bimfabrikhh_core_rs

A Rust library with Python bindings: parse CityGML 1.0 / 2.0 and CityJSON 1.1 and write IFC4 STEP.

About This Project

bimfabrikhh_core_rs is developed and maintained by Freie und Hansestadt Hamburg, Landesbetrieb Geoinformation und Vermessung (LGV), BIM-Leitstelle

This library is part of BIMFabrikHH. The core is implemented in Rust; Python bindings expose the conversion API. It converts urban building models to IFC4 STEP.

This project is in an early stage of development. The API and features will keep evolving.

Author: Ahmed Salem (BIM-Leitstelle, LGV Hamburg)

Quick Install

pip install bimfabrikhh_core_rs

Optional pydantic helpers for property-set specs, IFC type maps, and owner identity:

pip install "bimfabrikhh_core_rs[psets]"

Wheels are published for Linux x86_64, Windows x64, and macOS Apple Silicon. One wheel per platform works for Python 3.11+.

Overview

The conversion API has no Python runtime dependencies and does not use ifcopenshell. Geometry is parsed and written in Rust. You fetch the data; the package converts it. There is no HTTP inside the library.

• CityGML 1.0 / 2.0 and CityJSON 1.1 → one IFC4 STEP file
• mode="typed" — one product per semantic surface (IfcWall, IfcRoof, IfcSlab, …)
• mode="mesh" — one IfcBuildingElementProxy per building
• Property sets, colours, CAD layer, and owner history are opt-in
• Coordinates stay in the CRS they arrive in; epsg only tags the file

Public API

Conversion

• gml_to_ifc(path, ifc_path, ...) — CityGML file or list of files → IFC
• cityjson_to_ifc(source, ifc_path, ...) — CityJSON path, list of paths, or document text → IFC
• write_ifc(products, path, ...) — IFC from product dicts you built yourself
• trees_to_ifc(trees, ifc_path, ...) — trunk + crown meshes → IFC (TreesGenericApp path). Optional drape_vertices / drape_faces sets z = 0 from the DGM triangle (else nearest vertex).
• terrain_to_ifc(vertices, faces, ifc_path, ...) — triangulated DGM → IFC (TerrainGenericApp write path)

Parse

• parse_gml(path, ...) — CityGML → list of building dicts
• parse_cityjson(path, ...) — CityJSON file → building dicts
• parse_cityjson_text(text, ...) — CityJSON already in memory (after HTTP)

Optional helpers (bimfabrikhh_core_rs[psets])

• PsetSpec / Attr / Const / Quantity — declarative property-set rules
• hamburg_defaults() / attribute_mapping.specs() / tree_mapping.specs() / terrain_mapping.specs() — Hamburg property names
• hamburg_ifc_types() / IfcTypeMap — surface type → IFC entity
• OwnerIdentity / owner.defaults() — FILE_NAME + IfcPerson / IfcOrganization / IfcApplication

Key Features

• No HTTP — fetch tiles yourself; pass paths or CityJSON text
• Typed or mesh — per-surface products or one block per building
• Declarative psets — omit psets= and none are written
• Deterministic GUIDs — the same input produces a byte-identical IFC
• Hamburg-ready — optional type map, property mapping, _BIM_Stadtmodell layer, and Nullpunkt

Complete Working Example

from bimfabrikhh_core_rs import cityjson_to_ifc, gml_to_ifc, parse_cityjson_text

# CityGML tile → typed IFC (IfcWall / IfcRoof / IfcSlab per surface)
gml_to_ifc("tile.gml", "out.ifc", mode="typed", epsg=25833)

# CityJSON → one IfcBuildingElementProxy mesh per building
cityjson_to_ifc("page.city.json", "blocks.ifc", mode="mesh", epsg=25832)

# Inspect before writing
buildings = parse_cityjson_text(open("page.city.json").read())
print(len(buildings), buildings[0]["gml_name"])

Crop, colours, and Hamburg types:

from bimfabrikhh_core_rs import cityjson_to_ifc
from bimfabrikhh_core_rs.attribute_mapping import specs
from bimfabrikhh_core_rs.ifc_types import hamburg_ifc_types

cityjson_to_ifc(
        "page.city.json",
        "hamburg.ifc",
        mode="typed",
        epsg=25832,
        bbox_epsg=(565012.0, 5933559.0, 566570.0, 5934571.0),
        basepoint=(565012.0, 5933559.0),
        ifc_types=hamburg_ifc_types(),
        psets=specs(),
        cad_layer=True,
)

mode is "typed" or "mesh". Omit ifc_types and every product is IfcBuildingElementProxy. Omit psets and the IFC has no property sets. bbox_epsg is (min_x, min_y, max_x, max_y) in the projected CRS. Reprojection is out of scope — use pyproj on the caller side if you need to transform.

