Rename length() to size()

kuzudb · Oct 16, 2023 · d52b57a · d52b57a
1 parent c604332
commit d52b57a
Show file tree

Hide file tree

Showing 17 changed files with 137 additions and 137 deletions.
diff --git a/src/function/built_in_vector_functions.cpp b/src/function/built_in_vector_functions.cpp
@@ -447,7 +447,6 @@ void BuiltInVectorFunctions::registerStringFunctions() {
     vectorFunctions.insert({ENDS_WITH_FUNC_NAME, EndsWithVectorFunction::getDefinitions()});
     vectorFunctions.insert({LCASE_FUNC_NAME, LowerVectorFunction::getDefinitions()});
     vectorFunctions.insert({LEFT_FUNC_NAME, LeftVectorFunction::getDefinitions()});
-    vectorFunctions.insert({LENGTH_FUNC_NAME, LengthVectorFunction::getDefinitions()});
     vectorFunctions.insert({LOWER_FUNC_NAME, LowerVectorFunction::getDefinitions()});
     vectorFunctions.insert({LPAD_FUNC_NAME, LpadVectorFunction::getDefinitions()});
     vectorFunctions.insert({LTRIM_FUNC_NAME, LtrimVectorFunction::getDefinitions()});
@@ -512,7 +511,7 @@ void BuiltInVectorFunctions::registerCastFunctions() {
 void BuiltInVectorFunctions::registerListFunctions() {
     vectorFunctions.insert({LIST_CREATION_FUNC_NAME, ListCreationVectorFunction::getDefinitions()});
     vectorFunctions.insert({LIST_RANGE_FUNC_NAME, ListRangeVectorFunction::getDefinitions()});
-    vectorFunctions.insert({LIST_LEN_FUNC_NAME, ListLenVectorFunction::getDefinitions()});
+    vectorFunctions.insert({SIZE_FUNC_NAME, SizeVectorFunction::getDefinitions()});
     vectorFunctions.insert({LIST_EXTRACT_FUNC_NAME, ListExtractVectorFunction::getDefinitions()});
     vectorFunctions.insert({LIST_ELEMENT_FUNC_NAME, ListExtractVectorFunction::getDefinitions()});
     vectorFunctions.insert({LIST_CONCAT_FUNC_NAME, ListConcatVectorFunction::getDefinitions()});
@@ -559,7 +558,7 @@ void BuiltInVectorFunctions::registerMapFunctions() {
     vectorFunctions.insert({MAP_CREATION_FUNC_NAME, MapCreationVectorFunctions::getDefinitions()});
     vectorFunctions.insert({MAP_EXTRACT_FUNC_NAME, MapExtractVectorFunctions::getDefinitions()});
     vectorFunctions.insert({ELEMENT_AT_FUNC_NAME, MapExtractVectorFunctions::getDefinitions()});
-    vectorFunctions.insert({CARDINALITY_FUNC_NAME, ListLenVectorFunction::getDefinitions()});
+    vectorFunctions.insert({CARDINALITY_FUNC_NAME, SizeVectorFunction::getDefinitions()});
     vectorFunctions.insert({MAP_KEYS_FUNC_NAME, MapKeysVectorFunctions::getDefinitions()});
     vectorFunctions.insert({MAP_VALUES_FUNC_NAME, MapValuesVectorFunctions::getDefinitions()});
 }

diff --git a/src/function/vector_list_functions.cpp b/src/function/vector_list_functions.cpp
@@ -122,15 +122,17 @@ vector_function_definitions ListRangeVectorFunction::getDefinitions() {
     return result;
 }
 
-vector_function_definitions ListLenVectorFunction::getDefinitions() {
+vector_function_definitions SizeVectorFunction::getDefinitions() {
     vector_function_definitions result;
-    auto execFunc = UnaryExecFunction<list_entry_t, int64_t, ListLen>;
-    result.push_back(std::make_unique<VectorFunctionDefinition>(LIST_LEN_FUNC_NAME,
-        std::vector<LogicalTypeID>{LogicalTypeID::VAR_LIST}, LogicalTypeID::INT64, execFunc,
-        true /* isVarlength*/));
+    result.push_back(std::make_unique<VectorFunctionDefinition>(SIZE_FUNC_NAME,
+        std::vector<LogicalTypeID>{LogicalTypeID::VAR_LIST}, LogicalTypeID::INT64,
+        UnaryExecFunction<list_entry_t, int64_t, ListLen>, true /* isVarlength*/));
     result.push_back(std::make_unique<VectorFunctionDefinition>(CARDINALITY_FUNC_NAME,
-        std::vector<LogicalTypeID>{LogicalTypeID::MAP}, LogicalTypeID::INT64, execFunc,
-        true /* isVarlength*/));
+        std::vector<LogicalTypeID>{LogicalTypeID::MAP}, LogicalTypeID::INT64,
+        UnaryExecFunction<list_entry_t, int64_t, ListLen>, true /* isVarlength*/));
+    result.push_back(std::make_unique<VectorFunctionDefinition>(SIZE_FUNC_NAME,
+        std::vector<LogicalTypeID>{LogicalTypeID::STRING}, LogicalTypeID::INT64,
+        UnaryExecFunction<ku_string_t, int64_t, ListLen>, true /* isVarlength*/));
     return result;
 }
 

