sysmlpy is an open source pure Python library for constructing python-based classes consistent with the SysML v2.0 standard.
This project began as a fork of the sysml2py project by Christopher Cox. Since April 2026 Jon Fox decided to complete coverage of all SysMLv2 features over two months of weekends, and dropped the textX parser in favor of an ANTLR4 parser grammar and changed our unit library to pint. The project had diverged so much from sysml2py that a new name, sysmlpy, was selected.
v0.19.0: Semantic analysis engine with undefined symbol detection. Import resolution (namespace ::*, membership, recursive ::*::**). 530 tests passing. Symbol table with hierarchical scope resolution and qualified name lookup.
v0.17.0: 100% test suite pass rate (487/487). Cayley graph database storage backend via HTTP API. Full grammar round-trip coverage (56/56 tests). Programmatic API consistency fixes. NetworkXStore bug fix.
v0.16.0: 100% grammar round-trip test coverage (56/56). Analysis case usage, trade study, calculation redefinition, and case body member support. Import visibility defaults to private per SysML v2 spec.
v0.15.0: ISQ unit validation (300+ type-to-dimension mappings), US Customary unit support (21 custom definitions), PlantUML diagram generation with stereotype-based styling, and comprehensive API documentation.
sysmlpy requires the following Python packages:
- networkx — graph analysis backend (install with
pip install sysmlpy[graph]) - kuzu — embedded graph database with disk persistence and Cypher queries (install with
pip install sysmlpy[kuzu]) - cayley — graph database via HTTP API, supports BoltDB/LevelDB backends (install with
pip install sysmlpy[cayley]) - PlantUML v1.2020.0+ — diagram rendering (requires Java + PlantUML JAR or PlantUML server). The generator uses
<style>blocks andskinparamstereotype selectors introduced in v1.2020.
Multiple installation methods are supported by sysmlpy, including:
| Logo | Platform | Command |
|---|---|---|
 |
PyPI | python -m pip install sysmlpy |
|
PyPI | python -m pip install sysmlpy[graph] (with graph analysis) |
|
PyPI | python -m pip install sysmlpy[cayley] (with Cayley graph DB) |
 |
GitHub | python -m pip install https://github.com/mycr0ft/sysmlpy/archive/refs/heads/main.zip |
Documentation can be found here.
The code below will create a part called Stage 1, with a shortname of <'3.1'> referencing a specific requirement or document. It has a mass attribute of 100 kg. It has a thrust attribute of 1000 N. These attributes are created and placed as a child of the part. Next, we recall the part value for thrust and add 199 N. Finally, we can dump the output from this class.
from sysmlpy import Attribute, Part, ureg
a = Attribute(name='mass')
a.set_value(100 * ureg.kilogram)
b = Attribute(name='thrust')
b.set_value(1000 * ureg.newton)
c = Part(name="Stage_1", shortname="'3.1'")
c._set_child(a)
c._set_child(b)
v = "Stage_1.thrust"
c._get_child(v).set_value(c._get_child(v).get_value() + 199 * ureg.newton)
print(c.dump())
It will output the following:
part <'3.1'> Stage_1 {
attribute mass= 100 [kilogram];
attribute thrust= 1199.0 [newton];
}
The package is able to handle Items, Parts, and Attributes.
a = Part(name='camera')
b = Item(name='lens')
d = Attribute(name='mass')
c = Part(name='sensor')
c._set_child(a)
c._set_child(b)
a._set_child(d)
print(c.dump())
will return:
part sensor {
part camera {
attribute mass;
}
item lens;
}
Actions (activities) can be defined with input and output parameters::
from sysmlpy import Action
# Action definition with typed inputs/outputs
a = Action(definition=True, name='Focus')
a.add_input('scene', 'Scene')
a.add_output('image', 'Image')
print(a.dump())
# → action def Focus { in scene : Scene; out image : Image; }
# Action usage with references
b = Action(name='TakePicture')
b.add_input('scene')
b.add_output('picture')
print(b.dump())
# → action TakePicture { in scene; out picture; }
References can reference other elements::
from sysmlpy import Reference, Item
# Simple reference
r = Reference(name='driver')
print(r.dump())
# → ref driver;
# Reference with type
person = Item(name='Person')
r2 = Reference(name='driver')
r2.set_type(person)
print(r2.dump())
# → ref driver : Person;
# Reference redefinition
r3 = Reference(name='payload', redefines=True)
r3.set_type(person)
print(r3.dump())
# → ref :>> payload : Person;
loads() parses SysML v2 text and classtree() converts the result back to text. This round-trip is the basis for the grammar test suite.
