localportal 0.2.0

局域网语音输入中转工具 - 本地传送门

Local Portal (lportal)

📖 中文文档

A real-time LAN text relay tool —— the Local Portal

Features

• Phone Input → PC Clipboard Sync: Type on your phone, and the content instantly appears on your computer clipboard
• File Transfer: Send images and videos from your phone directly to your computer's download folder
• Copy Mode Switching: Supports cover mode (default) and append mode; multiple messages are automatically merged in append mode
• Text Beautification: Use LLM to convert colloquial input into structured documents, with streaming output and reasoning chain display
• Multi-Device Isolation: Each device registers independently with isolated sessions; messages and history are completely separated
• Device Session Mode: The server can establish dedicated sessions with specific devices for bidirectional messaging
• Pure Browser Solution: No app installation required on the phone; just open the browser and go
• Interactive CLI: The computer side provides a CLI terminal with slash command support
• Pairing Code Security: 4-digit numeric pairing code prevents unauthorized access

Use Cases:

• Quickly send URLs, verification codes, or long text to your computer
• Transfer photos and videos directly from your phone to your computer
• Temporary note and to-do transmission
• Type on your phone, paste on your computer
• Convert voice input into professionally structured documents with one click
• Multiple devices connected simultaneously with completely isolated data
• Server actively pushes messages and notifications to specified devices

Installation

Recommended: Install from PyPI

pip install localportal

Then run:

lportal

Alternative: Global Install with uv

If you use uv, you can also install globally as a tool:

uv tool install localportal

uv tool installs the package into an isolated environment (e.g. %APPDATA%\uv\tools\lportal\ on Windows, ~/.local/share/uv/tools/lportal/ on Linux/macOS) and creates an executable in your user's script directory, so it won't pollute your system Python.

Upgrade / Uninstall:

uv tool upgrade localportal   # upgrade
uv tool uninstall localportal # uninstall

Development

If you want to run from source or contribute:

With uv:

uv sync        # install dependencies
uv run lportal # run

With pip (editable):

# Create virtual environment (recommended)
python -m venv .venv

# Activate virtual environment
# Windows:
.venv\Scripts\activate
# Linux/Mac:
source .venv/bin/activate

# Install in editable mode
pip install -e .

# Run
lportal

Build & install a local wheel:

python -m build
pip install dist/localportal-*.whl
lportal

Usage

# Default start (port 14554)
lportal

# Specify port
lportal --port 8080

# Disable auto-copy
lportal --no-auto-copy

# Set max history entries
lportal --max-history 20

Set Download Directory

Files (images, videos) are saved to the system download directory (~/Downloads) by default. You can customize the download path via the LPORTAL_DOWNLOAD_DIR environment variable:

# Windows PowerShell
$env:LPORTAL_DOWNLOAD_DIR="D:\\MyDownloads"
lportal

# Windows CMD
set LPORTAL_DOWNLOAD_DIR=D:\\MyDownloads
lportal

# Linux / Mac
export LPORTAL_DOWNLOAD_DIR=/home/user/MyDownloads
lportal

LLM Configuration (Text Beautification)

The /beauty command requires an OpenAI-compatible API. Create a .env file in one of the following locations:

Option 1: User Config Directory (Recommended for global install users)

# Windows
mkdir "%APPDATA%\localportal"
echo OPENAI_BASE_URL=https://api.openai.com/v1 > "%APPDATA%\localportal\.env"
echo OPENAI_API_KEY=sk-xxxxxx >> "%APPDATA%\localportal\.env"
echo OPENAI_MODEL=gpt-4o-mini >> "%APPDATA%\localportal\.env"

# macOS
mkdir -p ~/Library/Application\ Support/localportal
cat > ~/Library/Application\ Support/localportal/.env << EOF
OPENAI_BASE_URL=https://api.openai.com/v1
OPENAI_API_KEY=sk-xxxxxx
OPENAI_MODEL=gpt-4o-mini
EOF

