Compare raw PDF draw order against XY-Cut++ reading order to spot multi-column and layout-related extraction issues
Details
What this tool helps you do
Run the same PDF through OpenDataLoader with readingOrder=off and readingOrder=xycut, then inspect the per-page text difference. This is useful for multi-column reports, scientific papers, brochures, and any layout where plain text draw order may scramble reading flow.
Execution
Run this tool
Fill in the form, run the tool, and review the result in one place.
The PDF Reading Order Debugger is a specialized tool for developers and data scientists to identify and resolve text extraction issues caused by complex document layouts. By comparing the raw draw order of a PDF against the advanced XY-Cut++ reading order algorithm, it provides a side-by-side HTML report that highlights where multi-column layouts, sidebars, or headers might scramble the logical flow of text.
When to use
When extracted text from multi-column reports or academic papers appears out of order.
Before configuring RAG pipelines to ensure document context and citations remain logically sequenced.
When debugging whether structural tags or layout-aware algorithms are necessary for a specific set of documents.
How it works
1
Use cases
Validating the extraction logic for two-column scientific journals to prevent sentence fragmentation.
Optimizing financial report processing by identifying where sidebars interfere with table data.
Troubleshooting layout-related extraction errors in brochures and marketing materials with non-linear text flows.
Compare extraction with and without repeated page furniture to spot header/footer noise before using PDF text in RAG, summarization, or editing workflows
Compare StructTree-enabled and plain PDF extraction to see whether a document behaves like a tagged PDF and how much semantic structure it exposes
Compare local and hybrid OpenDataLoader extraction to identify PDF pages where formulas, charts, or dense visuals may need AI-assisted parsing
Who it's for
Best for
Convert PDF files in the browser.
The “Debug a multi-column report before building citations” workflow, done in one pass.
Repeatable jobs with PDF files and the same settings every time.
Not ideal for
Fully automated batch jobs — this page is built for interactive use; use the API endpoint for scripts.
Edge cases & common failures
Password-protected PDFs
The file cannot be read until the password is removed — this tool has no password field.
Scanned PDFs with no text layer
Pages that are only images contain no selectable text, so text output comes back empty — run OCR on the file first.
Very large files
Large files take longer and can hit browser memory limits — try a smaller sample first.
Examples
1. Multi-Column Research Paper Debugging
AI Engineer
Background
An engineer is building a citation-aware RAG system but finds that text from the left column is merging with the right column.
Problem
The raw extraction order is reading horizontally across the entire page width instead of following the columns.
How to use
Upload the research PDF, set the page range to '1-5', and run the debugger with XY-Cut++ enabled.
Outcome
The HTML report shows that XY-Cut++ correctly isolates the columns, while the raw order fails, confirming that layout-aware extraction is required.
2. Financial Report Header Interference
Data Analyst
FAQ
What is XY-Cut++?
It is a layout-analysis algorithm that recursively partitions a page into horizontal and vertical blocks to determine the correct human reading sequence.
Why does the raw draw order often look scrambled?
PDFs store text based on the order it was added to the file, which frequently differs from the visual layout of columns and sidebars.
Can I test how headers and footers affect extraction?
Yes, you can toggle the 'Include Header/Footer' option to see if these elements disrupt the main content flow.
What does the 'Use Struct Tree' option do?
It attempts to use the internal structural tags (if present in the PDF) to determine the reading order instead of relying solely on visual layout.
What format is the final report?
The tool outputs an HTML file that provides a visual comparison of the text order differences for every processed page.
Order
21
3212
593
471
103
239
333
Image, audio, and video processing, conversion, and optimization tools
Calculators, numerics, date logic, statistics, and finance tools
Color, layout, graphics, visual styling, and design helper tools
Writing, cleanup, formatting, extraction, and text analysis tools
Developer utilities, networking, web debugging, and automation helpers
Format, file, archive, unit, and encoding conversion tools
PDF Samples
Generated PDF samples from tools dated 2026-02-01 to 2026-02-10
Markdown Slide Deck Samples
Remark/Marp style Markdown slide decks for testing PDF export layouts
Time Zone Workflow Scheduler ICS Samples
ICS files generated in the same structure returned by the Time Zone Workflow Scheduler, with multiple VEVENT meeting candidates exported from overlap windows
OFD Samples
Sample OFD (Open Fixed-layout Document) files for GB/T 33190 parsing and preview tests