Discussion
One of several things which specifically appealed to me about Papyrix is that it includes code specifically for rendering Arabic text! So how does it do so?
There's support routines for checking whether a character is in certain ranges, or branching over the char to decide on which side it joins with other chars.
There's a (redundant) support routine for decoding UTF-8 text & iterating over its chars.
1/2?
There's a couple support routines for scanning a lookuptable & combining with data on adjacent chars to determine how chars should be joined. With another lookuptable being used to determine how to handle Lam-Alef characters.
There's utils for branching over joining types. There's a util for branching over a char to return the mirror-image char.
Then there's the core routine which validates that it has text & decodes the UTF-8 before really getting into it!
2/3!
An initial iteration looks for Lam-Alef pairs to combine into a ligature. Once it has found a Lam char it looks for an arabic diacritic & iterates further to find one which joins with the Lam char, looks up the ligature via a helper function, & copies the modified char(s) to a newly-allocated array.
A 2nd iteration applies contextual forms by copying base-chars over, gathering any left-join chars followed by right-joins, & looks up the corresponding glyph.
3/4!!
And the final passes implements the Unicode Bidirectional algorithm. With one pass gathering a table of text directions, a 2nd pass resolving neutral directions based on context, an optional pass gathering an array of same-direction runs, & a final pass reversing appropriate runs.
3.5/3.5 Fin for today! Tomorrow: EPD Fonts.
Install bonfire.cafe
Get the full app experience
