Skip to main content

​
Overview

The configuration module provides utilities for reading and writing TOML configuration files, particularly for managing Flower and Syft-Flwr project metadata.

​
Functions

​
load_flwr_pyproject

Load and validate a Flower project’s pyproject.toml file. 
def load_flwr_pyproject(
    path: Path,
    check_module: bool = True
) -> dict

​
Parameters

​
path
Path
required
Path to the project directory or pyproject.toml file. If a directory is provided, it will automatically look for pyproject.toml inside.
​
check_module
bool
default:"True"
Whether to validate that the module references in the configuration exist. Set to False to skip module validation (useful during testing or parallel execution).

​
Returns

​
pyproject
dict
Parsed and validated pyproject.toml configuration as a dictionary.

​
Usage Example

From run.py:22-25: 
from syft_flwr.config import load_flwr_pyproject
from pathlib import Path

flower_project_dir = Path("./fl-diabetes-prediction")
pyproject_conf = load_flwr_pyproject(flower_project_dir)

# Access configuration values
client_ref = pyproject_conf["tool"]["flwr"]["app"]["components"]["clientapp"]
server_ref = pyproject_conf["tool"]["flwr"]["app"]["components"]["serverapp"]
app_name = pyproject_conf["tool"]["syft_flwr"]["app_name"]
datasites = pyproject_conf["tool"]["syft_flwr"]["datasites"]
aggregator = pyproject_conf["tool"]["syft_flwr"]["aggregator"]

​
Exceptions

​
Exception
Raised if the configuration is invalid or module references cannot be found (when check_module=True).

​
load_toml

Load any TOML file into a dictionary. 
def load_toml(path: str) -> dict

​
Parameters

​
path
str
required
Path to the TOML file to load.

​
Returns

​
config
dict
Parsed TOML file as a dictionary.

​
Usage Example

from syft_flwr.config import load_toml

config = load_toml("./config.toml")
print(config["section"]["key"])

​
write_toml

Write a dictionary to a TOML file. 
def write_toml(path: str, val: dict) -> None

​
Parameters

​
path
str
required
Path to the TOML file to write.
​
val
dict
required
Dictionary to serialize and write to the TOML file.

​
Returns

None - The function writes to the file in place.

​
Usage Example

From bootstrap.py:83: 
from syft_flwr.config import write_toml

pyproject_conf = {
    "tool": {
        "syft_flwr": {
            "app_name": "ds@example.com_my-project_1234567890",
            "datasites": ["do1@example.com", "do2@example.com"],
            "aggregator": "ds@example.com",
            "transport": "syftbox"
        }
    }
}

write_toml("./pyproject.toml", pyproject_conf)

​
Configuration Structure

After bootstrapping, your pyproject.toml will contain both Flower and Syft-Flwr configurations:

​
Flower Configuration

[tool.flwr.app.components]
serverapp = "your_project.server_app:app"
clientapp = "your_project.client_app:app"

[tool.flwr.app.config]
partition-id = 0
num-partitions = 1

​
Syft-Flwr Configuration

[tool.syft_flwr]
app_name = "ds@openmined.org_fl-diabetes-prediction_1753435643"
datasites = [
    "do1@openmined.org",
    "do2@openmined.org",
]
aggregator = "ds@openmined.org"
transport = "syftbox"  # or "p2p"

​
Project Dependencies

[project]
name = "fl-diabetes-prediction"
dependencies = [
    "flwr>=1.0.0",
    "syft_flwr==0.1.0",  # Added automatically by bootstrap
    # ... other dependencies
]

​
Transport Types

​
syftbox
string
Default for local/server environmentsUses local SyftBox file synchronization with RPC and end-to-end encryption. Best for:
  • Local development and testing
  • Server deployments with SyftBox installed
  • Maximum security with E2E encryption
​
p2p
string
Default for Google ColabUses peer-to-peer synchronization via cloud storage (Google Drive/OneDrive). Best for:
  • Google Colab environments
  • Remote collaboration without SyftBox
  • Cloud-based deployments

​
Validation

The load_flwr_pyproject function validates:
  1. File existence - pyproject.toml must exist
  2. Required sections - Must have [tool.flwr.app] configuration
  3. Component references - serverapp and clientapp must be valid Python module paths
  4. Module availability - Referenced modules must be importable (when check_module=True)

​
Notes

  • Configuration files are written in binary mode for proper TOML encoding
  • Validation warnings are logged but don’t prevent loading
  • Module checking can be disabled for faster loading during testing

Build docs developers (and LLMs) love

Get started for freeTalk to us