Pdfium Impl Page Manipulation

pdfium-render: Page Manipulation

Mutating whole pages of a PdfDocument: rotation, deletion, flattening, page-box geometry (cropping), and watermarking. This skill is for operations ON a page as a unit. For editing the objects INSIDE a page (text, paths, images) use pdfium-impl-page-objects-edit. For navigating the page collection use pdfium-syntax-pages.

All API names below are verified against docs.rs/pdfium-render 0.9.1. The default target is the 0.9.x API; 0.8.x differences are flagged inline.

Quick Reference

PdfPageRenderRotation variants: None, Degrees90, Degrees180, Degrees270. PdfPageBoundaryBoxType variants: Media, Crop, Bleed, Trim, Art, Bounding.

Full signatures with version annotations: see references/methods.md.

How to Get a Mutable Page

Page mutation needs &mut PdfPage or an owned PdfPage. PdfPages::get(index) returns an OWNED PdfPage<'a>, so binding it mut is enough:

let mut page = document.pages().get(0)?;   // owned PdfPage<'a>
page.set_rotation(PdfPageRenderRotation::Degrees90);

ALWAYS bind the page mut when you will rotate, flatten, regenerate, or call boundaries_mut(). NEVER try to mutate a PdfPage reference yielded by pages().iter() while the iterator borrow is live: the iterator holds &PdfPages and the mutation needs an owned page. Iterate by index instead when mutating.

Decision Tree: which operation

Need to change page orientation?
  -> set_rotation(PdfPageRenderRotation::DegreesNN)   [does NOT change MediaBox]

Need to remove a page from the document?
  -> get(index).delete()   [delete() CONSUMES the page]
  -> deleting MANY pages: delete highest index FIRST (indices shift down)

Need annotations / form fields baked into the page content?
  -> flatten()   [0.8.19+ reloads the page so the effect is visible]

Need to crop / resize the visible area?
  -> boundaries_mut().set_crop(PdfRect)   [cropping == setting the CropBox]

Need print production boxes (bleed/trim/art)?
  -> boundaries_mut().set_bleed/set_trim/set_art(PdfRect)

Need a stamp on every page?
  -> document.pages().watermark(closure)

Pattern: Rotate a Page

set_rotation takes a PdfPageRenderRotation value. Rotation is page metadata; it does NOT change the MediaBox dimensions, only how viewers and render_with_config present the page.

use pdfium_render::prelude::*;

let pdfium = Pdfium::default();
let mut document = pdfium.load_pdf_from_file("input.pdf", None)?;

let mut page = document.pages().get(0)?;
page.set_rotation(PdfPageRenderRotation::Degrees90);   // 90deg clockwise
drop(page);                                            // release the page borrow

document.save_to_file("rotated.pdf")?;

PdfPageRenderRotation::as_degrees() and as_radians() (both const fn) convert a variant to a numeric clockwise rotation when you need the value.

ALWAYS use PdfPageRenderRotation. The old PdfBitmapRotation enum was deprecated in 0.8.6 and REMOVED in 0.9.0; code referencing it will not compile against 0.9.x.

Pattern: Delete a Page

delete() consumes the PdfPage by value (pub fn delete(self) -> Result<...>). You MUST own the page; you cannot delete through a shared reference.

// Delete the third page (zero-based index 2).
document.pages().get(2)?.delete()?;
document.save_to_file("output.pdf")?;

Deleting a page shifts every later page's index down by one. When deleting MULTIPLE pages, ALWAYS process indices from highest to lowest so earlier deletions never invalidate the indices you have not reached yet:

let mut to_delete = vec![1, 3, 5];
to_delete.sort_unstable();
for index in to_delete.into_iter().rev() {        // 5, 3, 1
    document.pages().get(index)?.delete()?;
}

delete() was added in 0.7.30. Requesting an out-of-range index returns PdfiumError::PageIndexOutOfBounds (a recoverable error). See pdfium-errors-runtime.

Pattern: Flatten a Page

flatten() merges annotation and form-field appearances into the page's content stream, producing a static page. It needs &mut self.

let mut page = document.pages().get(0)?;
page.flatten()?;
drop(page);
document.save_to_file("flattened.pdf")?;

Version behaviour (issue #140): before 0.8.19, rendering a page right after flatten() still showed the unflattened content because the in-memory page was stale; you had to drop and reload the page manually. From 0.8.19 flatten() reloads the page itself, so the flattened result is immediately visible to render_with_config(). ALWAYS target 0.8.19 or later when you flatten then render in the same run. A failed flatten returns PdfiumError::PageFlattenFailure.

