Why Urdu Text Copied from PDFs Looks Broken or Reversed

You find the perfect paragraph in an Urdu PDF: a gazette notification, a research paper, a court order or a book. You select it, copy it, paste it, and get something like this:

ےہ باتک یریم ہی

That should read یہ میری کتاب ہے. Or the letters are right but your search cannot find the word. Or every ی has become ي. Copying Urdu out of PDFs is one of the most common frustrations for students, journalists and researchers, and the cause is almost never your computer.

This guide explains what is going on inside the PDF and how to repair each kind of damage.

A PDF stores appearance, not always meaning

A Word document stores text: a sequence of letters, in reading order, that the program lays out when you open it. A PDF stores a finished page: which shape to draw, in which font, at which position. Sometimes it also stores a map from each shape back to the letter it represents. When you copy text from a PDF, your viewer rebuilds the text from those shapes and maps.

For English, this mostly works, because one shape is one letter and lines run left to right. Urdu is harder in three ways:

1. Letters change shape. ب has a different form at the start, middle and end of a word. The PDF may record the shape rather than the letter.
2. Text runs right to left. The PDF may record shapes in the order they appear across the page, left to right, which is backwards for Urdu.
3. Urdu shares a script with Arabic. Some PDF tools map Urdu shapes to the nearest Arabic letters.

Each of these produces a different kind of broken text.

Problem 1: joined letter shapes (presentation forms)

Unicode has two ways to represent Arabic-script letters:

• Ordinary letters (U+0600–U+06FF): ب is ب, and the font decides which shape to draw.
• Presentation forms (U+FB50–U+FEFF): a separate code for every shape, such as “beh, initial form” or “beh, final form”.

Presentation forms exist for compatibility with old systems. Many PDF creators still use them. The text looks right but is made of different characters, so:

• searching for پاکستان does not find ﭘﺎﻛﺴﺘﺎﻥ written in shape codes,
• spell checkers mark every word wrong,
• word counters and sorting go haywire,
• some apps show the letters disconnected.

Fix: map each shape back to its base letter. Our Fix Broken Urdu Text tool does this with a targeted table and highlights every changed character. It does not use blanket “compatibility normalisation” (NFKC), because that also expands meaningful symbols. The honorific ﷺ, for example, would turn into four separate words.

Problem 2: reversed lines (visual order)

Some PDF generators write each line as the glyphs appear on screen, from left to right. For Urdu, that means the last letter comes first. When you copy, every word and every line comes out backwards:

PDF gives you	Should be
ےہ باتک یریم ہی	یہ میری کتاب ہے
۔ںیہ ےتھڑپ ہو	وہ پڑھتے ہیں۔

You can often spot reversal at a glance: words start with ے or ں, which never happens in correct Urdu, because those letters only end words.

Fix: reverse each line, but carefully. A naïve reversal also flips English words and numbers, which were already left to right, and detaches diacritics from their letters. The Fix Broken Urdu Text tool detects likely reversal, then asks before acting. When you confirm, it reverses only the Urdu parts, keeps English and numbers such as “Word 2016” intact, reattaches zabar, zer and pesh, and mirrors brackets.

Problem 3: Arabic letters instead of Urdu letters

Urdu and Arabic share most letters but differ in a few that look alike in some fonts:

Arabic	Urdu	Name
ي	ی	yeh
ك	ک	kaf
ه	ہ or ھ	heh
ة	ۃ	teh marbuta

PDFs made with Arabic fonts, and text typed on Arabic keyboards, use the left column. In Nastaliq fonts the difference is obvious and ugly, and searching for کتاب will not find كتاب.

The tricky letter is ه, because Urdu has two letters where Arabic has one: gol hay (ہ), as in ہے، کہا، بہت, and do-chashmi hay (ھ), which marks aspirated sounds as in بھی، تھا، کچھ. No simple rule separates them. کہا (said) and کھا (eat) differ only in this letter.

Our tool uses gol hay by default, since it is far more common, and switches to do-chashmi hay for words found in a built-in list of common aspirated words and verb roots (دیکھ، لکھ، رکھ، سمجھ، پڑھ…). For example:

Rare words may still need a manual touch, which is why changed characters are highlighted for review.

Problem 4: invisible characters and spacing

PDF extraction often adds characters you cannot see: zero-width spaces, left-to-right and right-to-left marks, byte-order marks and soft hyphens. They break searches (“پاکستان” with a hidden character in the middle is a different word), confuse word counts and make cursor movement jumpy.

There are also visible spacing problems: double spaces, spaces before ۔ and ،, and English full stops and question marks after Urdu words.

Fix: remove invisible characters, but keep the zero-width non-joiner (ZWNJ) where it sits between two letters, because there it is doing a job: stopping two letters from joining inside a compound word. Collapse repeated spaces, remove spaces before Urdu punctuation, and convert English punctuation after Urdu words to ۔ ، ؟ ؛. The repair tool does all of this, and each step can be switched off under Options.

Problem 5: the PDF has no real text at all

Sometimes copying gives random English letters, empty boxes or nothing. That usually means the PDF contains pictures of letters rather than text:

• scanned pages,
• PDFs exported from InPage, which often embed shapes from InPage’s fonts without a usable text map,
• PDFs where text was converted to outlines for printing.

No text repair can help here, because there is no text to repair. Your options are:

1. Find the source file. If it came from InPage, copy from InPage and use the InPage to Unicode converter.
2. Use OCR (optical character recognition). Our free Urdu OCR reads scanned pages and images in your browser. It is very accurate on Naskh-style Urdu but only gives a rough draft for Nastaliq, so proofread carefully.
3. Retype short passages with the Urdu Typing Pad.

If the PDF does have real text and you need the whole document rather than a paragraph, open it in Urdu PDF to Word: it applies all the repairs above to every page and gives you an editable Word file. To set the text in InPage instead, use PDF to InPage.

A quick diagnosis guide

Symptom	Cause	Action
Words start with ے or ں; everything reads backwards	Visual order	Fix Broken Urdu Text → Reverse each line
Looks right, but search and spell check fail	Presentation forms	Fix Broken Urdu Text (joined letter shapes)
ي ك ه visible, looks odd in Nastaliq	Arabic letters	Fix Broken Urdu Text (Arabic → Urdu letters)
١٢٣ instead of ۱۲۳	Arabic digits	Fix Broken Urdu Text, or the Urdu Digits Converter
Random English letters or boxes	No real text in the PDF	Find the source file or use Urdu OCR
ƒœ’„-style symbols	InPage text, not a PDF problem	InPage to Unicode

If you make PDFs: how to avoid the problem

If you publish Urdu PDFs, a little care makes your documents searchable and quotable:

• Create PDFs from Unicode sources such as Word, LibreOffice or InDesign with Unicode Urdu, not from legacy InPage text.
• Use Save as PDF or Export rather than “Print to PDF”, which more often loses the text layer.
• Embed fonts, and make sure the document language is set to Urdu.
• Test your own PDF: copy a paragraph into a plain text box and check that it reads correctly.

Your readers, and search engines, will be able to find, quote and share your work.