Add new_from_strings to create MapArrays

[viirya]

Which issue does this PR close?

Closes #1158.

Rationale for this change

What changes are included in this PR?

Are there any user-facing changes?

[alamb]

Thanks @viirya -- this is certainly looking like a much better API

I looked at arrow2 as well and it didn't look like it had anything nicer https://github.com/jorgecarleitao/arrow2/blob/f7f6186ab69c086b013d9f454c91bf3f9438436d/src/array/map/mod.rs

[alamb]

What would you think about using a values: &dyn Array here? I think that might be more natural

Also while reviewing this PR, it seems like keys and entry_offsets are connected to be matched -- I wonder if it would make more sense to have a function like

    pub fn new_keys_and_offsets<'a>(
        keys_and_offsets: impl Iterator<Item = (&'a str, usize)>,
        values: &dyn Array,
    )
...

So that the user would provide some way to provide the key names and offsets into values directly 🤔

[viirya]

Hmm, if we make it values: &dyn Array, how can we call StructArray::from or make StructArray from it? We cannot move values.

[viirya]

For new_keys_and_offsets, the keys and offsets are not 1:1 matched.

For example, in the test, the keys is vec!["a", "b", "c", "d", "e", "f", "g", "h"]; and the offsets are [0, 3, 6, 8]. So it represents [[a, b, c], [d, e, f], [g, h]].

Otherwise, do you mean ("a", 0), ("b", 0), ("c", 0), ("d", 3), ("e", 3), ("f", 3), ("g", 6), ("h", 6) as keys_and_offsets?

[alamb]

Hmm, if we make it values: &dyn Array, how can we call StructArray::from or make StructArray from it? We cannot move values.

I guess I was thinking we could use array.data() to get the ArrayData needed to construct the StructArray

For example, in the test, the keys is vec!["a", "b", "c", "d", "e", "f", "g", "h"]; and the offsets are [0, 3, 6, 8]. So it represents [[a, b, c], [d, e, f], [g, h]].

Ah, sorry, I missed that point.

[viirya]

I guess I was thinking we could use array.data() to get the ArrayData needed to construct the StructArray

Got it. data() returns &ArrayData, so we need a clone here, but I think it should be fine as it won't copy array data.

[codecov-commenter]

Codecov Report

Merging #1507 (a3922af) into master (c5442cf) will increase coverage by 0.07%.
The diff coverage is 95.65%.

@@            Coverage Diff             @@
##           master    #1507      +/-   ##
==========================================
+ Coverage   82.68%   82.76%   +0.07%     
==========================================
  Files         188      190       +2     
  Lines       54361    54748     +387     
==========================================
+ Hits        44951    45313     +362     
- Misses       9410     9435      +25     

Impacted Files	Coverage Δ
arrow/src/array/array_map.rs	84.89% <95.65%> (+3.38%)	⬆️
arrow/src/array/equal/utils.rs	74.45% <0.00%> (-3.24%)	⬇️
parquet_derive/src/parquet_field.rs	65.75% <0.00%> (-0.46%)	⬇️
arrow/src/datatypes/field.rs	53.79% <0.00%> (-0.31%)	⬇️
arrow/src/array/transform/mod.rs	86.46% <0.00%> (-0.12%)	⬇️
arrow/src/array/builder.rs	86.68% <0.00%> (-0.05%)	⬇️
arrow/src/compute/kernels/filter.rs	88.00% <0.00%> (ø)
arrow/src/compute/kernels/length.rs	100.00% <0.00%> (ø)
arrow/src/ffi_stream.rs	79.03% <0.00%> (ø)
arrow/src/array/equal/union.rs	100.00% <0.00%> (ø)
... and 9 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update c5442cf...a3922af. Read the comment docs.

[alamb]

Thanks @viirya

[viirya]

Thanks @alamb !