Add INT8 type

CMakeLists.txt

@@ -1,6 +1,6 @@
 cmake_minimum_required(VERSION 3.11)
 
-project(Kuzu VERSION 0.0.8.5 LANGUAGES CXX)
+project(Kuzu VERSION 0.0.8.6 LANGUAGES CXX)
 
 find_package(Threads REQUIRED)
 

dataset/tinysnb/eStudyAt.csv

@@ -1,4 +1,4 @@
-from,to,YEAR,Places,length
-0,1,2021,"[wwAewsdndweusd,wek]",5
-2,1,2020,"[anew,jsdnwusklklklwewsd]",55
-8,1,2020,"[awndsnjwejwen,isuhuwennjnuhuhuwewe]",22
+from,to,YEAR,Places,length,level
+0,1,2021,"[wwAewsdndweusd,wek]",5,5
+2,1,2020,"[anew,jsdnwusklklklwewsd]",55,120
+8,1,2020,"[awndsnjwejwen,isuhuwennjnuhuhuwewe]",22,2

dataset/tinysnb/schema.cypher

@@ -1,8 +1,8 @@
 create node table person (ID INt64, fName StRING, gender INT64, isStudent BoOLEAN, isWorker BOOLEAN, age INT64, eyeSight DOUBLE, birthdate DATE, registerTime TIMESTAMP, lastJobDuration interval, workedHours INT64[], usedNames STRING[], courseScoresPerTerm INT64[][], grades INT64[4], height float, PRIMARY KEY (ID));
 create node table organisation (ID INT64, name STRING, orgCode INT64, mark DOUBLE, score INT64, history STRING, licenseValidInterval INTERVAL, rating DOUBLE, state STRUCT(revenue INT16, location STRING[], stock STRUCT(price INT64[], volume INT64)), info UNION(price FLOAT, movein DATE, note STRING),PRIMARY KEY (ID));
-create node table movies (name STRING, length INT32, note STRING, description STRUCT(rating DOUBLE, views INT64, release TIMESTAMP, film DATE), content BYTEA, audience MAP(STRING, INT64), grade union(credit boolean, grade1 double, grade2 int64), PRIMARY KEY (name));
+create node table movies (name STRING, length INT32, note STRING, description STRUCT(rating DOUBLE, stars INT8, views INT64, release TIMESTAMP, film DATE), content BYTEA, audience MAP(STRING, INT64), grade union(credit boolean, grade1 double, grade2 int64), PRIMARY KEY (name));
 create rel table knows (FROM person TO person, date DATE, meetTime TIMESTAMP, validInterval INTERVAL, comments STRING[], MANY_MANY);
-create rel table studyAt (FROM person TO organisation, year INT64, places STRING[], length INT16,MANY_ONE);
+create rel table studyAt (FROM person TO organisation, year INT64, places STRING[], length INT16, level INT8, MANY_ONE);
 create rel table workAt (FROM person TO organisation, year INT64, grading DOUBLE[2], rating float, MANY_ONE);
 create rel table meets (FROM person TO person, location FLOAT[2], times INT, data BYTEA, MANY_ONE);
 create rel table marries (FROM person TO person, usedAddress STRING[], address INT16[2], note STRING, ONE_ONE);

dataset/tinysnb/vMovies.csv

@@ -1,3 +1,3 @@
-Sóló cón tu párejâ,126, this is a very very good movie,"{rating: 5.3, views: 152, release: 2011-08-20 11:25:30, film: 2012-05-11}","\\xAA\\xABinteresting\\x0B","{audience1= 52,audience53= 42}",true
-The 😂😃🧘🏻‍♂️🌍🌦️🍞🚗 movie,2544, the movie is very very good,"{rating: 7, views: 982, release: 2018-11-13 13:33:11, film: 2014-09-12}","\\xAB\\xCD",{audience1= 33},8.989
-Roma,298,the movie is very interesting and funny,"{rating: 1223, views: 10003, release: 2011-02-11 16:44:22, film: 2013-02-22}","pure ascii characters","{}",254
+Sóló cón tu párejâ,126, this is a very very good movie,"{rating: 5.3, stars:2, views: 152, release: 2011-08-20 11:25:30, film: 2012-05-11}","\\xAA\\xABinteresting\\x0B","{audience1= 52,audience53= 42}",true
+The 😂😃🧘🏻‍♂️🌍🌦️🍞🚗 movie,2544, the movie is very very good,"{rating: 7, stars:10, views: 982, release: 2018-11-13 13:33:11, film: 2014-09-12}","\\xAB\\xCD",{audience1= 33},8.989
+Roma,298,the movie is very interesting and funny,"{rating: 1223, stars:100, views: 10003, release: 2011-02-11 16:44:22, film: 2013-02-22}","pure ascii characters","{}",254

