umlmeta
Clean-room Python implementation of the UML 2.5.1 metamodel and the SysML v1 → v2 transformation, generated from OMG-published XMI and validated against the sysmlpy v2 reader.
No vendor tooling, no UML/SysML libraries — the generated packages are pure stdlib, and every semantic in them is traceable to an OMG document.
What's inside
| Piece | What it is | Checks |
|---|---|---|
umlmeta.uml25 (src/umlmeta/uml25.py) |
UML 2.5.1 metamodel as Python classes — generated by generate.py from the OMG UML XMI: 242 metaclasses, 620 reference descriptors, 22 derived unions computed at read time, 328 hidden association ends (the subset/redefine lattice is closed by the generator), 449 OCL constraints carried as metadata |
check.py 45 |
umlmeta.standard_profile, umlmeta.sysml, umlmeta.uaf, umlmeta.uaf13 |
StandardProfile + SysML v1 + UAF 1.2 (256) + UAF 1.3 (272) profile stereotypes, generated by generate_profiles.py; stereotypes fold their base metaclasses via Python inheritance, including cross-profile folds (UAF generalizes SysML stereotypes, e.g. Capability(sysml.Block, ...)); 1.3 adds the Mission-domain stereotypes and carries the http://www.omg.org/spec/UAF/20241101/UAF URI |
check_profiles.py 60 |
umlmeta.v1_to_v2 |
Clean-room SysML v1 → v2 textual-notation emitter anchored to OMG SysML v2.0 Beta 4, Part 2: SysML v1 to SysML v2 Transformation (ptc/2025-04-07). Every emitted form is calibrated against sysmlpy; anything without a normative mapping raises UnmappedFeature instead of guessing |
check_v1_to_v2.py 47 |
umlmeta.xmi21 |
XMI 2.1/EMF instance reader (dialect-tolerant: unprefixed containment, EMF feature defaults, cross-file href markers, stereotype applications) → feeds the emitter end-to-end |
check_dodaf.py 32 |
umlmeta.mdzip |
MagicDraw/Cameo .mdzip scrape + forensics: member inventory, model-snapshot extraction with Nomagic→OMG namespace aliasing, proxy/project-usages (local:/PROJECT-… dependencies), version-era fingerprints (ID prefixes like _2021x_2_…), and hex/base64 payload extraction (SVG/PNG/JPEG/ZIP) with a manifest |
check_mdzip.py 28 |
umlmeta.mdzip_import |
.mdzip → umlmeta.uml25 objects: merges delta-snapshot members (keep-last dedup), retags Nomagic's XMI-2.0/xsi spellings to the reader's dialect, injects xmi:type from tags/features and coerces raw-ID types (metamodel-as-map + property signature), scrubs cross-project/name-typed/profile-layer refs into synthetic external markers, and delegates to read_xmi21 — provenance (usages, eras) travels with the model |
check_mdzip_import.py 26 |
umlmeta.mdzip_diff |
Semantic diff of two .mdzip files: flatten to (metaclass, qualified-path, feature) records with xmi:ids discarded entirely, then match by qualified path → metaclass+name moved → same-owner similarity → cross-owner same-name similarity, and report added/removed/renamed/moved/feature-changed. Survives TeamworkCloud-style id churn by construction | check_mdzip_diff.py 12 |
umlmeta.xmi_write |
XMI 2.1/EMF instance writer — the read/write pair round-trips both OMG dialects (DoDAFLibrary 2.1 form and MeasurementsLibrary 2.5.1-era form) with canonical structural parity (canon.py); writes stereotype applications (both forms, tags, _URI/_EXTENSIONS-driven) and profile applications; write_cameo_xmi(...) adds the Cameo/MagicDraw 2024x dialect (UML 2.5.1 namespace pair, uml:Model project wrapper, P10 applications — see docs/CAMEO_XMI.md) |
check_xmi_write.py 48, check_xmi_cameo.py 15 |
