How to map links in pdf to the bookmark format? - pdf

Question

A pdf file contains links like the following.
<</05_preface [15 0 R /XYZ 27.999998 763.99994 0]
/sec1 [343 0 R /XYZ 27.999998 393.1575 0]
/fn1 [343 0 R /XYZ 204.5918 254.82751 0]
...
I want to set bookmarks pointing to these locations using cpdf.
I guess 343 is a page number. What is the meaning of 0 after 343? What does "R" mean?
How can I convert this data to a format understandable by cpdf to set pdf bookmarks? Solution in pdftk is also welcome.
EDIT: mkl commented on my error about "343". I see the following contents in the pdf referring to "343 0". How to figure out the page number of "343 0" then? What is the meaning of R?
343 0 obj
<</Type /Annot
/Subtype /Link
/F 4
/Border [0 0 0]
/Rect [34.005386 592.09595 226.92856 604.85718]
/Dest /sec1
/StructParent 100300>>
endobj
...
100300 0 obj
<</Type /StructElem
/S /NonStruct
/P 100299 0 R
/K [<</Type /MCR
/Pg 16055 0 R
/MCID 28>>]
/ID (node00083500)>>
endobj
...
16055 0 obj
<</Type /Page
/Resources <</ProcSet [/PDF /Text /ImageB /ImageC /ImageI]
/ExtGState <</G3 3 0 R>>
/Font <</F7 7 0 R>>>>
/MediaBox [0 0 612 792]
/Contents 16056 0 R
/StructParents 1308
/Parent 16810 0 R>>
endobj
...
16810 0 obj
<</Type /Pages
/Count 8
/Kids [16047 0 R 16049 0 R 16051 0 R 16053 0 R 16055 0 R 16057 0 R 16059 0 R 16061 0 R]
/Parent 16868 0 R>>
endobj
...
16868 0 obj
<</Type /Pages
/Count 64
/Kids [16807 0 R 16808 0 R 16809 0 R 16810 0 R 16811 0 R 16812 0 R 16813 0 R 16814 0 R]
/Parent 16875 0 R>>
endobj
...
16875 0 obj
<</Type /Pages
/Count 512
/Kids [16864 0 R 16865 0 R 16866 0 R 16867 0 R 16868 0 R 16869 0 R 16870 0 R 16871 0 R]
/Parent 16877 0 R>>
endobj
...
16877 0 obj
<</Type /Pages
/Count 1604
/Kids [16873 0 R 16874 0 R 16875 0 R 16876 0 R]>>
endobj
...
I tried to use cpdf to get all objects of the type of /Pages. Why only the last one is printed?
$ cpdf -print-dict-entry /Pages chrome2pdf.pdf
{"/Type":{"N":"/Pages"},"/Count":{"I":1604},"/Kids":[16873,16874,16875,16876]}

Answer 1

Deep PDF surgery, this.
You say that you want to use cpdf to add some bookmarks which point to some destinations which already exist in the file. For example...
[15 0 R /XYZ 27.999998 763.99994 0]
... is a destination on the page whose page tree node has internal PDF object number 15.
So it sounds like you want a correspondence between page numbers and page tree node object numbers. This is not trivial. When you use -list-bookmarks or -list-bookmarks-json in cpdf, the format substitutes the internal PDF object number (say, 15, here) for the actual page number (say 10). It does this to make roundtripping possible - i.e save bookmarks, modify a file in some way, add the bookmarks back. So you are not exposed to the internal object numbers, which are subject to change.
Use
cpdf -output-json in.pdf -o out.json
to output the PDF file as JSON. Now you can read the page tree from that JSON, by following /Pages in the document catalog, as referenced from the document root. The pages in the page tree, when read in an in-order tree-traversal represent the pages 1...n of the file.
Now you have a mapping from object numbers back to page numbers e.g [(15, 1), (27, 2), (9, 3) ... (m, n)].
Now, you can build a cpdf bookmarks file using those page numbers, and the destination coordinates e.g /XYZ 27.999998 763.99994 0 that you have already extracted.
(Side note: the -print-dict-entry example you give does look like it might be a bug. Can you report it, and supply the file, please?)