Ghostcript used to find the number of pages on output could fail

Hello,

I have experienced ghostscript failure to produce expected output for my specific input document.

Error: /rangecheck in --getinterval--                                                                         
Operand stack:                                                                                                
   ClippedLabel   ()   0   -1                                                                                 
Execution stack:                                                                                              
   %interp_exit   .runexec2   --nostringval--   --nostringval--   --nostringval--   2   %stopped_push   --nost
ringval--   --nostringval--   --nostringval--   false   1   %stopped_push   1933   1   3   %oparray_pop   1932
   1   3   %oparray_pop   1917   1   3   %oparray_pop   1787   1   3   %oparray_pop   --nostringval--   %error
exec_pop   .runexec2   --nostringval--   --nostringval--   --nostringval--   2   %stopped_push   --nostringval
--   --nostringval--   --nostringval--   --nostringval--   --nostringval--   --nostringval--   --nostringval--
   %loop_continue                                                                                             
Dictionary stack:                                                                                             
   --dict:745/1123(ro)(G)--   --dict:0/20(G)--   --dict:88/200(L)--   --dict:34/100(L)--   --dict:31/100(L)-- 
Current allocation mode is local                                                                              
Current file position is 15757                                                                                GPL Ghostscript 10.05.1: Unrecoverable error, exit code 1

It seems it is unable to generate output which cfFilterGhostscript() uses to cound the number of pages in the input document.

cupsfilters/ghostscript.c:
...
1032       // Ghostscript runs too long on files converted from djvu files                                    
1033       // Using -dDEVICEWIDTHPOINTS -dDEVICEHEIGHTPOINTS params solves the                                
1034       // problem                                                                                         
1035       snprintf(gscommand, 65536, "%s -q -dNOPAUSE -dBATCH -sDEVICE=bbox -dDEVICEWIDTHPOINTS=1 -dDEVICEHEI     GHTPOINTS=1 %s 2>&1 | grep -c HiResBoundingBox",                                                         
1036                CUPS_GHOSTSCRIPT, filename);
...

I have found well working alternative to this potentially problematic approach. It uses inkcov gs device. See the attached patch and consider merging it.

Thank you,
m.

gs.patch

Start in cupsfilters/ghostscript.c around lines 1032-1036 and review how the Ghostscript bbox command is used to count pages. Compare the attached inkcov-based patch with the current approach, then verify that the failing input no longer triggers the rangecheck error and produces the expected page count.