umlmeta.bpmn |
BPMN 2.0.2 metamodel — OMG publishes BPMN as CMOF XMI (BPMN20.cmof), not a UML profile; generate_bpmn.py is a clean-room CMOF generator: 137 classes, 9 enums, 193 associations, opposite wiring, deliberately NOT UML Elements |
check_bpmn.py 25 |
umlmeta.mm_write |
CMOF writer — reconstructs OMG BPMN20.cmof from the generated metamodel with canonical structural parity (canon.py, member order as set); _SYNTH carries the raw serialization forms the runtime tables don't. write_emof adds the EMOF dialect (MOF 2.0/XMI §6.5.2 rules: derived and defaults not serialized, no Associations) |
check_cmof_write.py 20, check_emof_write.py 18 |
umlmeta.derived |
UML 2.5.1 normative derivations over umlmeta.uml25 objects (spec OCL anchors): ownership chains, qualified names (template-aware), generalization closures, normative member (inherited/imported), redefinition-aware derived-union evaluation |
check_derived.py 42 |
umlmeta.query |
Query helpers over loaded instance graphs: deterministic containment walk, find (metaclass by class or name, stereotype labels, qualified names, predicates), stereotypes label resolution |
check_query.py 19 |
umlmeta.sysml2 |
SysML v2 abstract syntax (KerML + SysML), generated by generate_sysml2.py from the OMG normative CMOF XMI (conformance R2): 175 classes, KerML ownership/membership wiring at runtime (OwningMembership/FeatureMembership containment, FeatureTyping typing, Subclassification specialization), derived-union reads (ownedElement, feature) computed live |
check_sysml2.py 26 |
umlmeta.v1_to_v2_as |
SysML v1 → v2 abstract-syntax transformer (conformance R3, ptc/2025-04-07 §2): transforms umlmeta.uml25+umlmeta.sysml objects (R1 artifacts) into umlmeta.sysml2 objects (R2 artifacts) — Block→PartDefinition, Property→PartUsage/AttributeUsage/OccurrenceUsage via FeatureMembership + FeatureTyping, Generalization→Subclassification, Comment→Documentation; one AS object per v1 element, same honesty rule as the textual emitter |
check_v1_to_v2_as.py 33 |
481 checks, all green. REPORT.md is the detailed build journal: dialect facts,
mapping tables with spec anchors, calibration evidence, and honest elisions.
Quick start
The check suites for the metamodel and profiles are stdlib-only; the emitter
suites validate with sysmlpy (the
authoritative SysML v2 reader), and the generator/reader need lxml:
python check.py # UML metamodel: 45 checks (stdlib only)
python check_profiles.py # profiles (Std + SysML + UAF 1.2 + UAF 1.3): 60 checks (stdlib only)
python check_bpmn.py # BPMN 2.0.2 CMOF metamodel: 25 checks (stdlib only)
python check_sysml2.py # SysML v2 abstract syntax: 26 checks (stdlib only)
python check_emof_write.py # EMOF writer: 18 checks (stdlib only)
python check_ocl.py # OCL translation: 19 checks (stdlib only)
# these need the sysmlpy reader and/or the OMG corpora
$SYSMLPY_PYTHON check_v1_to_v2.py # v1->v2 emitter: 47 checks
$SYSMLPY_PYTHON check_dodaf.py # XMI end-to-end: 32 checks
# (needs lxml + DoDAFLibrary.xmi, see REPORT Part 4)
# the generators need lxml and write into src/umlmeta/
$LXML_PYTHON generate.py # regenerate src/umlmeta/uml25.py
$LXML_PYTHON generate_profiles.py # regenerate the profiles
The check scripts print every assertion; a failing assertion names the mapping.
check_v1_to_v2.py also writes vehicle_example_v2.sysml, wave3_example_v2.sysml,
and wave4_example_v2.sysml — worked examples of the emitted notation.
