Am 27.05.2015 um 20:02 schrieb Balaji Venkatamohan:
Thanks Tilman for letting the website developer know about the shortcomings
of their compression technique.
The PDF owner did not share with us the information about which website
they used for compressing the PDF. My teammates helped in identifying this
website. I will let the customer know about this particular website and
will leave it to them regarding continuing to use this website for their
PDF documents.
Could you also answer the following question please?
Would Pdfbox API change its code to accommodate the incorrect condition
that annotation fields (editable fields) are outside acro form fields as
well? I know the PDF compressed by the website is incorrect and hence I
would understand if you don't go ahead with this.
As I said, I'm not the acroform specialist here. So I can't tell if it
is possible to repair these, and if there aren't side effects that e.g.
PDFs with annotations end up being forms. Yes, we've done all sorts of
things to accomodate broken PDFs. But here the fault is known, it is a
website that deletes data from PDFs to "compress" them. The better
solution would be to have this guy fix his website, i.e. allow options
to decide what is to be removed, and what not. Another solution (which I
mentioned before) would be to have your customer compress his PDFs with
the method I mentioned in this thread, i.e. if this customer of yours
generates PDFs, but doesn't have the knowledge to compress the streams.
He could of course look into our source code (FlateFilter.java, it is
just 10 lines) and see how to compress himself.
Tilman
Thanks,
Balaji
On Tue, May 26, 2015 at 10:45 PM, Tilman Hausherr <[email protected]>
wrote:
I just tested it. It also removes /Outlines and /Metadata and more
important data from PDF files.
So your client can't share the PDF with us, but he shared it some website.
A little research shows that this website is owned by Lauri Lehtinen from
Talinn, Estonia.
http://www.checkdomain.com/cgi-bin/checkdomain.pl?domain=pdfcompress.com
https://www.linkedin.com/in/laurilehtinen
https://twitter.com/laurii
I also tweeted him.
Tilman
Am 27.05.2015 um 03:06 schrieb Balaji Venkatamohan:
Okay, I found out the online tool used by the customer to compress their
PDF.
It is : https://www.pdfcompress.com/
I don't need to rely on the PDF sent by the customer because all PDFs that
are available on the web, are compressed in the same manner by this tool,
that is, it gets rid of all acro form fields during compression.
For example, the f941 govt form available at this site:
http://www.irs.gov/pub/irs-pdf/f941.pdf
If we compress this using the online tool, the resultant file size is very
low, which is good. However, there are no acro form fields in the
compressed PDF.
Thanks,
Balaji
On Sun, May 24, 2015 at 2:38 AM, Maruan Sahyoun <[email protected]>
wrote:
Hi,
Am 23.05.2015 um 16:37 schrieb Balaji Venkatamohan <[email protected]
:
Hi,
So AcroForms/Fields is an empty Array?
Yes, in the filled interview_compressed.pdf, the acroforms are not null
but
empty. Size of array is zero.
Also, I tried qpdf command line tool to compress the file interview.pdf
and
the resultant compressed file size of 1.6MB was no way near the file
size
of interview_compressed.pdf (21 KB).
would you think it's possible to get a similar PDF file or permission to
use it internally so we have a sample to look at a potential fix.
Although the PDF is not inline with the spec as Acrobat is able to handle
it we could look into getting a similar result.
BR
Maruan
Thanks,
Balaji
On Fri, May 22, 2015 at 11:58 PM, Maruan Sahyoun <
[email protected]
wrote:
Hi,
Am 22.05.2015 um 23:00 schrieb Balaji Venkatamohan <
[email protected]
:
I opened the interview_compressed in notepad++ and did not see any
'Acroform' text anywhere.
However, as Maruan suggested, I entered some data into what looks like
form
fields of interview_compressed.pdf and saved it. When I opened this
file
in
notepad++, I did see 'Acroform' text in it. I also noticed an increase
in
file size from 21 KB to ~530 KB.
I then ran this filled saved compressed PDF in pdfdebugger.java and
saw
that the field values were getting stored but not under Acroform
fields
but
under Annotations.
So AcroForms/Fields is an empty Array?
