From 52038a21fc8c874a765f34f19237fdcfee10ad4a Mon Sep 17 00:00:00 2001
From: metsw24-max <metsw24@gmail.com>
Date: Mon, 8 Jun 2026 14:15:41 +0530
Subject: [PATCH] cast to unsigned char before ctype calls in gandiva
 precompiled

---
 cpp/src/gandiva/precompiled/string_ops.cc | 16 +++++++++-------
 cpp/src/gandiva/precompiled/time.cc       |  6 +++---
 2 files changed, 12 insertions(+), 10 deletions(-)

diff --git a/cpp/src/gandiva/precompiled/string_ops.cc b/cpp/src/gandiva/precompiled/string_ops.cc
index 035d3c8c62e1..b6d6fed236e1 100644
--- a/cpp/src/gandiva/precompiled/string_ops.cc
+++ b/cpp/src/gandiva/precompiled/string_ops.cc
@@ -2300,7 +2300,8 @@ const char* binary_string(gdv_int64 context, const char* text, gdv_int32 text_le
         (text[i + 1] == 'x' || text[i + 1] == 'X')) {
       char hd1 = text[i + 2];
       char hd2 = text[i + 3];
-      if (isxdigit(hd1) && isxdigit(hd2)) {
+      if (isxdigit(static_cast<unsigned char>(hd1)) &&
+          isxdigit(static_cast<unsigned char>(hd2))) {
         // [a-fA-F0-9]
         ret[j] = to_binary_from_hex(hd1) * 16 + to_binary_from_hex(hd2);
         i += 3;
@@ -2345,7 +2346,7 @@ const char* binary_string(gdv_int64 context, const char* text, gdv_int32 text_le
     int read_index = 0;                                                                \
     while (read_index < in_len) {                                                      \
       char c1 = in[read_index];                                                        \
-      if (isxdigit(c1)) {                                                              \
+      if (isxdigit(static_cast<unsigned char>(c1))) {                                  \
         digit = to_binary_from_hex(c1);                                                \
                                                                                        \
         OUT_TYPE next = result * 16 - digit;                                           \
@@ -2916,7 +2917,8 @@ const char* from_hex_utf8(int64_t context, const char* text, int32_t text_len,
   for (int32_t i = 0; i < text_len; i += 2) {
     char b1 = text[i];
     char b2 = text[i + 1];
-    if (isxdigit(b1) && isxdigit(b2)) {
+    if (isxdigit(static_cast<unsigned char>(b1)) &&
+        isxdigit(static_cast<unsigned char>(b2))) {
       // [a-fA-F0-9]
       ret[j++] = to_binary_from_hex(b1) * 16 + to_binary_from_hex(b2);
     } else {
@@ -2984,9 +2986,9 @@ 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(in[i]) > 0) {
+    if (isalpha(static_cast<unsigned char>(in[i])) > 0) {
       // Retain the first letter
-      ret[0] = toupper(in[i]);
+      ret[0] = toupper(static_cast<unsigned char>(in[i]));
       start_idx = i + 1;
       break;
     }
@@ -3002,8 +3004,8 @@ 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(in[i]) > 0) {
-      c = toupper(in[i]) - 65;
+    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++;
diff --git a/cpp/src/gandiva/precompiled/time.cc b/cpp/src/gandiva/precompiled/time.cc
index 8414d0ed37cf..d85e448f42aa 100644
--- a/cpp/src/gandiva/precompiled/time.cc
+++ b/cpp/src/gandiva/precompiled/time.cc
@@ -643,7 +643,7 @@ gdv_date64 castDATE_utf8(int64_t context, const char* input, gdv_int32 length) {
   int dateIndex = 0, index = 0, value = 0;
   int year_str_len = 0;
   while (dateIndex < 3 && index < length) {
-    if (!isdigit(input[index])) {
+    if (!isdigit(static_cast<unsigned char>(input[index]))) {
       dateFields[dateIndex++] = value;
       value = 0;
     } else {
@@ -713,7 +713,7 @@ gdv_timestamp castTIMESTAMP_utf8(int64_t context, const char* input, gdv_int32 l
   int year_str_len = 0, sub_seconds_len = 0;
   int ts_field_index = TimeFields::kYear, index = 0, value = 0;
   while (ts_field_index < TimeFields::kMax && index < length) {
-    if (isdigit(input[index])) {
+    if (isdigit(static_cast<unsigned char>(input[index]))) {
       value = (value * 10) + (input[index] - '0');
       if (ts_field_index == TimeFields::kYear) {
         year_str_len++;
@@ -842,7 +842,7 @@ gdv_time32 castTIME_utf8(int64_t context, const char* input, int32_t length) {
 
   bool has_invalid_digit = false;
   while (time_field_idx < TimeFields::kDisplacementHours && index < length) {
-    if (isdigit(input[index])) {
+    if (isdigit(static_cast<unsigned char>(input[index]))) {
       value = (value * 10) + (input[index] - '0');
 
       if (time_field_idx == TimeFields::kSubSeconds) {