pdfbox-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Vomlel Jan <Jan.Vom...@aipsafe.cz>
Subject RE: problem with pdf eof
Date Mon, 20 Oct 2014 11:50:03 GMT
Hi Maruan,

I create patch for bug PDFBOX-2436.

After %%EOF it skips data to next object.

I don´t know, if such data are allowed by specification, but some czech portal create them
and acrobat have no problem with them.

I changed org.apache.pdfbox.pdfparser.PDFParser near line 584, branch 1.8. Can you commit
it and fix this bug?


                            pdfSource.unread(eof.getBytes("ISO-8859-1"));
                        }
                    }
                }
                isEndOfFile = true;

                //PDFBOX-2436 - some files contain binary data after %%EOF.
                skipToNextObj();
            }
        }
        //we are going to parse an normal object
        Else

Thank you, Jan

-----Original Message-----
From: Vomlel Jan
Sent: Friday, October 17, 2014 9:12 AM
To: users@pdfbox.apache.org; brzrkr@pobox.com
Subject: RE: problem with pdf eof

I reported parsing error for load function:
https://issues.apache.org/jira/browse/PDFBOX-2436
Jan

-----Original Message-----
From: Maruan Sahyoun [mailto:sahyoun@fileaffairs.de]
Sent: Thursday, October 16, 2014 8:23 PM
To: users@pdfbox.apache.org; brzrkr@pobox.com
Subject: Re: problem with pdf eof

sorry if that has been unclear - as of now if you’d like to sign you have to use load()
loadNonSeq() is not an option!

For all other cases use loadNonSeq() and if that fails load() as a fallback.

We are working on getting the missing signing support into nonSeq() but that will probably
be after 2.0.

Now if you have parsing issues with load() please open an issue in Jira and attach the PDFs
together with code to reproduce it. Same if you have parsing issues with loadNonSeq().

Of course if someone is willing to help getting that in … patches are welcome.

Maruan

Am 16.10.2014 um 20:13 schrieb Brzrk One <brzrk1@gmail.com>:

