Some PDF forms don't save entered information

[Robin Munn]

Binary package hint: evince

When trying to fill out certain PDF forms, such as the PDF tax forms available form irs.gov, with evince, I noticed a problem. Some forms could be filled out and saved just fine, and when those forms were reloaded the entered form data was still there. But other forms didn't work properly: filling in the form and saving it appeared to work, but when I loaded the form again the entered data was blank.

One form that worked properly was the 1040 form, http://www.irs.gov/pub/irs-pdf/f1040.pdf
One form that failed to work was the 1040 Schedule SE form, http://www.irs.gov/pub/irs-pdf/f1040sse.pdf

Also, when clicking on the rows of the SE that I had previously entered data into (and that now appeared blank), I noticed that evince no longer gave me an editing cursor for those rows. It was as if the form's text-entry box for those rows (and ONLY those rows) had been replaced by a blank, white, non-interactive rectangle. All other rows on the form (that I had not previously entered data into) were still interactive, but the rows that *should* have contained data were blank and unresponsive.

To reproduce the problem:

1) Download the 1040 and 1040 SE forms from the above links.
2) Open the 1040 in evince. Enter some data. Save the form. Close evince.
3) Re-open the 1040. Verify that the data entered previously is still there, and can be edited.
4) Open the 1040 SE in evince. Enter some data on rows 1a and 2 of the form, leaving row 1b blank. Save the form. Close evince.
5) Re-open the 1040 SE. Notice that rows 1a and 2 are blank and no longer respond to mouse clicks or keypresses, but row 1b (and other rows) can still be used.

This happens on Ubuntu Karmic (9.10), using version 2.28.1-0ubuntu1.2 of the evince package (from karmic-updates).

[Robin Munn (rmunn)]

Note that this does not appear to be the same bug as #492959 or #477644: the saved forms are not 0-byte files as in bug #492959, and the PDFs are not encrypted as in bug #477644.

[Robin Munn (rmunn)]

The URLs in the bug description link to the 2009 U.S. tax forms. In case the bug is dependent on a particular quirk of the forms that will go away in 2010 (when those URLs no longer point to the 2009 forms), I've attached year-2009 copies of these two forms to this bug report.

[madbiologist (me-again)]

Confirmed on Lucid 10.04 alpha 2 with poppler updated to the latest Ubuntu package version.

Uname: Linux 2.6.32-10-generic i686
Packages: evince 2.29.5-0ubuntu1
                 poppler 0.12.3-0ubuntu1

[John T. Anderson (john-skylions)]

This bug also applies to the following IRS form:

http://www.irs.gov/pub/irs-pdf/f1040x.pdf
(form attached)

From the properties window in evince for this form:

Title: Form 1040X (Rev. January 2010)
Producer: Adobe LiveCycle Designer ES 8.2
Creator: Adobe LiveCycle Designer ES 8.2
Created: Tue 28 Jul 2009 12:18:08 PM PDT
Modified: Mon 25 Jan 2010 03:54:07 PM PST
Format: PDF-1.7

evince 2.28.1 on karmic (2.6.31-20-generic #58-Ubuntu SMP Fri Mar 12 04:38:19 UTC 2010 x86_64 GNU/Linux)

It looks like the problem persists into 2010.

I don't have Acrobat so I can't test if the Adobe product works correctly.

[Roland Dreier (roland.dreier)]

FWIW, I notice that evince prints

    Error: Reference in Fields array to an invalid or non existant object

a few times on the console when trying to save a form where it doesn't work properly.

[Martin Mai (mrkanister-deactivatedaccount-deactivatedaccount)]

The remote bug points to the Debian bug tracker, not Gnome.

[Tristan Schmelcher (tschmelcher)]

I have observed the exact same thing here on Karmic amd64, also while preparing my taxes. :(

Another form that exhibits the saving bug is the Schedule D form: http://www.irs.gov/pub/irs-pdf/f1040sd.pdf

[Tristan Schmelcher (tschmelcher)]

FYI, for anyone in need of a workaround, <email address hidden> describes a workaround in the Debian bug tracker entry:

"Running the original pdf through either pdfopt or pdfedit's delinearize
function creates a form that can be correctly saved, at least in this
case."

I can attest that this works. Just open a terminal and type "pdfopt <original pdf file> <new file>", then open <new file> in Evince and make all your edits with that version of the file. Saving that file will work.

