Copilot commented on code in PR #49813:
URL: https://github.com/apache/arrow/pull/49813#discussion_r3291445715


##########
cpp/src/gandiva/gdv_string_function_stubs.cc:
##########
@@ -751,6 +800,11 @@ const char* translate_utf8_utf8_utf8(int64_t context, 
const char* in, int32_t in
             break;
           }
 
+          // Updating len to char in this position
+          len_char_from = gdv_fn_utf8_char_length(from[from_for]);
+          // Making copy to std::string with length for this char position
+          std::string copy_from_compare(from + from_for, len_char_from);
+

Review Comment:
   The inner multibyte loop over `from` increments by `len_char_from` 
(`from_for += len_char_from`), but `len_char_from` is computed via 
`gdv_fn_utf8_char_length(from[from_for])` which returns 0 for invalid leading 
bytes. If `from` contains an invalid byte with the high bit set, this loop can 
become non-terminating. Add validation for `len_char_from <= 0` (and similarly 
for `len_char_to`) to avoid infinite loops and to report invalid UTF-8.



##########
cpp/src/gandiva/gdv_string_function_stubs.cc:
##########
@@ -367,6 +396,15 @@ const char* gdv_fn_substring_index(int64_t context, const 
char* txt, int32_t txt
     return "";
   }
 
+  if (ARROW_PREDICT_FALSE(txt_len < 0)) {
+    *out_len = 0;
+    return "";
+  }
+  if (ARROW_PREDICT_FALSE(pat_len < 0)) {

Review Comment:
   `gdv_fn_substring_index` now guards against negative `txt_len` / `pat_len` 
by returning an empty string, but it doesn't set an error message. This differs 
from other negative-length checks in this file (e.g., cast macros / 
`is_datalen_valid`) that set a context error, and makes invalid inputs fail 
silently. Consider calling `gdv_fn_context_set_error_msg` for negative lengths 
so callers can detect misuse.
   



##########
cpp/src/gandiva/precompiled/string_ops.cc:
##########
@@ -2444,182 +2462,178 @@ void concat_word(char* out_buf, int* out_idx, const 
char* in_buf, int in_len,
   *out_idx += in_len;
 }
 
-FORCE_INLINE
-const char* concat_ws_utf8_utf8(int64_t context, const char* separator,
-                                int32_t separator_len, bool separator_validity,
-                                const char* word1, int32_t word1_len, bool 
word1_validity,
-                                const char* word2, int32_t word2_len, bool 
word2_validity,
-                                bool* out_valid, int32_t* out_len) {
-  *out_len = 0;
-  int numValidInput = 0;
-  // If separator is null, always return null
-  if (!separator_validity) {
-    *out_len = 0;
-    *out_valid = false;
-    return "";
-  }
+// Helper structure to maintain state during safe length accumulation
+struct SafeLengthState {
+  int32_t total_len = 0;
+  int32_t num_valid = 0;
+  bool overflow = false;
+};
+
+// Helper to safely add a word length
+static inline bool safe_accumulate_word(int64_t context, SafeLengthState& 
state,
+                                        int32_t word_len, bool word_validity) {
+  if (!word_validity) return true;
 
-  if (word1_validity) {
-    *out_len += word1_len;
-    numValidInput++;
+  if (word_len < 0) {
+    gdv_fn_context_set_error_msg(context, "Invalid word length.");
+    return false;
   }
-  if (word2_validity) {
-    *out_len += word2_len;
-    numValidInput++;
+
+  int32_t temp = 0;
+  if (ARROW_PREDICT_FALSE(
+          arrow::internal::AddWithOverflow(state.total_len, word_len, &temp))) 
{
+    gdv_fn_context_set_error_msg(context, "Overflow in addition detected.");
+    state.overflow = true;
+    return false;
   }
+  state.total_len = temp;
+  state.num_valid++;
+  return true;
+}
 
