Face only) Interactively select .gguf files or series.
• -debug: Enable debug logging to log.log.
• -update: Self-update the tool.
• -t: Show system hardware info.
• install <app_name>: Install a pre-built llama.cpp binary (see below).
• update <app_name>: Update a llama.cpp binary.
• remove <app_name>: Remove a llama.cpp binary.
• model search <query>: Search Hugging Face models from the command line. Can be used with --token.

---

Model Registry

You can use the -m flag with the following aliases to quickly download popular models:

qwen3-4b, qwen3-8b, qwen3-14b, qwen3-32b, qwen3-30b-moe, gemma3-27b

---

Model Search

Search for models on Hugging Face directly from the command line:

./dl model search llama 7b gguf

---

Llama.cpp App Management

Install, update, or remove official pre-built llama.cpp binaries for your platform from github:

./dl install llama-mac-arm
./dl update llama
./dl remove llama-win-cuda

---

System Info

Show system hardware information:

./dl -t

---

Self-Update

Update the tool to the latest version:

./dl -update

---

Build

To build the tool for all supported platforms, run:

./build.sh

This will produce binaries for macOS (Intel/ARM), Windows (x64/ARM), and Linux (x64/ARM) in the build/ directory.

---

License

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