pygments-step

Pygments lexers for EXPRESS (ISO 10303-11) and STEP Part 21 (ISO 10303-21).

Additional aliases: exp, iso-10303-11 for EXPRESS; p21, step, stp, spf, iso-10303-21 for Part 21.

Both lexers are schema-agnostic. They lex the languages, not any single application protocol, so they work equally well for IFC, AP203, AP214 or any other STEP-based schema without carrying a schema-specific keyword table.

Install

pip install pygments-step

Or, with uv:

uv add pygments-step

The only runtime dependency is Pygments>=2.14. Python 3.9+ is supported.

Installing is all you need to do. The lexers register themselves with Pygments through pygments.lexers entry points, so every tool that renders code through Pygments — MkDocs, Sphinx, pygmentize, Rich, Jupyter — picks them up automatically. There is no configuration step.

Usage

Markdown code fences

Tag a fence with express or step21:

```express
ENTITY cartesian_point
  SUBTYPE OF (geometric_representation_item);
    coordinates : LIST [1:3] OF length_measure;
  DERIVE
    dim : INTEGER := HIINDEX(coordinates);
  WHERE
    valid_dim : SIZEOF(coordinates) <= 3;
END_ENTITY;
```

```step21
ISO-10303-21;
HEADER;
FILE_SCHEMA(('SOME_APPLICATION_PROTOCOL'));
ENDSEC;
DATA;
#1= CARTESIAN_POINT('',(0.,0.,0.));
#4= AXIS2_PLACEMENT_3D('',#1,#3,$);
#6= NAMED_UNIT(.T.,.F.,.UNSPECIFIED.);
ENDSEC;
END-ISO-10303-21;
```

This works out of the box in MkDocs (including Material for MkDocs, whose pymdownx.highlight extension delegates to Pygments) and in any other Markdown pipeline backed by Pygments — no mkdocs.yml changes required.

Python API

from pygments import highlight
from pygments.formatters import HtmlFormatter
from pygments.lexers import get_lexer_by_name

lexer = get_lexer_by_name("step21")
print(highlight("#1= CARTESIAN_POINT('',(0.,0.,0.));", lexer, HtmlFormatter()))

The classes can also be imported directly:

from pygments_step import ExpressLexer, StepFileLexer

Dispatch by filename works as well:

from pygments.lexers import get_lexer_for_filename

get_lexer_for_filename("schema.exp")   # EXPRESS
get_lexer_for_filename("model.stp")    # STEP Part 21

Command line

pygmentize -l express schema.exp
pygmentize -l step21 -f html -O full,style=friendly -o model.html model.stp

Because the file extensions are registered, -l can usually be omitted:

pygmentize schema.exp

What the lexers cover

EXPRESS (ISO 10303-11)

• Case-insensitive keywords, per clause 6.1 and annex A.1.1.
• Nested (* ... (* ... *) ... *) embedded remarks and -- ... tail remarks.
• Declaration heads (SCHEMA, ENTITY, TYPE, FUNCTION, PROCEDURE, RULE, CONSTANT, SUBTYPE_CONSTRAINT) highlight the declared name distinctly from the keyword.
• Literals: doubled-quote string escapes ('it''s'), binary literals (%10110), encoded string literals ("000000E9"), reals and integers.
• Built-in functions (SIZEOF, HIINDEX, EXISTS, TYPEOF, …) and built-in types (INTEGER, LIST, SELECT, LOGICAL, …).

STEP Part 21 (ISO 10303-21)

• Exchange structure keywords: ISO-10303-21, END-ISO-10303-21, HEADER, DATA, ENDSEC, ANCHOR, REFERENCE, SIGNATURE.
• Entity instance definitions (#1=) are tokenised differently from references (,#1,), so the two read differently in rendered output.
• Enumerations (.T., .F., .UNSPECIFIED.), unset values ($) and derived values (*).
• String control directives: \S\, \P?\, \X\, \X2\...\X0\, \X4\...\X0\, plus '' escapes.
• Print control directives \N\ and \F\, both inside strings and at any position where a token separator may appear.
• Binary literals ("0F3A"), user-defined keywords (!USER_DEFINED_KEYWORD) and /* ... */ comments.

A note on IFC

IFC files are Part 21 files, so StepFileLexer handles them — but ifc and *.ifc are deliberately not claimed as an alias or filename pattern. IFC is only one of many SPF-based formats, and privileging it in a standards-level lexer would be arbitrary. Use step21 (or p21 / spf) for IFC content.

Test

Clone the repository and install with the test extra:

git clone https://github.com/AECVerge/pygments-step.git
cd pygments-step
uv sync --extra test
uv run pytest

Or with pip:

pip install -e ".[test]"
pytest

The suite also runs standalone, without pytest:

python tests/test_lexers.py

Tests cover alias and filename registration, token-level regressions for both lexers, and assert that neither fixture produces a single Error token.

tests/test_docs_pages.py does the same for the rendered docs/express and docs/step pages. Those pages are an exhaustive, category-by-category corpus for the two lexers (keywords, operators, types, literals, comments, …); the test parses every code fence on them and asserts that the lexer produces zero Error tokens, and that each page really covers the whole family of tokens it claims to. The docs build in CI runs with mkdocs build --strict, so a snippet that stops lexing cleanly blocks the deploy as well.

License

MIT