Sinhala Font Design

Design research, design processes and design-side tooling for Sinhala type, by mooniak.

This repo answers how to design a Sinhala typeface — anatomy, what order to draw glyphs in, how spacing inherits, which forms are composites, what to test with, and the browser tools to look at all of it. It deliberately stops where engineering begins.

Renaming. This repository is being renamed from sinhala-font-tools to sinhala-font-design to match that scope. Existing clone URLs will keep redirecting, but update your remotes when you next pull.

Where the boundary is

Design lives here. Font engineering — glyph inventories, glyph naming, encoding, OpenType features, building, QA and release — lives in three other places, and this repo defers to them rather than duplicating them:

Standard / project	What it governs	Use it for
lanka-glyphsets	The glyphset standard for Sri Lankan fonts: cumulative Sinhala levels (Kernel / Core / Plus / Pro) mapped to ICTA Levels 1–3, plus a Tamil glyphset, and the normative glyph-naming standard (visual-form names such as kRa-sinh). Ships @mooniak/lankaglyphsets (npm).	What a font must contain and what each glyph is called.
SLS 1134	The Sri Lanka Standard for Sinhala character encoding, published by the Sri Lanka Standards Institution — the national standard behind Sinhala in Unicode: character repertoire, code points and how Sinhala text is encoded.	Encoding, cmap, codepoint questions, conformance claims.
mooniak-font-tools (mnik)	The production toolchain across all mooniak font repos: building with gftools, QA/fontspector, proof PDFs, build manifests, release channels, native Glyphs export, and the font directory/dashboard.	Building, QA-ing, proofing, releasing and tracking a font project.

Glyph names used throughout this repo are lanka-glyphsets names. If a model here disagrees with lanka-glyphsets, lanka-glyphsets wins.

What's here

Models (the design data)

• sinhala-design-stages/ — a dependency-based development model. Every design decision (a metric, a skeleton, a spacing class, a mark-join convention) is introduced once by one glyph and inherited by every glyph that depends on it; stages, drills and coverage are computed from the graph. Orthogonal to lanka-glyphsets levels: levels say what a font must contain, stages say in what order to draw it.
• sinhala-spacing.yaml — the spacing scheme: sidebearing relationships derived from the stages model, applied to sources as Glyphs metrics keys.
• sinhala-composites.yaml — which glyphs should be assembled from components rather than drawn as merged outlines.

Tools

• tools/glyph-chart/ — browser font tester: a grid of base letters × vowel marks/compound signs. Drop in a .ttf/.otf/.woff.
• tools/spacing-map/ — builds sinhala-spacing.yaml from the stages model and audits a font's sidebearings against it (writing metrics keys into .glyphspackage sources is on its roadmap).
• tools/composite-map/ — builds the composite map and rebuilds atomic glyphs into proper composites in Glyphs sources; its Sinhala Glyph Builder Glyphs plugin adds glyphs from a visual catalogue of every stage glyph and attested grapheme (draw, composite or features), parents first.
• tools/glyphs-filters/ — generates Glyphs 3 sidebar filters from these models, so a working font's sidebar is a live view of the model. Output in filters/.
• tools/test-generator/ — adhesiontext-style test text generator: list the graphemes a font has (ටු, කෝ, ක්‍රි… as units, drawn or composed) and get real words and running text that use only those.
• test-texts/corpus-build/ — builds, from the raw UCSC Sinhala corpora (not in the repo), the data bundle test-generator loads: cleaned words and frequencies, word pairs, a word-building model and ~150k real sentences sharded by rarest grapheme.
• tools/font-comparison/ — side-by-side rendering comparison across fonts. Ships with just the system fonts; see its README to set up your own default collection.
• tools/feature-generator/ — lankafea (PyPI mnik-lankafea): generates Sinhala/Tamil OpenType features from the lanka-glyphsets standard, gated on the glyphs a font actually has, and reports dependency blockers and orphans. Its Glyphs 3 plugin, Lanka Features, is in glyphs-plugin/. Moved here from lanka-glyphsets so that repo holds only the standard; it reads the glyphset YAMLs from a sibling lanka-glyphsets checkout.
• tools/glyphname-unicode-converter/ — glyph name ↔ Unicode converter: paste Sinhala clusters to get LankaGlyphset names, or names to get the Unicode. Vendored from lanka-glyphsets, web tool and engine; the engine (lankaglyphset-map.js) is also used by glyph-chart, glyphs-filters and test-generator.

Reference material

• test-texts/ — the canonical Sinhala test-text corpus: display paragraphs, kern pairs, conjuncts, marks, numerals, punctuation, classical/Pali/Sanskrit text, pangrams, multiscript samples. Vendored into mnik's proof PDFs and into polytypetester — see test-texts/README.md for the sync mapping.
• resources/ — Sinhala anatomy notes, character-frequency analysis, reference fonts.

Getting started

Most of the browser tools are static — open the index.html and drop a font in, no build step. The Python tools (spacing-map, composite-map) each have their own README with setup instructions. tools/glyphs-filters reuses js-yaml from sinhala-design-stages/node_modules.

All the browser tools share one black-and-white palette, tools/_shared/bw.css. Use its tokens rather than colours: show differences with fill, weight and solid/dashed borders, not hue.

If you are starting a Sinhala font, the order is roughly:

1. Pick a target level in lanka-glyphsets — that's your inventory and your glyph names.
2. Design in the order given by sinhala-design-stages/, using the Glyphs filters in filters/ to see what should exist by now.
3. Apply spacing and composites with the tools above.
4. Generate features and validate with lanka-glyphsets (mnik-lankafea).
5. Build, QA, proof and release with mnik.

Licence

The models and tools authored in this repo are GPL-3.0.

Some vendored and reference content carries its own licence instead:

Content	Licence
test-texts/ (most files)	CC-BY-SA-4.0, per each file's license: frontmatter
test-texts/libreoffice-ui-strings.md	Extracted from LibreOffice's Sinhala .po files; frontmatter says CC-BY-SA-4.0, but upstream LibreOffice is MPL-2.0 — confirm before reuse
resources/Hodipotha-9.ttf	© ICTA 2014, CC BY-ND 4.0 — see resources/README.md
resources/sinhala-frequency-analysis/sinhala/Corpus/	Sinhala newspaper text (e.g. Silumina, ANCL, 2004); no licence recorded — check redistribution terms before reuse
sinhala-design-stages/docs/sinhala.min.yaml	Extracted from Noto Sans Sinhala (OFL-1.1); regenerated locally, not vendored
tools/_generated/lanka-glyph-data.js, shaping-exceptions.js, tools/glyphname-unicode-converter/ (index.html, converter.js, lankaglyphset-map.js)	Vendored from lanka-glyphsets; see that repo's licence

When in doubt about a specific file, check its own header/frontmatter first — it overrides this table.