from sysmlpy import loads
from sysmlpy.formatting import classtree
text = """package 'Action Example' {
action def Focus { in scene : Scene; out image : Image; }
action def Shoot { in image: Image; out picture : Picture; }
action def TakePicture {
in item scene : Scene;
out item picture : Picture;
bind focus.scene = scene;
action focus : Focus { in scene; out image; }
flow focus.image to shoot.image;
first focus then shoot;
action shoot : Shoot { in image; out picture; }
bind shoot.picture = picture;
}
}"""
model = loads(text)
tree = classtree(model)
print(tree.dump())100% of the 56 grammar round-trip tests pass (56/56), covering packages, parts, items, ports, interfaces, binding connectors, flow connections, all action forms (definition, shorthand, succession, decomposition), expressions, calculations, constraints, state definitions, requirements, analysis cases, and trade studies.
sysmlpy includes a semantic analysis engine that detects undefined symbol references in parsed models. It builds a hierarchical symbol table from the model tree, resolves imports, and cross-references all type/subsetting/redefinition references.
from sysmlpy import loads, analyze
model = loads("""
package Types {
part def Engine;
}
package Vehicle {
import Types::*;
part myCar : Engine; # resolved via import
part myWheel : Wheel; # undefined!
}
""")
issues = analyze(model)
for issue in issues:
print(f"[{issue.severity}] {issue.code}: {issue.message}")
# → [error] UNDEFINED_SYMBOL: Undefined symbol 'Wheel' referenced in Part 'myWheel'The analyzer resolves three import patterns:
| Pattern | Example | Effect |
|---|---|---|
| Namespace import | import Types::* |
Makes all symbols from Types visible |
| Membership import | import Types::Engine |
Makes only Engine visible |
| Recursive import | import Types::*::** |
Imports from Types and all nested packages |
- Hierarchical scopes: Each package and definition creates a child scope
- Parent chain lookup: References resolve through parent scopes
- Qualified names:
P::AandOuter::Inner::DeepPartresolve correctly - Cross-package:
import A::B::*works across arbitrary package depth - Library whitelist: 80+ standard library symbols (
ScalarValues::Integer,ISQ::LengthValue, etc.) are pre-recognized
sysmlpy provides a unified Store protocol with multiple backend implementations. All backends support the same API: put, get, delete, children, parents, relationships, query, has, ids, clear, plus graph traversal methods (descendants, ancestors, path).
from sysmlpy.store import create_store
# In-memory (default, zero dependencies)
store = create_store("memory")
# NetworkX graph (analysis, shortest paths, centrality)
store = create_store("networkx")
# Kuzu embedded graph DB (disk persistence, Cypher queries)
store = create_store("kuzu", database="/tmp/model.db")
# Cayley remote graph DB (HTTP API, BoltDB/LevelDB backends)
store = create_store("cayley", host="localhost", port=64210)Dict-based backend with O(1) lookups. Zero external dependencies. Ideal for testing and small models.
Graph backend using NetworkX MultiDiGraph. Enables graph analysis algorithms:
from sysmlpy.store import NetworkXStore
store = NetworkXStore()
store.put(eid, {"name": "Engine", "sysml_type": "part"})
# Graph analysis
components = store.connected_components()
centrality = store.centrality()
cycles = store.cycles()
stats = store.stats() # nodes, edges, density, avg_degree
subgraph = store.subgraph([eid1, eid2])
store.export_graphml("model.graphml")Embedded graph database with disk persistence. Uses Cypher for queries. Data survives across process restarts.