Trees (same look as BIMFabrik TreesGenericApp; Python still prepares records):

from bimfabrikhh_core_rs import trees_to_ifc
from bimfabrikhh_core_rs.tree_mapping import specs

trees_to_ifc(
    [
        {
            "name": "Baum_001",
            "position": (565100.0, 5933600.0, 0.0),
            "trunk_radius": 0.3,
            "trunk_height": 4.0,
            "crown_radius": 2.5,
            "detail": 1,
            "segments": 8,
            "attributes": {"baumnummer": "001", "gattung_deutsch": "Eiche", "art_baum": "Quercus robur"},
        }
    ],
    "trees.ifc",
    epsg=25832,
    psets=specs(),
    basepoint=(565100.0, 5933600.0),
    # drape_vertices=mesh.vertices,  # optional: lift z=0 from nearest DGM sample
)

Terrain (Python still samples / Delaunay; Rust only writes IFC):

from bimfabrikhh_core_rs import terrain_to_ifc
from bimfabrikhh_core_rs.terrain_mapping import specs

terrain_to_ifc(vertices, faces, "dgm.ifc", epsg=25832, psets=specs(), basepoint=(565000.0, 5933000.0))

Property sets

Pass psets= a list of specs describing rules rather than values. The spec is validated once and evaluated in Rust for every building and surface.

from bimfabrikhh_core_rs import cityjson_to_ifc
from bimfabrikhh_core_rs.psets import Attr, Const, PsetSpec, Quantity

building = PsetSpec(
        name="Pset_Objektinformation",
        scope="building",
        properties={
                "_IDEbene1": Const("Stadtmodell"),
                "_StadtmodellLoD": Attr("lod"),
                "_Dachform": Attr("roof_type", catalog="dachform"),
                "_GrundhoeheNN": Attr("extra:Grundhöhe NN"),
        },
)
surface = PsetSpec(
        name="BIMFabrikHH_Quantities",
        scope="surface",
        properties={
                "GrossArea": Quantity("area", decimals=4),
                "SurfaceType": Quantity("surface_type"),
        },
)

cityjson_to_ifc("page.city.json", "out.ifc", epsg=25832, psets=[building, surface])

A missing source field omits that property. A plain dict works without pydantic.

Other options

from bimfabrikhh_core_rs import cityjson_to_ifc
from bimfabrikhh_core_rs.ifc_types import IfcTypeMap
from bimfabrikhh_core_rs.owner import defaults

cityjson_to_ifc(
        "page.city.json",
        "out.ifc",
        epsg=25832,
        ifc_types=IfcTypeMap(RoofSurface="IfcRoof", WallSurface="IfcWall"),
        colors={"RoofSurface": (1.0, 0.0, 0.0)},
        owner=defaults(),
)

Default colours match BIMFabrikHH: roofs red, everything else yellow. styles=False writes no colour. Owner history defaults to Ahmed Salem / BIM-Leitstelle, LGV Hamburg; owner_history=False skips it.

Examples

Additional examples are available in the GitHub repository.

Links

• PyPI: https://pypi.org/project/bimfabrikhh-core-rs/
• GitHub: https://github.com/LGV-BIM-Leitstelle/bimfabrikhh_core_rs
• Issues: https://github.com/LGV-BIM-Leitstelle/bimfabrikhh_core_rs/issues

License

This project is licensed under the GNU Lesser General Public License v2.1 (LGPL-2.1).

Copyright (C) 2026 Freie und Hansestadt Hamburg, Landesbetrieb Geoinformation und Vermessung BIM-Leitstelle, Ahmed Salem