Cdtjkyj

By getting metadata from some pdf files with bookmarks by pdftk, I observe the format of bookmarks in pdf metadata. So I am guessing that with pdftk, it is possible to add and edit bookmarks of a pdf file.

Here are three steps that I follow:

1. I firstly get the metadata (including bookmarks if any) from a pdf
   file into a text file by pdftk in.pdf dump_data > in.info.


2. 
   

   Next I add some bookmarks into the metadata text file in.info, changing it
   from

   
   
   

   InfoKey: Creator
   
   InfoValue: PScript5.dll Version 5.2
   
   InfoKey: Title
   
   InfoValue: SSReader Print.
   
   InfoKey: Producer
   
   InfoValue: Acrobat Distiller 7.0 (Windows)
   
   InfoKey: Author
   
   InfoValue: Administrator
   
   InfoKey: ModDate
   
   InfoValue: D:20050605073244+08'00'
   
   InfoKey: CreationDate
   
   InfoValue: D:20050605073244+08'00'
   
   PdfID0: 591a87c91dc76881fdf2ccf3811e72a5
   
   PdfID1: 6b6ab11de8824e438e4f5eb1d85ec72
   
   NumberOfPages: 400
   
   PageLabelNewIndex: 1
   
   PageLabelStart: 1
   
   PageLabelNumStyle: DecimalArabicNumerals
   
   

   
   
   

   to

   
   
   

   InfoKey: Creator
   
   InfoValue: PScript5.dll Version 5.2
   
   InfoKey: Title
   
   InfoValue: SSReader Print.
   
   InfoKey: Producer
   
   InfoValue: Acrobat Distiller 7.0 (Windows)
   
   InfoKey: Author
   
   InfoValue: Administrator
   
   InfoKey: ModDate
   
   InfoValue: D:20050605073244+08'00'
   
   InfoKey: CreationDate
   
   InfoValue: D:20050605073244+08'00'
   
   PdfID0: 591a87c91dc76881fdf2ccf3811e72a5
   
   PdfID1: 6b6ab11de8824e438e4f5eb1d85ec72
   
   NumberOfPages: 400
   
   BookmarkBegin
   
   BookmarkTitle: Front cover
   
   BookmarkLevel: 1
   
   BookmarkPageNumber: 1
   
   BookmarkBegin
   
   BookmarkTitle: About the Author
   
   BookmarkLevel: 1
   
   BookmarkPageNumber: 5
   
   BookmarkBegin
   
   BookmarkTitle: Title page
   
   BookmarkLevel: 1
   
   BookmarkPageNumber: 6
   
   BookmarkBegin
   
   BookmarkTitle: Copyright page
   
   BookmarkLevel: 1
   
   BookmarkPageNumber: 7
   
   BookmarkBegin
   
   BookmarkTitle: Foreword
   
   BookmarkLevel: 1
   
   PageLabelNewIndex: 1
   
   PageLabelStart: 1
   
   PageLabelNumStyle: DecimalArabicNumerals
   
   

   
   


3. Then I try to put the edited metadata back to the pdf file by pdftk
in.pdf update_info in.info output out.pdf.

But when I open the new pdf file out.pdf in evince or adobe reader, there is no bookmark in the new pdf file.

If I get the metadata from the new pdf file out.pdf by pdftk out.pdf dump_data > out.info, there will be no bookmarks in out.info. It looks like the bookmarks were not added successfully.

I was wondering if there are some mistakes? How shall I add and edit bookmarks of a pdf file, not necessarily by pdftk?

It looks to me like from version 1.45 (2012-12-06), pdftk does allow modification of bookmarks via the update_info command:

You can now add or change a PDF’s bookmarks using update_info.

via:
http://www.pdflabs.com/docs/pdftk-version-history/

jPDFtweak (Java, so runs on Unix/Linux) can alter bookmarks, but I don't know if you can script anything with it.

For scripting, I'd guess your only native Unix/Linux option would be pdflatex with the pdfpages package. But that's a learning curve if you're not already a LaTeX user.

EDIT: Actually it may be possible with ghostscript: See here or here or
here

If you still stick with those unix scripts, then

1. extract bookmark data dumped from pdftk
   


2. write one extra script to convert dumped bookmark data to pdfmarks format, which ghostscript command gs is accepted.


3. use gs script to merge them together with pdfmarks

Take a look at http://blog.tremily.us/posts/PDF_bookmarks_with_Ghostscript/ the script pdf-merge.py do exactly what you (or I) want.

pdf-merge.py --output=merged.pdf input1.pdf input2.pdf

Some minor improvements could be done in his script

1. unicode handling


2. output bookmark files, so people can adjust it as well

Anyway, it should work

pdftk is definetly the right tool (with the right syntax):

BookmarkBegin

BookmarkTitle: Chapter 1

BookmarkLevel: 1

BookmarkPageNumber: 1

BookmarkBegin

BookmarkTitle: Paragraph 1.1

BookmarkLevel: 2

BookmarkPageNumber: 1

Caleb,

As far as I understand, bookmarks as per the PDF spec can't be injected via a diagnostic tool like pdftk. Updating the metadata to refer to chapters and bookmark landing anchors that don't exist will definitely not work, it may even make your PDF inconsistent or unopenable.

I ended up using java+iText libraries to do what you're suggesting as per this tutorial, but we're dealing with hundreds of pdfs daily, so it needed to be automated. If you're doing a one-off, Adobe Acrobat should be able to do this.

To summarise all this good answers:

There is bmconverter project on GitHub that can convert various PDF bookmark formats. It can convert pdftk output to jpdftweak format, although if you will use jpdftweak then you wont need to convert pdftk output to csv, as you can do all work from within jpdftweak.

Unfortunately pdfmarks is not supported by the project, but fortunately someone posted script in bmconverter issues that can convert pdftk output to pdfmarks. So ghostscript batch option is an option

pdftk bookmark format is a little bit tedious to write. Instead I created my own script using bash, sed, pdftk and python3. Check it out at this repo: https://github.com/SiddharthPant/booky

So now I can create a text file(bkmrks.txt) like this which takes just 5 minutes to write even for a 1000 page pdf.

{

  Title1, 1

  Title2, 2

  {

    Subtitle1, 3

    Subtitle2, 4

    {

      SubSubtitle1, 5

      ...

    }

  }

}

and then use my script

./booky.sh pdf_file.pdf bkmrks.txt

this automatically creates a pdf(pdf_file_new.pdf) that has my bookmarks in it.