src/common/types/types.cpp

@@ -22,6 +22,8 @@
         return "INT32";
     case PhysicalTypeID::INT16:
         return "INT16";
+    case PhysicalTypeID::INT8:
+        return "INT8";
     case PhysicalTypeID::DOUBLE:
         return "DOUBLE";
     case PhysicalTypeID::FLOAT:
@@ -53,6 +55,8 @@
         return sizeof(int32_t);
     case PhysicalTypeID::INT16:
         return sizeof(int16_t);
+    case PhysicalTypeID::INT8:
+        return sizeof(int8_t);
     case PhysicalTypeID::DOUBLE:
         return sizeof(double_t);
     case PhysicalTypeID::FLOAT:
@@ -334,6 +338,9 @@
     case LogicalTypeID::INT16: {
         physicalType = PhysicalTypeID::INT16;
     } break;
+    case LogicalTypeID::INT8: {
+        physicalType = PhysicalTypeID::INT8;
+    } break;
     case LogicalTypeID::DOUBLE: {
         physicalType = PhysicalTypeID::DOUBLE;
     } break;
@@ -405,6 +412,8 @@
         return LogicalTypeID::INT16;
     } else if ("INT" == upperDataTypeIDString) {
         return LogicalTypeID::INT32;
+    } else if ("INT8" == upperDataTypeIDString) {
+        return LogicalTypeID::INT8;
     } else if ("DOUBLE" == upperDataTypeIDString) {
         return LogicalTypeID::DOUBLE;
     } else if ("FLOAT" == upperDataTypeIDString) {
@@ -480,6 +489,7 @@
     case LogicalTypeID::INT64:
     case LogicalTypeID::INT32:
     case LogicalTypeID::INT16:
+    case LogicalTypeID::INT8:
     case LogicalTypeID::DOUBLE:
     case LogicalTypeID::FLOAT:
     case LogicalTypeID::DATE:
@@ -514,6 +524,8 @@
         return "INT32";
     case LogicalTypeID::INT16:
         return "INT16";
+    case LogicalTypeID::INT8:
+        return "INT8";
     case LogicalTypeID::DOUBLE:
         return "DOUBLE";
     case LogicalTypeID::FLOAT:

src/common/types/value.cpp

@@ -50,6 +50,8 @@
         return Value((int32_t)0);
     case LogicalTypeID::INT16:
         return Value((int16_t)0);
+    case LogicalTypeID::INT8:
+        return Value((int8_t)0);
     case LogicalTypeID::BOOL:
         return Value(true);
     case LogicalTypeID::DOUBLE:
@@ -106,6 +108,11 @@
     val.booleanVal = val_;
 }
 
