syft-objects 0.3.4

Share files with explicit mock vs private control

🔐 Syft Objects

Share files with explicit mock vs private control

Quick Start

import syft_objects as syo

# Create an object with demo and real content
obj = syo.syobj(
    name="AI Results",
    mock_contents="Model achieved good performance",
    private_contents="Accuracy: 94.5%, Cost: $127"
)

# Browse all your objects interactively
syo.objects

# Search for specific objects
syo.objects.search("financial")

What It Does

Mock vs Private Pattern: Every object has two versions:

• Mock: What everyone sees (demo/sample data)
• Private: What authorized users see (real data)

Example:

obj = syo.syobj(
    name="Customer Analysis", 
    mock_contents="Sample: 100 customers, avg age 42, 60% retention rate",
    private_contents="Full: 47,293 customers, avg age 41.7, 58.3% retention, avg LTV $1,247"
)

Interactive Object Browser

The syo.objects collection provides a beautiful interactive interface:

# Browse all objects with search and selection
syo.objects

# Search by name, email, description, or metadata
syo.objects.search("financial")
syo.objects.search("customer")

# Filter by email
syo.objects.filter_by_email("andrew")

# Get specific objects
selected = [syo.objects[i] for i in [0, 1, 5]]

# Refresh after creating new objects
syo.objects.refresh()

Interactive Features

• 🔍 Real-time search across names, emails, descriptions, and metadata
• ☑️ Multi-select with checkboxes
• 📋 Code generation - click "Generate Code" to get copy-paste Python
• 📊 10-column table with all object details
• 🔄 Auto-refresh - new objects appear immediately
• 📱 Responsive design with horizontal scrolling

Rich Object Display

Each object shows beautifully in Jupyter with:

• 📁 File information with availability status
• 🎯 Permission levels with color-coded badges
• 📋 Metadata including creation/update times
• 🖱️ Interactive buttons to view content inline
• 📄 Smart truncation for long content (first 1000 chars)

Permission Control

obj = syo.syobj(
    name="Financial Report",
    mock_contents="Q4 Summary: Revenue up 10%", 
    private_contents="Q4: $2.5M revenue, $400K profit, 23.7% margin",
    discovery_read=["public"],           # Who knows it exists
    mock_read=["[e-mail]"],  # Who sees demo
    private_read=["[e-mail]"]     # Who sees real data
)

File-Based Objects

# Use existing files
obj = syo.syobj(
    name="Dataset Analysis",
    mock_file="sample_100_rows.csv",      # Demo file
    private_file="full_50k_rows.csv"      # Real file  
)

Installation

pip install syft-objects

For SyftBox integration:

pip install syft-objects[syftbox]

🚀 SyftBox Auto-Installation

When you import syft-objects, it automatically checks if you have SyftBox installed locally and sets up the syft-objects app for you:

import syft_objects as syo  # Auto-installs to SyftBox if detected

What happens automatically:

• ✅ Detects SyftBox: Checks for ~/SyftBox/ directory
• ✅ Verifies app: Looks for existing ~/SyftBox/apps/syft-objects/
• ✅ Auto-clones: If missing, runs git clone https://github.com/OpenMined/syft-objects.git
• ✅ Silent operation: Only shows messages during installation
• ✅ Graceful fallback: Works fine without SyftBox too

Requirements for auto-installation:

• SyftBox installed locally (creates ~/SyftBox/ directory)
• Git available in your system PATH
• Internet connection for cloning

Manual installation (if needed):

cd ~/SyftBox/apps/
git clone https://github.com/OpenMined/syft-objects.git

Key Features

• 🎯 One function: syo.syobj() - simple and clean
• 🔒 Explicit control: You decide what goes in mock vs private
• 🎨 Beautiful display: Rich HTML widgets in Jupyter
• 🔍 Interactive browsing: Search and select objects easily
• 🆔 Unique filenames: No collisions with UID-based naming
• ⚡ Real-time updates: New objects appear immediately
• 📊 Comprehensive table: 10 columns with all object details
• 🖱️ Inline content: View file contents directly in notebook
• 🎯 Permission system: Fine-grained access control

Quick Reference

import syft_objects as syo

# Create objects
obj = syo.syobj(name="My Data", mock_contents="Demo", private_contents="Real")

# Browse interactively
syo.objects                    # Show interactive table
syo.objects[0]                 # Get first object  
syo.objects[:3]                # Get first 3 objects
len(syo.objects)               # Count objects

# Search and filter
syo.objects.search("keyword")           # Search everywhere
syo.objects.filter_by_email("user")    # Filter by email
syo.objects.get_by_indices([0,1,5])    # Get specific objects

# Utilities
syo.objects.list_unique_emails()       # List all emails
syo.objects.refresh()                  # Refresh collection

License

Apache License 2.0

Syft Objects

A distributed file discovery and addressing system for SyftBox.

Syft Objects allows you to discover, share, and access data objects across the SyftBox network through a simple Python interface and web UI.

Features

• 🔍 Object Discovery: Automatically find and catalog distributed syft objects
• 🌐 Web Interface: Modern, responsive UI for browsing and managing objects
• 🐍 Python API: Simple and intuitive Python interface
• 🔒 Privacy-First: Respects SyftBox privacy and permission models
• ⚡ Real-time: Live updates and synchronization
• 📱 Mobile-Friendly: Works seamlessly on all devices

Installation

Option 1: Direct Installation

pip install syft-objects

Option 2: Auto-Installation via SyftBox

If you have SyftBox installed, syft-objects will automatically install itself as a SyftBox app when you first import it:

import syft_objects  # This triggers auto-installation if SyftBox is present

Auto-Installation Details

When you import syft-objects, it will:

1. Check for SyftBox: Detect if SyftBox is installed on your system
2. Locate Apps Directory: Find your ~/SyftBox/apps/ directory
3. Auto-Clone: Automatically clone the syft-objects repository if not present
4. Graceful Fallback: Work normally even if SyftBox is not available

Requirements for Auto-Installation:

• SyftBox must be installed and configured
• Git must be available in your system PATH
• Internet connection for cloning the repository

Manual Installation (if auto-install fails):

cd ~/SyftBox/apps/
git clone https://github.com/OpenMined/syft-objects.git

The auto-installation feature makes it seamless to get started with syft-objects in SyftBox environments while maintaining compatibility with standalone usage.

Port Discovery

When syft-objects runs as a SyftBox app, it uses dynamic port assignment via the SYFTBOX_ASSIGNED_PORT environment variable. The system automatically:

1. Port Assignment: SyftBox assigns a port via SYFTBOX_ASSIGNED_PORT
2. Port Persistence: The run.sh script saves the assigned port to a .port file
3. Dynamic Discovery: The Python API automatically discovers the port by checking:
   • Current working directory: ./.port
   • SyftBox apps directory: ~/SyftBox/apps/syft-objects/.port
   • Package directory: <package_path>/.port
4. Fallback: Uses port 8003 if no port file is found

This ensures the widget and API calls always connect to the correct server instance.

Usage:

import syft_objects as syo

# Widget automatically uses the correct port
syo.objects.widget()  # Uses dynamic port discovery

# Manual port/URL access
port = syo.get_syft_objects_port()  # Get current port
url = syo.get_syft_objects_url("api/objects")  # Get API URL