fix

apache · Mar 7, 2025 · 92e0daa · 92e0daa
1 parent 739fc59
commit 92e0daa
Show file tree

Hide file tree

Showing 2 changed files with 34 additions and 14 deletions.
diff --git a/be/src/vec/columns/column_string.cpp b/be/src/vec/columns/column_string.cpp
@@ -40,8 +40,14 @@ template <typename T>
 void ColumnStr<T>::sanity_check() const {
 #ifndef NDEBUG
     sanity_check_simple();
-    auto count = cast_set<int>(offsets.size());
-    for (int i = 0; i < count; ++i) {
+    auto count = offsets.size();
+    if (count == 1) {
+        if (offsets[0] < offsets[-1]) {
+            throw Exception(Status::InternalError("row count: {}, offsets[{}]: {}, offsets[{}]: {}",
+                                                  count, 0, offsets[9], -1, offsets[-1]));
+        }
+    }
+    for (size_t 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 +59,19 @@ void ColumnStr<T>::sanity_check() const {
 template <typename T>
 void ColumnStr<T>::sanity_check_simple() const {
 #ifndef NDEBUG
-    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(), count - 1, offsets[count - 1]));
+    auto count = offsets.size();
+    if (count == 0) {
+        if (chars.size() != offsets[-1]) {
+            throw Exception(Status::InternalError("row count: {}, chars.size(): {}, offset[{}]: {}",
+                                                  count, chars.size(), -1, offsets[-1]));
+        }
+    } else {
+        auto size_may_overflow = cast_set<UInt32>(chars.size());
+        if (size_may_overflow != 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]));
@@ -668,11 +683,7 @@ void ColumnStr<T>::compare_internal(size_t rhs_row_id, const IColumn& rhs, int n
 
 template <typename T>
 ColumnPtr ColumnStr<T>::convert_column_if_overflow() {
-#ifdef BE_TEST
-    if (std::is_same_v<T, UInt32> && chars.size() > 10) {
-#else
     if (std::is_same_v<T, UInt32> && chars.size() > config::string_overflow_size) {
-#endif
         auto new_col = ColumnStr<uint64_t>::create();
 
         const auto length = offsets.size();

diff --git a/be/test/vec/columns/column_string_test.cpp b/be/test/vec/columns/column_string_test.cpp
@@ -878,13 +878,22 @@ TEST_F(ColumnStringTest, convert_column_if_overflow) {
     }
     {
         auto tmp_col = column_str32->clone();
+        auto* tmp_col_str32 = assert_cast<ColumnString*>(tmp_col.get());
         auto src_size = column_str32->size();
-        auto* tmp_col_str = assert_cast<ColumnString*>(tmp_col.get());
-        EXPECT_GT(tmp_col_str->get_chars().size(), 10);
+        auto chars_size = column_str32->get_chars().size();
+        auto max_chars_size = config::string_overflow_size;
+        while (chars_size < max_chars_size) {
+            tmp_col->insert_range_from_ignore_overflow(*column_str32, 0, column_str32->size());
+            chars_size = tmp_col_str32->get_chars().size();
+        }
+        tmp_col->insert_range_from_ignore_overflow(*column_str32, 0, column_str32->size());
+        auto tmp_col_row_count = tmp_col->size();
+        chars_size = tmp_col_str32->get_chars().size();
+        EXPECT_GT(chars_size, max_chars_size);
         auto tmp_col_converted = tmp_col->convert_column_if_overflow();
         EXPECT_TRUE(tmp_col_converted->is_column_string64());
-        for (size_t i = 0; i < src_size; ++i) {
-            EXPECT_EQ(tmp_col_converted->get_data_at(i), column_str32->get_data_at(i));
+        for (size_t i = 0; i < tmp_col_row_count; ++i) {
+            EXPECT_EQ(tmp_col_converted->get_data_at(i), column_str32->get_data_at(i % src_size));
         }
     }
 }