Getting started

You need Python 3.9+ and pip. Basic pandas usage is enough to run rules.

Install

pip install open-fdd

Included with pip install open-fdd: pytest (running tests from a clone). The engine does not bundle rdflib; Brick .ttl → column_map is implemented in open-fdd-afdd-stack (openfdd_stack.platform.brick_ttl_resolver).

Install separately if you need them: matplotlib (notebooks / plotting), python-docx (Word export in report helpers), black / pre-commit (formatting — see repo config).

BACnet, FastAPI, and other platform dependencies live in open-fdd-afdd-stack, not in this package.

Run rules on a DataFrame

Rules are YAML (type, inputs, params, flag, …). In YAML, inputs are keyed by Brick-style names (e.g. Outside_Air_Temperature_Sensor). Your DataFrame almost certainly uses different column names — pass column_map so each rule input points at the right column.

import pandas as pd
from open_fdd import RuleRunner

df = pd.DataFrame(
    {
        "timestamp": pd.date_range("2024-01-01", periods=24, freq="h"),
        "oat": [72.0, 74.0, 95.0] + [55.0] * 21,  # one hot stretch
    }
)

runner = RuleRunner(
    rules=[
        {
            "name": "oat_too_hot",
            "type": "expression",
            "flag": "oat_too_hot_flag",
            "inputs": {
                "Outside_Air_Temperature_Sensor": {
                    "brick": "Outside_Air_Temperature_Sensor",
                },
            },
            "params": {"hi": 90.0},
            "expression": "Outside_Air_Temperature_Sensor > hi",
        }
    ]
)

df_out = runner.run(
    df,
    timestamp_col="timestamp",
    column_map={"Outside_Air_Temperature_Sensor": "oat"},
)
# df_out includes oat_too_hot_flag (and original columns)

Load rules from disk instead of inline dicts:

runner = RuleRunner("/path/to/dir/of/yaml/rules")
df_out = runner.run(df, timestamp_col="timestamp", column_map={...})

If a rule references columns you do not have, use skip_missing_columns=True so those rules are skipped instead of raising. More mapping options: Column map & resolvers. For CSV → DataFrame, use pandas.read_csv then runner.run(df, ...); sample data and notebooks: Examples.

The same YAML runs in the full Docker platform; there the stack builds column_map from the data model. Here you supply the map (dict or load_column_map_manifest / ManifestColumnMapResolver).

Clone and run tests (contributors)

git clone https://github.com/bbartling/open-fdd.git
cd open-fdd
python3 -m venv env && source env/bin/activate
pip install -U pip && pip install -e .
pytest

More detail: TESTING.md.

Full platform operators

Compose services, Caddy, BACnet, REST APIs, and the React UI are documented in open-fdd-afdd-stack:

bbartling.github.io/open-fdd-afdd-stack

PyPI releases (maintainers)

Version bumps, tags, and the companion openfdd-engine package (same code, alternate install name) are covered in the stack docs:

PyPI releases (open-fdd)