Prioritize print-color-mode-supported over PWG when creating color mo…

[zdohnal]

…dels

See apple/cups#5722

[michaelrsweet]

Not sure about the reasons for this change, as print-color-mode doesn't actually convey color space support for a printer, just the basic mono-vs-color choice at print time. Why was this change made in cups-filters?

urf-supported is an AirPrint attribute and pwg-raster-document-type-supported is for IPP Everywhere/Wi-Fi Direct/Mopria support - seems like only having print-color-mode-supported will be the odd use case.

[zdohnal]

@michaelrsweet It was introduced due problems with at least two Canon printers (Canon ImageRunner Advance at the office and Canon MF645Cx from apple/cups#5713), where pwg-raster-document-type-supported attribute contained only srgb_8, resulting into the problem created PPD didn't contain an option to change color mode into grayscale.
I was considering creating a 'template', meaning if pwg-raster-document-type-supported does provide colored option, but does not grayscale one, just add a grayscale one automatically, but we ended up with this solution.

[michaelrsweet]

Given the description of the issue in the conversation, I think I would prefer having the code look for sgray_8, black_1, or black_8 when it sees srgb_8, and synthesize sgray_8 if there is no grayscale color space listed. The reason is that print-color-mode is effectively a rendering intent while pwg-raster-document-type-supported lists supported color spaces like AdobeRGB, sRGB, CMYK, etc. which are important for photo printing in particular.

[zdohnal]

The code is updated - I used ippContainsString() for checking for grayscale color models, and used ippGetCount() to check whether there is only one keyword. I thought it could speed up the execution if there is only one keyword.

Our Canon printer at the office got firmware update, which fixed the issue (got now srgb_8 is listed twice in pwg-document-format-supported :( , but I would let it go and report it to Canon), so I'm not able to verify whether it works when only srgb_8 is defined, but it doesn't generate a duplicate Grayscale entry at least and IMO there can be still printers in the wild with this error, so merging it makes sense.

There was an other reporter, which hit the issue with Canon, so I'll try to synch with him and see if we can verify the fix.

[michaelrsweet]

I think I'll drop the check for count == 1 but otherwise I like the new fix much better!