Merge pull request #1328 from kuzudb/int-types

Add INT16,INT32 datatype
kuzudb · Mar 1, 2023 · d151554 · d151554
2 parents f33e003 + 67f1c81
commit d151554
Show file tree

Hide file tree

Showing 53 changed files with 815 additions and 246 deletions.
diff --git a/dataset/copy-fault-tests/invalid-number/schema.cypher b/dataset/copy-fault-tests/invalid-number/schema.cypher
@@ -0,0 +1,2 @@
+create node table person (ID INT64, gender INT32, PRIMARY KEY (ID));
+create node table movie (ID INT64, length INT32, PRIMARY KEY (ID));
diff --git a/dataset/copy-fault-tests/invalid-number/vMovie.csv b/dataset/copy-fault-tests/invalid-number/vMovie.csv
@@ -0,0 +1 @@
+0,312abc
diff --git a/dataset/copy-fault-tests/invalid-number/vPerson.csv b/dataset/copy-fault-tests/invalid-number/vPerson.csv
@@ -0,0 +1,2 @@
+0,4
+2,2147483650
diff --git a/dataset/tinysnb/eMarries.csv b/dataset/tinysnb/eMarries.csv
@@ -1,3 +1,3 @@
-0,2,"[toronto]",
-3,5,,"long long long string"
-7,8,"[vancouver]","short str"
+0,2,"[toronto]","[4,5]",
+3,5,,"[2,5]","long long long string"
+7,8,"[vancouver]","[3,9]","short str"
diff --git a/dataset/tinysnb/eMeets.csv b/dataset/tinysnb/eMeets.csv
@@ -1,7 +1,7 @@
-0,2,"[7.82,3.54]"
-2,5,"[2.87,4.23]"
-3,7,"[3.65,8.44]"
-7,3,"[2.11,3.1]"
-8,3,"[2.2,9.0]"
-9,3,"[3,5.2]"
-10,2,"[3.5,1.1]"
+0,2,"[7.82,3.54]",5
+2,5,"[2.87,4.23]",2
+3,7,"[3.65,8.44]",3
+7,3,"[2.11,3.1]",7
+8,3,"[2.2,9.0]",9
+9,3,"[3,5.2]",11
+10,2,"[3.5,1.1]",13
diff --git a/dataset/tinysnb/eStudyAt.csv b/dataset/tinysnb/eStudyAt.csv
@@ -1,4 +1,4 @@
-from,to,YEAR,Places
-0,1,2021,"[wwAewsdndweusd,wek]"
-2,1,2020,"[anew,jsdnwusklklklwewsd]"
-8,1,2020,"[awndsnjwejwen,isuhuwennjnuhuhuwewe]"
+from,to,YEAR,Places,length
+0,1,2021,"[wwAewsdndweusd,wek]",5
+2,1,2020,"[anew,jsdnwusklklklwewsd]",55
+8,1,2020,"[awndsnjwejwen,isuhuwennjnuhuhuwewe]",22
diff --git a/dataset/tinysnb/schema.cypher b/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, PRIMARY KEY (ID));
-create node table movies (name STRING, PRIMARY KEY (name));
+create node table movies (name STRING, length INT32, 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[], MANY_ONE);
+create rel table studyAt (FROM person TO organisation, year INT64, places STRING[], length INT16,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], MANY_ONE);
-create rel table marries (FROM person TO person, usedAddress STRING[], note STRING, ONE_ONE);
+create rel table meets (FROM person TO person, location FLOAT[2], times INT, MANY_ONE);
+create rel table marries (FROM person TO person, usedAddress STRING[], address INT16[2], note STRING, ONE_ONE);
diff --git a/dataset/tinysnb/vMovies.csv b/dataset/tinysnb/vMovies.csv
@@ -1,3 +1,3 @@
-Sóló cón tu párejâ
-The 😂😃🧘🏻‍♂️🌍🌦️🍞🚗 movie
-Roma
+Sóló cón tu párejâ,126
+The 😂😃🧘🏻‍♂️🌍🌦️🍞🚗 movie,2544
+Roma,298
diff --git a/src/binder/expression_binder.cpp b/src/binder/expression_binder.cpp
@@ -88,7 +88,8 @@ std::shared_ptr<Expression> ExpressionBinder::implicitCast(
         auto uniqueName = ScalarFunctionExpression::getUniqueName(functionName, children);
         return std::make_shared<ScalarFunctionExpression>(functionName, FUNCTION,
             DataType{targetType.typeID}, std::move(children),
-            VectorCastOperations::bindExecFunc(expression->dataType.typeID, targetType.typeID),
+            VectorCastOperations::bindImplicitCastFunc(
+                expression->dataType.typeID, targetType.typeID),
             nullptr /* selectFunc */, std::move(uniqueName));
     } else {
         throw common::BinderException("Expression " + expression->toString() + " has data type " +