+Value::Value(int8_t val_) : isNull_{false} {
+    dataType = std::make_unique<LogicalType>(LogicalTypeID::INT8);
+    val.int8Val = val_;
+}
+
 Value::Value(int16_t val_) : isNull_{false} {
     dataType = std::make_unique<LogicalType>(LogicalTypeID::INT16);
     val.int16Val = val_;
@@ -192,6 +199,9 @@
     case LogicalTypeID::INT16: {
         val.int16Val = *((int16_t*)value);
     } break;
+    case LogicalTypeID::INT8: {
+        val.int8Val = *((int8_t*)value);
+    } break;
     case LogicalTypeID::BOOL: {
         val.booleanVal = *((bool*)value);
     } break;
@@ -255,6 +265,9 @@
     case PhysicalTypeID::INT16: {
         val.int16Val = other.val.int16Val;
     } break;
+    case PhysicalTypeID::INT8: {
+        val.int8Val = other.val.int8Val;
+    } break;
     case PhysicalTypeID::DOUBLE: {
         val.doubleVal = other.val.doubleVal;
     } break;
@@ -297,6 +310,8 @@
         return TypeUtils::toString(val.int32Val);
     case LogicalTypeID::INT16:
         return TypeUtils::toString(val.int16Val);
+    case LogicalTypeID::INT8:
+        return TypeUtils::toString(val.int8Val);
     case LogicalTypeID::DOUBLE:
         return TypeUtils::toString(val.doubleVal);
     case LogicalTypeID::FLOAT:
@@ -502,6 +517,9 @@
     case PhysicalTypeID::INT16: {
         SerDeser::serializeValue(val.int16Val, fileInfo, offset);
     } break;
+    case PhysicalTypeID::INT8: {
+        SerDeser::serializeValue(val.int8Val, fileInfo, offset);
+    } break;
     case PhysicalTypeID::DOUBLE: {
         SerDeser::serializeValue(val.doubleVal, fileInfo, offset);
     } break;
@@ -549,6 +567,9 @@
     case PhysicalTypeID::INT16: {
         SerDeser::deserializeValue(val->val.int16Val, fileInfo, offset);
     } break;
+    case PhysicalTypeID::INT8: {
+        SerDeser::deserializeValue(val->val.int8Val, fileInfo, offset);
+    } break;
     case PhysicalTypeID::DOUBLE: {
         SerDeser::deserializeValue(val->val.doubleVal, fileInfo, offset);
     } break;

src/common/vector/value_vector.cpp

@@ -140,6 +140,9 @@
     case PhysicalTypeID::INT16: {
         memcpy(dstValue, &value.val.int16Val, numBytesPerValue);
     } break;
+    case PhysicalTypeID::INT8: {
+        memcpy(dstValue, &value.val.int8Val, numBytesPerValue);
+    } break;
     case PhysicalTypeID::DOUBLE: {
         memcpy(dstValue, &value.val.doubleVal, numBytesPerValue);
     } break;
@@ -226,6 +229,9 @@
     case PhysicalTypeID::INT16: {
         value->val.int16Val = getValue<int16_t>(pos);
     } break;
+    case PhysicalTypeID::INT8: {
+        value->val.int8Val = getValue<int8_t>(pos);
+    } break;
     case PhysicalTypeID::DOUBLE: {
         value->val.doubleVal = getValue<double_t>(pos);
     } break;
@@ -345,6 +351,7 @@
 template void ValueVector::setValue<int64_t>(uint32_t pos, int64_t val);
 template void ValueVector::setValue<int32_t>(uint32_t pos, int32_t val);
 template void ValueVector::setValue<int16_t>(uint32_t pos, int16_t val);
+template void ValueVector::setValue<int8_t>(uint32_t pos, int8_t val);
 template void ValueVector::setValue<double_t>(uint32_t pos, double_t val);
 template void ValueVector::setValue<float_t>(uint32_t pos, float_t val);
 template void ValueVector::setValue<hash_t>(uint32_t pos, hash_t val);

src/include/c_api/kuzu.h

@@ -165,12 +165,12 @@ KUZU_C_API typedef enum {
     KUZU_INT64 = 23,
     KUZU_INT32 = 24,
     KUZU_INT16 = 25,
-    KUZU_DOUBLE = 26,
-    KUZU_FLOAT = 27,
-    KUZU_DATE = 28,
-    KUZU_TIMESTAMP = 29,
-    KUZU_INTERVAL = 30,
-    KUZU_FIXED_LIST = 31,
+    KUZU_DOUBLE = 27,
+    KUZU_FLOAT = 28,
+    KUZU_DATE = 29,
+    KUZU_TIMESTAMP = 30,
+    KUZU_INTERVAL = 31,
+    KUZU_FIXED_LIST = 32,
     KUZU_INTERNAL_ID = 40,
     KUZU_ARROW_COLUMN = 41,
     // variable size types

