Edit Arabic text in a PDF: joined letters, right to left

What goes wrong in most PDF editors

Arabic letters change shape with their neighbours, and a line runs from right to left, with numbers and Latin words keeping their own direction inside it. A PDF editor that writes characters one at a time, unjoined and left to right, turns a word into separate letters or into its mirror image.

The reverse problem exists too: some Arabic PDFs store their letters so that copying them gives nonsense. In both cases the text looks wrong, or cannot be changed at all, in an editor that was built for Latin letters.

How GenPDF writes Arabic

GenPDF shapes each word as Arabic is written: letters take their joined forms and the line is set right to left, with digits and Latin words in their own direction. It uses the document's own font when the PDF contains the letters, so an edited word matches the words around it.

When a PDF stores unreadable characters for one of its fonts, GenPDF recovers the text, so it can be read, searched and edited, and tells you that it did. If the embedded font lacks a letter you type, that letter is drawn with a matching Arabic font and you are told which.

Scans, Word and mixed languages

A scanned Arabic page is first read, which needs your agreement: the page's image goes to our server, is read, and is deleted as soon as the text is back. Then its lines can be edited like typed text.

A line can mix Arabic, numbers and French or English words. Each part keeps its direction, as in a document typed in Word. And PDF to Word marks Arabic paragraphs as right to left, so they read in order in the .docx.

Where it stops

Good to know before you start:

• Reading a scan is best on clear, straight pages. Handwriting is not reliably read, and turned text is left out.
• A decorative font that the PDF does not contain is replaced by the closest font GenPDF has, and its name is shown before you save.
• A PDF of more than 25 MB does not open in the editor, and a password-protected one has to be unlocked first.
• Editing runs in your browser, so the PDF is not uploaded to change its words.