diff --git a/src/common/csv_reader/csv_reader.cpp b/src/common/csv_reader/csv_reader.cpp
@@ -218,12 +218,12 @@ bool CSVReader::hasNextTokenOrError() {
 
 int64_t CSVReader::getInt64() {
     setNextTokenIsProcessed();
-    return TypeUtils::convertToInt64(line + linePtrStart);
+    return TypeUtils::convertStringToNumber<int64_t>(line + linePtrStart);
 }
 
 double_t CSVReader::getDouble() {
     setNextTokenIsProcessed();
-    return TypeUtils::convertFloatingPointNumber<double_t>(line + linePtrStart, DOUBLE);
+    return TypeUtils::convertStringToNumber<double_t>(line + linePtrStart);
 }
 
 uint8_t CSVReader::getBoolean() {

diff --git a/src/common/type_utils.cpp b/src/common/type_utils.cpp
@@ -1,26 +1,11 @@
 #include "common/type_utils.h"
 
-#include <cerrno>
-#include <climits>
-
 #include "common/exception.h"
 #include "common/utils.h"
 
 namespace kuzu {
 namespace common {
 
-int64_t TypeUtils::convertToInt64(const char* data) {
-    char* eptr;
-    errno = 0;
-    auto retVal = strtoll(data, &eptr, 10);
-    throwConversionExceptionIfNoOrNotEveryCharacterIsConsumed(data, eptr, INT64);
-    if ((LLONG_MAX == retVal || LLONG_MIN == retVal) && errno == ERANGE) {
-        throw ConversionException(
-            prefixConversionExceptionMessage(data, INT64) + " Input out of range.");
-    }
-    return retVal;
-}
-
 uint32_t TypeUtils::convertToUint32(const char* data) {
     std::istringstream iss(data);
     uint32_t val;
@@ -88,23 +73,5 @@ std::string TypeUtils::prefixConversionExceptionMessage(const char* data, DataTy
            Types::dataTypeToString(dataTypeID) + ".";
 }
 
-void TypeUtils::throwConversionExceptionIfNoOrNotEveryCharacterIsConsumed(
-    const char* data, const char* eptr, DataTypeID dataTypeID) {
-    if (data == eptr) {
-        throw ConversionException(prefixConversionExceptionMessage(data, dataTypeID) +
-                                  ". Invalid input. No characters consumed.");
-    }
-    if (*eptr != '\0') {
-        throw ConversionException(prefixConversionExceptionMessage(data, dataTypeID) +
-                                  " Not all characters were read. read from character " + *data +
-                                  " up to character: " + *eptr + ".");
-    }
-}
-
-void TypeUtils::throwConversionExceptionOutOfRange(const char* data, DataTypeID dataTypeID) {
-    throw ConversionException(
-        prefixConversionExceptionMessage(data, dataTypeID) + " Input out of range.");
-}
-
 } // namespace common
 } // namespace kuzu
diff --git a/src/common/types/types.cpp b/src/common/types/types.cpp
@@ -40,11 +40,13 @@ DataType::DataType(const DataType& other) {
     case INTERNAL_ID:
     case BOOL:
     case INT64:
+    case INT32:
+    case INT16:
     case DOUBLE:
+    case FLOAT:
     case DATE:
     case TIMESTAMP:
     case INTERVAL:
-    case FLOAT:
     case STRING: {
         typeID = other.typeID;
     } break;
@@ -58,13 +60,13 @@ DataType::DataType(DataType&& other) noexcept
       fixedNumElementsInList{other.fixedNumElementsInList} {}
 
 std::vector<DataTypeID> DataType::getNumericalTypeIDs() {
-    return std::vector<DataTypeID>{INT64, DOUBLE, FLOAT};
+    return std::vector<DataTypeID>{INT64, INT32, INT16, DOUBLE, FLOAT};
 }
 
 std::vector<DataTypeID> DataType::getAllValidTypeIDs() {
     // TODO(Ziyi): Add FIX_LIST type to allValidTypeID when we support functions on VAR_LIST.
-    return std::vector<DataTypeID>{
-        INTERNAL_ID, BOOL, INT64, DOUBLE, STRING, DATE, TIMESTAMP, INTERVAL, VAR_LIST, FLOAT};
+    return std::vector<DataTypeID>{INTERNAL_ID, BOOL, INT64, INT32, INT16, DOUBLE, STRING, DATE,
+        TIMESTAMP, INTERVAL, VAR_LIST, FLOAT};
 }
 
