Package Pre-download
Suitable for environments without network connection and without Docker support.
This method requires downloading all dependency packages in an environment with network access, then transferring them to the offline environment for installation.
Step 1: Download Packages in Network-Connected Environment
We provide a wheel downloader tool to prepare all dependency packages in advance:
Download Core Dependencies
# Download packages for Linux environment
python demo/petsard_wheel_downloader.py --branch main --python-version 3.11 --os linux
# Download packages for Windows environment
python demo/petsard_wheel_downloader.py --branch main --python-version 3.11 --os windows
# Download packages for macOS Intel environment
python demo/petsard_wheel_downloader.py --branch main --python-version 3.11 --os macos
# Download packages for macOS Apple Silicon environment
python demo/petsard_wheel_downloader.py --branch main --python-version 3.11 --os macos-armDownload Additional Dependency Groups
# Download data science related packages
python demo/petsard_wheel_downloader.py --branch main --python-version 3.11 --os linux --groups ds
# Download complete packages (including all features)
python demo/petsard_wheel_downloader.py --branch main --python-version 3.11 --os linux --groups ds all
# Download development tools
python demo/petsard_wheel_downloader.py --branch main --python-version 3.11 --os linux --groups devParameter Descriptions:
| Parameter | Description | Examples |
|---|---|---|
--branch | Git branch name | main, dev |
--python-version | Python version | 3.10, 3.11, 3.11.5 |
--os | Target operating system | linux, windows, macos, macos-arm |
--groups | Optional dependency groups | ds, all, dev (can specify multiple separated by spaces) |
Operating System Support:
linux: Linux 64-bitwindows: Windows 64-bitmacos: macOS Intelmacos-arm: macOS Apple Silicon
Dependency Group Descriptions:
- Default (no specification needed): Core functionality for configuration, data loading, synthesis, and evaluation
ds: Data science features + Jupyter Notebook supportall: Complete features + extended support (benchmark datasets, Excel file support)dev: Testing and development tools
Download Location
Downloaded packages are saved in the wheels/ directory by default, structured as follows:
wheels/
├── petsard-x.y.z-py3-none-any.whl
├── pandas-x.y.z-cp311-cp311-linux_x86_64.whl
├── numpy-x.y.z-cp311-cp311-linux_x86_64.whl
└── ... (other dependency packages)Step 2: Transfer Package Files
Transfer the entire wheels/ directory to the offline environment via USB drive, internal network, or other means.
Step 3: Install in Offline Environment
Method A: Direct Installation with pip (Recommended)
# Navigate to wheels directory
cd wheels/
# Install all packages
pip install *.whl
# Or install PETsARD and its dependencies
pip install --no-index --find-links=. petsardMethod B: Using pip with Local Directory
# Install from wheels directory
pip install --no-index --find-links=./wheels petsard
# Install specific group packages
pip install --no-index --find-links=./wheels petsard[ds]Verify Installation
# Check if PETsARD is correctly installed
python -c "import petsard; print('✅ PETsARD installed, version:', petsard.__version__)"
# Check if Executor is available
python -c "from petsard.executor import Executor; print('✅ Executor available')"Troubleshooting
No Matching Package Version Found
Issue: Installation prompts no matching package version found
Solution:
- Ensure the Python version used for download matches the offline environment
- Ensure the operating system used for download matches the offline environment
- Re-download the correct version of packages
Dependency Conflicts
Issue: Installation prompts package dependency conflicts
Solution:
# Force installation, ignoring dependency checks (use with caution)
pip install --no-deps --no-index --find-links=./wheels petsard
# Then manually install missing dependencies
pip install --no-index --find-links=./wheels <missing-package-name>Permission Issues
Issue: No permission to install to system Python
Solution:
# Install to user directory
pip install --user --no-index --find-links=./wheels petsard
# Or use virtual environment (recommended)
python -m venv petsard-env
source petsard-env/bin/activate # Linux/macOS
# or
petsard-env\Scripts\activate # Windows
pip install --no-index --find-links=./wheels petsardAdvanced Usage
Download Specific PETsARD Version
# Download packages from specific branch
python demo/petsard_wheel_downloader.py --branch v1.0.0 --python-version 3.11 --os linux
# Download development version
python demo/petsard_wheel_downloader.py --branch dev --python-version 3.11 --os linuxBatch Download for Multiple Platforms
# Download packages for multiple platforms
for os in linux windows macos macos-arm; do
python demo/petsard_wheel_downloader.py \
--branch main \
--python-version 3.11 \
--os $os \
--groups ds
doneNext Steps
After installation, you can:
- Check the Getting Started guide for detailed examples
- Visit the PETsARD YAML documentation to learn about configuration
- Explore benchmark datasets for testing
- Review example configurations in the GitHub repository