PDF splitter & page organiser

What it does

Organise PDF pages, inspect and reduce supported embedded images, and review bookmarks before a local export.

How to use it

1. Choose an unencrypted PDF, then select pages using checkboxes or ranges such as 1, 3-5. Arrows change the export order; rotation turns a page clockwise.
2. Inspect indirect image objects by stored bytes, dimensions and encoding. Select supported images to resize or re-encode as JPEG; compare reduced previews and the whole-file size before downloading.
3. Review bookmark titles, hierarchy and original page targets. Retain, delete or retarget entries when pages are removed; unsupported destinations require a decision. Create one PDF or split selected pages into a ZIP.
4. Up to 25 MiB input, 200 pages and 50 MiB export, with a cancellable 15-second worker operation. Signed, certified, encrypted and XFA PDFs are rejected. Ordinary form fields are flattened when organising pages; page-copy exports omit document metadata and attachments. Keep the original.

Example

Select pages 1 and 3 from a three-page PDF to create a two-page document.

Methodology

pdf-lib inspects indirect image objects by stored bytes and resource references, including nested Forms. Supported RGB/gray images can be resized without upscaling and re-encoded to JPEG, retaining the original payload when it would grow; reduced before/after previews and whole-file byte counts support review. Page-copy exports apply order/rotation and rebuild reviewed bookmark titles, hierarchy and targets against original pages. Removed parents reparent retained children; split exports keep relevant headings and targets. PDF.js renders previews and fflate packages split pages.

Limitations

25 MiB input / 200 pages / 50 MiB export, including ZIP overhead; split input bytes times selected pages must fit. Cancellable 15-second worker operations. Up to 500 images, 500 bookmarks / 12 levels / 500-character titles. Select at most 25 images / 40 MP combined; 16 MP and 16,384 pixels per source side. JPEG quality 20–95, maximum output side 256–8,192. Only supported 8-bit DeviceRGB/DeviceGray JPEG or raw/Flate images are changed; masks, ICC/indexed/CMYK, decode arrays, alternate/external images and unsupported codecs/filter chains remain unchanged. Resource references are not painted occurrence counts; inline images are unlisted and shared-image edits affect all references. Text/vectors remain intact; colors can change and PDF overhead can still grow. Signed/certified/signature-field, encrypted and XFA inputs are rejected. Up to 300 ordinary fields flatten on page organisation. Bookmark views reset to Fit and actions are never copied. Page-copy exports omit document metadata/attachments; image-only optimization preserves other structures and is not sanitization.

Automation

This tool runs in your browser. No upload API or MCP endpoint is available.