Add empty2null Spark function (facebookincubator#10616)

Summary:
A function converts empty string input to null value.

Spark implementation: https://github.com/apache/spark/blob/34a52ad2ce377dd077c4cf70740ba3ab6c18c739/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/stringExpressions.scala#L3486-L3508

Pull Request resolved: facebookincubator#10616

Reviewed By: DanielHunte

Differential Revision: D60599503

Pulled By: kagamiori

fbshipit-source-id: 3082d1877cccfc2ce3c36a7d4385811d7932d214

Author: Yohahaha

velox/docs/functions/spark/string.rst

@@ -53,6 +53,15 @@ Unless specified otherwise, all functions return NULL if at least one of the arg
         SELECT conv("11abc", 10, 10); -- '11'
         SELECT conv('H016F', 16, 10); -- '0'
 
+.. spark:function:: empty2null(input) -> varchar
+
+    Returns NULL if ``input`` is empty. Otherwise, returns ``input``.
+    Note: it's an internal Spark function used to convert empty value of a partition column,
+    which is then converted to Hive default partition value ``__HIVE_DEFAULT_PARTITION__``. ::
+
+        SELECT empty2null(''); -- NULL
+        SELECT empty2null('abc'); -- 'abc'
+
 .. spark:function:: endswith(left, right) -> boolean
 
     Returns true if 'left' ends with 'right'. Otherwise, returns false. ::

velox/functions/sparksql/Register.cpp

@@ -242,6 +242,8 @@ void registerFunctions(const std::string& prefix) {
   registerFunction<Sha2HexStringFunction, Varchar, Varbinary, int32_t>(
       {prefix + "sha2"});
   registerFunction<CRC32Function, int64_t, Varbinary>({prefix + "crc32"});
+  registerFunction<Empty2NullFunction, Varchar, Varchar>(
+      {prefix + "empty2null"});
 
   exec::registerStatefulVectorFunction(
       prefix + "regexp_extract", re2ExtractSignatures(), makeRegexExtract);

velox/functions/sparksql/String.h

@@ -1431,4 +1431,26 @@ struct LevenshteinDistanceFunction {
   }
 };
 
+/// empty2null(input) -> varchar
+///
+///    Returns NULL when the input is empty,
+///    otherwise, it returns the input itself.
+template <typename T>
+struct Empty2NullFunction {
+  VELOX_DEFINE_FUNCTION_TYPES(T);
+
+  // Results refer to strings in the first argument.
+  static constexpr int32_t reuse_strings_from_arg = 0;
+
+  FOLLY_ALWAYS_INLINE bool call(
+      out_type<Varchar>& result,
+      const arg_type<Varchar>& input) {
+    if (input.empty()) {
+      return false;
+    }
+    result.setNoCopy(input);
+    return true;
+  }
+};
+
 } // namespace facebook::velox::functions::sparksql

velox/functions/sparksql/tests/StringTest.cpp

@@ -995,5 +995,14 @@ TEST_F(StringTest, trim) {
       trimWithTrimStr("\u6570", "\u6574\u6570 \u6570\u636E!"),
       "\u6574\u6570 \u6570\u636E!");
 }
+
+TEST_F(StringTest, empty2Null) {
+  const auto empty2Null = [&](const std::optional<std::string>& a) {
+    return evaluateOnce<std::string>("empty2null(c0)", a);
+  };
+
+  EXPECT_EQ(empty2Null(""), std::nullopt);
+  EXPECT_EQ(empty2Null("abc"), "abc");
+}
 } // namespace
 } // namespace facebook::velox::functions::sparksql::test