Pattern: Crop and Page-Box Geometry

A PDF page carries up to five nested boxes. pdfium-render exposes them through PdfPageBoundaries (the collection added in 0.5.2):

Cropping a page means SETTING the CropBox. It does not delete content; it changes the region viewers and renderers display.

let mut page = document.pages().get(0)?;

// PdfRect::new_from_values order is (bottom, left, top, right) in PdfPoints units.
let crop = PdfRect::new_from_values(72.0, 72.0, 720.0, 540.0);
page.boundaries_mut().set_crop(crop)?;

drop(page);
document.save_to_file("cropped.pdf")?;

CRITICAL: PdfRect::new and PdfRect::new_from_values take arguments in the order (bottom, left, top, right), NOT (left, top, right, bottom). Passing them in the intuitive order silently produces an inverted or zero-area rectangle. Use PdfRect::new(PdfPoints, PdfPoints, PdfPoints, PdfPoints) for typed values or new_from_values(f32, f32, f32, f32) for raw points.

Each getter returns a PdfPageBoundaryBox with two public fields: box_type: PdfPageBoundaryBoxType and bounds: PdfRect.

let media = page.boundaries().media()?;
println!("MediaBox: {:?}", media.bounds);   // PdfRect

PdfPageBoundaries::get(PdfPageBoundaryBoxType) and set(box_type, rect) are the generic equivalents of the named methods.

Pattern: Watermark Every Page

PdfPages::watermark applies a closure to every page. The closure receives a mutable PdfPageGroupObject to fill, the page index, and the page width and height as PdfPoints:

PdfPages::watermark<F>(&self, watermarker: F) -> Result<(), PdfiumError>
where
    F: Fn(&mut PdfPageGroupObject<'a>, PdfPageIndex, PdfPoints, PdfPoints)
        -> Result<(), PdfiumError>

document.pages().watermark(|group, _index, width, height| {
    // Build watermark objects (text / image) and add them to `group`.
    // Object construction is covered by pdfium-impl-page-objects-edit.
    // `width` and `height` are PdfPoints; use them to centre the stamp.
    let _ = (group, width, height);
    Ok(())
})?;
document.save_to_file("watermarked.pdf")?;

ALWAYS build the watermark's page objects inside the closure with the techniques in pdfium-impl-page-objects-edit; this skill owns the watermark() entry point and its closure contract, not object construction.

Content Regeneration

Page mutations (rotation, box changes, object edits) must be committed to the underlying document before saving. PdfPageContentRegenerationStrategy controls when that happens:

With the default strategy, set_rotation and set_crop commit immediately and a following save_to_file is correct. If you switch to Manual (with set_content_regeneration_strategy) for batched edits, ALWAYS call regenerate_content() before dropping the page or saving, or the edits are lost.

let mut page = document.pages().get(0)?;
page.set_content_regeneration_strategy(PdfPageContentRegenerationStrategy::Manual);
page.set_rotation(PdfPageRenderRotation::Degrees180);
page.boundaries_mut().set_crop(PdfRect::new_from_values(0.0, 0.0, 600.0, 800.0))?;
page.regenerate_content()?;                 // commit the batch
drop(page);
document.save_to_file("output.pdf")?;

Full working programs: see references/examples.md.

Version Traps

Anti-Patterns (summary)

• Calling delete() on a page reference from pages().iter(): delete() consumes self; you need an owned page from pages().get(index).
• Deleting multiple pages low-index-first: indices shift; delete highest first.
• PdfRect argument order assumed (left, top, right, bottom): it is (bottom, left, top, right).
• Expecting set_rotation to change page dimensions: it does not touch the MediaBox.
• Switching to Manual regeneration and forgetting regenerate_content().

Each anti-pattern with the failure message and fix: see references/anti-patterns.md.

Cross-References

• pdfium-syntax-pages: the PdfPages collection, get, iter, len, indexing.
• pdfium-impl-multi-page: appending, importing, reordering pages between documents.
• pdfium-impl-page-objects-edit: constructing the objects used inside watermark.
• pdfium-core-coordinates: PdfPoints, PdfRect, the bottom-left PDF origin.
• pdfium-impl-saving: save_to_file, save_to_bytes, save_to_writer.
• pdfium-errors-runtime: PageIndexOutOfBounds, PageFlattenFailure.

Reference Files

• references/methods.md: complete API signatures with version annotations.
• references/examples.md: verified, runnable Rust programs.
• references/anti-patterns.md: real failures, why they fail, the fix.

Source

Verified 2026-05-20 against https://docs.rs/pdfium-render/latest/pdfium_render/ (prelude struct pages for PdfPage, PdfPages, PdfPageBoundaries, PdfPageBoundaryBox, PdfRect, PdfPoints, and the PdfPageRenderRotation / PdfPageBoundaryBoxType / PdfPageContentRegenerationStrategy enums).