Interestingly, the pdfopt workaround will NOT work on a file that has _already_ been saved by Evince. Instead, it outputs a fatal error:

$ pdfopt f1040sd-borked.pdf f1040sd-borked-out.pdf
Error: /undefined in --get--
Operand stack:
   --nostringval-- FirstPageXrefLength --nostringval-- --dict:9/9(ro)(G)-- filetype
Execution stack:
   %interp_exit .runexec2 --nostringval-- --nostringval-- --nostringval-- 2 %stopped_push --nostringval-- --nostringval-- --nostringval-- false 1 %stopped_push 1862 1 3 %oparray_pop 1861 1 3 %oparray_pop 1845 1 3 %oparray_pop 1739 1 3 %oparray_pop --nostringval-- %errorexec_pop .runexec2 --nostringval-- --nostringval-- --nostringval-- 2 %stopped_push --nostringval-- --nostringval-- --nostringval-- --nostringval-- --nostringval-- --nostringval-- --nostringval-- --nostringval-- --dict:13/14(L)-- --nostringval-- 12 %dict_continue --nostringval-- --nostringval--
Dictionary stack:
   --dict:1150/1684(ro)(G)-- --dict:1/20(G)-- --dict:109/200(L)-- --dict:76/200(G)-- --dict:285/300(ro)(G)-- --dict:18/25(L)-- --dict:46/80(L)--
Current allocation mode is local
Last OS error: 2
Current file position is 36516
GPL Ghostscript 8.70: Unrecoverable error, exit code 1
$

So evidently Evince is actually saving corrupt data.

[Donald Henry (donaldhenry6)]

Donald Henry wrote on 2010-04-14

My IRS Form 1040A appeared as a print preview of the actual document to send to the IRS. But neither printing it nor saving it on my computer and printer cannot be done. I already know that printing to paper and to file cannot be done from previous experience. So, once this problem report is complete and sent the developers, I will attempt to troubleshoot that issue using a technique sent to me by a Ubuntu volunteer (engineer?). External to this issue is that the IRS electronic forms program known as FreeFile produced by a contractor has yet to reveal to me how to e-file my completed tax return to the IRS. It seems to me that completing a simple tax return by pen on a paper IRS form and mailing it via USPS is a lot easier than doing it via computer and electronic filing over telecommunications lines. A lesson here for all of us is that using a computer to prepare and print/save/file/send anything one time takes much longer to learn and perform than doing it by pre-computer hand method. Tax return programs, for example, are only useful for persons who have no arithmetic ability and/or have a need for the program to interpret the tax code and fill in a space on a form for an uneducated taxpayer. That is not an issue with Linux Ubuntu, I will admit, but I am spending a lot of time using this platform after recently have been persuaded that Linux Ubuntu is better that Windows XP. ( as wll as learning how to complete a simple form according to a government contractor's computerized program.)

[Catalin Patulea (cpatulea)]

I was running into this same bug with "GNOME Document Viewer 2.30.3". The pdfopt workaround worked for me.

[madbiologist (me-again)]

This is fixed in Ubuntu 12.04 "Precise Pangolin" beta 1. The files attached to comments #2, #3 and #5 all work properly.

Packages: evince 3.3.90-0ubuntu2
          libpoppler19 0.18.4-1ubuntu1
          libcairo2 1.10.2-6.1ubuntu2

[Patrick Burger (iceman6602)]

I still experience this bug with ubuntu 12.04 regarding this file. and others like it. although it I have not tried it with the IRS tax forms. I can upload the form to a site of your choosing if you wish. The workaround also did not work. It accepts the data, but save the data but the data is not viewable unless it is reentered. Even saving it again did not work.

[Patrick Burger (iceman6602)]

This is one of the files. sorry for the content of the file, but this is sanitized. If you enter information in ANYWHERE but the ability score modifier column and any other place on that sheet where it factors in those numbers those boxes are left blank unless you click on them to see the contents. reapplying those numbers will result in a viewable sheet again, until it is once again saved.

Columns affected:
Ability score modifiers, Ability modifiers and Skill modifiers (In the Skills section) Dex modifier and AC in the AC section.

placing numbers here will result in blank columns where the modifiers should go and blank total modifiers.