PDF Adding Electronic Signature - pdf

Question

We have an API for adding electronic signature to PDF documents. PDF documents signed by our API generally have the following structure:
/Type/Annot/
/Type/Sig/
/Type/Font/BaseFont/Helvetica
/Type/Font/BaseFont/ZapfDingbats
/Type/XObject/Subtype/Form
/Type/Page/
/Type/Catalog/
xref
0 1
0000000000 65535 f
57 1
0000088682 00000 n
221 7
0000088804 00000 n
0000088450 00000 n
0000054970 00000 n
0000088289 00000 n
0000054837 00000 n
0000088111 00000 n
0000088211 00000 n
trailer
<</Size 228/Root 221 0 R/Info 222 0 R/ID [<e8f997fdc4d9ee619c59add6882586d8><d13406b4c1e0cb3de6c1e8dd21d6be13>]/Prev 50291>>
startxref
89038
%%EOF
Yet, when we analyzed other API outputs, we also saw structure as follows:
/Type/Sig
/Type/XObject/Subtype/Form
/Type/Metadata/Subtype/XML
/Type/Catalog
/Type/ObjStm/N 4
34 0 obj
<</Length 52/Filter/FlateDecode/Size 35/Root 8 0 R/Info 6 0 R/ID [<4dc91a1875a6d707aec203bb021c93a0><b6fc5ae423a75860537207ff441de268>]/W[1 2 2]/Type/XRef/Index[0 2 6 1 8 2 28 7]/Prev 116>>stream
xœc``øÿŸq‘-ãB9 ±Şš‰A™QÈ]Pá%AXŒ ‚‰¤ =› ñ
endstream
endobj
startxref
45383
%%EOF
We could not find any reference to this difference in the PDF Standarts document nor Adding e-Signature to PDF documents guidelines. Can you please give information on when to use which format.
Thanks in advance

Answer 1

Your question is totally unrelated to iText, and even unrelated to digital signatures.
Let's start with PDF's that end like this:
xref
0 1
0000000000 65535 f
57 1
0000088682 00000 n
221 7
0000088804 00000 n
0000088450 00000 n
0000054970 00000 n
0000088289 00000 n
0000054837 00000 n
0000088111 00000 n
0000088211 00000 n
trailer
<</Size 228/Root 221 0 R/Info 222 0 R/ID [<e8f997fdc4d9ee619c59add6882586d8><d13406b4c1e0cb3de6c1e8dd21d6be13>]/Prev 50291>>
startxref
89038
%%EOF
You see the end-of-file marker (%%EOF) preceded with the starting positing of the cross-reference table (startxref, in this case the xref table starts at position 89038).
The cross-reference table defines the byte offsets of every object inside the PDF. In PDF versions prior to PDF 1.5, the Cross-reference table is added in plain text, and each byte offset is defined using ten digits. Hence the size of a file will be limited to 10 to the tenth bytes (approximately 10 gigabytes).
The maximum size of a PDF with a version older than PDF 1.5 is about 10 gigabytes.
Starting with PDF 1.5, you can also have this:
34 0 obj
<</Length 52/Filter/FlateDecode/Size 35/Root 8 0 R/Info 6 0 R/ID [<4dc91a1875a6d707aec203bb021c93a0><b6fc5ae423a75860537207ff441de268>]/W[1 2 2]/Type/XRef/Index[0 2 6 1 8 2 28 7]/Prev 116>>stream
xœc``øÿŸq‘-ãB9 ±Şš‰A™QÈ]Pá%AXŒ ‚‰¤ =› ñ
endstream
endobj
startxref
45383
%%EOF
We still have the end-of-file marker, and we still point to the starting point of the cross-reference table, but the entries of the cross-reference table are no longer visible in clear text. They are compressed: xœc``øÿŸq‘-ãB9 ±Şš‰A™QÈ]Pá%AXŒ ‚‰¤ =› ñ
This allows you to create files with a file size higher than 10 gigabytes. All of this is explained in ISO 32000-1 and ISO 32000-2.
This is a screen shot of the PDF specification:
If you have software that digitally signs documents, but only supports the format that predates PDF 1.5, I would seriously question whether that software supports digital signatures that meet today's standards. I recently saw a project where a developer still used SHA-1 as the hashing algorithm. Fortunately, I was able to explain that this was wrong by referring him to a blog about SHA-1, otherwise he'd probably have been fired if his employer found out (or sued if his customer discovered that he was paying for signatures that are no longer considered being safe).