Practical guide
How PDFManage checks Arabic PDF conversions
A practical quality checklist based on regression cases for editable Arabic text, right-to-left order, table cells and preserved images.
First identify what the PDF page contains
A PDF may contain selectable text, a scan, or both. When the characters are already text, extraction can preserve more of the source than OCR. OCR is for text painted into a page image. Mixed pages need both paths, so check a page with text and a figure before converting a whole document.
Check Arabic as complete phrases
Compare a whole line rather than isolated characters. A useful synthetic check is: علا لا لأ لإ لآ — سنة ٢٠٢٦ — PDFManage. Confirm the words read right to left, while dates and English terms remain in their intended order. Also inspect punctuation, dots and line endings; OCR can confuse similar letter shapes in a faint scan.
Verify the table structure, not just its appearance
Follow one row from its headings to its values. Check that a blank cell stays blank, an identifier such as 0012 keeps its leading zeros, and a value such as 2.0 keeps the displayed precision. Look for notes above or below the grid, because a table is not always the entire page. For Excel, inspect both the cell values and the right-to-left sheet view.
Keep photographs separate from editable text
A diagram or photograph should retain its proportions; recognized captions should remain text when the source makes them distinguishable. A page-sized image may look faithful but its words are not editable. Choose a page-image output only when exact visual appearance matters more than editing individual words.
What our regression checks cover
The project includes synthetic test PDFs for a damaged Arabic Unicode map with intact embedded font data, right-to-left Arabic table cells, drawn and filled table borders, notes outside a grid, numeric formatting, and a photograph with an editable caption. The tests check whether outputs keep expected text, cells and image proportions. They are safeguards for known cases, not a guarantee that every source file will convert perfectly.
A careful workflow for important documents
Keep the original PDF. Convert one representative page first, compare names, dates, totals and headings, then continue only if the sample is usable. If a word or number is unclear in the source itself, rescan or check it manually instead of trusting OCR. For confidential records, use a redacted or invented example when reporting a problem.