 DataType& DataType::operator=(const DataType& other) {
@@ -81,11 +83,13 @@ DataType& DataType::operator=(const DataType& other) {
     case INTERNAL_ID:
     case BOOL:
     case INT64:
+    case INT32:
+    case INT16:
     case DOUBLE:
+    case FLOAT:
     case DATE:
     case TIMESTAMP:
     case INTERVAL:
-    case FLOAT:
     case STRING: {
         typeID = other.typeID;
     } break;
@@ -110,11 +114,13 @@ bool DataType::operator==(const DataType& other) const {
     case INTERNAL_ID:
     case BOOL:
     case INT64:
+    case INT32:
+    case INT16:
     case DOUBLE:
+    case FLOAT:
     case DATE:
     case TIMESTAMP:
     case INTERVAL:
-    case FLOAT:
     case STRING:
         return typeID == other.typeID;
     default:
@@ -147,12 +153,14 @@ std::unique_ptr<DataType> DataType::copy() {
     case INTERNAL_ID:
     case BOOL:
     case INT64:
+    case INT32:
+    case INT16:
     case DOUBLE:
+    case FLOAT:
     case DATE:
     case TIMESTAMP:
     case INTERVAL:
     case STRING:
-    case FLOAT:
         return std::make_unique<DataType>(typeID);
     default:
         throw InternalException("Unsupported DataType: " + Types::dataTypeToString(typeID) + ".");
@@ -193,8 +201,16 @@ DataTypeID Types::dataTypeIDFromString(const std::string& dataTypeIDString) {
         return INTERNAL_ID;
     } else if ("INT64" == dataTypeIDString) {
         return INT64;
+    } else if ("INT32" == dataTypeIDString) {
+        return INT32;
+    } else if ("INT16" == dataTypeIDString) {
+        return INT16;
+    } else if ("INT" == dataTypeIDString) {
+        return INT32;
     } else if ("DOUBLE" == dataTypeIDString) {
         return DOUBLE;
+    } else if ("FLOAT" == dataTypeIDString) {
+        return FLOAT;
     } else if ("BOOLEAN" == dataTypeIDString) {
         return BOOL;
     } else if ("STRING" == dataTypeIDString) {
@@ -205,8 +221,6 @@ DataTypeID Types::dataTypeIDFromString(const std::string& dataTypeIDString) {
         return TIMESTAMP;
     } else if ("INTERVAL" == dataTypeIDString) {
         return INTERVAL;
-    } else if ("FLOAT" == dataTypeIDString) {
-        return FLOAT;
     } else {
         throw InternalException("Cannot parse dataTypeID: " + dataTypeIDString);
     }
@@ -225,12 +239,14 @@ std::string Types::dataTypeToString(const DataType& dataType) {
     case INTERNAL_ID:
     case BOOL:
     case INT64:
+    case INT32:
+    case INT16:
     case DOUBLE:
+    case FLOAT:
     case DATE:
     case TIMESTAMP:
     case INTERVAL:
     case STRING:
-    case FLOAT:
         return dataTypeToString(dataType.typeID);
     default:
         throw InternalException("Unsupported DataType: " + Types::dataTypeToString(dataType) + ".");
@@ -251,8 +267,14 @@ std::string Types::dataTypeToString(DataTypeID dataTypeID) {
         return "BOOL";
     case INT64:
         return "INT64";
+    case INT32:
+        return "INT32";
+    case INT16:
+        return "INT16";
     case DOUBLE:
         return "DOUBLE";
+    case FLOAT:
+        return "FLOAT";
     case DATE:
         return "DATE";
     case TIMESTAMP:
@@ -265,8 +287,6 @@ std::string Types::dataTypeToString(DataTypeID dataTypeID) {
         return "VAR_LIST";
     case FIXED_LIST:
         return "FIXED_LIST";
-    case FLOAT:
-        return "FLOAT";
     default:
         throw InternalException(
             "Unsupported DataType: " + Types::dataTypeToString(dataTypeID) + ".");
@@ -301,8 +321,14 @@ uint32_t Types::getDataTypeSize(DataTypeID dataTypeID) {
         return sizeof(uint8_t);
     case INT64:
         return sizeof(int64_t);
+    case INT32:
+        return sizeof(int32_t);
+    case INT16:
+        return sizeof(int16_t);
     case DOUBLE:
         return sizeof(double_t);
+    case FLOAT:
+        return sizeof(float_t);
     case DATE:
         return sizeof(date_t);
     case TIMESTAMP:
@@ -313,8 +339,6 @@ uint32_t Types::getDataTypeSize(DataTypeID dataTypeID) {
         return sizeof(ku_string_t);
     case VAR_LIST:
         return sizeof(ku_list_t);
-    case FLOAT:
-        return sizeof(float_t);
     default:
         throw InternalException(
             "Cannot infer the size of dataTypeID: " + dataTypeToString(dataTypeID) + ".");
@@ -328,13 +352,15 @@ uint32_t Types::getDataTypeSize(const DataType& dataType) {
     case INTERNAL_ID:
     case BOOL:
     case INT64:
+    case INT32:
+    case INT16:
     case DOUBLE:
+    case FLOAT:
     case DATE:
     case TIMESTAMP:
     case INTERVAL:
     case STRING:
     case VAR_LIST:
-    case FLOAT:
         return getDataTypeSize(dataType.typeID);
     default:
         throw InternalException(