from sysmlpy.store import KuzuStore
# Persistent database
store = KuzuStore(database="/path/to/model.db")
# In-memory mode
store = KuzuStore()Remote graph database backend communicating with a Cayley server over HTTP. Supports any Cayley backend (BoltDB, LevelDB, in-memory). Uses the quad model (subject, predicate, object, label) for flexible data representation.
from sysmlpy.store import CayleyStore
# Connect to local Cayley server
store = CayleyStore()
# Custom host/port with namespace isolation
store = CayleyStore(host="cayley.example.com", port=64210, label="my_project")
# Graph analysis
store.put(eid, {"name": "Wheel", "sysml_type": "part"})
descendants = store.descendants(root_id)
ancestors = store.ancestors(leaf_id)
path = store.path(source_id, target_id)
components = store.connected_components()
cycles = store.cycles()
centrality = store.centrality()
store.export_graphml("model.graphml")Running Cayley with Docker:
# In-memory backend
docker run -p 64210:64210 --rm cayley/cayley
# Persistent BoltDB backend
docker run -p 64210:64210 -v /data:/data --rm cayley/cayley -db boltdb -dbpath /data/cayley.dbQuad Model: Elements are stored as quads where the subject is the element UUID, predicates are property names (e.g., name, sysml_type), and objects are property values. Relationships are stored as quads where the predicate is the relationship type (e.g., parent_child, typed_by). Labels provide namespace isolation for multi-tenant scenarios.
Generate SysML v2 structure diagrams from parsed models using the built-in PlantUML generator. Definitions render with sharp corners and usage elements with rounded corners. Relationships are differentiated by arrow style, thickness, and color — following the official SysML v2 Pilot Implementation approach.
from sysmlpy import loads
from sysmlpy.plantuml import PlantUMLGenerator
text = """package Vehicle {
part def Wheel {
attribute radius : LengthValue;
attribute pressure : PressureValue;
}
part def BrakeSystem {
attribute padThickness : LengthValue;
}
part def VehicleAssembly {
part frontLeft : Wheel;
part frontRight : Wheel;
part brakes : BrakeSystem;
}
part myVehicle : VehicleAssembly;
}"""
model = loads(text)
gen = PlantUMLGenerator(model, title="Vehicle Structure")
print(gen.generate())Produces PlantUML source that renders as:
@startuml
skinparam RoundCorner 0
skinparam rectangle<<(D,#8B4513) part def>> {
RoundCorner 0
BackgroundColor #FFF8F0
BorderColor #8B4513
}
skinparam rectangle<<(P,#32CD32) part>> {
RoundCorner 15
BackgroundColor #F0FFF0
BorderColor #32CD32
}
title Vehicle Structure
rectangle "Wheel" as Wheel <<(D,#8B4513) part def>>
rectangle "BrakeSystem" as BrakeSystem <<(D,#8B4513) part def>>
rectangle "VehicleAssembly" as VehicleAssembly <<(D,#8B4513) part def>>
rectangle "myVehicle" as myVehicle <<(P,#32CD32) part>>
VehicleAssembly *-- frontLeft : owns
VehicleAssembly *-- frontRight : owns
VehicleAssembly *-- brakes : owns
myVehicle --:|> VehicleAssembly : types
legend right
<b>Legend</b>
|= Element |= Notation |
| <<(D,#8B4513) part def>> | Definition (type) |
| <<(P,#32CD32) part>> | Usage (instance) |
| --:|> | Feature typing |
| *-- | Composite containment |
endlegend
@endumlThe generator supports filtering to highlight specific elements:
# Show only elements related to 'myVehicle'
gen = PlantUMLGenerator(model, focus="myVehicle")
# Show a custom set of elements
gen = PlantUMLGenerator(model, elements=["Wheel", "BrakeSystem"])
# Limit nesting depth
gen = PlantUMLGenerator(model, max_depth=2)See docs/plantuml-examples/ for 9 rendered examples covering usage vs definition, relationships, vehicle structure, requirements, interconnections, state machines, and activity diagrams.
100% of 123 OMG XPect conformance tests pass (123/123).
sysmlpy is released under the MIT license, hence allowing commercial use of the library.