PDF is printed incorrectly with 'number-up' and 'page-ranges'

[michaelrsweet]

Version: 1.2.3
CUPS.org User: drakamakafon

PDF is printed incorrectly when 'number-up' and 'page-ranges' options are specified simultaneously.

Tested OS:
1.1.23 (SuSE 9.3, package name: cups-1.1.23-7.1)
1.1.23 (Mandriva 2006, package name: cups-1.1.23-17mdk)
1.2.3 (Mandriva 2007, package name: cups-1.2.3-5mdv2007.0)
1.2.4 (Ubuntu 6.10, package name: 1.2.4-2ubuntu3)

After investigation I found that 'pstops' from different version of cups produces different output when 'number-up' and 'page-ranges' are specified simultaneously.

test7.ps - postscript file (7 pages)

/usr/lib/cups/filter/pstops 12345 user title 1 "-o page-ranges=1-3 -o number-up=2" test.ps > pstops-out

1.1.23 (SuSE 9.3, Mandriva 2006)
######### #########
#1 # #3 #

# #

######### #########
#2 # #blank #

# #

######### #########
Looks like operations are performed in the following order:

1. Select pages according to 'page-ranges' value
2. Layout selected pages according to 'number-up' value

1.2.3 (Mandriva 2007) 1.2.4 (Ubuntu 6.10)
######### ######### #########
#1 # #3 # #5 #

# # # #

######### ######### #########
#2 # #4 # #6 #

# # # #

######### ######### #########
Looks like the same operations are performed in reverse order, i.e:

1. Layout ALL pages according to 'number-up' value
2. Select pages from new page sequence according to 'page-ranges' value
   The same behaviour in later versions (v1.3 Fedora 8). Is it the feature of newer 'pstops' or bug?.
   ASAIK 'pdftops' doesn't take into account 'number-up' option so after http://cups.org/str.php?L273 is resolved the 'pdftops|pstops' pipe may produce strange output.

Example:

lpr "-o page-ranges=1,3-4,6 -o number-up=2" test.pdf

1. 'pdftops' output:

   ### ######### ######### ######### #########

   # # # # # # # # # # #

   # # # # # # # # # # #

   # # # # # # # # # # #

   # # # # # # # # # # #

   Missing links and out-of-date systems overviews on documentation page #1 # #blank # connect from cups client to bsd lpd server fails : Remote host did not accept control file (100) #3 # CUPS displays PPD options without "*OpenUI" ... "*CloseUI" #4 # #blank # ESP GhostScript: "psmono" and "psgray" produce shrinked pages #6 # #blank

   ### ######### ######### ######### #########

2. 'pstops' output:

2.1. Layout ALL pages according to 'number-up' value
######### ######### ######### #########
#1 # #3 # #blank # #blank #

# # # # # #

######### ######### ######### #########
#blank # #4 # #6 # #blank #

# # # # # #

######### ######### ######### #########

2.2. Select pages from new page sequence according to 'page-ranges' value (1,3-4,6)
######### ######### #########
#1 # #blank # #blank #

# # # #

######### ######### #########
#blank # #6 # #blank #

# # # #

######### ######### #########
that are 1,3-4 from the previous step (page 6 became out of page range after n-up).
So totally 3 pages are printed (3rd page is empty). It looks pretty strange in respect that page-ranges=1,3-4,6 was requested. Either 'pstops' should process options in the v1.1 order or 'pdftops' should understand 'number-up' option.

[michaelrsweet]

CUPS.org User: mike

The 1.1.x (and 1.0.x) behavior was incorrect. According to the IPP specification, number-up is applied first, and page-ranges refers to the imposed pages. Similarly, page-set is applied after imposition.

[michaelrsweet]

CUPS.org User: mike

Reopening - the issue is that STR #273 changed the PDF filter to omit pages that were not needed, but that filtering isn't accounting for number-up processing.

[michaelrsweet]

CUPS.org User: mike

Fixed in Subversion repository.