Fix out-of-bounds read in Gandiva soundex_utf8

cpp/src/gandiva/precompiled/string_ops.cc

@@ -3125,9 +3125,14 @@ const char* soundex_utf8(gdv_int64 context, const char* in, gdv_int32 in_len,
 
   int start_idx = 0;
   for (int i = 0; i < in_len; ++i) {
-    if (isalpha(static_cast<unsigned char>(in[i])) > 0) {
+    unsigned char ch = static_cast<unsigned char>(in[i]);
+    if ((ch >= 'A' && ch <= 'Z') || (ch >= 'a' && ch <= 'z')) {
       // Retain the first letter
-      ret[0] = toupper(static_cast<unsigned char>(in[i]));
+      if (ch >= 'a') {
+        ret[0] = static_cast<char>(ch - 'a' + 'A');
+      } else {
+        ret[0] = static_cast<char>(ch);
+      }
       start_idx = i + 1;
       break;
     }
@@ -3143,12 +3148,17 @@ const char* soundex_utf8(gdv_int64 context, const char* in, gdv_int32 in_len,
   soundex[0] = '\0';
   // Replace consonants with digits and special letters with 0
   for (int i = start_idx; i < in_len; i++) {
-    if (isalpha(static_cast<unsigned char>(in[i])) > 0) {
-      c = toupper(static_cast<unsigned char>(in[i])) - 65;
-      if (mappings[c] != soundex[si - 1]) {
-        soundex[si] = mappings[c];
-        si++;
-      }
+    unsigned char ch = static_cast<unsigned char>(in[i]);
+    if (ch >= 'A' && ch <= 'Z') {
+      c = ch - 'A';
+    } else if (ch >= 'a' && ch <= 'z') {
+      c = ch - 'a';
+    } else {
+      continue;
+    }
+    if (mappings[c] != soundex[si - 1]) {
+      soundex[si] = mappings[c];
+      si++;
     }
   }
 

cpp/src/gandiva/precompiled/string_ops_test.cc

@@ -18,10 +18,12 @@
 #include <gmock/gmock.h>
 #include <gtest/gtest.h>
 
+#include <cctype>
 #include <cstring>
 #include <limits>
 #include <memory>
 
+#include "arrow/testing/gtest_util.h"
 #include "gandiva/execution_context.h"
 #include "gandiva/precompiled/types.h"
 
@@ -2929,6 +2931,38 @@ TEST(TestStringOps, TestFromHex) {
   EXPECT_EQ(out_valid, false);
 }
 
+static void CheckSoundexNonAsciiBytes() {
+  for (int byte = 128; byte <= 255; ++byte) {
+    const std::string non_ascii(1, static_cast<char>(byte));
+    for (const auto& input : {non_ascii + "Robert", "R" + non_ascii + "obert"}) {
+      gandiva::ExecutionContext ctx;
+      auto ctx_ptr = reinterpret_cast<int64_t>(&ctx);
+      int32_t out_len = -1;
+      bool validity = false;
+      const auto* out =
+          soundex_utf8(ctx_ptr, input.data(), static_cast<int32_t>(input.size()), true,
+                       &validity, &out_len);
+      ASSERT_TRUE(validity);
+      ASSERT_EQ(out_len, 4);
+      EXPECT_EQ(std::string(out, out_len), "R163");
+      EXPECT_FALSE(ctx.has_error());
+    }
+  }
+}
+
+TEST(TestStringOps, TestSoundexNonAsciiBytes) {
+  arrow::LocaleGuard locale_guard("C");
+  CheckSoundexNonAsciiBytes();
+}
+
+TEST(TestStringOps, TestSoundexNonAsciiLocale) {
+  arrow::LocaleGuard locale_guard("en_US.ISO-8859-1");
+  if (std::isalpha(static_cast<unsigned char>(0xe9)) == 0) {
+    GTEST_SKIP() << "Requires an ISO-8859-1 locale classifying high bytes as letters";
+  }
+  CheckSoundexNonAsciiBytes();
+}
+
 TEST(TestStringOps, TestSoundex) {
   gandiva::ExecutionContext ctx;
   auto ctx_ptr = reinterpret_cast<int64_t>(&ctx);