fix(latex): render unbraced greek sub/sup symbols

[tianbaiting]

Summary

• normalize unbraced sub/sup style-symbol lookups so _\alpha and ^\alpha resolve through the existing subscript/superscript glyph maps
• fall back to regular symbol glyphs for unbraced style-symbols when no dedicated Unicode sub/sup form exists, so cases like _\delta and _\theta still render as symbols instead of raw TeX
• derive parser fake-preview allowlists from markview.symbols.entries to keep parser and renderer behavior aligned
• add regression examples for unbraced and braced greek sub/sup forms in test/latex.md

Reproduction

Before this change, plain \alpha rendered correctly, but unbraced forms such as q_\alpha and q^\alpha stayed as raw TeX in Markview preview. The same issue affected other greek symbol commands in unbraced sub/sup form.

Verification

• headless render check now emits ₐ for _\alpha, δ for _\delta, and θ for _\theta
• headless render check now emits ᵅ for ^\alpha, ᵟ for ^\delta, and ᶿ for ^\theta
• braced forms _{\alpha} and ^{\alpha} still render correctly

[Copilot]

Pull request overview

Fixes Markview’s LaTeX preview rendering so unbraced Greek subscript/superscript forms (e.g. _\alpha, ^\alpha) resolve through the existing subscript/superscript glyph maps, avoiding raw-TeX fallback in previews.

Changes:

• Normalize unbraced sub/sup symbol lookups in the LaTeX renderer so _\alpha / ^\alpha map to ₐ / ᵅ.
• Derive parser fake-preview allowlists from markview.symbols.subscripts / markview.symbols.superscripts to reduce parser/renderer drift.
• Add regression examples covering unbraced and braced Greek sub/sup usage.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.

File	Description
test/latex.md	Adds regression examples for unbraced/braced Greek sub/superscripts and a mixed expression.
lua/markview/renderers/latex.lua	Normalizes unbraced sub/sup keys (e.g. strips leading \) before glyph-map lookup.
lua/markview/parsers/latex.lua	Generates supported “fake preview” command allowlists directly from the glyph maps.

Comments suppressed due to low confidence (2)

lua/markview/parsers/latex.lua:327

• supported_symbols is rebuilt (including iterating + sorting) on every subscript capture. Since markview.symbols.subscripts is effectively static, consider computing/caching the command list once at module scope (or memoizing per map) to avoid repeated work during parsing large documents.

---@param range markview.parsed.range
latex.subscript = function (_, TSNode, text, range)

lua/markview/parsers/latex.lua:366

• supported_symbols is rebuilt (including iterating + sorting) on every superscript capture. Since markview.symbols.superscripts is effectively static, consider computing/caching the command list once at module scope (or memoizing per map) to reduce repeated work during parsing.

---@param range markview.parsed.range
latex.superscript = function (_, TSNode, text, range)

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[OXY2DEV]

@tianbaiting Sir, did you actually read what the original code did?

The whole reason a SUPPORTED_SYMBOLS table exists is so that an unsupported symbol can be shown as ↑(<symbol>)/↓(<symbol>) instead of breaking style of text that can be shown as superscript/subscript text.

I also don't understand why you bothered adding a new function to check for supported symbols. You could've simply removed the original piece of code and it should've been enough.

[tianbaiting]

You're right — I expanded the fix too far and changed the intended role of SUPPORTED_SYMBOLS.
I misunderstood that table as just a preview allowlist, while it actually also preserves the fallback behavior for unsupported symbols.

The real bug is narrower: for unbraced forms like _\alpha / ^\alpha, the renderer strips _ / ^ but still keeps the leading backslash, so it looks up "\alpha" instead of "alpha" in symbols.subscripts / symbols.superscripts.

[OXY2DEV]

@tianbaiting I have added direct support for symbols inside subscript/superscript(for some reason it is parsed differently to normal symbols). It should now fix all edge cases.

[OXY2DEV]

The real bug is narrower: for unbraced forms like _\alpha / ^\alpha, the renderer strips _ / ^ but still keeps the leading backslash, so it looks up "\alpha" instead of "alpha" in symbols.subscripts / symbols.superscripts.

Fixing this by itself isn't enough btw(cause the function needs to register a style region which is used for symbol lookup). But the commit I made also carries additional changes to make it work correctly.

[tianbaiting]

Thanks for the fix!

And thanks for the explanation about the style-region registration and symbol lookup — that helped me understand the rendering pipeline better.

I also realized that more complex subscripts like \mu_{\text{pair}} probably can't be rendered cleanly with Unicode alone, since there isn't a full set of subscript glyphs for arbitrary text.