Core: Request/Response models and parsers for REST Scan Planning

[singhpk234]

About the change

Solves #13005

PR for resuming the work for scan planning (previous pr : #11180), presently since the client is not ready it can't be used even if IRC Server (for ex Apache Polaris ) supports it. Scan planning unblocks many use-cases such as inter-op.

This is Part 1 of changes #13004 (comment) have rest 2 in my local will send them over part by part !

Parsers :
[1] ContentFileParser for Delete and DataFile is re-used, there is one API change though that now the fromJson and toJson require specsById, as unless the file is parsed we don;t know which spec it belongs to in rest case
[2] TableScanResponseParser is introduced which is a stateful parser, introduced via #13191

[3] Request (and its corresponding models)
[a] PlanTableScanRequestParser : parser for https://github.com/apache/iceberg/blob/main/open-api/rest-catalog-open-api.yaml#L4370
[b] FetchScanTasksRequestParser : parser for https://github.com/apache/iceberg/blob/main/open-api/rest-catalog-open-api.yaml#L4434
4 Response (and its corresponding models)

[a] FetchPlanningResultResponseParser : parser for https://github.com/apache/iceberg/blob/main/open-api/rest-catalog-open-api.yaml#L4700
[b] FetchScanTasksResponseParser : parser for https://github.com/apache/iceberg/blob/main/open-api/rest-catalog-open-api.yaml#L4707
[c] PlanTableScanResponseParser : parser for https://github.com/apache/iceberg/blob/main/open-api/rest-catalog-open-api.yaml#L4693.

Acknowledgement

I connected with the author i.e @rahil-c to see if he plans to resume it, apparently he doesn't have cycles and was kind enough to allow me to push this to the end.

I just rebased the pr, on the current main and go break the pr into smaller chunks while addressing the pending feedback from review, I have something working locally, will publish them soon

(SideNote : I was one of the reviewers to the PR too, will be interesting to see the other side 😌)

Co-author: @rahil-c

[singhpk234]

Breaking this change in to 3 logical changes, have all 3 working locally, will send these as we go !

• Request / Response Models, Parsers
• Plumbing (RestTable /RestScan) to core module
• Spark Integ

Have couple of things to propose to spec as well, will do it in the Plumbing and Spark Integ phase of the PR

[amogh-jahagirdar]

Thanks @singhpk234! I think this is close, just one thing that I think should not be public

[amogh-jahagirdar]

Thanks @singhpk234! Minor style comment but overall looks good to me. I will hold for a bit in case @danielcweeks or any others have any remaining comments.

[amogh-jahagirdar]

Minor followup: Just noticed this but in the new structure we should probably have a clearer message rather than a potential NPE in case the file somehow references a spec ID that doesn't exist for some reason. It's unlikely but I think it's generally better to have clearer error messages so that if it does happen a user knows what to check rather than having to go deeper in the implementation. We can also probably extract specsById.get(specId) into a separate variable since I see it's called three times.

[danielcweeks]

Minor comments, but +1. Thanks @singhpk234 !

[amogh-jahagirdar]

Thanks @singhpk234 @rahil-c for the implementation and patience in following through on this. Thanks @danielcweeks for the review!

[rahil-c]

Thanks @singhpk234 for pushing this thru, happy to see it finally get merged!
Thanks @amogh-jahagirdar reviewing this for such a long time!