src/include/common/type_utils.h

@@ -23,7 +23,14 @@ class StringCastUtils {
         if (iss.str().empty()) {
             throw ConversionException{"Empty string."};
         }
-        iss >> result;
+
+        if constexpr (std::is_same_v<int8_t, T>) {
+            int val;
+            iss >> val; // C++ will recognize int8 as char if we don't separate this case.
+            result = val;
+        } else
+            iss >> result;
+
         if (iss.fail() || !iss.eof()) {
             return false;
         }
@@ -51,6 +58,7 @@ class TypeUtils {
     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(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) {

src/include/common/types/types.h

@@ -78,12 +78,13 @@ KUZU_API enum class LogicalTypeID : uint8_t {
     INT64 = 23,
     INT32 = 24,
     INT16 = 25,
-    DOUBLE = 26,
-    FLOAT = 27,
-    DATE = 28,
-    TIMESTAMP = 29,
-    INTERVAL = 30,
-    FIXED_LIST = 31,
+    INT8 = 26,
+    DOUBLE = 27,
+    FLOAT = 28,
+    DATE = 29,
+    TIMESTAMP = 30,
+    INTERVAL = 31,
+    FIXED_LIST = 32,
 
     INTERNAL_ID = 40,
 
@@ -105,9 +106,10 @@ enum class PhysicalTypeID : uint8_t {
     INT64 = 2,
     INT32 = 3,
     INT16 = 4,
-    DOUBLE = 5,
-    FLOAT = 6,
-    INTERVAL = 7,
+    INT8 = 5,
+    DOUBLE = 6,
+    FLOAT = 7,
+    INTERVAL = 8,
     INTERNAL_ID = 9,
     ARROW_COLUMN = 10,
 

src/include/common/types/value.h

@@ -51,6 +51,8 @@ class Value {
      * @param val_ the int16_t value to set.
      * @return a Value with INT16 type and val_ value.
      */
+    KUZU_API explicit Value(int8_t val_);
+
     KUZU_API explicit Value(int16_t val_);
     /**
      * @param val_ the int32_t value to set.
@@ -200,6 +202,7 @@ class Value {
         int64_t int64Val;
         int32_t int32Val;
         int16_t int16Val;
+        int8_t int8Val;
         double doubleVal;
         float floatVal;
         interval_t intervalVal;
@@ -363,6 +366,15 @@ inline bool Value::getValue() const {
     return val.booleanVal;
 }
 
+/**
+ * @return int8 value.
+ */
+KUZU_API template<>
+inline int8_t Value::getValue() const {
+    assert(dataType->getLogicalTypeID() == LogicalTypeID::INT8);
+    return val.int8Val;
+}
+
 /**
  * @return int16 value.
  */
@@ -463,6 +475,15 @@ inline bool& Value::getValueReference() {
     return val.booleanVal;
 }
 
+/**
+ * @return the reference to the int8 value.
+ */
+KUZU_API template<>
+inline int8_t& Value::getValueReference() {
+    assert(dataType->getLogicalTypeID() == LogicalTypeID::INT8);
+    return val.int8Val;
+}
+
 /**
  * @return the reference to the int16 value.
  */
@@ -562,6 +583,11 @@ inline Value Value::createValue(bool val) {
     return Value(val);
 }
 
+KUZU_API template<>
+inline Value Value::createValue(int8_t val) {
+    return Value(val);
+}
+
 /**
  * @param val the int16 value
  * @return a Value with INT16 type and val value.

src/include/processor/operator/order_by/order_by_key_encoder.h

@@ -29,6 +29,8 @@ namespace processor {
 
 #define BSWAP16(x) ((uint16_t)((((uint16_t)(x)&0xff00) >> 8) | (((uint16_t)(x)&0x00ff) << 8)))
 
