Converting Amiga and Harikrishna documents

1One table, twenty fonts

Amiga fonts ship under many names — Arati, Ashok, Bhakti, Bipin, Chitra, Deep, Drupad, Girdhar, Gokul, Hanuman, Haresh, Hum, India, Jarna, Mahesh, Mani, Mohan, Muralidhar, Patrika — and Harikrishna, which is not called Amiga at all.

All twenty use the same byte layout. Each was verified individually by rendering test strings and reading what appeared, rather than assumed from the family name.

If your documents use any of them, select the Amiga family in the converter and they will convert.

2The names that look right and are not

Three faces we ship carry Amiga-like names and do not use the Amiga encoding. One of them is Devanagari — Hindi, not Gujarati — and applying a Gujarati table to it would produce confident, completely wrong output.

This is why conversion is keyed to the font file rather than to the font’s name. A name is a claim; the rendering is evidence.

3What makes these fonts tractable

These fonts encode strokes rather than letters. A consonant is drawn as a shape plus a vertical stem, and the font splits them at that seam — so one byte can be half a letter, and the byte after it completes the letter.

That is why converting is not a simple character swap, and why a converter that treats it as one gets conjuncts and vowel signs wrong in exactly the places a legal document cannot afford them: names, dates and amounts.