[
https://issues.apache.org/jira/browse/PDFBOX-4499?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16804205#comment-16804205
]
Michael Klink edited comment on PDFBOX-4499 at 3/29/19 9:19 AM:
----------------------------------------------------------------
The cause is that the signature field is not referenced (neither directly nor
indirectly) from the *AcroForm* form definition *Fields* dictionary, it is only
referenced from the the page's *Annots* array.
PDFBox counts on all fields being referenced from the *AcroForm* form
definition *Fields*, and that is ok because according to the specification
|*Fields*|array|_(Required)_ An array of references to the document’s root
fields (those with no ancestors in the field hierarchy).|
I.e. all the document's root fields have to be in that array and all non-root
fields have ancestors up to a root field which (see before) is in that array.
In case of your document exactly one field is not referenced from the *Fields*
array:
{noformat}
38 0 obj
<<
/Rect [ 335.353 332.67 575.797 351.39 ]
/Border [ 0 0 0 ]
/T (EMPLOYEE SIGNATURE)
/F 4
/Subtype /Widget
/TU (EMPLOYEE SIGNATURE)
/DA (/Helvetica 12 Tf 0 g)
/AP 115 0 R
/Contents ()
/AAPL:AKExtras 116 0 R
/FT /Sig
/Type /Annot
/Ff 0
/DR 117 0 R
>>
{noformat}
As you see it doesn't have a *Parent* entry. Thus, it is a root field. On the
other hand the *AcroForm* dictionary looks like this:
{noformat}
2 0 obj
<<
/Fields [ 33 0 R 34 0 R 35 0 R 41 0 R 42 0 R 36 0 R 39 0 R 40 0 R 22 0 R
23 0 R 14 0 R 16 0 R 18 0 R 19 0 R 20 0 R 24 0 R 25 0 R 29 0 R 37 0 R 15 0 R
17 0 R 21 0 R 26 0 R 27 0 R 28 0 R 209 0 R 30 0 R ]
/NeedAppearances true
>>
endobj
{noformat}
As you see, no reference {{38 0 R}}
----
{quote}The signature field contained in the document can be signed on Adobe
Reader and in a other close sourced commercial library.
{quote}
Some PDF processors don't only look for fields in the *AcroForm* *Fields* but
also in the page *Annots*. In particular PDF viewers and libraries derived from
such viewers do so, especially when displaying the page on which the orphaned
field is an annotation.
While this might be appropriate for a viewer - it would be weird if the viewer
displayed the field as annotation but wouldn't allow to fill it - it is not
necessarily a good idea for automatic processing. After all, just as it is
possible that an orphaned field was forgotten to be added to the *AcroForm*
*Fields*, it is also possible that it was forgotten to be removed from the page
*Annots*. A human viewer of the form might decide based on plausibility, an
automatic form filler most likely not.
was (Author: mkl):
The cause is that the signature field is not referenced (neither directly nor
indirectly) from the *AcroForm* form definition *Fields* dictionary, it is only
referenced from the the page's *Annots* array.
PDFBox counts on all fields being referenced from the *AcroForm* form
definition *Fields*, and that is ok because according to the specification
|*Fields*|array|_(Required)_ An array of references to the document’s root
fields (those with no ancestors in the field hierarchy).|
I.e. all the document's root fields have to be in that array and all non-root
fields have ancestors up to a root field which (see before) is in that array.
In case of your document exactly one field is not referenced from the *Fields*
array:
{noformat}
38 0 obj
<<
/Rect [ 335.353 332.67 575.797 351.39 ]
/Border [ 0 0 0 ]
/T (EMPLOYEE SIGNATURE)
/F 4
/Subtype /Widget
/TU (EMPLOYEE SIGNATURE)
/DA (/Helvetica 12 Tf 0 g)
/AP 115 0 R
/Contents ()
/AAPL:AKExtras 116 0 R
/FT /Sig
/Type /Annot
/Ff 0
/DR 117 0 R
>>
{noformat}
As you see it doesn't have a *Parent* entry. Thus, it is a root field. On the
other hand the *AcroForm* dictionary looks like this:
{noformat}
2 0 obj
<<
/Fields [ 33 0 R 34 0 R 35 0 R 41 0 R 42 0 R 36 0 R 39 0 R 40 0 R 22 0 R
23 0 R 14 0 R 16 0 R 18 0 R 19 0 R 20 0 R 24 0 R 25 0 R 29 0 R 37 0 R 15 0 R
17 0 R 21 0 R 26 0 R 27 0 R 28 0 R 209 0 R 30 0 R ]
/NeedAppearances true
>>
endobj
{noformat}
As you see, no reference {{38 0 R}}
> PDDocument.getSignatureFields() does not return all existing signature fields
> -----------------------------------------------------------------------------
>
> Key: PDFBOX-4499
> URL: https://issues.apache.org/jira/browse/PDFBOX-4499
> Project: PDFBox
> Issue Type: Bug
> Components: AcroForm, Signing
> Affects Versions: 2.0.14
> Reporter: Noé Beuret
> Priority: Major
> Attachments: Form_example.pdf
>
>
> Hello,
> I'm having an issue with a PDF while trying to retrieve an existing signature
> field in the document.
> I'm using the method {{PDDocument.getSignatureFields()}} which return
> {{PDSignatureField}} objects on some PDF, but nothing on others which I'm
> sure they have signature fields.
> {noformat}
> PDDocument document = PDDocument.load(getDocumentData("Form_example.pdf"));
> assertNotEmpty(document.getSignatureFields());{noformat}
> One of those PDF is attached to the ticket. The signature field contained in
> the document can be signed on Adobe Reader and in a other close sourced
> commercial library. But on PDFBox, the signature field does not exist.
> Is it an issue or a known limitation of the library or I did I miss something
> ?
>
> Thanks
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org
For additional commands, e-mail: dev-h...@pdfbox.apache.org
