Missing text after redaction

[NoraishaYusuf]

I'm currently redacting sensitive information in resumes.
There are two major issues i'm facing as follows:

1. I found chunk of texts & headings, and sometimes photos missing for several of the redacted resumes.
   For one of the resumes, these were the warnings thrown (not in the right order):
   mupdf: unknown colorspace: R26
   mupdf: unknown colorspace: R28
   mupdf: syntax error in content stream
   mupdf: unknown keyword: 'EI'
   mupdf: unknown colorspace: R27
   mupdf: unknown keyword: 'kd'
   mupdf: array not closed before end of file

I have attached the relevant part of my codes for reference.

2. The redacted word overlaps with words located above and below it (as shown in the images)

I appreciate it very much if anyone can help me solve these two issues.

[JorjMcKie]

Problem 1:

Hard to tell without looking at that file (probably confidential anyway). But there arethings like damaged PDFs ...
You could try cleaning the file / the page before processing to reveal / remove any errors.

• cleaning the file e.g. by mutool clean -gggsc file.pdf
• cleaning the page inside the script: page.clean_contents(sanitize=True)

Problem 2:

MuPDF normally uses the full font-defined line height when identifying the hits of search. If the PDF is made with smaller distances between lines, then adjacent lines may overlap somewhat. The redaction logic of MuPDF in turn removes every character overlaping the redaction rectangle - the result of this is what you saw.

• set a global parameter that reduces character bbox heights to the fontsize value: fitz.TOOLS.set_small_glyph_heights(True) - right after importing fitz.

[mani2106]

Does the clean_contents method exist for a Page object as per fitz version 1.17.0?

[JorjMcKie]

Yes! Should have been callled page.cleanContents() then.