[Chore](column) fix wrong chars size on TransformerToStringOneArgument and add more check

be/src/vec/columns/column_string.cpp

@@ -40,8 +40,8 @@ template <typename T>
 void ColumnStr<T>::sanity_check() const {
 #ifndef NDEBUG
     sanity_check_simple();
-    auto count = offsets.size();
-    for (size_t i = 0; i < count; ++i) {
+    auto count = cast_set<int>(offsets.size());
+    for (int i = 0; i < count; ++i) {
         if (offsets[i] < offsets[i - 1]) {
             throw Exception(Status::InternalError("row count: {}, offsets[{}]: {}, offsets[{}]: {}",
                                                   count, i, offsets[i], i - 1, offsets[i - 1]));
@@ -53,10 +53,10 @@ void ColumnStr<T>::sanity_check() const {
 template <typename T>
 void ColumnStr<T>::sanity_check_simple() const {
 #ifndef NDEBUG
-    auto count = offsets.size();
+    auto count = cast_set<int>(offsets.size());
     if (chars.size() != offsets[count - 1]) {
-        throw Exception(Status::InternalError("row count: {}, chars.size(): {}, offset[{}]: ",
-                                              count, chars.size(), offsets[count - 1]));
+        throw Exception(Status::InternalError("row count: {}, chars.size(): {}, offset[{}]: {}",
+                                              count, chars.size(), count - 1, offsets[count - 1]));
     }
     if (offsets[-1] != 0) {
         throw Exception(Status::InternalError("wrong offsets[-1]: {}", offsets[-1]));
@@ -639,6 +639,7 @@ template <typename T>
 void ColumnStr<T>::compare_internal(size_t rhs_row_id, const IColumn& rhs, int nan_direction_hint,
                                     int direction, std::vector<uint8>& cmp_res,
                                     uint8* __restrict filter) const {
+    sanity_check_simple();
     auto sz = offsets.size();
     DCHECK(cmp_res.size() == sz);
     const auto& cmp_base =

be/src/vec/columns/column_string.h

@@ -131,11 +131,13 @@ class ColumnStr final : public COWHelper<IColumn, ColumnStr<T>> {
 
     Field operator[](size_t n) const override {
         assert(n < size());
+        sanity_check_simple();
         return Field(String(reinterpret_cast<const char*>(&chars[offset_at(n)]), size_at(n)));
     }
 
     void get(size_t n, Field& res) const override {
         assert(n < size());
+        sanity_check_simple();
         if (res.get_type() == Field::Types::JSONB) {
             // Handle JsonbField
             res = JsonbField(reinterpret_cast<const char*>(&chars[offset_at(n)]), size_at(n));
@@ -146,6 +148,7 @@ class ColumnStr final : public COWHelper<IColumn, ColumnStr<T>> {
 
     StringRef get_data_at(size_t n) const override {
         DCHECK_LT(n, size());
+        sanity_check_simple();
         return StringRef(&chars[offset_at(n)], size_at(n));
     }
 

be/src/vec/exprs/vexpr_context.cpp

@@ -61,6 +61,12 @@ Status VExprContext::execute(vectorized::Block* block, int* result_column_id) {
     RETURN_IF_CATCH_EXCEPTION({
         st = _root->execute(this, block, result_column_id);
         _last_result_column_id = *result_column_id;
+        if (_last_result_column_id != -1) {
+            if (const auto* column_str = check_and_get_column<ColumnString>(
+                        block->get_by_position(*result_column_id).column.get())) {
+                column_str->sanity_check();
+            }
+        }
     });
     return st;
 }

be/src/vec/functions/date_time_transforms.h

@@ -317,6 +317,7 @@ struct TransformerToStringOneArgument {
                     cast_set<UInt32>(Transform::execute(date_time_value, res_data, offset));
             null_map[i] = !date_time_value.is_valid_date();
         }
+        res_data.resize(res_offsets[res_offsets.size() - 1]);
     }
 
     static void vector(FunctionContext* context,
@@ -336,6 +337,7 @@ struct TransformerToStringOneArgument {
                     cast_set<UInt32>(Transform::execute(date_time_value, res_data, offset));
             DCHECK(date_time_value.is_valid_date());
         }
+        res_data.resize(res_offsets[res_offsets.size() - 1]);
     }
 };
 

be/test/vec/function/function_test_util.h

@@ -317,6 +317,9 @@ Status check_function(const std::string& func_name, const InputTypeSet& input_ty
     // 3. check the result of function
     ColumnPtr column = block.get_columns()[result];
     EXPECT_TRUE(column);
+    if (const auto* column_str = check_and_get_column<ColumnString>(column.get())) {
+        column_str->sanity_check();
+    }
 
     for (int i = 0; i < row_size; ++i) {
         // update current line