perf: improve performance of translate by reusing buffers (#19533)

## Which issue does this PR close?

<!--
We generally require a GitHub issue to be filed for all bug fixes and
enhancements and this helps us generate change logs for our releases.
You can link an issue to this PR using the GitHub syntax. For example
`Closes #123` indicates that this PR will close issue #123.
-->

- Closes #.

## Rationale for this change

<!--
Why are you proposing this change? If this is already explained clearly
in the issue then this section is not needed.
Explaining clearly why changes are proposed helps reviewers understand
your changes and offer better suggestions for fixes.
-->

## What changes are included in this PR?

<!--
There is no need to duplicate the description in the issue here but it
is sometimes worth providing a summary of the individual changes in this
PR.
-->

Optimize translate function by reusing HashMap and Vec buffers across
all rows instead of allocating new ones for each row.

Changes:
- Moved HashMap and Vec allocations outside the main loop
- Clear and reuse buffers for each row instead of reallocating
- Use explicit loops instead of chained iterators for better control
- Added benchmark to measure performance improvements

Optimization:
- Before: Allocated HashMap + 4 Vecs for every row
- After: Single set of reusable buffers cleared for each row

Benchmark Results:
- size=1024, str_len=8:  234.6 µs → 147.9 µs (37% faster)
- size=1024, str_len=32: 628.6 µs → 394.2 µs (37% faster)
- size=4096, str_len=8:  964.4 µs → 575.2 µs (40% faster)
- size=4096, str_len=32: 2.54 ms → 1.56 ms (39% faster)

The optimization shows exceptional 37-40% performance improvements
across all test cases. The HashMap reuse is particularly impactful since
HashMap creation/destruction has significant overhead with hashing,
bucket allocation, and internal bookkeeping. Combined with eliminating 4
Vec allocations per row, this becomes the most significant optimization
in this series.

## Are these changes tested?

<!--
We typically require tests for all PRs in order to:
1. Prevent the code from being accidentally broken by subsequent changes
2. Serve as another way to document the expected behavior of the code

If tests are not included in your PR, please explain why (for example,
are they covered by existing tests)?
-->

## Are there any user-facing changes?

<!--
If there are user-facing changes then we may require documentation to be
updated before approving the PR.
-->

<!--
If there are any breaking changes to public APIs, please add the `api
change` label.
-->

Author: viirya

datafusion/functions/Cargo.toml

@@ -270,6 +270,11 @@ harness = false
 name = "ends_with"
 required-features = ["string_expressions"]
 
+[[bench]]
+harness = false
+name = "translate"
+required-features = ["unicode_expressions"]
+
 [[bench]]
 harness = false
 name = "levenshtein"

