Loading page
About this tool
Compare a local PDF text layer's extracted sequence with a bounded Y-then-X visual-order estimate for manual review.
PDF Reading-Order Preview extracts a PDF's text layer in the sequence the file actually stores it, then builds a separate Y-then-X visual estimate — grouping items into lines by vertical position, reading top to bottom and left to right — and shows you where the two disagree. Those divergences are where a screen reader is most likely to read a page out of the order a sighted reader sees, the concern behind WCAG Success Criterion 1.3.2 Meaningful Sequence. It is a manual review aid: it reads coordinates, not the PDF tag tree, and does not predict what any particular assistive technology will do.
Open PDF Reading-Order Preview on AltFTool — it loads instantly in your browser.
Drop a file on the Upload PDF Document zone, click Browse PDF File, or paste one with Ctrl+V — the picker accepts .pdf only, up to 100 MB and 100 pages, otherwise it reports "Choose a PDF no larger than 100 MB." or "Choose a PDF with 1 to 100 pages." Try Demo PDF loads a built-in sample instead.
The Open PDF button reads Analyzing... while the text layer is parsed, then the dashboard shows Accessibility Score, Reading Blocks and Detected Defects. Choose a Visual Overlay View Mode — All Overlays, Reading Numbers Only, Bounding Boxes Only, Direction Flow Arrows Only, Accessibility Heatmap or Color-coded by Tag Type — and press Play Sequence to step through the page block by block.
Open the Issues tab for Detected Accessibility Defects; clicking one jumps to its page and highlights the block, while the Screen Reader Stream lists the same blocks in order and Outline shows each block's tag type. The status bar keeps Score and Defects in view, and the Reset inspection button clears the file.
It shows the stored source sequence and the coordinate-derived visual estimate together, so the disagreement itself is the finding.
Repeating horizontal gaps are detected as possible columns and the affected blocks are marked ambiguous rather than silently reordered.
Rotated items are flagged as cases the simple Y-then-X estimate may misread, instead of being folded into a confident answer.
No. It compares stored text order with a visual-order estimate to help you spot likely sequence problems; it does not read the structure tag tree, and reading order is only one part of PDF accessibility. Treat the output as evidence for a manual review against WCAG 1.3.2, not as a compliance verdict.
Items are grouped into a line when their vertical positions are within a tolerance derived from the median glyph height — 45% of it, clamped between 2 and 18 points — then sorted left to right within each line and top to bottom across lines.
Up to 100 MB and 100 pages, with per-page ceilings of 3,500 text items and 75,000 characters, and overall ceilings of 50,000 items and 750,000 characters. Anything past those bounds is reported as truncated.
A scan is an image with no text layer, so there is no stored sequence to compare. Run OCR to add a text layer first — the approach described in W3C technique PDF7 for scanned documents.