Why Urdu breaks when you copy it
Urdu is a joined, right-to-left script. Each letter takes a different shape depending on where it sits in a word: ب looks different at the start, middle and end. Modern software stores only the letter and lets the font choose the shape. Older PDF makers, scanned-document tools and some websites took shortcuts instead. Those shortcuts come back to bite you when you copy text out.
The four problems we see most often are:
- Presentation forms. Each letter was saved as its shape, for example “meem, initial form”, using special Unicode characters from the range U+FB50–U+FEFF. It may look fine, but search, spell-check and word counts fail, and some apps show it disjointed.
- Visual order. The PDF stored each line from left to right as it appears on screen, so the copied Urdu comes out backwards: ےہ باتک instead of کتاب ہے.
- Arabic letters. The text uses Arabic ي، ك and ه where Urdu needs ی، ک, and ہ or ھ. It can look almost right in some fonts and completely wrong in Nastaliq.
- Invisible junk. Zero-width spaces, direction marks, byte-order marks and double spaces that break searching and line wrapping.
This tool repairs each problem in a careful order and shows you exactly which characters it changed.
What each repair does
| Fix | Example before | After |
|---|---|---|
| Joined letter shapes | ﭘﺎﻛﺴﺘﺎﻥ (shape codes) | پاکستان |
| Arabic → Urdu letters | يه كتاب | یہ کتاب |
| ه → ہ or ھ | مجهے بهت | مجھے بہت |
| Arabic digits | ١٩٤٧ | ۱۹۴۷ |
| Punctuation | ہے . | ہے۔ |
| Reversed line | ےہ باتک یریم ہی | یہ میری کتاب ہے |
Joined-letter shapes are converted with a targeted map, not blanket Unicode NFKC normalisation. NFKC would also expand honorifics such as ﷺ into several words and change other symbols you meant to keep. We keep ﷺ, ﷽ and ornate brackets exactly as they are.
Worked example
This is the kind of text you might copy from an older government PDF:
مجهے يه كتاب بهت پسند هے ، اور ميں اسے روز پڑهتا هوں ?
مجھے یہ کتاب بہت پسند ہے، اور میں اسے روز پڑھتا ہوں؟
Nine letters changed: ي → ی, ك → ک, and every ه resolved by word (مجھے، پڑھتا with ھ; یہ، بہت، ہے، ہوں with ہ). Three punctuation fixes followed: the stray spaces before the comma and the question mark went, and the question mark became ؟.
Common problems and fixes
The text still reads backwards
Click Reverse each line. If only some lines are reversed (common when a PDF mixes columns), paste those lines separately. English words and numbers inside a reversed line are kept in their normal left-to-right order.
Letters look fine but search still fails
That is the classic sign of presentation forms. Make sure Joined letter shapes is on under Options. Afterwards, every changed character shows in green.
Some words have the wrong ہ or ھ
Uncommon words containing do-chashmi ھ default to ہ. Fix them by hand, and let us know the word so we can add it to the list.
The copied text is random English letters
The PDF has no real Urdu text inside, only drawings of letters, which is common with InPage exports. Repair cannot help there. If you have the original InPage file, copy from InPage and use the InPage to Unicode converter instead.
Who uses this tool
- Researchers and students quoting Urdu from journal articles, gazettes and theses saved as PDF.
- Journalists lifting quotes from press releases and court documents.
- Web editors cleaning text pasted from old websites that used Arabic fonts for Urdu.
- Translators preparing source text for translation memory tools, which need clean Unicode.
Many of them use it alongside other tools: Urdu PDF to Word when they need the whole document rather than a quote, Urdu OCR when the PDF turns out to be a scan with no text inside, and the Urdu word counter to check the length of the cleaned text. The guide why Urdu text copied from a PDF looks broken explains the causes step by step.
Your text stays on your device
Every repair runs in your browser. Nothing is uploaded or stored, so you can safely clean confidential reports, legal documents and unpublished research.