openbim-ifc (Python)

Read, edit and write IFC STEP files from Python. A thin binding over the openbim-ifc Rust crates: one abi3 wheel for CPython 3.9 and later.

pip install openbim-ifc

Documentation: Python guide · source and issues

from openbim_ifc import IfcModel, Text

with open("model.ifc", "rb") as f:
    model = IfcModel.parse(f.read())

print(model.schema, len(model))
for id in model.ids_of_type_including_subtypes("IfcWall"):
    print(id, model.attribute_by_name(id, "Name"))  # e.g. Text('Wall')

model.set_attribute_by_name(id, "Name", Text("Renamed"))
with open("out.ifc", "wb") as f:
    f.write(model.write())

Names resolve against the release the file's header declares, inherited attributes included; attribute_names(id) lists them in slot order with their optional and derived flags. attribute(id, index) and set_attribute(id, index, value) address a slot by position.

Entities, plain values, property sets

A pure-Python layer over those calls reads like IfcOpenShell:

import openbim_ifc
from openbim_ifc import Text

model = openbim_ifc.open("model.ifc")
for wall in model.by_type("IfcWall"):          # subtypes included
    print(wall.Name, wall.psets["Pset_WallCommon"]["IsExternal"])
wall.Name = Text("Renamed")                     # set_attribute_by_name
frame = model.to_dataframe("IfcWall")           # pip install 'openbim-ifc[pandas]'

Reads give plain Python values; a typed wrapper's type name, * and .U. (both None) and the text/enum distinction are lost there, and wall.raw("Name") keeps them. wall.psets and wall.qtos are read-only mappings with type values inherited; write with wall.set_property(...). The package is typed (py.typed) and checked with mypy --strict.

Values

Attribute values are frozen dataclasses, one per STEP form: Null ($), Derived (*), Bool, Unknown (.U.), Integer, Real, Text, Binary, Enum, Ref (#42), List, and Typed (IFCLENGTHMEASURE(2.5)). They keep every distinction IFC makes, so a file read and written back through Python is unchanged.

Bare Python values are refused with TypeError: 3 could be an Integer or a Real, "x" a Text or an Enum, and the binding does not guess.

Errors

Every failure raises IfcError with a stable code (parse, missing-entity, invalid-value, unsupported-schema, ...; the Python guide lists them all). The codes are shared with the JavaScript and C bindings; a code is never renamed or reused.

Threads

parse releases the GIL while it reads. A model can be passed between threads; the GIL serialises calls on it.

Scope

The record model -- entities, attributes, the STEP codec, exact and subtype queries -- plus lenient reads (ParseOptions), the STEP header, validation, ifcXML, the reachability lint and the read-only domain views (property sets and quantities, the spatial tree, classification, materials, systems, cost, georeferencing; #123), each as frozen dataclasses, writing property sets and quantities as one checked transaction (set_properties), the Pythonic layer over them (#332), and creating entities as one checked transaction (author with AuthorOps: entities by type and named attributes, the spatial structure, placed, contained and typed products, removal with relationships; #330), and each product's world placement and Body (product_placements; meshes from a --features mesh wheel, #328).

Build from source

uv venv && . .venv/bin/activate
uv pip install maturin
maturin develop --release     # or: scripts/check-python.sh to build + test

Faster reading (opt-in)

maturin build --release --features rusty_alloc

Uses the pure-Rust rusty_alloc allocator for the extension's own memory: reading STEP into a model takes 18-35% less CPU time on seven real IFC files, at 1-7% less peak memory (#49). Python's own allocator is not replaced. Off by default because a global allocator is a build-time choice; the version is pinned exactly.