GitHub - muonium-ai/pygments-swift: A swift port of pygments library for syntax highlighting, built for qw code review tool

Pygments Port

Vibe coded by Senthil Nayagam (@senthilnayagam) from Muonium AI Venture Studios.

This repo ports the Python Pygments syntax highlighting library into other languages.

Swift port

The Swift implementation lives in pygments-swift as a SwiftPM package.

What’s included

• A Pygments-inspired RegexLexer engine (state machine with push/pop, include/inherit/combined states, byGroups, using, and default transitions).
• A lexer registry for selecting lexers by language name or filename extension.
• Parity tests for selected lexers against the in-repo Python Pygments source.

Supported lexers (Swift)

Strict parity (Swift tests compare token streams to Python Pygments for chosen samples):

• Swift
• JSON
• JSON-LD

Pragmatic (smoke-test level highlighting for common code):

• Python, JavaScript, TypeScript, Java
• C, C++, C#, Go, Rust
• D, Crystal
• Kotlin, Ruby, PHP
• BibTeX, ASN.1, CDDL, Devicetree (DTS), PromQL, Rego, JMESPath, PRQL, Typst, Smithy
• Bash/Shell
• Tcl, Awk, Sed
• Windows Batchfile, VBScript
• Pascal
• Racket, Scheme, Common Lisp, Emacs Lisp
• Elm, Haxe, V
• Nix
• Fish, Nushell
• Raku
• CUE
• Scala
• R
• reStructuredText, LaTeX
• GitIgnore, EditorConfig, Properties, CSV
• Graphviz (DOT), PlantUML, Mermaid
• ApacheConf
• Ada, COBOL, Prolog, Smalltalk, Eiffel, Visual Basic
• SystemVerilog, VHDL, LLVM IR, GLSL
• JSON5, Jsonnet, YARA, YANG, WGSL, WebAssembly (WAT), WebIDL, Meson, GDScript, Teal
• AsciiDoc, Org-mode, Kconfig, Caddyfile, SPARQL, Turtle, Thrift, Cap'n Proto, QML, HLSL

Build & test

cd pygments-swift
swift test

Basic usage

Use the registry to pick a lexer by language or filename:

import PygmentsSwift

let lexer = LexerRegistry.makeLexer(languageName: "swift")!
let tokens = lexer.getTokens("let x = 1")

OCR-based render regression checks

To catch rasterization/layout regressions in PNG output (for example, missing tail lines), this repo includes an OCR-based checker:

• Script: tools/ocr_compare_png.py
• Make targets: make ocr-check and make ocr-check-pdf

Dependencies (macOS)

• python3 (system Python is fine; no virtualenv required)
• tesseract (installed via Homebrew)
• sips (built into macOS; required only for the PDF baseline mode)

Install Tesseract (Homebrew)

Verify:

which tesseract
tesseract --version

Verify sips is available

sips is included with macOS:

which sips
sips --version

Run the checks

Generate sample renders first:

make render-samples-all CONFIG=release

Run PNG OCR checks (fails if any PNG is flagged CROPPED?):

Optional: also OCR-check the matching PDFs via sips as a baseline (this may show false positives depending on OCR noise/theme contrast):

To make the PDF baseline strict:

make ocr-check-pdf FAIL=1

To print source/OCR tails for flagged cases:

make ocr-check SHOW_TAILS=1
make ocr-check-pdf SHOW_TAILS=1