+#define BSWAP8(x) ((uint8_t)(x))
+
 // The OrderByKeyEncoder encodes all columns in the ORDER BY clause into a single binary sequence
 // that, when compared using memcmp will yield the correct overall sorting order. On little-endian
 // hardware, the least-significant byte is stored at the smallest address. To encode the sorting

src/include/storage/storage_info.h

@@ -12,11 +12,11 @@ using storage_version_t = uint64_t;
 
 struct StorageVersionInfo {
     static std::unordered_map<std::string, storage_version_t> getStorageVersionInfo() {
-        return {{"0.0.8.5", 19}, {"0.0.8.4", 19}, {"0.0.8.3", 19}, {"0.0.8.2", 19}, {"0.0.8.1", 18},
-            {"0.0.8", 17}, {"0.0.7.1", 16}, {"0.0.7", 15}, {"0.0.6.5", 14}, {"0.0.6.4", 13},
-            {"0.0.6.3", 12}, {"0.0.6.2", 11}, {"0.0.6.1", 10}, {"0.0.6", 9}, {"0.0.5", 8},
-            {"0.0.4", 7}, {"0.0.3.5", 6}, {"0.0.3.4", 5}, {"0.0.3.3", 4}, {"0.0.3.2", 3},
-            {"0.0.3.1", 2}, {"0.0.3", 1}};
+        return {{"0.0.8.6", 20}, {"0.0.8.5", 19}, {"0.0.8.4", 19}, {"0.0.8.3", 19}, {"0.0.8.2", 19},
+            {"0.0.8.1", 18}, {"0.0.8", 17}, {"0.0.7.1", 16}, {"0.0.7", 15}, {"0.0.6.5", 14},
+            {"0.0.6.4", 13}, {"0.0.6.3", 12}, {"0.0.6.2", 11}, {"0.0.6.1", 10}, {"0.0.6", 9},
+            {"0.0.5", 8}, {"0.0.4", 7}, {"0.0.3.5", 6}, {"0.0.3.4", 5}, {"0.0.3.3", 4},
+            {"0.0.3.2", 3}, {"0.0.3.1", 2}, {"0.0.3", 1}};
     }
 
     static storage_version_t getStorageVersion();

src/include/storage/storage_structure/column.h

@@ -218,6 +218,7 @@ class ColumnFactory {
         case common::LogicalTypeID::INT64:
         case common::LogicalTypeID::INT32:
         case common::LogicalTypeID::INT16:
+        case common::LogicalTypeID::INT8:
         case common::LogicalTypeID::DOUBLE:
         case common::LogicalTypeID::FLOAT:
         case common::LogicalTypeID::BOOL:

src/include/storage/storage_structure/lists/lists.h

@@ -250,6 +250,7 @@ class ListsFactory {
         case common::LogicalTypeID::INT64:
         case common::LogicalTypeID::INT32:
         case common::LogicalTypeID::INT16:
+        case common::LogicalTypeID::INT8:
         case common::LogicalTypeID::DOUBLE:
         case common::LogicalTypeID::FLOAT:
         case common::LogicalTypeID::BOOL:

src/processor/operator/order_by/order_by_key_encoder.cpp

@@ -216,6 +216,10 @@ void OrderByKeyEncoder::getEncodingFunction(PhysicalTypeID physicalType, encode_
         func = encodeTemplate<int16_t>;
         return;
     }
+    case PhysicalTypeID::INT8: {
+        func = encodeTemplate<int8_t>;
+        return;
+    }
     case PhysicalTypeID::DOUBLE: {
         func = encodeTemplate<double_t>;
         return;
@@ -237,6 +241,15 @@ void OrderByKeyEncoder::getEncodingFunction(PhysicalTypeID physicalType, encode_
     }
 }
 
+template<>
+void OrderByKeyEncoder::encodeData(int8_t data, uint8_t* resultPtr, bool swapBytes) {
+    if (swapBytes) {
+        data = BSWAP8(data);
+    }
+    memcpy(resultPtr, (void*)&data, sizeof(data));
+    resultPtr[0] = flipSign(resultPtr[0]);
+}
+
 template<>
 void OrderByKeyEncoder::encodeData(int16_t data, uint8_t* resultPtr, bool swapBytes) {
     if (swapBytes) {