bijoy2unicode

Convert legacy Bijoy / Sutonny MJ Bangla text and Word documents to Unicode Bengali, in the browser or on Node.

• Pure TypeScript, zero native deps. Runs in browsers, Node 18+, Deno, Bun, Edge runtimes.
• String API for converting individual strings.
• File API for .docx, .odt, .rtf, .html, .txt. Uses JSZip only when you import bijoy2unicode/docx.
• Strict font-aware detection — English text in Latin fonts is never corrupted, even when sitting in the same paragraph as Bijoy.
• Handles inherited fonts via paragraph styles, character styles, and docDefaults — same way Word resolves them.
• Battle-tested on real-world Bangladesh government bio-data forms with thousands of mixed runs.

Live demo

Try it without installing anything:

• 🌐 https://bijoy2unicode.jehadurre.me (primary)
• 🌐 https://bijoy2unicodex.vercel.app (Vercel mirror)

Drag-and-drop a .docx, .doc, .odt, .rtf, .html, or .txt file. Conversion runs entirely in your browser; nothing is uploaded.

Install

npm install bijoy2unicode
# or
pnpm add bijoy2unicode
# or
yarn add bijoy2unicode

Quick start

String conversion

import { convertBijoyToUnicode } from "bijoy2unicode";

convertBijoyToUnicode("Avgvi †mvbvi evsjv");
// → "আমার সোনার বাংলা"

convertBijoyToUnicode("Kg©KZ©vi e¨w³MZ I PvKzix msµvšÍ Z_¨vw`");
// → "কর্মকর্তার ব্যক্তিগত ও চাকুরী সংক্রান্ত তথ্যাদি"

Word document conversion (.docx)

import { convertDocx } from "bijoy2unicode/docx";

const fileInput = document.querySelector<HTMLInputElement>("input")!;
fileInput.addEventListener("change", async () => {
  const file = fileInput.files![0];
  const blob = await convertDocx(file, {
    onProgress: ({ stage, percent }) => {
      console.log(`${percent}% ${stage}`);
    },
  });
  const url = URL.createObjectURL(blob);
  Object.assign(document.createElement("a"), {
    href: url,
    download: file.name.replace(/\.docx$/, "-unicode.docx"),
  }).click();
});

Auto-route by extension

import { convertFile } from "bijoy2unicode/docx";

const result = await convertFile(file);
// result.blob is the converted file
// result.filename is suggested output name

API

bijoy2unicode (default entry)

| export | description | |---|---| | convertBijoyToUnicode(src: string): string | Core converter. Always converts. | | looksLikeBijoy(text: string): boolean | Heuristic — does the text contain Bijoy-specific high-byte chars? | | hasBengaliUnicode(text: string): boolean | Does the text already contain Unicode Bengali? | | shouldConvertAsBijoy(text, fontIsBijoy?) | Combine the heuristics into a single decision. | | scanUnmapped(unicodeText): Map<string, number> | Find leftover suspicious chars (likely unmapped Bijoy bytes). | | isSuspiciousLeftover(ch): boolean | Single-char version of the leftover check. | | describeCodepoint(ch): string | Human-readable codepoint label, e.g. "U+09C7". |

bijoy2unicode/docx

All file-format helpers accept Blob | File. They return a Blob (or ConversionResult for convertFile).

| export | description | |---|---| | convertDocx(file, opts?) | Microsoft Word .docx | | convertOdt(file, opts?) | OpenDocument Text .odt | | convertRtf(file, opts?) | Rich Text Format .rtf | | convertHtml(file, opts?) | HTML / .htm (browser only — uses DOMParser) | | convertTxt(file, opts?) | Plain text | | convertFile(file, opts?) | Auto-route by extension | | detectExt(name) / suggestOutputName(name, ext) | Filename helpers |

ConvertOptions

interface ConvertOptions {
  onProgress?: (info: { stage: string; percent: number }) => void;
  /** Treat every run as Bijoy regardless of font / heuristics. Default: false. */
  force?: boolean;
}

How the font detection works (the important part)

When converting .docx, the package does not blindly run the converter on every text run. It walks the OOXML and resolves the effective font of each <w:r>:

