[Epic] OMML Math + SmartArt [STRATEGIC]

[MHoroszowski]

Problem

Two of the modern hard XML schemas are unmodelled: Office Math Markup Language (OMML) for equations, and DrawingML-Diagram for SmartArt. Both are STRATEGIC DEBT — neither is small, and both have niche-but-loyal user demand.

OMML / Equations. scanny/python-pptx#126 ("maths formula or LaTeX support", 14c) is the canonical issue; equations cannot even be extracted from text frames (scanny/python-pptx#947). The good news: there is a fork (OscarPellicer/python-pptx) with a working MathML → LaTeX translation (281 symbols mapped, +1241 lines in oxml/text.py), so a translator-style API is available as a starting point.

SmartArt / Diagrams. scanny/python-pptx#83 (4c) is the SmartArt anchor. The four-part bundle (diagram1.xml, diagram1_colors.xml, diagram1_quickStyle.xml, diagram1_layout.xml) round-trips today as a verbatim graphicFrame blob, but there is no API for create / edit / re-style. Jona210399/python-pptx has exploratory parsing ("kind of working"), and Martin005/python-pptx-ng has a pptx/parts/smartart.py skeleton that's a viable scaffold. The right scope here is "build the data model, let PowerPoint relayout on open" — full SmartArt layout regeneration in Python is a tar pit.

Sub-features

OMML / Equations

• OMML XML element classes for <m:oMath>, <m:f> (fraction), <m:rad> (radical), <m:sub>/<m:sup>, <m:nary> (sum/integral), <m:m> (matrix), at minimum
• Equation.from_latex(latex_string) — convert via existing pylatexenc-style translator
• Equation.to_latex() — reverse direction (port OscarPellicer 281-symbol map)
• Run.equation = Equation(...) — embed an equation inside a text run
• Read-API: extract equations as text/latex from existing presentations

SmartArt

• Slide.shapes.add_smartart(layout_id, data) where layout_id selects from a set of supported layouts (organization chart, bullet list, process flow first)
• Data-model only: write <dgm:dataModel> with the user's hierarchy; let PowerPoint regenerate the visual layout on open
• Round-trip preservation of all SmartArt parts (verify no corruption)
• Read-API: traverse the data model of an existing SmartArt and extract the hierarchy

Prior art

• OMML. Open PRs: scanny/python-pptx#528 inserting OMML, scanny/python-pptx#892 parse equations. Forks: OscarPellicer/python-pptx MathML→LaTeX translator (cherry-pick basis).
• SmartArt. Open PRs: scanny/python-pptx#1100 chart+SmartArt parsing (SergeevAI 1168 LOC). Forks: Jona210399/python-pptx WIP parsing, Martin005/python-pptx-ng pptx/parts/smartart.py + pptx/oxml/diagrams/data.py skeleton.
• User issues this would close: OMML: #83, #126, #528, #892, #947. SmartArt: #83 (shared anchor).
• OOXML: Math Markup §22.1 <m:oMath> element tree, lives inside text run as <a:r><m:oMath>.... SmartArt §21.4 <dgm:dataModel> four-part bundle.
• Code paths: new src/pptx/equations/ subpackage (port OscarPellicer translator), new src/pptx/diagrams/ subpackage (port python-pptx-ng skeleton).

Acceptance criteria

OMML

• Run.equation = Equation.from_latex(r"\frac{a+b}{c}") produces a deck where the equation renders correctly in PowerPoint.
• Reverse: equation.to_latex() for an existing OMML run returns reconstructable LaTeX.
• 30+ symbol-mapping unit tests.

SmartArt

• slide.shapes.add_smartart("OrgChart", root) with a 5-level hierarchy renders correctly when opened in PowerPoint.
• Round-trip of SmartArt-containing decks does not corrupt the diagram.

Effort: L (combined)

Strategic debt. Recommend separate sub-epics in execution if either ships independently. OMML phase ~6 weeks (mostly translator porting). SmartArt phase ~10 weeks (data-model writing, layout-regen contract testing).

Labels

status:strategic-debt, priority:P2.