[BugFix] Fix UDTF wrong result when miss multibyte UTF-8 (#51232)

Signed-off-by: stdpain <[email protected]>
(cherry picked from commit 266bfc8)

be/src/udf/java/java_udf.cpp

@@ -406,14 +406,11 @@ jobject JVMFunctionHelper::newString(const char* data, size_t size) {
     return nstr;
 }
 
-size_t JVMFunctionHelper::string_length(jstring jstr) {
-    return _env->GetStringUTFLength(jstr);
-}
-
 Slice JVMFunctionHelper::sliceVal(jstring jstr, std::string* buffer) {
-    size_t length = this->string_length(jstr);
-    buffer->resize(length);
-    _env->GetStringUTFRegion(jstr, 0, length, buffer->data());
+    const size_t utf_length = _env->GetStringUTFLength(jstr);
+    buffer->resize(utf_length);
+    const size_t string_length = _env->GetStringLength(jstr);
+    _env->GetStringUTFRegion(jstr, 0, string_length, buffer->data());
     return {buffer->data(), buffer->length()};
 }
 

be/src/udf/java/java_udf.h

@@ -127,7 +127,6 @@ class JVMFunctionHelper {
     DECLARE_NEW_BOX(double, double, Double)
 
     jobject newString(const char* data, size_t size);
-    size_t string_length(jstring jstr);
 
     Slice sliceVal(jstring jstr, std::string* buffer);
     jclass string_clazz() { return _string_class; }

test/sql/test_udf/R/test_jvm_udf

@@ -90,6 +90,34 @@ select count(udtfdouble) from t0, udtfdouble(c1);
 -- result:
 81920
 -- !result
+select * from TABLE(udtfstring(""));
+-- result:
+-- !result
+select * from TABLE(udtfstring("▁▂▃▄▅▆▇█"));
+-- result:
+▁▂▃▄▅▆▇█
+▁▂▃▄▅▆▇█
+-- !result
+select * from TABLE(udtfstring("中文测试"));
+-- result:
+中文测试
+中文测试
+-- !result
+select * from TABLE(udtfstring("∀∂∈ℝ∧∪≡∞ ↑↗↨↻⇣ ┐┼╔╘░►☺♀ fi�⑀₂ἠḂӥẄɐː⍎אԱა"));
+-- result:
+∀∂∈ℝ∧∪≡∞ ↑↗↨↻⇣ ┐┼╔╘░►☺♀ fi�⑀₂ἠḂӥẄɐː⍎אԱა
+∀∂∈ℝ∧∪≡∞ ↑↗↨↻⇣ ┐┼╔╘░►☺♀ fi�⑀₂ἠḂӥẄɐː⍎אԱა
+-- !result
+select * from TABLE(udtfstring("2H₂ + O₂ ⇌ 2H₂O"));
+-- result:
+2H₂ + O₂ ⇌ 2H₂O
+2H₂ + O₂ ⇌ 2H₂O
+-- !result
+select * from TABLE(udtfstring("ᚻᛖ ᚳᚹᚫᚦ ᚦᚫᛏ ᚻᛖ ᛒᚢᛞᛖ ᚩᚾ ᚦᚫᛗ ᛚᚪᚾᛞᛖ ᚾᚩᚱᚦᚹᛖᚪᚱᛞᚢᛗ ᚹᛁᚦ ᚦᚪ ᚹᛖᛥᚫ"));
+-- result:
+ᚻᛖ ᚳᚹᚫᚦ ᚦᚫᛏ ᚻᛖ ᛒᚢᛞᛖ ᚩᚾ ᚦᚫᛗ ᛚᚪᚾᛞᛖ ᚾᚩᚱᚦᚹᛖᚪᚱᛞᚢᛗ ᚹᛁᚦ ᚦᚪ ᚹᛖᛥᚫ
+ᚻᛖ ᚳᚹᚫᚦ ᚦᚫᛏ ᚻᛖ ᛒᚢᛞᛖ ᚩᚾ ᚦᚫᛗ ᛚᚪᚾᛞᛖ ᚾᚩᚱᚦᚹᛖᚪᚱᛞᚢᛗ ᚹᛁᚦ ᚦᚪ ᚹᛖᛥᚫ
+-- !result
 set streaming_preaggregation_mode="force_streaming";
 -- result:
 -- !result

test/sql/test_udf/T/test_jvm_udf

@@ -64,6 +64,14 @@ select count(udtfint) from t0, udtfint(c1);
 select count(udtfbigint) from t0, udtfbigint(c1);
 select count(udtffloat) from t0, udtffloat(c1);
 select count(udtfdouble) from t0, udtfdouble(c1);
+-- test udtf with utf8 case
+select * from TABLE(udtfstring(""));
+select * from TABLE(udtfstring("▁▂▃▄▅▆▇█"));
+select * from TABLE(udtfstring("中文测试"));
+select * from TABLE(udtfstring("∀∂∈ℝ∧∪≡∞ ↑↗↨↻⇣ ┐┼╔╘░►☺♀ fi�⑀₂ἠḂӥẄɐː⍎אԱა"));
+select * from TABLE(udtfstring("2H₂ + O₂ ⇌ 2H₂O"));
+select * from TABLE(udtfstring("ᚻᛖ ᚳᚹᚫᚦ ᚦᚫᛏ ᚻᛖ ᛒᚢᛞᛖ ᚩᚾ ᚦᚫᛗ ᛚᚪᚾᛞᛖ ᚾᚩᚱᚦᚹᛖᚪᚱᛞᚢᛗ ᚹᛁᚦ ᚦᚪ ᚹᛖᛥᚫ"));
+
 
 -- test group by limit case:
 set streaming_preaggregation_mode="force_streaming";