Skip to content

limitations.md

Latest commit

 

History

History
94 lines (89 loc) · 5.23 KB

limitations.md

File metadata and controls

94 lines (89 loc) · 5.23 KB
Raw

Limitations

For an overview of what each platform supports, see the capability matrix.

  • Not implemented yet: editing annotations (forms are filled in, not restructured); facing pages in horizontal or paged layouts, or right to left.
  • Dark pages on iOS: annotations keep their own colours (see Dark pages).
  • Forms: JavaScript actions and calculations in a form do not run; signature fields cannot be signed (their places are reported); XFA forms are not supported. iOS writes exports of a changed form anew with PDFKit (field hierarchy flattened into dotted names, field order may change; content PDFKit cannot read is not kept; a protected document gets a new owner password, see Forms; seconds for long documents, in the background) and keeps one selected option per list box. Android: filling in needs Android 15+, or 12–14 with SDK extension 13 (elsewhere fields are read from the file, setFormFieldValues rejects and exportDocument copies the file); a choice's custom text cannot be typed by touch; a radio group that cannot be turned off cannot be cleared; screen readers do not reach form fields.
  • Several sources are shown for reading only: no form filling, printing or export (see Several documents as one). Links from one source to another file are not followed.
  • Page overlays are not clipped to their page, and are not drawn in page images or printing. On iOS an overlay's drawing is scaled with the page, so its text can soften at high zoom (sharp at 2.5× in tests); Android draws it again at the new size.
  • Text selection: iOS uses PDFKit's selection, which can span pages. Android draws its own: a selection stays on one page, Select all selects that page, copying ends the selection (as in a text field), dragging a handle to the edge does not scroll, and a handle dragged past the other end stops at the first character.
  • Outline (getOutline): on Android, encrypted documents reject with not-supported (their titles are encrypted). iOS (PDFKit) leaves PDF 1.1 named destinations (/Dest /name) without a page.
  • Metadata (PdfDocument.getMetadata): on Android, encrypted documents reject, like the outline. XMP metadata is not read, only the information dictionary.
  • Android: several features come from the platform renderer's newer API — on Android 15+, and on Android 12–14 when the device has SDK extension 13 (the PDF module updated through Google Play system updates; the library then uses PdfRendererPreV). "Android 15+" below includes those devices.
    • Annotations: the platform renderer draws highlights, sticky notes and free text from Android 15+, stamps from Android 17, form fields always — and never squares, circles, lines, polygons, ink, underline or strike-out (measured; see docs/api-design.md "Annotations"). iOS draws all of them.
    • Password-protected PDFs require Android 15+; other devices report unsupported-pdf.
    • Text selection requires Android 15+; other devices ignore long presses.
    • Search requires Android 15+; other devices report isSupported: false.
    • Links work on every Android version, in every form (read from the file). In password-protected documents they come from PdfRenderer instead: Android 15+ (or 12–14 with SDK extension 13), and of the internal links only GoTo actions with an XYZ destination; other links there behave like the rest of the page and are reported through onPageTap.
    • PdfRenderer renders one page at a time per process, so several visible viewers share one render pipeline.
    • Page sizes are discovered as pages are first rendered (reading them up front costs memory equal to the file size in PdfRenderer). Documents with mixed page sizes may re-anchor slightly the first time a page far from the viewport is visited.
    • Documents are parsed in-process; the platform recommends an isolated process for untrusted input, which this library does not provide.
    • A page that cannot be rendered, or running out of memory while rendering, is reported through onError after onLoad (with error.page), and the document stays open (see Errors). iOS reports load failures only.
  • iOS: PDFKit does not expose cache statistics; getStats() reports -1 for them.
  • A base64 source's text is held twice while the document is shown (your string and its native copy, each a third larger than the document), so large documents are better opened as files. data: URIs are not accepted (see Base64 documents).
  • Downloads use GET only, run only while the app runs, and are not limited in total size (see Offline reading and caching).

What's next

Being considered for later releases:

  • Password-protected documents, search and text selection on Android versions without the newer renderer API, with the library's own PDF reading.
  • Editing annotations.
  • Downloads with POST and a body.
  • Screen reader access to form fields on Android, and text selections across pages.