Merge pull request #2141 from kuzudb/copy-to-plan

Refactor toString()
kuzudb · Oct 4, 2023 · af6bf97 · af6bf97
2 parents 37800fe + 988c398
commit af6bf97
Show file tree

Hide file tree

Showing 4 changed files with 68 additions and 44 deletions.
diff --git a/src/common/type_utils.cpp b/src/common/type_utils.cpp
@@ -54,6 +54,37 @@ std::string TypeUtils::castValueToString(
     }
 }
 
+template<>
+std::string TypeUtils::toString(const bool& val, void* valueVector) {
+    return val ? "True" : "False";
+}
+
+template<>
+std::string TypeUtils::toString(const internalID_t& val, void* valueVector) {
+    return std::to_string(val.tableID) + ":" + std::to_string(val.offset);
+}
+
+template<>
+std::string TypeUtils::toString(const date_t& val, void* valueVector) {
+    return Date::toString(val);
+}
+
+template<>
+std::string TypeUtils::toString(const timestamp_t& val, void* valueVector) {
+    return Timestamp::toString(val);
+}
+
+template<>
+std::string TypeUtils::toString(const interval_t& val, void* valueVector) {
+    return Interval::toString(val);
+}
+
+template<>
+std::string TypeUtils::toString(const ku_string_t& val, void* valueVector) {
+    return val.getAsString();
+}
+
+template<>
 std::string TypeUtils::toString(const list_entry_t& val, void* valueVector) {
     auto listVector = (ValueVector*)valueVector;
     if (val.size == 0) {
@@ -77,6 +108,7 @@ std::string TypeUtils::toString(const list_entry_t& val, void* valueVector) {
     return result;
 }
 
+template<>
 std::string TypeUtils::toString(const struct_entry_t& val, void* valVector) {
     auto structVector = (ValueVector*)valVector;
     auto fields = StructType::getFields(&structVector->dataType);

diff --git a/src/include/common/type_utils.h b/src/include/common/type_utils.h
@@ -16,28 +16,15 @@ class TypeUtils {
 
 public:
     static uint32_t convertToUint32(const char* data);
-
-    static inline std::string toString(bool boolVal) { return boolVal ? "True" : "False"; }
-    static inline std::string toString(int64_t val) { return std::to_string(val); }
-    static inline std::string toString(int32_t val) { return std::to_string(val); }
-    static inline std::string toString(int16_t val) { return std::to_string(val); }
-    static inline std::string toString(int8_t val) { return std::to_string(val); }
-    static inline std::string toString(uint64_t val) { return std::to_string(val); }
-    static inline std::string toString(uint32_t val) { return std::to_string(val); }
-    static inline std::string toString(uint16_t val) { return std::to_string(val); }
-    static inline std::string toString(uint8_t val) { return std::to_string(val); }
-    static inline std::string toString(double_t val) { return std::to_string(val); }
-    static inline std::string toString(float_t val) { return std::to_string(val); }
-    static inline std::string toString(const internalID_t& val) {
-        return std::to_string(val.tableID) + ":" + std::to_string(val.offset);
+    template<typename T>
+    static inline std::string toString(const T& val, void* valueVector = nullptr) {
+        static_assert(std::is_same<T, int64_t>::value || std::is_same<T, int32_t>::value ||
+                      std::is_same<T, int16_t>::value || std::is_same<T, int8_t>::value ||
+                      std::is_same<T, uint64_t>::value || std::is_same<T, uint32_t>::value ||
+                      std::is_same<T, uint16_t>::value || std::is_same<T, uint8_t>::value ||
+                      std::is_same<T, double_t>::value || std::is_same<T, float_t>::value);
+        return std::to_string(val);
     }
-    static inline std::string toString(const date_t& val) { return Date::toString(val); }
-    static inline std::string toString(const timestamp_t& val) { return Timestamp::toString(val); }
-    static inline std::string toString(const interval_t& val) { return Interval::toString(val); }
-    static inline std::string toString(const ku_string_t& val) { return val.getAsString(); }
-    static inline std::string toString(const std::string& val) { return val; }
-    static std::string toString(const list_entry_t& val, void* valueVector);
-    static std::string toString(const struct_entry_t& val, void* valueVector);
 
     static inline void encodeOverflowPtr(
         uint64_t& overflowPtr, page_idx_t pageIdx, uint16_t pageOffset) {
@@ -57,5 +44,23 @@ class TypeUtils {
     static std::string castValueToString(const LogicalType& dataType, uint8_t* value, void* vector);
 };
 