-  *out_len += separator_len * (numValidInput > 1 ? numValidInput - 1 : 0);
-  if (*out_len == 0) {
-    *out_valid = true;
-    return "";
+// Helper to safely add separators based on number of valid words
+static inline bool safe_add_separators(int64_t context, SafeLengthState* state,
+                                       int32_t separator_len) {
+  if (state->num_valid <= 1) return true;
+
+  int32_t sep_total = 0;
+  int32_t temp = 0;
+
+  if (ARROW_PREDICT_FALSE(arrow::internal::MultiplyWithOverflow(
+          separator_len, state->num_valid - 1, &sep_total))) {
+    gdv_fn_context_set_error_msg(context, "Overflow in multiplication 
detected.");
+    state->overflow = true;
+    return false;
   }
 
-  char* out = reinterpret_cast<char*>(gdv_fn_context_arena_malloc(context, 
*out_len));
-  if (out == nullptr) {
-    gdv_fn_context_set_error_msg(context, "Could not allocate memory for 
output string");
-    *out_len = 0;
-    *out_valid = false;
-    return "";
+  if (ARROW_PREDICT_FALSE(
+          arrow::internal::AddWithOverflow(state->total_len, sep_total, 
&temp))) {
+    gdv_fn_context_set_error_msg(context, "Overflow in addition detected.");
+    state->overflow = true;
+    return false;
   }
 
-  char* tmp = out;
-  int out_idx = 0;
-  bool seenAnyValidInput = false;
+  state->total_len = temp;
+  return true;
+}
 
-  concat_word(tmp, &out_idx, word1, word1_len, word1_validity, separator, 
separator_len,
-              &seenAnyValidInput);
-  concat_word(tmp, &out_idx, word2, word2_len, word2_validity, separator, 
separator_len,
-              &seenAnyValidInput);
+// Helper to handle overflow failure (sets output parameters and returns empty 
string)
+static inline const char* handle_overflow_failure(bool* out_valid, int32_t* 
out_len) {
+  *out_len = 0;
+  *out_valid = false;
+  return "";
+}
 
+// Helper to handle empty result (all words invalid)
+static inline const char* handle_empty_result(bool* out_valid, int32_t* 
out_len) {
+  *out_len = 0;
   *out_valid = true;
-  *out_len = out_idx;
-  return out;
+  return "";
 }
 
-FORCE_INLINE
-const char* concat_ws_utf8_utf8_utf8(
-    int64_t context, const char* separator, int32_t separator_len,
-    bool separator_validity, const char* word1, int32_t word1_len, bool 
word1_validity,
-    const char* word2, int32_t word2_len, bool word2_validity, const char* 
word3,
-    int32_t word3_len, bool word3_validity, bool* out_valid, int32_t* out_len) 
{
+struct WordArg {
+  const char* data;
+  int32_t len;
+  bool valid;
+};
+
+static inline const char* concat_ws_impl(int64_t context, const char* 
separator,
+                                         int32_t separator_len, bool 
separator_validity,
+                                         bool* out_valid, int32_t* out_len,
+                                         std::initializer_list<WordArg> words) 
{
   *out_len = 0;
-  int numValidInput = 0;
-  // If separator is null, always return null
+
+  // Separator validity check
   if (!separator_validity) {
-    *out_len = 0;
     *out_valid = false;
     return "";
   }
-
-  if (word1_validity) {
-    *out_len += word1_len;
-    numValidInput++;
-  }
-  if (word2_validity) {
-    *out_len += word2_len;
-    numValidInput++;
+  if (separator_len < 0) {

Review Comment:
   `concat_ws_impl` treats a negative `separator_len` as invalid and returns 
`out_valid=false`, but it doesn't set an error message on the execution 
context. This makes negative-length failures silent, unlike the other 
invalid-length/overflow paths here that call `gdv_fn_context_set_error_msg`. 
Consider setting a clear error (and keeping `*out_len = 0`) when `separator_len 
< 0` so callers/tests can distinguish invalid input from NULL separator 
handling.
   



-- 
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.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to