diff --git a/src/common/types/value.cpp b/src/common/types/value.cpp
@@ -39,7 +39,11 @@ Value Value::createNullValue(DataType dataType) {
 Value Value::createDefaultValue(const DataType& dataType) {
     switch (dataType.typeID) {
     case INT64:
-        return Value(0);
+        return Value((int64_t)0);
+    case INT32:
+        return Value((int32_t)0);
+    case INT16:
+        return Value((int16_t)0);
     case BOOL:
         return Value(true);
     case DOUBLE:
@@ -69,8 +73,12 @@ Value::Value(bool val_) : dataType{BOOL}, isNull_{false} {
     val.booleanVal = val_;
 }
 
-Value::Value(int32_t val_) : dataType{INT64}, isNull_{false} {
-    val.int64Val = (int64_t)val_;
+Value::Value(int16_t val_) : dataType{INT16}, isNull_{false} {
+    val.int16Val = val_;
+}
+
+Value::Value(int32_t val_) : dataType{INT32}, isNull_{false} {
+    val.int32Val = val_;
 }
 
 Value::Value(int64_t val_) : dataType{INT64}, isNull_{false} {
@@ -136,6 +144,12 @@ void Value::copyValueFrom(const uint8_t* value) {
     case INT64: {
         val.int64Val = *((int64_t*)value);
     } break;
+    case INT32: {
+        val.int32Val = *((int32_t*)value);
+    } break;
+    case INT16: {
+        val.int16Val = *((int16_t*)value);
+    } break;
     case BOOL: {
         val.booleanVal = *((bool*)value);
     } break;
@@ -186,6 +200,12 @@ void Value::copyValueFrom(const Value& other) {
     case INT64: {
         val.int64Val = other.val.int64Val;
     } break;
+    case INT32: {
+        val.int32Val = other.val.int32Val;
+    } break;
+    case INT16: {
+        val.int16Val = other.val.int16Val;
+    } break;
     case DOUBLE: {
         val.doubleVal = other.val.doubleVal;
     } break;
@@ -238,6 +258,10 @@ std::string Value::toString() const {
         return TypeUtils::toString(val.booleanVal);
     case INT64:
         return TypeUtils::toString(val.int64Val);
+    case INT32:
+        return TypeUtils::toString(val.int32Val);
+    case INT16:
+        return TypeUtils::toString(val.int16Val);
     case DOUBLE:
         return TypeUtils::toString(val.doubleVal);
     case DATE:
@@ -307,6 +331,12 @@ std::vector<std::unique_ptr<Value>> Value::convertKUFixedListToVector(
     case common::DataTypeID::INT64: {
         putValuesIntoVector<int64_t>(fixedListResultVal, fixedList, numBytesPerElement);
     } break;
+    case common::DataTypeID::INT32: {
+        putValuesIntoVector<int32_t>(fixedListResultVal, fixedList, numBytesPerElement);
+    } break;
+    case common::DataTypeID::INT16: {
+        putValuesIntoVector<int16_t>(fixedListResultVal, fixedList, numBytesPerElement);
+    } break;
     case common::DataTypeID::DOUBLE: {
         putValuesIntoVector<double_t>(fixedListResultVal, fixedList, numBytesPerElement);
     } break;