Bug 425791

Summary: Box characters in contents panel
Product: [Applications] okular Reporter: itisme1997
Component: PDF backendAssignee: Okular developers <okular-devel>
Status: RESOLVED UPSTREAM    
Severity: normal CC: aacid, yurchor
Priority: NOR    
Version First Reported In: 1.7.3   
Target Milestone: ---   
Platform: Ubuntu   
OS: Linux   
Latest Commit: Version Fixed/Implemented In:
Sentry Crash Report:
Attachments: Rendering with Okular 1.11 (poppler 0.89)

Description itisme1997 2020-08-25 18:48:56 UTC 
SUMMARY

There are box characters in the contents panel of a pdf that is being rendered. pdftk does not extract any characters where the boxes are being displayed.

STEPS TO REPRODUCE
1. Open FoundationsOfMachineLearning_Mohri_Rostamizadeh_Talwalkar.pdf
2. Examine the contents panel.

OBSERVED RESULT
Chapter headings without suffixed boxes

EXPECTED RESULT
Some chapter headings with suffixed boxes.

SOFTWARE/OS VERSIONS
Linux/KDE Plasma: Ubuntu 19.10 5.3.0-64-generic
KDE Frameworks Version: 5.62.0
Qt Version: 5.12.4

ADDITIONAL INFORMATION
`pdftk FoundationsOfMachineLearning_Mohri_Rostamizadeh_Talwalkar.pdf dump_data | grep BookmarkTitle | cat -vTE` produces does not show characters after the section headers.
The file renders correctly in the firefox pdf renderer and evince.
Comment 1 Yuri Chornoivan 2020-08-25 19:04:40 UTC 
Created attachment 131179 [details]
Rendering with Okular 1.11 (poppler 0.89)

No problems here. Can you give us an address for your file?

Thanks in advance for your answer.
Comment 2 Yuri Chornoivan 2020-08-25 19:05:23 UTC 
Change status.
Comment 3 itisme1997 2020-08-25 19:11:54 UTC 
I can't upload the file, as it's too big. In attempting to create a smaller file with pdftk, I discovered that extracting and updating the info fixes the issue.

pdftk $DOCUMENT.pdf dump_data > in.info
pdftk $DOCUMENT.pdf update_info in.info output out.pdf

There is a diff between these documents, but I don't know why. The diff is also too large to upload.
Comment 4 itisme1997 2020-08-25 19:12:41 UTC 
Here is a link to the pdf.

https://www.dropbox.com/s/7voitv0vt24c88s/10290.pdf?dl=1
Comment 5 Yuri Chornoivan 2020-08-25 19:22:40 UTC 
Boxes can be seen in Okular, but invisible in Evince.
Comment 6 Albert Astals Cid 2020-08-25 21:08:02 UTC 
I would argue the PDF is actually broken, it just has NULL characters in the text, but since poppler was already clearing 1 NULL character from the end if it was there, i've changed it to clear as many NULL characters from the end of strings

https://gitlab.freedesktop.org/poppler/poppler/-/merge_requests/619