Convert Unicode to Bijoy

What the conversion does to the text

Unicode stores letters in logical order and Bijoy in visual order, so the conversion is a reordering before it is a lookup. A pre-base vowel sign (ি, ে, ৈ) is moved in front of the consonant cluster it is attached to; a reph (র্) is moved after the cluster it sits above, past a vowel sign that has already been moved; the two-part vowels ো and ৌ are split into the ে that goes before the consonant and the া or ৗ that goes after. Then every conjunct is matched longest first — ক্ষ্ম before ক্ষ before ক — against the table of SutonnyMJ ligatures, and the glyph forms that depend on the neighbouring letter are chosen: the u-kar is one key after most letters and another after ক, চ, ট and the others whose foot it would collide with; the ra-phala has three shapes; the e-kar at the start of a word is a different character from the one in the middle.

In a Word file the converted runs are set to SutonnyMJ in all four font slots, and because Word keeps the size and weight of complex-script text separately and the Latin glyphs now use the ordinary ones, the complex-script values are copied across where the ordinary ones were missing — otherwise a document set at 13 point comes back at Word's default. Neighbouring runs with the same formatting are converted together, since a word Word split at a spelling mark would otherwise be converted in two halves with the vowel sign stranded on the wrong side.

Bangla digits ০–৯ become 0–9, which SutonnyMJ draws as Bangla digits; the danda (।) becomes the "|" key, and the dash and the curly quotes become their Bijoy keys. Everything else — Latin letters, ASCII punctuation, line breaks — passes through unchanged, and a text file is saved as UTF-8 so it opens in any editor; apply SutonnyMJ there.

When you want something else

If the destination is a web page, a phone, a search index or anything that will be read on a device you do not control, do not convert at all. Bijoy text is readable only where SutonnyMJ is installed, which is a Windows machine with Bijoy Bayanno on it — on everything else it is "evsjv‡`k", and it is unsearchable even where it displays. Unicode is the format that works everywhere; Bijoy is the format one specific office needs.

For a layout that has already been set in Unicode and only needs a Bijoy font for the press, ask the press first whether they can take Unicode. Most presses running a current InDesign can, since Adobe's World-Ready composer has handled Bangla for a decade, and the converted file loses the ability to be searched or edited with a modern keyboard. Convert the copy they will typeset; keep the Unicode original.