diff --git a/src/function/vector_string_functions.cpp b/src/function/vector_string_functions.cpp
@@ -5,7 +5,6 @@
 #include "function/string/functions/contains_function.h"
 #include "function/string/functions/ends_with_function.h"
 #include "function/string/functions/left_operation.h"
-#include "function/string/functions/length_function.h"
 #include "function/string/functions/lpad_function.h"
 #include "function/string/functions/regexp_extract_all_function.h"
 #include "function/string/functions/regexp_extract_function.h"
@@ -168,14 +167,6 @@ vector_function_definitions LeftVectorFunction::getDefinitions() {
     return definitions;
 }
 
-vector_function_definitions LengthVectorFunction::getDefinitions() {
-    vector_function_definitions definitions;
-    definitions.emplace_back(make_unique<VectorFunctionDefinition>(LENGTH_FUNC_NAME,
-        std::vector<LogicalTypeID>{LogicalTypeID::STRING}, LogicalTypeID::INT64,
-        UnaryExecFunction<ku_string_t, int64_t, Length>, false /* isVarLength */));
-    return definitions;
-}
-
 vector_function_definitions LpadVectorFunction::getDefinitions() {
     vector_function_definitions definitions;
     definitions.emplace_back(make_unique<VectorFunctionDefinition>(LPAD_FUNC_NAME,

diff --git a/src/include/common/expression_type.h b/src/include/common/expression_type.h
@@ -47,7 +47,6 @@ const std::string LIST_CREATION_FUNC_NAME = "LIST_CREATION";
 const std::string LIST_RANGE_FUNC_NAME = "RANGE";
 const std::string LIST_EXTRACT_FUNC_NAME = "LIST_EXTRACT";
 const std::string LIST_ELEMENT_FUNC_NAME = "LIST_ELEMENT";
-const std::string LIST_LEN_FUNC_NAME = "LEN";
 const std::string LIST_CONCAT_FUNC_NAME = "LIST_CONCAT";
 const std::string LIST_CAT_FUNC_NAME = "LIST_CAT";
 const std::string ARRAY_CONCAT_FUNC_NAME = "ARRAY_CONCAT";
@@ -174,6 +173,7 @@ const std::string REGEXP_MATCHES_FUNC_NAME = "REGEXP_MATCHES";
 const std::string REGEXP_REPLACE_FUNC_NAME = "REGEXP_REPLACE";
 const std::string REGEXP_EXTRACT_FUNC_NAME = "REGEXP_EXTRACT";
 const std::string REGEXP_EXTRACT_ALL_FUNC_NAME = "REGEXP_EXTRACT_ALL";
+const std::string SIZE_FUNC_NAME = "SIZE";
 
 // Date functions.
 const std::string DATE_PART_FUNC_NAME = "DATE_PART";

diff --git a/src/include/function/list/functions/list_len_function.h b/src/include/function/list/functions/list_len_function.h
@@ -4,16 +4,39 @@
 #include <cstring>
 
 #include "common/types/ku_list.h"
+#include "utf8proc.h"
 
 namespace kuzu {
 namespace function {
 
 struct ListLen {
 public:
-    static inline void operation(common::list_entry_t& input, int64_t& result) {
+    template<typename T>
+    static inline void operation(T& input, int64_t& result) {
         result = input.size;
     }
 };
 
+template<>
+inline void ListLen::operation(common::ku_string_t& input, int64_t& result) {
+    auto totalByteLength = input.len;
+    auto inputString = input.getAsString();
+    for (auto i = 0; i < totalByteLength; i++) {
+        if (inputString[i] & 0x80) {
+            int64_t length = 0;
+            // Use grapheme iterator to identify bytes of utf8 char and increment once for each
+            // char.
+            utf8proc::utf8proc_grapheme_callback(
+                inputString.c_str(), totalByteLength, [&](size_t start, size_t end) {
+                    length++;
+                    return true;
+                });
+            result = length;
+            return;
+        }
+    }
+    result = totalByteLength;
+}
+
 } // namespace function
 } // namespace kuzu
diff --git a/src/include/function/list/vector_list_functions.h b/src/include/function/list/vector_list_functions.h
@@ -203,7 +203,7 @@ struct ListRangeVectorFunction : public VectorListFunction {
         const binder::expression_vector& arguments, FunctionDefinition* definition);
 };
 
