Where Man Wins over ... PDF

non-editable or "secured" PDF

Recently (end 2018) however I heard some junior colleagues talking about non-editable PDF being requested in support for some job applications. So I decided to go a little deep in the matter.

• In fact one job announcement required the application form (if not sent in hardcopy) to be sent by "certified electronic e-mail" (an idiosyncrasy of the public administration in this country) as a "non-modifiable PDF".
  Luckily the request was just for the application form, not for any other attachment, and not for the publications (journal papers).
• A later job announcement instead requires the application form in "PDF/A format".

So I had a look about the way to produce such beasts.

• I started from a very simple PDF file, this one, created via ps2pdf from a PostScript file created by hand
  If you issue the command pdfinfo or open it with Acrobat Reader and go to Document -> Security -> Show Security Properties you will see there is no protection of any sort (Show Details is dimmed out).
• Of course if you open the file with a text editor, you would be able to change it, but it is unlikely you'd know what kind of edit to apply since the file is a mixture of directives in plain ASCII and binary (compressed ?) data.

• Now to produce a non-editable version (just for fun I made it also non-printable) I used the command
  qpdf --encrypt '' ownpwd 40 --modify=n --print=n -- start.pdf secured.pdf
  The result is here.
• Such file, when opened with Acrobat Reader, and going to Document -> Security -> Show Security Properties -> Show Details, says "Changing the document: not allowed". Similarly pdfinfo says "Encrypted: yes (print:no copy:yes change:no addNotes:yes algorithm:RC4)"
  Of course you can open the file with a text editor; you will note the usual mixture of text and binary, and you will note that it is different from the input. The difference is that it is encrypted so, even if you'd know how to edit a normal PDF, here you have an additional layer of difficulty: to decrypt the content you have to know the owner password (in this case lucio).
  So a "not-editable" file actually is a "file editable by the owner only".

• If in the qpdf invocation I'd specified a non-blank user password before the owner password, the resulting file would require the user password (here lucio1) to be viewed and the owner password (here lucio2) to be edited.

• Now to be really sure that what I think is a non-editable file cannot really be edited, I need a PDF editor to open it. It should not open it at all, unless I know the user password.
  Such an editor is ... Libre Office !
  If you open a (textual, not scanned) PDF with Libre Office, it will open it in Draw. You can then modify it, but you shall not Save it (it will save as an .odg drawing), but Export it as PDF.
• The Export menu of Libre Office (which is also available when one is composing a text document within Libre Office itself) has a Security tab.
  If a file is unprotected you have to Set Passwords first (at least an owner password, which Libre Office calls permission password), and then you will be allowed to set e.g. Changes to Not permitted.
  The General tab allows to set the format to PDF/A but this is incompatible with encryption and protection.

• Here is a file PDF/A saved from Libre Office after an editing of the original start file. Of course it is unprotected.
• Here is a file protected and edited within Libre Office (the owner password is the usual lucio).
• Here is a file created directly by Libre Office, and protected by both an user (lucio1) and owner (lucio2) password.

• Summarising ...
  • a normal PDF file can be opened with Acrobat Reader, gv or Libre Office
  • a secured PDF file can be opened with Acrobat Reader, gv but requires the owner passowrd for Libre Office
  • a protected PDF file can be opened with Acrobat Reader specifying either password, is opened as blank by gv, and requires the owner password only with Libre Office (the user password opens as uneditable)
  • a PDF/A file by Libre Office can be opened as a normal file
  • a secured PDF file by Libre Office can be opened as any secured file
  • a protected PDF file by Libre Office can be opened as any protected file
  As a crazy example on can produce (with qpdf) this file which has both password blank, so it is encrypted, requests a password to be opened with Libre Office but reacts to a carriage return.

Note also that a "secured" file (not modifiable unless one knows the owner password, but with blank user password) can easily be unsecured (decrypted) with qpdf. Only a "protected" file (with two passwords) cannot be decrypted unless one supplies the owner password. But of course, while an user password is OK for a private exchange, one cannot distribute a public file with an user password. It will be just a nuisance. So a "secured" file is not really protected, it is just a psychologic effect.

Finally qpdf --qdf produces a PDF file with almost no binary information, which, knowing the PDF syntax, can be edited with a text editor. Here I edited by hand the original file.