> I hear dual advice here...
> - don't use NonSeq for signatures
> - but use NonSeq for multiple EOFs
> Files with both multiple EOFs and signatures will have problems...
> unless you mean we should parse 2x?
>
> On Thu, Oct 16, 2014 at 12:12 PM, Maruan Sahyoun
> <sahyoun@fileaffairs.de>
> wrote:
>
>> depends on the parser being used. NonSeq does follow the Xref
>> information and handles multiple EOFs (incremental updates) when parsing.
>>
>> BR
>> Maruan
>>
>> Am 16.10.2014 um 17:01 schrieb Brzrk One <brzrk1@gmail.com>:
>>
>> I've noticed that when there are multiple EOFs in the file, PDFBox
>> parsing is less reliable.
>>
>>
>> On Thu, Oct 16, 2014 at 9:44 AM, Vomlel Jan <Jan.Vomlel@aipsafe.cz> wrote:
>>
>> When I use load insted of loadNoSeq, signatures are in this case  valid.
>>
>> But for some documents load function doesnot read complete document.
>> That is why I used loadNoSeq. Some signatures are then missing.
>>
>> Viz:
>> http://leteckaposta.cz/831516385
>> h1.pdf - original file (signature and timestamp) h2.pdf - add first
>> signature by pdfbox (timestamp is missing) h3.pdf - add second
>> signature by pdfbox (timestamp and previous signature is missing)
>>
>> Jan
>>
>> -----Original Message-----
>> From: Maruan Sahyoun [mailto:sahyoun@fileaffairs.de]
>> Sent: Thursday, October 16, 2014 2:37 PM
>> To: users@pdfbox.apache.org
>> Subject: Re: problem with pdf eof
>>
>> when signing please make sure that you load the pdf using
>> PDDocument.load instead of PDDocument.loadNonSeq.
>>
>>
>> Am 16.10.2014 um 11:57 schrieb Vomlel Jan <Jan.Vomlel@aipsafe.cz>:
>>
>>
>>
>> -----Original Message-----
>> From: Maruan Sahyoun [mailto:sahyoun@fileaffairs.de]
>> Sent: Thursday, October 16, 2014 11:55 AM
>> To: users@pdfbox.apache.org
>> Subject: Re: problem with pdf eof
>>
>> when you say invalid do you mean it’s corrupted or e.g. you get a
>>
>> warning sign in Adobe Reader? Would you have a sample PDF?
>>
>>
>> When you sign a document and sign it again the first signature points
>> to
>>
>> a different document revision as you have changed the documents
>> content afterwards. So invalid in that context could mean that the
>> warning you might be getting is only reflecting that fact. Would need
>> to see the document to  understand what’s going on.
>>
>>
>> BR
>>
>> Maruan
>>
>> Am 16.10.2014 um 11:48 schrieb Vomlel Jan <Jan.Vomlel@aipsafe.cz>:
>>
>> Hi Maruan and others,
>>
>> I created signature and it seems OK.
>> But when I create second signature (loadNonSeq, addSignature,
>>
>> saveIncremental again), the first signature becomes invalid.
>>
>> I think that there can be problem, that first page is updated
>> (signatur
>>
>> is invisible), but I dont understand it enough.
>>
>>
>> Jan
>>
>> -----Original Message-----
>> From: Maruan Sahyoun [mailto:sahyoun@fileaffairs.de]
>> Sent: Monday, October 13, 2014 4:09 PM
>> To: users@pdfbox.apache.org
>> Subject: Re: problem with pdf eof
>>
>> Hi Jan,
>>
>> there are sample in the examples package for various ways to sign a
>>
>> document [1]. Signing a document needs incremental saving.
>>
>>
>> OTOH choosing the right solution should not be made on the base if
>>
>> there is a license fee or not.
>>
>>
>> Maruan Sahyoun
>>
>> [1]
>>
>>
>> http://svn.apache.org/viewvc/pdfbox/trunk/examples/src/main/java/org/
>> apache/pdfbox/examples/signature/
>>
>>
>>
>> Am 13.10.2014 um 16:02 schrieb Vomlel Jan <Jan.Vomlel@aipsafe.cz>:
>>
>> Hi Maruan (and others),
>>
>> I would like to use pdfbox and bouncycastle for managing pdf
>>
>> signatures. Parsing, validation, timestamping (PADES LTV) .
>>
>> We used itext for it, but it is under commercial licence.
>> Parsing signatures seems to be working (thanks to your advice). So I
>>
>> will try to create timestamp.
>>
>> Is it possible with pdfbox?  I found save method on PDDocument, but
>>
>> Iˇm afraid, that it can change bite representation of pdf, and
>> signatures become invalid. Is it true? What is right way to create
>> signature or timestamp with pdfbox?
>>
>>
>> Jan
>>
>>
>> -----Original Message-----
>> From: Maruan Sahyoun [mailto:sahyoun@fileaffairs.de]
>> Sent: Friday, October 10, 2014 10:44 AM
>> To: users@pdfbox.apache.org
>> Subject: Re: problem with pdf eof
>>
>> Hi Jan,
>>
>> choosing the right technology is very important so I do understand
>>
>> your concerns. I had to make such decision about using PDFBox in the
>> past too.
>>
>> It can
>> If you have specific issues I can answer I’m happy to try to do so.
>> As
>>
>> a general statement PDFBox is used in production environments today
>> (as an example we ourselves are using it for a banking customer to
>> process account statements, an airline company to preprocess
>> archiving documents and various other customers).
>>
>>
>> PDFBox is continuously enhancing the parsing as we try to deal with
>>
>> real world PDF files which are not always inline with the the PDF
>> specification. Currently the best approach is to use
>> PDDocument.loadNonSeq (which parses documents according to the Xref
>> information) and in case of an exception PDDocument.load (which
>> parses sequentially). The Apache Tika project, which uses PDFBox for
>> parsing PDF’s, is running the parsing and text extraction against 50k
>> PDFs being made available via http://digitalcorpora.org
>>
>>
>> What is the application you would like to be using PDFBox for? Text
>>
>> Extraction, image conversion …. - I might be able to give you more
>> specific information for your use case.
>>
>>
>> BR
>>
>> Maruan
>>
>> Am 10.10.2014 um 10:10 schrieb Vomlel Jan <Jan.Vomlel@aipsafe.cz>:
>>
>> Thank you Maruan, this function loads document.
>>
>> I have read https://pdfbox.apache.org/ideas.html "Replace/Enhance
>>
>> PDF parsing". I think correct parsing is very important, and I have
>> some doubts, if I can use pdfbox in production. Can you say something
>> to rest me :-).
>>
>>
>> Jan
>>
>> -----Original Message-----
>> From: Maruan Sahyoun [mailto:sahyoun@fileaffairs.de]
>> Sent: Friday, October 10, 2014 9:25 AM
>> To: users@pdfbox.apache.or
>> Subject: Re: problem with pdf eof
>>
>> Hi
>>
>> you can try PDDocument.loadNonSeq(InputStream is, null)
>>
>> BR
>>
>> Maruan
>>
>> Am 10.10.2014 um 09:09 schrieb Vomlel Jan <Jan.Vomlel@aipsafe.cz>:
>>
>> Hello,
>> I use PDFBox 1.8.7  PDDocument.load(InputStream is) method to parse
>>
>> PDF document in attachement.
>>
>> Method return without exception, but document model is incomplete.
>>
>> Problem is in characters after EOF (ofset 22939):
>> startxref
>> 22449
>> %%EOF
>> @
>> 16 0 obj
>> <<
>> /Type /Catalog
>>
>> PDFBox create internal IOException and ignore it with comment:
>>               /*
>>                * PDF files may have random data after the EOF
>>
>> marker. Ignore errors if
>>
>>                * last object processed is EOF.
>>                */
>>
>> Is this PDF construction valid?
>> Which parser in PDFBox is correct? I tried ConformingPDParser, but
>>
>> another error occured.
>>
>>
>> Jan
>>
>>
>>
>>
>> Tento e-mail ani žádný z připojených souborů nejsou přijetím návrhu
>>
>> na uzavření smlouvy, ledaže je to v nich výslovně uvedeno. Pokud tomu
>> tak není, nelze je považovat za jednání, které by zakládalo jakékoliv
>> nároky vůči společnosti AiP Safe. Tento e-mail je určen pouze
>> uvedenému příjemci a dalším osobám, které jsou jmenovitě uvedeny jako
>> příjemci, a jeho obsah, včetně obsahu všech připojených souborů, je
>> důvěrný. Jestliže nejste oprávněný příjemce, zdržte se, prosím,
>> jakékoliv formy zveřejnění, reprodukce, kopírování, distribuce nebo
>> šíření jeho obsahu, včetně obsahu všech připojených souborů. Pokud
>> jste obdržel tento e-mail omylem, oznamte to, prosím, neprodleně jeho
>> odesilateli a e-mail, včetně všech připojených souborů, vymažte.
>> Všechny e maily adresované, přijímané nebo posílané AiP Safe s.r.o.
>> nebo zaměstnanci AiP Safe s.r.o. jsou považovány za zásadně pracovní
>> e-maily. V souladu s tím odesilatel nebo příjemce těchto e mailů
>> souhlasí, že mohou být čteny jinými zaměstnanci AiP Safe s.r.o., než
>> je daný příjemce nebo odesilatel, proto aby byla zajištěna kontinuita pracovních
aktivit a byla umožněna jejich kontrola..
>>
>>
>>
>>
>>
>>
>>
>>
>>