-struct ListLenVectorFunction : public VectorListFunction {
+struct SizeVectorFunction : public VectorListFunction {
     static vector_function_definitions getDefinitions();
 };
 

diff --git a/src/include/function/string/functions/array_extract_function.h b/src/include/function/string/functions/array_extract_function.h
@@ -4,7 +4,7 @@
 #include <cstring>
 
 #include "common/types/ku_string.h"
-#include "length_function.h"
+#include "function/list/functions/list_len_function.h"
 #include "substr_function.h"
 
 namespace kuzu {
@@ -19,7 +19,7 @@ struct ArrayExtract {
         }
         auto stringVal = str.getAsString();
         int64_t strLen;
-        Length::operation(str, strLen);
+        ListLen::operation(str, strLen);
         auto idxPos = idx > 0 ? std::min(idx, strLen) : std::max(strLen + idx, (int64_t)0) + 1;
         auto startPos = idxPos - 1;
         auto endPos = startPos + 1;

diff --git a/src/include/function/string/functions/left_operation.h b/src/include/function/string/functions/left_operation.h
@@ -4,7 +4,7 @@
 #include <cstring>
 
 #include "common/types/ku_string.h"
-#include "length_function.h"
+#include "function/list/functions/list_len_function.h"
 #include "substr_function.h"
 
 namespace kuzu {
@@ -15,7 +15,7 @@ struct Left {
     static inline void operation(common::ku_string_t& left, int64_t& right,
         common::ku_string_t& result, common::ValueVector& resultValueVector) {
         int64_t leftLen;
-        Length::operation(left, leftLen);
+        ListLen::operation(left, leftLen);
         int64_t len =
             (right > -1) ? std::min(leftLen, right) : std::max(leftLen + right, (int64_t)0);
         SubStr::operation(left, 1, len, result, resultValueVector);

diff --git a/src/include/function/string/functions/length_function.h b/src/include/function/string/functions/length_function.h
diff --git a/src/include/function/string/functions/right_function.h b/src/include/function/string/functions/right_function.h
@@ -4,7 +4,7 @@
 #include <cstring>
 
 #include "common/types/ku_string.h"
-#include "length_function.h"
+#include "function/list/functions/list_len_function.h"
 #include "substr_function.h"
 
 namespace kuzu {
@@ -15,7 +15,7 @@ struct Right {
     static inline void operation(common::ku_string_t& left, int64_t& right,
         common::ku_string_t& result, common::ValueVector& resultValueVector) {
         int64_t leftLen;
-        Length::operation(left, leftLen);
+        ListLen::operation(left, leftLen);
         int64_t len =
             (right > -1) ? std::min(leftLen, right) : std::max(leftLen + right, (int64_t)0);
         SubStr::operation(left, leftLen - len + 1, len, result, resultValueVector);

diff --git a/src/include/function/string/vector_string_functions.h b/src/include/function/string/vector_string_functions.h
@@ -71,10 +71,6 @@ struct LeftVectorFunction : public VectorStringFunction {
     static vector_function_definitions getDefinitions();
 };
 
-struct LengthVectorFunction : public VectorStringFunction {
-    static vector_function_definitions getDefinitions();
-};
-
 struct LowerVectorFunction : public VectorStringFunction {
     static inline vector_function_definitions getDefinitions() {
         return getUnaryStrFunctionDefintion<Lower>(common::LOWER_FUNC_NAME);

diff --git a/test/test_files/tinysnb/exception/list.test b/test/test_files/tinysnb/exception/list.test
@@ -25,6 +25,6 @@ Binder exception: Cannot bind LIST_CONCAT with parameter type INT64[] and STRING
 Binder exception: Cannot bind LIST_CREATION with parameter type INT64 and STRING.
 
 -CASE ListPrepareError
--STATEMENT MATCH (a:person) RETURN len($1)
+-STATEMENT MATCH (a:person) RETURN size($1)
 ---- error
 Binder exception: Cannot resolve recursive data type for expression $1.
diff --git a/test/test_files/tinysnb/function/case.test b/test/test_files/tinysnb/function/case.test
@@ -30,7 +30,7 @@ Hubert Blaine Wolfeschlegelsteinhausenbergerdorff
 7.100000
 
 -LOG CaseListTest
--STATEMENT MATCH (a:person) WHERE a.ID > 6 RETURN CASE WHEN len(a.workedHours) = 1 THEN a.courseScoresPerTerm END
+-STATEMENT MATCH (a:person) WHERE a.ID > 6 RETURN CASE WHEN size(a.workedHours) = 1 THEN a.courseScoresPerTerm END
 ---- 4