Pagecraft

redact

Redact PDF

Remove text and pictures from a document so that they are gone from the file.

Reads
a PDF — application/pdf
Writes
one file — PDFs
Tool page
/redact-pdf

Example

Take a name out of a document, rather than draw a box over it.

Given

  • statement.pdf — a PDF of written text

From code

            import { loadDocument, runOperation } from '@pagecraft/core';

const statement = await loadDocument({ bytes, name: 'statement.pdf' });

const [output] = await runOperation('redact', [statement], {
  find: 'Priya Nair',
  caseSensitive: false,
});

From a shell

            pagecraft redact statement.pdf --find 'Priya Nair'

What comes back

One PDF of 1 page, no longer containing “Priya Nair”.

This example is run by the test suite. If it stopped being true, the build would fail.

Options

find--find

string

Remove every

Text to remove wherever it appears in the document.

regex--regex

boolean

Regular expression

Treat what was typed as a pattern rather than as the text to find.

Default: false

caseSensitive--case-sensitive

boolean

Match case

Default: false

wholeWord--whole-word

boolean

Whole words only

So that 'ann' does not take the middle out of 'annual'.

Default: false

regions--regions

object[]

Areas

Rectangles to clear, in points from the bottom-left corner of the page as it is shown.

color--color

string, '#rrggbb'

Box colour

The colour of the box drawn where the content was.

Default: '#000000'

padding--padding

number (pt)

Margin

How far past the text to remove, so that no part of a letter is left behind.

Default: 1

What it cannot do

  • A scanned page holds a picture of words, and nothing here can read it. Run OCR over it first, or give the area as a rectangle.
  • A rectangle that reaches into a word takes the rest of that word — up to eight letters of it, or two where the file does not say which of its codes is a space — because the rectangle a search works out is right to within a letter rather than to the glyph, and half a name left on the page is worse than a whole one removed.
  • Text that is not upright on the page — a rotated watermark, a sideways heading — is removed a run at a time rather than a letter at a time, because a reader reports no position inside a turned run.
  • Text is searched over the whole document. Redacting a name on one page and leaving it on the next is not something this Operation will do; to clear part of a page and nothing else, give the area as a rectangle.
  • Drawings are not removed: a chart or a signature under a rectangle is covered by the box rather than taken out of the file. Text and pictures are.
  • A picture stored as a fax, a JBIG2 scan or a JPEG 2000 cannot be edited by this build, so one lying under a rectangle is removed whole rather than in part.
  • What the document no longer refers to is dropped, so a file that had been saved incrementally comes back without its earlier versions.
  • The title, the author and every other document detail are removed, because they routinely restate what was redacted.
  • An annotation or a form field whose own text is what is being removed is deleted wherever on the page it sits, not only where it lies under a rectangle: a value a reader hands back on demand is a copy of the thing that was redacted.
  • A bookmark, a file attached to the document or a named destination is left as it was. What is cleared is the page content, the annotations over it, the tagging that described it and the document details.
  • A protected document is written back out in the clear. Protect it again if it needs a password.