Please refer to this image:
http://imageshack.com/a/img540/9951/QGLDtS.jpg
So, whatever the compression technique was, it simply made all the
Acroform
fields disappear from the original PDF but retained all annotations
which
also contain the interactive forms and this helped reduce the file size
so
much? If this is the case, can pdfbox API also use similar compression
technique to compress such a a huge file into a smaller one?
On Fri, May 22, 2015 at 1:25 PM, Maruan Sahyoun <
[email protected]>
wrote:
Hi,
Am 22.05.2015 um 21:54 schrieb Tilman Hausherr <
[email protected]
:
Am 22.05.2015 um 17:53 schrieb Balaji Venkatamohan:
Hello,
I used PdfDebugger to make the internal PDF structure of the two
files
(1)
interview.pdf and (2) interview_compressed.pdf visually available
and I
have uploaded my images to imageshack. Here are the four links:
http://imageshack.com/a/img538/8277/JghCpG.jpg
http://imageshack.com/a/img909/6140/KsYNGR.jpg
http://imageshack.com/a/img903/8644/mk15As.jpg
http://imageshack.com/a/img901/8610/NXe3mJ.jpg
http://imageshack.com/a/img673/8633/0GMdjQ.jpg
The first two links are from the internal structure of
interview.pdf
(original uncompressed file)
The third and fourth links are from the internal structure of
interview_compressed.pdf (compressed file)
The fifth link compares the file sizes of the two files and as you
can
also
see, the difference is huge.
As you might notice, the file interview_compressed.pdf has no
acroform
Indeed... but this is needed - from the spec:
"The contents and properties of a document’s interactive form shall
be
defined by an interactive form dictionary that shall be referenced
from
the
AcroForm entry in the document catalogue (see 7.7.2, “Document
Catalog”).
Table 218 shows the contents of this dictionary."
correct
fields listed even though opening the PDF in pdf reader allows me to
enter
values in places which look like AcroForm fields and also save them.
Are
there any other PDF 'types' similar to Acroform fields which would
enable
users to fill data and which can be accessed in PdfBox APIs without
having
to go through PDAcrofield?
Yes, annotations... there are some common parts, but this is just a
vague observation from me, I'm not the acroform specialist.
from a first glance it looks like there are all entries necessary to
(re-)
generate the form fields. That's what's likely happening for this
document
in Adobe Reader. Would be interesting to see what's being save after
the
forms has been filled out and saved using Acrobat. We'd need a test
form to
come up with an enhancement like this.
BR
Maruan
What you should do: use NOTEPAD++ to look whether there's
"/AcroForm"
in
the "compressed" file.
- if it is missing, tell the client (or your boss) just that
- if it isn't missing, then there's some problem in PDFBox (try also
the
loadNonSeq I mentioned earlier)
Tilman
You can use qpdf , then use these options:
I will now try using this link to compress the original file.
Another strategy to think about - can your client generate a
non-confidential file, so that you can share it, and the
"compressed"
file?
I wish I had direct communication with the clients but due to
bureaucracy,
I am having to go through multiple layers to get my message across
to
them.
I will share more information as soon as I have them.
PS: i sent these image links to my personal email first to make
sure
that I
can open them. I could and so I am hoping you all could too. If you
are
unable to open them, please let me know.
Thanks,
Balaji
On Fri, May 22, 2015 at 6:45 AM, Tilman Hausherr <
[email protected]
wrote:
Am 22.05.2015 um 08:28 schrieb Andreas Lehmkühler:
Hi,
Balaji Venkatamohan <[email protected]> hat am 20. Mai 2015 um
03:24
geschrieben:
Thank you for your pointers and sorry about the image. I am
attaching it
with this email.
The point I am trying to make is that the PDF, which was
decompressed
using
WriteDecodedDoc, is smaller in size than the original PDF given
to
us by
our customers.
Also, the decompressed PDF generated by WriterDecodedDoc of
PDFBox
did
not
have any PDAcroform fields whereas the decompressed PDF given to
us
by
the
customers does contain Acroform fields. Hence I wanted to know
how
to
properly decompress the PDF using pdfbox APIs. The reason why I
was
analyzing COSStream was to check if the decompression of the
compressed
PDF
was happening correctly while using PDFBox APIs.