+// Forward declaration of template specializations.
+template<>
+std::string TypeUtils::toString(const bool& val, void* valueVector);
+template<>
+std::string TypeUtils::toString(const internalID_t& val, void* valueVector);
+template<>
+std::string TypeUtils::toString(const date_t& val, void* valueVector);
+template<>
+std::string TypeUtils::toString(const timestamp_t& val, void* valueVector);
+template<>
+std::string TypeUtils::toString(const interval_t& val, void* valueVector);
+template<>
+std::string TypeUtils::toString(const ku_string_t& val, void* valueVector);
+template<>
+std::string TypeUtils::toString(const list_entry_t& val, void* valueVector);
+template<>
+std::string TypeUtils::toString(const struct_entry_t& val, void* valueVector);
+
 } // namespace common
 } // namespace kuzu
diff --git a/src/include/processor/operator/persistent/csv_file_writer.h b/src/include/processor/operator/persistent/csv_file_writer.h
@@ -23,9 +23,6 @@ class CSVFileWriter : public FileWriter {
     template<typename T>
     void writeToBuffer(common::ValueVector* vector, bool escapeStringValue = false);
 
-    template<typename T>
-    void writeListToBuffer(common::ValueVector* vector);
-
     inline void writeToBuffer(const std::string& value) { buffer << value; }
     inline void writeToBuffer(const char value) { buffer << value; }
 

diff --git a/src/processor/operator/persistent/csv_file_writer.cpp b/src/processor/operator/persistent/csv_file_writer.cpp
@@ -1,7 +1,5 @@
 #include "processor/operator/persistent/csv_file_writer.h"
 
-#include <fstream>
-
 #include "common/constants.h"
 #include "common/string_utils.h"
 #include "common/type_utils.h"
@@ -57,23 +55,15 @@ void CSVFileWriter::writeValues(std::vector<common::ValueVector*>& outputVectors
 template<typename T>
 void CSVFileWriter::writeToBuffer(common::ValueVector* vector, bool escapeStringValue) {
     auto selPos = vector->state->selVector->selectedPositions[0];
-    auto value = vector->isNull(selPos) ? "" : TypeUtils::toString(vector->getValue<T>(selPos));
+    auto value = vector->isNull(selPos) ? "" :
+                                          TypeUtils::toString(vector->getValue<T>(selPos),
+                                              reinterpret_cast<void*>(vector));
     if (escapeStringValue) {
         escapeString(value);
     }
     writeToBuffer(value);
 }
 
-template<typename T>
-void CSVFileWriter::writeListToBuffer(common::ValueVector* vector) {
-    // vectors are always flat
-    auto selPos = vector->state->selVector->selectedPositions[0];
-    auto value =
-        vector->isNull(selPos) ? "" : TypeUtils::toString(vector->getValue<T>(selPos), vector);
-    escapeString(value);
-    writeToBuffer(value);
-}
-
 void CSVFileWriter::escapeString(std::string& value) {
     StringUtils::replaceAll(value, "\"", "\"\"");
     value = "\"" + value + "\"";
@@ -95,20 +85,20 @@ void CSVFileWriter::writeValue(common::ValueVector* vector) {
     case LogicalTypeID::FLOAT:
         return writeToBuffer<float>(vector);
     case LogicalTypeID::DATE:
-        return writeToBuffer<date_t>(vector, true);
+        return writeToBuffer<date_t>(vector, true /* escapeStringValue */);
     case LogicalTypeID::TIMESTAMP:
-        return writeToBuffer<timestamp_t>(vector, true);
+        return writeToBuffer<timestamp_t>(vector, true /* escapeStringValue */);
     case LogicalTypeID::INTERVAL:
-        return writeToBuffer<interval_t>(vector, true);
+        return writeToBuffer<interval_t>(vector, true /* escapeStringValue */);
     case LogicalTypeID::STRING:
-        return writeToBuffer<ku_string_t>(vector, true);
+        return writeToBuffer<ku_string_t>(vector, true /* escapeStringValue */);
     case LogicalTypeID::INTERNAL_ID:
-        return writeToBuffer<internalID_t>(vector, true);
+        return writeToBuffer<internalID_t>(vector, true /* escapeStringValue */);
     case LogicalTypeID::VAR_LIST:
     case LogicalTypeID::FIXED_LIST:
-        return writeListToBuffer<list_entry_t>(vector);
+        return writeToBuffer<list_entry_t>(vector, true /* escapeStringValue */);
     case LogicalTypeID::STRUCT:
-        return writeListToBuffer<struct_entry_t>(vector);
+        return writeToBuffer<struct_entry_t>(vector, true /* escapeStringValue */);
     default: {
         throw NotImplementedException("CSVFileWriter::writeValue");
     }