[GitHub] [arrow] edponce commented on a change in pull request #10317: ARROW-12713 [C++] String reverse kernel

[GitBox <gi...@apache.org>]

edponce commented on a change in pull request #10317:
URL: https://github.com/apache/arrow/pull/10317#discussion_r632626972



##########
File path: cpp/src/arrow/compute/kernels/scalar_string.cc
##########
@@ -266,6 +266,52 @@ void EnsureLookupTablesFilled() {}
 
 #endif  // ARROW_WITH_UTF8PROC
 
+template <typename Type>
+struct AsciiReverse : StringTransform<Type, AsciiReverse<Type>> {
+  using Base = StringTransform<Type, AsciiReverse<Type>>;
+  using offset_type = typename Base::offset_type;
+
+  bool Transform(const uint8_t* input, offset_type input_string_ncodeunits,
+                 uint8_t* output, offset_type* output_written) {
+    uint8_t utf8_char_found = 0;
+    for (offset_type i = 0; i < input_string_ncodeunits; i++) {
+      // if a utf8 char is found, report to utf8_char_found
+      utf8_char_found |= input[i] & 0x80;

Review comment:
       Consider the case where you have a long string, would it be better to have the check inside the loop so that it stops as soon as an invalid character is found?  Perhaps using `ARROW_PREDICT_FALSE`. I understand a conditional check in a loop may break any compiler vectorization, so I would suggest verifying if this is the case.




-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org