Arabic text copied from a PDF is reversed or garbled: how to fix it

Why Arabic changes when you copy it from a PDF

A PDF does not store sentences the way a Word file does. It stores glyphs (the drawn shapes of letters) and their positions on the page, plus a table that maps each glyph back to a character. When you copy, the viewer reads that table and tries to rebuild the text.

Arabic is drawn with joined letter shapes, and that is where things go wrong. Some PDF producers store the joined shapes themselves instead of the underlying letters, and some store them in the order they appear on the page, left to right. Copy from such a file and you get what the file recorded, not what the author typed.

Unicode is clear that the Arabic presentation forms exist for historical compatibility and that using them in data does not guarantee integrity or interoperability; normal Arabic text is stored in logical (typing) order using the main Arabic block. So copied text has to be turned back into ordinary letters in logical order before you search it, store it or translate it.

Identify the problem from what you pasted

Paste a short paragraph into any plain-text editor and compare it with this table:

To confirm presentation forms, paste a word or two into Unicode normalization and choose NFKC. Code points starting with U+FE or U+FB are presentation forms; after normalization you should see ordinary letters from the U+06 range:

ﻻ ﺗﻨﺲ

لا تنس

U+FEFB U+0020 U+FE97 U+FEE8 U+FEB2

U+0644 U+0627 U+0020 U+062A U+0646 U+0633

Notice that “لا” was a single character in the copied text and becomes two, lam and alef. That one-character ligature is exactly why a search for the normal spelling fails.

Repair reversed words, step by step

1. Copy a section of the PDF. Start with one paragraph so you can check the result.
2. Open Reversed Arabic and set Goal to Repair reversed text copied from a file.
3. Paste the text. The tool converts presentation forms back to ordinary letters, splits lam-alef into two letters and restores each line to typing order.
4. Read the result, paying particular attention to numbers, punctuation and English words.
5. Copy it or download it as a .txt file, then carry on cleaning if needed.

ﻦﻴﺋﺪﺘﺒﻤﻠﻟ ﺔﻴﺑﺮﻌﻟﺍ ﺔﻐﻠﻟﺍ ﻢﻴﻠﻌﺗ

تعليم اللغة العربية للمبتدئين

Numbers and Latin words stay left to right inside the line and return, together with brackets, to their place in the sentence:

ﻙ.ﺩ 12.500 :(ﺔﺒﻳﺮﻀﻟﺍ ﻊﻣ) ﻲﻟﺎﻤﺟﻹﺍ

الإجمالي (مع الضريبة): 12.500 د.ك

The tool handles up to 20,000 text units per run and works only on text you paste. It does not upload or open the PDF itself, so copy a long document in sections.

When the order is right but the letters are odd

Sometimes the words come out in the right order but the letters are stored as presentation forms. The text looks fine, yet a search for “مرحبا” finds nothing and a form or database may reject it. You do not need to reverse anything. Open the Cleaner and, under Arabic marks, tick Repair legacy Arabic presentation forms.

ﻣﺮﺣﺒﺎ ﺑﻜﻢ ﻓﻲ ﺍﻟﺪﻭﺭﺓ

مرحبا بكم في الدورة

Do not use that option on its own for reversed text. It fixes letter shapes but never changes order, so you end up with ordinary letters that still run backwards:

ﺔﻴﺑﺮﻌﻟﺍ ﺔﻐﻠﻟﺍ

ةيبرعلا ةغللا

If that is what you get, go back to the original copied text and use Repair reversed text copied from a file instead.

What no text tool can repair

If the pasted text shows random Latin letters, boxes, or Arabic letters that have nothing to do with the word, the PDF has no correct mapping from glyphs to characters. This is common with old fonts and files produced in non-standard ways. The copied text simply does not contain the right information, and no converter can guess it. Scanned PDFs are the same story: the page is an image, so there is no text to copy at all.

Your options:

• Ask the document's owner for the source file, such as Word or InDesign, and copy from that.
• Use optical character recognition (OCR). For example, Google Drive's help page explains how to open a PDF or image with Open with > Google Docs to convert it to text, and Arabic is among the supported languages. Proofread the result carefully, especially dots, hamzas and numbers.
• If the passage is short, retyping it may be faster and more accurate than any conversion.

After the repair: a quick check

• Line breaks: a PDF copies every printed line as a separate line. To rejoin a paragraph, use Join lines with a space under Spacing and lines in the Cleaner.
• Invisible characters: copied text may carry direction marks or zero-width spaces. Inspect it in the invisible character inspector before you delete anything.
• Numbers and percentages: a % sign or currency symbol can end up on the other side of its number. Fix those by hand.
• Comparison: if you have part of the original text, check the result against it in Compare.

تتناول هذه الدراسة أثر
القراءة اليومية في تحسين
الكتابة عند الطلاب.

تتناول هذه الدراسة أثر القراءة اليومية في تحسين الكتابة عند الطلاب.

For the wider cleanup routine, see cleaning copied Arabic text. For why Arabic breaks in design apps in the first place, read reversed Arabic in Photoshop and other apps.