Bijoy vs Unicode: Converting Legacy Bangla PDF Text

You open an old Bangla PDF, select a sentence, paste it into a message and see something like Avwg evsjv‡`‡ki. The page looked perfectly Bangla. The cause is almost always the same: the document was typed with a legacy Bangla font, and its text is not Unicode. This guide explains what that means and how to get real Bangla text out.

What a legacy Bangla font really stores

Before Unicode Bangla was usable on most computers, programs such as Bijoy used fonts (SutonnyMJ is the best-known) in which the glyph for আ sits where the English letters "Av" would be. The file stores English letters and the font turns them into Bangla shapes. Anything that does not have that font, such as a search engine, a database, a screen reader or a message box, sees only the English letters. That is also why a PDF can look right and still be impossible to search.

Two further details make conversion more than a table lookup. Legacy text is stored in visual order: the vowel signs ি, ে and ৈ are typed before the consonant they follow on screen, and the reph (র্) is typed after the whole letter group, while Unicode stores them in logical order, consonant first, reph before the consonant it sits on. And conjuncts such as ক্ষ, ন্ত্র or স্ক্র often have their own single glyph, so the converter has to recognise them as units.

How to recognise a legacy PDF

• Copied text is full of Latin letters and symbols such as ‡, ©, ª, ¨, Ñ, yet the page shows Bangla.
• The font list of the PDF names fonts ending in MJ, or Sutonny, or Bijoy. The PDF Font Checker lists them.
• If nothing can be copied at all, the pages are probably pictures (a scan) or the letters were converted to outlines. That is a different problem, covered below.

Converting it

The Bijoy / SutonnyMJ PDF to Unicode tool reads the PDF in your browser, keeps track of the font of each piece of text, converts the pieces set in legacy fonts, leaves ordinary English text alone, and shows the raw text beside the result so you can see what was read. If your viewer lets you copy the legacy text, you can also paste it into the tool and convert it without opening a PDF.

Use the result as a draft. Check conjuncts, the placement of ো and ৌ, and numbers against the PDF. Fix mistakes in the result box before you copy it.

Why it is never guaranteed

Legacy Bangla encodings were never standardised. Bijoy's layout is the most common, but other fonts map characters differently, some fonts add private symbols, and PDF producers can store text in odd orders or split words into pieces. A good converter handles the common cases and says what it did not recognise; none can promise that every legacy PDF converts perfectly.

When there is no text to convert

A PDF made by scanning paper contains pictures of pages, and one whose fonts were turned into outlines contains drawings of letters. Neither has text, so there is nothing to convert. Optical character recognition reads the shapes instead: try Bangla OCR and read Bangla OCR: tips for better results. The general difference between extracting text and recognising it is explained in PDF text extraction vs OCR.

After converting

Unicode Bangla displays in any modern program, but you need a font that contains the Bangla letters (Noto Sans Bengali, Nirmala UI, Kalpurush and others). To keep the text with the original pages, edit the PDF text only if you must; otherwise keep the PDF and the converted text side by side.