chore: formalize the default map field names to match default arrow spec

[rluvaton]

Which issue does this PR close?

N/A

Rationale for this change

In some places we write keys and values and in some cases key and value (without the s)

so matching the names to match the default arrow spec names

this is also important as this is a source of common pitfalls when comparing data types from schema and actual data (where actual data might coming from another arrow impl like java which uses key and value and not keys and values - https://github.com/apache/arrow-java/blob/a9f0086090e649b83c89ea864c82ed7fcc84ab2f/vector/src/main/java/org/apache/arrow/vector/complex/MapVector.java#L47-L49)

This is similar to what was done for list:

• Formalize the default nested list field name to item #6785

What changes are included in this PR?

add constants for entries, key and value field name in Field, replace manually every place, left out places that test different names specifically or made it more explicit

I did it manually since I don't trust LLM that will do it correctly and might change tests that check that case specifically and did not want to use a script to change that

Are these changes tested?

Existing tests

Are there any user-facing changes?

the default MapFieldNames is key and not keys, value and not values

[yonipeleg33]

Nice! Some cleanups of hard-coded strings, otherwise LGTM

[Jefffrey]

makes sense, thanks

[rluvaton]

@Jefffrey can you please merge?

[Jefffrey]

thanks @rluvaton

[alamb]

The actual Arrow spec uses key and value as well for the field ames

https://github.com/apache/arrow/blob/35c5ffd12173284406e4a2c86405415444e596d7/format/Schema.fbs#L120

/// A Map is a logical nested type that is represented as
///
/// List<entries: Struct<key: K, value: V>>

[alamb]

I hit an error due to this change when testing the upgrade to DataFusion:

• WIP: Update to arrow/parquet 59.2.0 datafusion#24030

I was working on

• Release arrow-rs / parquet Minor version 59.2.0 (July 2026) #9879

Here are the errors

• https://github.com/alamb/datafusion/actions/runs/30643722834/job/91201627268
• https://github.com/alamb/datafusion/actions/runs/30643722834/job/91201627088

thread 'tokio-rt-worker' (41727) panicked at /usr/local/cargo/git/checkouts/arrow-rs-583cca34693b79b8/1359cb9/arrow-array/src/array/struct_array.rs:91:46:
called `Result::unwrap()` on an `Err` value: InvalidArgumentError("Incorrect datatype for StructArray field \"metadata\", expected Map(\"entries\": non-null Struct(\"keys\": non-null Utf8, \"values\": Utf8), unsorted) got Map(\"entries\": non-null Struct(\"key\": non-null Utf8, \"value\": Utf8), unsorted)")
stack backtrace:

I think this PR is a good one to align to the spec, but I think we should probably treat it as a "breaking API change" and hold it for the next breaking release to minimize downstream churn on a minor release

I will prepare a PR

[alamb]

Here is my proposed revert PR:

• Revert "chore: formalize the default map field names to match default arrow spec (#10297)" #10506