# Linux
mkdir -p ~/.config/localportal
cat > ~/.config/localportal/.env << EOF
OPENAI_BASE_URL=https://api.openai.com/v1
OPENAI_API_KEY=sk-xxxxxx
OPENAI_MODEL=gpt-4o-mini
EOF

Option 2: Current Working Directory (Good for development & debugging)

Create a .env file in the project directory or any working directory:

OPENAI_BASE_URL=https://api.openai.com/v1/chat/completions
OPENAI_API_KEY=sk-xxxxxxxxxxxxxxxx
OPENAI_MODEL=gpt-4o-mini

Note: The .env in the current working directory takes precedence over the user config directory.

The system prompt is located at src/prompt/text-beauty.md; feel free to modify it as needed.

Device Registration

The mobile side must pass two verification steps after connecting:

1. Pairing Code Verification: Enter the 4-digit pairing code displayed on the computer
2. Device Registration: Enter a device name (must be unique among currently online devices). After successful registration, a unique login_id is assigned

• When the same device reconnects, as long as the server has not restarted, the login_id remains unchanged and history can be restored
• Use the /devices command to view all online devices

Pairing Code Security

A 4-digit pairing code is generated on startup; the mobile side must enter the correct code to connect.

• Scan to connect: Use the /qr command to display a QR code; scanning it auto-fills the pairing code
• Manual connect: Open the browser address and enter the pairing code
• Refresh pairing code: Use /refresh-qrcode (/rq) to generate a new pairing code; old connections will be disconnected
• Switch copy mode: Use /mode [cover|add] to switch between cover and append modes; changes are synchronized to all online devices
• Refresh session: In append mode, use /new-session to start a new session
• Text beautification: Use /beauty [N] to structure and polish the Nth history message via LLM

Commands

Command	Description
/auto [on|off]	Toggle auto-copy mode
/copy [N]	Copy history message (N=1-10, no arg = most recent)
/list (/ls)	List history messages
/status	Show service status (including pairing code)
/open	Open the main page in browser
/qrcode (/qr)	Display ASCII QR code (scan to connect)
/downloads	Open download folder
/refresh-qrcode (/rq)	Refresh pairing code (disconnects all connections)
/mode [cover|add]	Switch copy mode: cover = overwrite, add = append
/new-session	Refresh session in append mode (clear buffer)
/beauty [N]	Beautify the Nth history message via LLM (default: most recent)
/beauty-history	View the last 10 text beautification tasks
/beauty-copy [N]	Copy the Nth beautification result (default: most recent)
/devices	View all logged-in devices (name, login_id, login time)
/link <name|id>	Enter session mode with a specific device; prompt becomes lportal[device]>
/unlink	Exit device session mode
/send <filepath>	Send a file to the current session device (use after /link)
/help	Show help
/exit	Exit the program

Device Session Mode

After entering session mode, you can directly type text to send to the device without slash commands. You can also use /send to transfer files:

lportal> /link iPhone
[OK] Entered device session mode: iPhone (a3f9b2c1)

lportal[iPhone]> Hello, this is a server message
[11:30:15] -> iPhone: Hello, this is a server message...

lportal[iPhone]> /send C:\Users\xx\Documents\file.pdf
[OK] File sent to iPhone: file.pdf (123.4KB)

lportal[iPhone]> /unlink
[OK] Exited session mode with iPhone

Tech Stack

• Python 3.9+
• aiohttp (HTTP/WebSocket)
• typer (CLI)
• prompt_toolkit (Interactive terminal)
• rich (Terminal styling)
• pyperclip (Clipboard)

TODO

• Image transfer support
• Video transfer support
• Copy mode switching (cover / add)
• Session management (grouped display in append mode)
• Message persistent storage
• Multi-device online management
• Encrypted transmission support
• Web dark / light theme switching

License

MIT