I know it would have been difficult for you to help me without
the
actual
PDFs. For that, I would like to thank you for your time and
pointers.
Maybe it's worth to try to share the file "visually" with us. Open
both
files
(compressed and decompressed) with PDFDebugger [1] and post a
screenshot
of both
somehwere (dropbox etc.) and share the link with us. Maybe that
could
shed some
light on your issue.
@Balaji: here's an example on how such a screenshot would look
like:
http://home.snafu.de/tilman/tmp/pdfdebugger-screenshot.png
Tilman
BR
Andreas Lehmkühler
[1] http://pdfbox.apache.org/1.8/commandline.html#pdfDebugger
On Tue, May 19, 2015 at 2:57 PM, Tilman Hausherr <
[email protected]>
wrote:
Hi,
The image doesn't appear in the mailing list.
This is all very confusing... /acroform is in the document
catalog.
I
don't see how the page content stream is related to it. The best
is
that
you either go through the source code, or read the spec and then
look at
the pdf.
To find out what's going on, you'd have to start from that
/acroform
entry
and then compare the two files.
It is really difficult to help you without the files. The cause
could
be a
bug in pdfbox, or a malformed pdf...
Some more ideas:
- use loadNonSeq(file, null) instead of load(file)
- try the unreleased 2.0 version, that one has some
improvements
in
the
acroform stuff. Note that the API is different.
https://pdfbox.apache.org/download.cgi#scm
https://pdfbox.apache.org/2.0/getting-started.html
If you still need help, one possibility would be 1) post the
smallest
possible code that fails, and 2) post a small part of the raw
PDF,
i.e.
the
objects relevant to the field in your code.
Tilman
Am 19.05.2015 um 23:03 schrieb Balaji Venkatamohan:
Moreover, for every page of the compressed PDF (there are 3
pages), I
tried getting the COSStream for each of the page :
PDPage firstPage=(PDPage)
document.getDocumentCatalog().getAllPages().get(0);
pdStream=firstPage.getContents();
COSStream stream=pdStream.getStream();
In the above code snippet, the object stream, when analyzed in
debug
mode, has the following:
The line from the compressed PDF as opened with Notepad++ is :
<</Filter/FlateDecode/Length 5675>>stream
From this point on, using the COSStream object for every
page,
how
can I
decompress and find out the acroform fields given that the
unFilteredStream
object is null for COSStream?
On Tue, May 19, 2015 at 1:38 PM, Balaji Venkatamohan <
[email protected]
<mailto:[email protected]>> wrote:
Thank you for your response Tilman.
I had previously tried using the WriteDecodedDoc for my
compressed
PDF and I tried to get the number of acro form fields
present
in
the output file generated by WriteDecodedDoc. The API still
could
not find the acro form fields in the generated decompressed
file.
Also the decompressed file generated is 75 KB which is far
less
than the original decompressed file which I have (1.6 MB)
though I
could edit the acro form fields using acrobat reader.
Thanks,
Balaji
On Tue, May 19, 2015 at 1:18 PM, Tilman Hausherr
<[email protected] <mailto:[email protected]>>
wrote:
Am 19.05.2015 um 21:35 schrieb Balaji Venkatamohan:
My question is: how do I flatedecode a PDF so
that I
can
find all the
acroform fields within it. ANy help or pointers
would
be
highly appreciated.
You could try the WriteDecodedDoc option of the
command
line
app
https://pdfbox.apache.org/1.8/commandline.html#writeDecodeDoc
Maybe you can have further ideas by comparing the two
files
with NOTEPAD++.... however the two files might have
their
objects in different order.
Tilman
---------------------------------------------------------------------
To unsubscribe, e-mail:
[email protected]
<mailto:[email protected]>
For additional commands, e-mail:
[email protected]
<mailto:[email protected]>
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
<mailto:
[email protected]>
For additional commands, e-mail: [email protected]
<mailto:
[email protected]>
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
