Apply urlEncode to all partition values

Author: PingLiuPing

velox/connectors/hive/iceberg/IcebergPartitionIdGenerator.cpp

@@ -40,6 +40,26 @@ std::pair<std::string, std::string> makePartitionKeyValueString(
           partitionVector->as<SimpleVector<T>>()->valueAt(row)));
 }
 
+// Iceberg spec requires URL encoding in the partition path.
+// This function matches java.net.URLEncoder.encode(string, "UTF-8").
+std::string urlEncode(const StringView& data) {
+  std::ostringstream ret;
+
+  for (unsigned char c : data) {
+    // These characters are not encoded in Java's URLEncoder.
+    if (std::isalnum(c) || c == '-' || c == '_' || c == '.' || c == '*') {
+      ret << c;
+    } else if (c == ' ') {
+      ret << '+';
+    } else {
+      // All other characters are percent-encoded.
+      ret << fmt::format("%{:02X}", c);
+    }
+  }
+
+  return ret.str();
+}
+
 } // namespace
 
 IcebergPartitionIdGenerator::IcebergPartitionIdGenerator(
@@ -178,7 +198,7 @@ std::string IcebergPartitionIdGenerator::partitionName(
     if (partitionPathAsLowerCase_) {
       folly::toLowerAscii(key);
     }
-    ret << fmt::format("{}={}", urlEncode(key.data()), value);
+    ret << fmt::format("{}={}", urlEncode(key.data()), urlEncode(value.data()));
   }
 
   return ret.str();

velox/connectors/hive/iceberg/Transforms.cpp

@@ -45,26 +45,6 @@ const TypePtr findChildTypeKind(
   return currentType;
 }
 
-// Iceberg spec requires URL encoding in the partition path.
-// This function matches java.net.URLEncoder.encode(string, "UTF-8").
-std::string urlEncode(const StringView& data) {
-  std::ostringstream ret;
-
-  for (unsigned char c : data) {
-    // These characters are not encoded in Java's URLEncoder.
-    if (std::isalnum(c) || c == '-' || c == '_' || c == '.' || c == '*') {
-      ret << c;
-    } else if (c == ' ') {
-      ret << '+';
-    } else {
-      // All other characters are percent-encoded.
-      ret << fmt::format("%{:02X}", c);
-    }
-  }
-
-  return ret.str();
-}
-
 template <typename T>
 VectorPtr IdentityTransform<T>::apply(const VectorPtr& block) const {
   if constexpr (std::is_same_v<T, StringView>) {

velox/connectors/hive/iceberg/Transforms.h

@@ -29,8 +29,6 @@ const TypePtr findChildTypeKind(
     const RowTypePtr& inputType,
     const std::string& fullName);
 
-std::string urlEncode(const StringView& data);
-
 class Transform {
  public:
   Transform(TypePtr type, TransformType transformType, memory::MemoryPool* pool)
@@ -63,10 +61,6 @@ class Transform {
     return decimalToHumanString(value);
   }
 
-  std::string toHumanString(const StringView& value) const {
-    return urlEncode(value);
-  }
-
   std::string toHumanString(bool value) const {
     return value ? "true" : "false";
   }
@@ -80,7 +74,7 @@ class Transform {
     options.skipTrailingZeros = true;
     options.leadingPositiveSign = true;
     options.skipTrailingZeroSeconds = true;
-    return urlEncode(value.toString(options).data());
+    return value.toString(options);
   }
 
   std::string name() const {