________________________________

Tento e-mail ani žádný z připojených souborů nejsou přijetím návrhu na uzavření
smlouvy, ledaže je to v nich výslovně uvedeno. Pokud tomu tak není, nelze je považovat
za jednání, které by zakládalo jakékoliv nároky vůči společnosti AiP Safe. Tento
e-mail je určen pouze uvedenému příjemci a dalším osobám, které jsou jmenovitě uvedeny
jako příjemci, a jeho obsah, včetně obsahu všech připojených souborů, je důvěrný.
Jestliže nejste oprávněný příjemce, zdržte se, prosím, jakékoliv formy zveřejnění,
reprodukce, kopírování, distribuce nebo šíření jeho obsahu, včetně obsahu všech
připojených souborů. Pokud jste obdržel tento e-mail omylem, oznamte to, prosím, neprodleně
jeho odesilateli a e-mail, včetně všech připojených souborů, vymažte. Všechny e maily
adresované, přijímané nebo posílané AiP Safe s.r.o. nebo zaměstnanci AiP Safe s.r.o.
jsou považovány za zásadně pracovní e-maily. V souladu s tím odesilatel nebo příjemce
těchto e mailů souhlasí, že mohou být čteny jinými zaměstnanci AiP Safe s.r.o., než
je daný příjemce nebo odesilatel, proto aby byla zajištěna kontinuita pracovních aktivit
a byla umožněna jejich kontrola..
Mime
View raw message