Hamza

Hamza is a light-weight, fast and portable C99 OpenType shaping and rendering library. It's built alongside the stb_truetype.h TrueType library. It's designed to be a small, portable and optimized shaper that's easy to integrate into an existing project. Originally Hamza was a research project to learn about font shaping for an Arabic calligraphy software I had in mind. Later, with much research and work I have only developed more fascination with the general shaping process and possible strategies of optimization.

Objective

• Keep the library as a whole excluding backends and auto-generated tables under 10k SLOC.
• The goal is a non-pessimized, small and simple shaping library that can easily be included into any existing project.
• The intent is to optimize lookup application slow paths using SIMD + multi-threading while keeping the code as simple and clear as possible. Many think that shaping is a linear process and very difficult to multi-thread, this couldn't be further from the truth. GSUB table single substitution as an example is a perfect use-case for multi-threading, Unicode parsing can be vectorized and ligature matching can be optimized with SSE.
• Support for ASCII, Latin1, UTF-8, UTF-16, UTF-32, UCS2 and Johab encodings.
• For the renderer; an efficient MSDF dynamic cache on GPU using compute.
• Ability to render semi-transparent cursive text, soft shadows as well as proper outlining.

Features

• [x] Joining script support and RTL writing
• [x] Kerning
• [x] Ligatures
• [x] Support for new OpenType language tags (mixture of ISO 639-3 and ISO 639-2 codes)
• [ ] Support for old Two-Letter ISO 639-1:2002 language tags (same as HarfBuzz)
• [ ] Vertical layout (i.e. for Chinese)
• [ ] Emojis
• [ ] Multi-Threading

Script Support

• [x] Standard Scripts (Latin, Cyrillic, Greek, etc)
• [x] Arabic
• [ ] Buginese
• [ ] Hangul
• [ ] Hebrew
• [ ] Indic
  • [ ] Bengali
  • [ ] Devanagari
  • [ ] Gujarati
  • [ ] Gurmukhi
  • [ ] Kannada
  • [ ] Malayalam
  • [ ] Odia
  • [ ] Tamil
  • [ ] Telugu
• [ ] Javanese
• [ ] Khmer
• [ ] Lao
• [ ] Myanmar
• [ ] Sinhala
• [ ] Syriac
• [ ] Thaana
• [ ] Thai
• [ ] Tibetan

Usage

For usage of API functions, check hz.h and respective files. There is also documentation at https://saidwho12.github.io/hamza/.

Misc

Developed and maintained by Saïd Metiche.