[5.x]: Adding the ability to save geotiffs with a palette

[WeatherGod]

Description of Changes

Add support to save out a geotiff with supplied colortable

PR Checklist

• Indicate the version associated with this PR in the Title
  (e.g. "[5.x]: This is my PR title")
• Link to any issues that the PR addresses
• Add labels, especially if the PR should be ported to other versions
  (these labels start with "port: ")
• Open as a draft PR
  until ready for review
• Make sure GitHub tests pass
• Mark PR as "Ready for Review"

[CLAassistant]

All committers have signed the CLA.

[WeatherGod]

This is my first significant chunk of Java code, so don't assume I did any of this correctly.

The reason for this submission and its design is that I plan to also make a submission to TDS's WCS service to add a new Geotiff_Palette format that would then respect a dataset's "flag_values" and "flag_colors" attributes, much like how ncWMS does right now. We could also respect supplied styling options this way, too.

[WeatherGod]

And, this is partly in response to #1052

[WeatherGod]

I am just about done with a set of revisions that I think makes things more sane, but I needed to write a coerceInt() and a coerceByte() private methods to coerce an arbitrary Array into ArrayInt or ArrayByte copies, respectively. I feel like that is some sort of NiH syndrome. Is there already a "correct" way to do what in numpy would be like an a.astype(int)?

[tdrwenski]

Hi @WeatherGod, are you still working on this PR?

[WeatherGod]

Yes in the sense that it is a feature for a client of mine, so I've been maintaining it for them. There is a bug of sorts, but I think it is actually pre-existing. I would appreciate feedback on what I have so far and if there is anything else you all would like added to it.

…

On Thu, Apr 11, 2024 at 2:53 PM Tara Drwenski ***@***.***> wrote: Hi @WeatherGod <https://github.com/WeatherGod>, are you still working on this PR? — Reply to this email directly, view it on GitHub <#1066 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AACHF6CRDFEFHFNX3B5J32LY43L3DAVCNFSM57JF5HWKU5DIOJSWCZC7NNSXTN2JONZXKZKDN5WW2ZLOOQ5TEMBVGAZTCNZQGEZQ> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[tdrwenski]

Okay, thanks! Will try to have a look at your comments about the possible bug. To see the style issues you can run ./gradlew spotlessCheck or to automatically fix them you can run ./gradlew spotlessApply

[WeatherGod]

Crazy thought,, but what if we allowed color tables to work with greyscale==true? Right now, as coded, greyscale mode scales the given data to the range of 1-255 and encodes missing data as 0, effectively unsigned bytes, while the "colortable" mode doesn't do any scaling at all, and only works if the data type is unsigned bytes. But, it is getting tedious to document and code checks that the colortable can only be used when greyscale is false. Why not allow color tables to be used for greyscale mode?

It might be easier to explain the various features available as "greyscale normalizes to UBYTE", "color table works only for UBYTEs", and "Non-greyscale FLOATs encodes missing data as data minimum minus one".

Thoughts?

[WeatherGod]

Now that I've actually implemented that idea, I really like it. I'm running with it. It also addresses several other questions I had in my mind about possible use-cases and whether or not the setColorTable() was an anti-pattern.

[WeatherGod]

Test failures seem to be unrelated?

[tdrwenski]

Test failures seem to be unrelated?

Hi @WeatherGod, I think I know the issue there, should have a fix shortly!

[WeatherGod]

Is there anyplace where I should add some narrative docs explaining the changes to the geotiff module?