datafusion/functions/benches/translate.rs

@@ -0,0 +1,90 @@
+// Licensed to the Apache Software Foundation (ASF) under one
+// or more contributor license agreements.  See the NOTICE file
+// distributed with this work for additional information
+// regarding copyright ownership.  The ASF licenses this file
+// to you under the Apache License, Version 2.0 (the
+// "License"); you may not use this file except in compliance
+// with the License.  You may obtain a copy of the License at
+//
+//   http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing,
+// software distributed under the License is distributed on an
+// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+// KIND, either express or implied.  See the License for the
+// specific language governing permissions and limitations
+// under the License.
+
+extern crate criterion;
+
+use arrow::array::OffsetSizeTrait;
+use arrow::datatypes::{DataType, Field};
+use arrow::util::bench_util::create_string_array_with_len;
+use criterion::{Criterion, SamplingMode, criterion_group, criterion_main};
+use datafusion_common::DataFusionError;
+use datafusion_common::config::ConfigOptions;
+use datafusion_expr::{ColumnarValue, ScalarFunctionArgs};
+use datafusion_functions::unicode;
+use std::hint::black_box;
+use std::sync::Arc;
+use std::time::Duration;
+
+fn create_args<O: OffsetSizeTrait>(size: usize, str_len: usize) -> Vec<ColumnarValue> {
+    let string_array = Arc::new(create_string_array_with_len::<O>(size, 0.1, str_len));
+    // Create simple from/to strings for translation
+    let from_array = Arc::new(create_string_array_with_len::<O>(size, 0.1, 3));
+    let to_array = Arc::new(create_string_array_with_len::<O>(size, 0.1, 2));
+
+    vec![
+        ColumnarValue::Array(string_array),
+        ColumnarValue::Array(from_array),
+        ColumnarValue::Array(to_array),
+    ]
+}
+
+fn invoke_translate_with_args(
+    args: Vec<ColumnarValue>,
+    number_rows: usize,
+) -> Result<ColumnarValue, DataFusionError> {
+    let arg_fields = args
+        .iter()
+        .enumerate()
+        .map(|(idx, arg)| Field::new(format!("arg_{idx}"), arg.data_type(), true).into())
+        .collect::<Vec<_>>();
+    let config_options = Arc::new(ConfigOptions::default());
+
+    unicode::translate().invoke_with_args(ScalarFunctionArgs {
+        args,
+        arg_fields,
+        number_rows,
+        return_field: Field::new("f", DataType::Utf8, true).into(),
+        config_options: Arc::clone(&config_options),
+    })
+}
+
+fn criterion_benchmark(c: &mut Criterion) {
+    for size in [1024, 4096] {
+        let mut group = c.benchmark_group(format!("translate size={size}"));
+        group.sampling_mode(SamplingMode::Flat);
+        group.sample_size(10);
+        group.measurement_time(Duration::from_secs(10));
+
+        for str_len in [8, 32] {
+            let args = create_args::<i32>(size, str_len);
+            group.bench_function(
+                format!("translate_string [size={size}, str_len={str_len}]"),
+                |b| {
+                    b.iter(|| {
+                        let args_cloned = args.clone();
+                        black_box(invoke_translate_with_args(args_cloned, size))
+                    })
+                },
+            );
+        }
+
+        group.finish();
+    }
+}
+
+criterion_group!(benches, criterion_benchmark);
+criterion_main!(benches);

datafusion/functions/src/unicode/translate.rs

@@ -148,34 +148,48 @@ where
     let from_array_iter = ArrayIter::new(from_array);
     let to_array_iter = ArrayIter::new(to_array);
 
+    // Reusable buffers to avoid allocating for each row
+    let mut from_map: HashMap<&str, usize> = HashMap::new();
+    let mut from_graphemes: Vec<&str> = Vec::new();
+    let mut to_graphemes: Vec<&str> = Vec::new();
+    let mut string_graphemes: Vec<&str> = Vec::new();
+    let mut result_graphemes: Vec<&str> = Vec::new();
+
     let result = string_array_iter
         .zip(from_array_iter)
         .zip(to_array_iter)
         .map(|((string, from), to)| match (string, from, to) {
             (Some(string), Some(from), Some(to)) => {
-                // create a hashmap of [char, index] to change from O(n) to O(1) for from list
-                let from_map: HashMap<&str, usize> = from
-                    .graphemes(true)
-                    .collect::<Vec<&str>>()
-                    .iter()
-                    .enumerate()
-                    .map(|(index, c)| (c.to_owned(), index))
-                    .collect();
+                // Clear and reuse buffers
+                from_map.clear();
+                from_graphemes.clear();
+                to_graphemes.clear();
+                string_graphemes.clear();
+                result_graphemes.clear();
+
+                // Build from_map using reusable buffer
+                from_graphemes.extend(from.graphemes(true));
+                for (index, c) in from_graphemes.iter().enumerate() {
+                    from_map.insert(*c, index);
+                }
+
+                // Build to_graphemes
+                to_graphemes.extend(to.graphemes(true));
 
-                let to = to.graphemes(true).collect::<Vec<&str>>();
+                // Process string and build result
+                string_graphemes.extend(string.graphemes(true));
+                for c in &string_graphemes {
+                    match from_map.get(*c) {
+                        Some(n) => {
+                            if let Some(replacement) = to_graphemes.get(*n) {
+                                result_graphemes.push(*replacement);
+                            }
+                        }
+                        None => result_graphemes.push(*c),
+                    }
+                }
 
-                Some(
-                    string
-                        .graphemes(true)
-                        .collect::<Vec<&str>>()
-                        .iter()
-                        .flat_map(|c| match from_map.get(*c) {
-                            Some(n) => to.get(*n).copied(),
-                            None => Some(*c),
-                        })
-                        .collect::<Vec<&str>>()
-                        .concat(),
-                )
+                Some(result_graphemes.concat())
             }
             _ => None,
         })