Repo layout
pyproject.toml distribution "umlmeta", MPL-2.0, src layout
src/umlmeta/ the importable package
uml25.py generated UML 2.5.1 metamodel (stdlib-only, self-contained)
standard_profile.py generated StandardProfile stereotypes
sysml.py generated SysML v1 profile stereotypes
uaf.py generated UAF 1.2 UAFML stereotypes (folds sysml bases)
uaf13.py generated UAF 1.3 UAFML stereotypes
sysml2.py generated SysML v2 abstract syntax + KerML runtime (R2 artifact)
bpmn.py generated BPMN 2.0.2 metamodel (reuses uml25 descriptors)
v1_to_v2.py SysML v1 -> v2 textual emitter (mapping authority: ptc/2025-04-07)
v1_to_v2_as.py SysML v1 -> v2 abstract-syntax transformer (R3; constructs sysml2)
xmi21.py XMI 2.1/EMF instance reader -> uml25 objects
xmi_write.py XMI 2.1/EMF instance writer (canonical parity vs OMG corpus)
mm_write.py CMOF + EMOF metamodel writers (canonical parity; spec-rule checks)
derived.py UML 2.5.1 normative derivations (owners, qualified names, unions)
query.py query helpers (walk / find / stereotypes / exists / count)
canon.py shared canonical XML comparator (writer oracles)
ocl_translator.py clean-room OCL-to-Python translator
mdzip.py MagicDraw/Cameo .mdzip scrape + forensics
mdzip_import.py .mdzip -> uml25 objects (provenance travels with the model)
mdzip_diff.py semantic diff of two .mdzip files
boxes_view.py v1 data -> SysML v1 diagrams via diagramboxes
generate.py UML 2.5.1 metamodel generator (OMG UML.xmi -> src/umlmeta/uml25.py)
generate_profiles.py profile generator (StandardProfile.xmi, sysml.xmi, UAF.xmi -> src/umlmeta/*.py)
generate_sysml2.py SysML v2 AS generator (OMG normative CMOF XMI -> src/umlmeta/sysml2.py)
generate_bpmn.py CMOF generator (OMG BPMN20.cmof -> src/umlmeta/bpmn.py)
check*.py eighteen check suites (481 checks; corpus suites need /mnt/TBFox, mdzip suites need MDZIP_CORPUS / MDZIP_REVISIONS)
calibrate_v2.py textual-form calibration harness for sysmlpy
REPORT.md build journal: dialect facts, mapping tables, evidence
PLAN.md plan for the exporter/QoL waves (E1-E6: XMI/CMOF/EMOF writers, stereotype writing, derived unions, query helpers)
Install
pip install umlmeta # the generated metamodel and the library
pip install "umlmeta[xmi]" # + lxml, for the XMI/CMOF/EMOF readers and writers
The generated metamodel modules are pure stdlib; lxml is only needed by the
readers, writers and generators.
from umlmeta import uml25
from umlmeta import derived, query
Clean-room discipline
- The generator and emitter are written from normative OMG text only
(UML 2.5.1 XMI, profile XMI, the ptc/2025-04-07 transformation spec).
Reference extractions and example XMI corpora are kept out of the
repository (e.g. the local
transition.txtspec dump is git-ignored). - The only instance-model corpus used for reader tests is the
OMG-published
DoDAFLibrary.xmi(UPDM example). Third-party commercial book models are excluded on copyright grounds. - The emitter follows an honesty rule: elements with no normative
mapping raise
UnmappedFeaturenaming the v1 metaclass rather than guessing, and spec gaps (e.g. ProxyPort, SYSML2_-329) surface as explicit comments.
Related
- mycr0ft/sysmlpy — the SysML v2 reader used as the validation oracle for every emitted textual form
- mycr0ft/ams-gra — consumer of the emitter (AMS GRA v14.0 SysML v2 model, built via this toolchain)
- mycr0ft/ams-gra-sysml — the SysML v2 model itself
License
MPL-2.0 — see LICENSE.
Metadata
Release files for umlmeta 0.2.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| umlmeta-0.2.0.tar.gz | 365.1 kB | With provenance | Details for umlmeta-0.2.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| umlmeta-0.2.0-py3-none-any.whl | Python 3 | none | any | Details for umlmeta-0.2.0-py3-none-any.whl |
Total release size: 733.7 kB
Release files / umlmeta-0.2.0.tar.gz
| Download URL | umlmeta-0.2.0.tar.gz |
|---|---|
| Size | 365.1 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
115560b92aa906fa8fa4b5853cf458baa9c0f35dd706c0efc1b29ce079026ef1
|
|
BLAKE2b-256 checksum How to use checksums |
bed670a236debb8c2f3f08ed0e2a732b4cdb8a5c97e393810c467f2e70c2fa95
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 28, 2026.
Transparency logRelease files / umlmeta-0.2.0-py3-none-any.whl
| Download URL | umlmeta-0.2.0-py3-none-any.whl |
|---|---|
| Size | 368.6 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
df57c22c73d017552e06d8a25a2b68b696a8593c7981f4fd2e215631d463e205
|
|
BLAKE2b-256 checksum How to use checksums |
242cd20095f699d19e6334a29e06a26ba1cfda364bf5eb8e0f0895cc06496ff0
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 28, 2026.
Transparency logSecurity for umlmeta 0.2.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
Both files in this release have been uploaded with provenance metadata.
Attestation for umlmeta-0.2.0.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 28, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.