Add support for StringView and BinaryView statistics in StatisticsConverter

[alamb]

Part of #6163

Is your feature request related to a problem or challenge? Please describe what you are trying to do.
@efredine recently added support for extracting statistics from parquet files as arrays in #6046 using StatisticsConverter

During development we have also added support for StringViewArray and BinaryViewArray in #5374

Currently there is no way to read StringViewArray and BinaryViewArray statistics and it actually panics if you try to read data page level statistics as I found on apache/datafusion#11723

not implemented
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
External error: query failed: DataFusion error: Join Error
caused by

Describe the solution you'd like

1. Implement the ability to extract parquet statistics as StringView and BinaryView
2. Remove the panic caused by unimplemented! at

   arrow-rs/parquet/src/arrow/arrow_reader/statistics.rs

   Line 946 in 2905ce6

   _ => unimplemented!()

The code is in https://github.com/apache/arrow-rs/blob/master/parquet/src/arrow/arrow_reader/statistics.rs

Describe alternatives you've considered

You can avoid the panic by following the model of this:

arrow-rs/parquet/src/arrow/arrow_reader/statistics.rs

Lines 465 to 467 in 2905ce6

	let len = $iterator.count();
	// don't know how to extract statistics, so return a null array
	Ok(new_null_array($data_type, len))

Then, you can probably write a test followig the model of utf8 and binary

arrow-rs/parquet/src/arrow/arrow_reader/statistics.rs

Lines 1897 to 1917 in 2905ce6

	fn roundtrip_utf8() {
	Test {
	input: utf8_array([
	// row group 1
	Some("A"),
	None,
	Some("Q"),
	// row group 2
	Some("ZZ"),
	Some("AA"),
	None,
	// row group 3
	None,
	None,
	None,
	]),
	expected_min: utf8_array([Some("A"), Some("AA"), None]),
	expected_max: utf8_array([Some("Q"), Some("ZZ"), None]),
	}
	.run()
	}

arrow-rs/parquet/src/arrow/arrow_reader/statistics.rs

Lines 1956 to 1984 in 2905ce6

	fn roundtrip_binary() {
	Test {
	input: Arc::new(BinaryArray::from_opt_vec(vec![
	// row group 1
	Some(b"A"),
	None,
	Some(b"Q"),
	// row group 2
	Some(b"ZZ"),
	Some(b"AA"),
	None,
	// row group 3
	None,
	None,
	None,
	])),
	expected_min: Arc::new(BinaryArray::from_opt_vec(vec![
	Some(b"A"),
	Some(b"AA"),
	None,
	])),
	expected_max: Arc::new(BinaryArray::from_opt_vec(vec![
	Some(b"Q"),
	Some(b"ZZ"),
	None,
	])),
	}
	.run()
	}

And then implement the missing pieces of code (use StringViewBuilder / BinaryViewBuilder instead of StringBuilder / BinaryBuilder)

I have a hacky version in apache/datafusion#11753 that looks something like

            DataType::Utf8View => {
                let iterator = [<$stat_type_prefix ByteArrayStatsIterator>]::new($iterator);
                let mut builder = StringViewBuilder::new();
                for x in iterator {
                    let Some(x) = x else {
                        builder.append_null(); // no statistics value
                        continue;
                    };

                    let Ok(x) = std::str::from_utf8(x) else {
                        log::debug!("Utf8 statistics is a non-UTF8 value, ignoring it.");
                        builder.append_null();
                        continue;
                    };

                    builder.append_value(x);
                }
                Ok(Arc::new(builder.finish()))
            },

Additional context

[alamb]

I think this would be a good first issue as it has a detailed description and existing patterns to follow

[Kev1n8]

take

[yashathwani]

Can i work on this issue?

[alamb]

Can i work on this issue?

Hi @yashathwani -- it looks like @Kev1n8 has already created a PR: #6181 -- perhaps you can help review it?

[alamb]

label_issue.py automatically added labels {'parquet'} from #6187