1. Run's own <w:rFonts> (highest priority)
2. Run's <w:rStyle> referenced character style
3. Paragraph's <w:pPr><w:rPr><w:rFonts>
4. Paragraph's <w:pStyle> referenced paragraph style (with basedOn chain resolution)
5. <w:rPrDefault> document default

If the resolved font is a known Bijoy font (any …MJ, BijoyEkattor, Sulekha, Boishakhi, …) the run is converted. If it's a known Unicode Bangla font (Nikosh, SolaimanLipi, Vrinda, Noto Bengali, …) the run is left alone. For ambiguous runs, it uses character-level heuristics (high-byte detection, Bijoy bigram density, English word lookup, vowel ratio) to decide.

This means English text in Calibri / Times New Roman / Arial stays untouched, even in a Bijoy-heavy document.

Supported Bijoy variants

Wide coverage of real-world Sutonny MJ / Bijoy text, including alternate encodings other libraries miss:

• Trailing-© reph reorder (Kg©KZ©vi → কর্মকর্তার)
• Í (U+00CD) prefix-pair → ন্ত / স্ত / ক্ত conjuncts (we¯ÍvwiZ → বিস্তারিত)
• iæ → রু (ordered before bare æ → ু to handle name forms like নূরুল correctly)
• ÿ → ক্ষ alternate kkho encoding
• Precomposed nukta forms (ড়/ঢ়/য়) preserved per Word/Office convention
• Multi-character sequences (Mø → গ্ল, cø → প্ল, eø → ব্ল, kø → শ্ল, etc.)
• Document-default font inheritance for runs that omit <w:rFonts> entirely

Limitations

• Legacy .doc (binary OLE format) is not supported by this package. Convert via LibreOffice or CloudConvert first; this package will then handle the resulting .docx perfectly.
• .html conversion requires DOMParser (browser environment). For server-side HTML, run the string converter on extracted text.
• Pure-ASCII Bijoy text (bvg wcZvi bvg with no high-byte chars) is genuinely ambiguous in .txt/.rtf. Set force: true to convert anyway.

Repo & contributions

Source, issues, and the demo app live at github.com/JehadurRE/Bijoy2Unicode. PRs welcome — especially samples of Bijoy .docx files that don't convert cleanly. The leftover-character report flow in the demo app makes it easy to send those reports.

Author

Md. Jehad (Jehadur Rahman Emran)

Full Stack Developer & System Architect · Cloud Connect AI

Professional Profile

Md. Jehad is a passionate Full Stack Developer and System Architect specializing in building scalable, cross-platform applications. With expertise spanning mobile, web, and backend technologies, he brings a comprehensive approach to software development.

Core Competencies

Mobile Development

• 🎯 Flutter & Dart — cross-platform mobile applications
• 📱 React Native & Expo — native mobile experiences
• 🎨 Material Design 3 & Adaptive UI/UX

Backend Engineering

• ⚡ FastAPI & Python — high-performance REST APIs
• 🗄️ PostgreSQL & Database Design — normalized schemas & optimization
• 🔄 Redis & Caching Strategies — performance optimization
• 🔐 JWT Authentication & Security — OWASP compliance

System Architecture

• 🏗️ Domain-Driven Design (DDD) — clean architecture patterns
• 📊 Database Schema Design — ERD modeling & optimization
• 🔗 API Integration — third-party services & payment gateways
• 📈 Scalable Systems — microservices & distributed architecture

DevOps & Tools

• 🐳 Docker & Containerization
• 🔄 CI/CD Pipelines — automated deployment
• 📊 Monitoring & Analytics — Firebase, Crashlytics
• 🔧 Git & Version Control

Inspiration

Standing on the shoulders of earlier open-source Bijoy converters that paved the way:

• Mad-FOX/bijoy2unicode — early Python conversion algorithm
• almehady/Bijoy-to-Unicode-File-Converter — Python file-wrapper that demonstrated the use case
• nishiafia/bijoytounicodeconverter — PHP/JS plugin with helpful alternate mappings

This package is an original TypeScript implementation with significant corrections and additions for real-world Sutonny MJ documents (see the Supported Bijoy variants section for the diff).

License

MIT