Implement Commenting on Tables

dataset/ldbc-sf01/copy.cypher

@@ -1,4 +1,4 @@
-COPY Comment FROM "dataset/ldbc-sf01/Comment.csv" (HEADER=true, DELIM='|');
+COPY `Comment` FROM "dataset/ldbc-sf01/Comment.csv" (HEADER=true, DELIM='|');
 COPY Forum FROM "dataset/ldbc-sf01/Forum.csv" (HEADER=true, DELIM='|');
 COPY Organisation FROM "dataset/ldbc-sf01/Organisation.csv" (HEADER=true, DELIM='|');
 COPY Person FROM "dataset/ldbc-sf01/Person.csv" (HEADER=true, DELIM='|');

dataset/ldbc-sf01/schema.cypher

@@ -1,16 +1,16 @@
-create node table Comment (id int64, creationDate INT64, locationIP STRING, browserUsed STRING, content STRING, length INT32, PRIMARY KEY (id));
+create node table `Comment` (id int64, creationDate INT64, locationIP STRING, browserUsed STRING, content STRING, length INT32, PRIMARY KEY (id));
 create node table Forum (id INT64, title STRING, creationDate INT64, PRIMARY KEY (id));
 create node table Organisation (id INT64, label STRING, name STRING, url STRING, PRIMARY KEY (id));
 create node table Person (id INT64, firstName STRING, lastName STRING, gender STRING, birthday INT64, creationDate INT64, locationIP STRING, browserUsed STRING, PRIMARY KEY (id));
 create node table Place (id INT64, name STRING, url STRING, label STRING, PRIMARY KEY (id));
 create node table Post (id INT64, imageFile STRING, creationDate INT64, locationIP STRING, browserUsed STRING, language STRING, content STRING, length INT32, PRIMARY KEY (id));
 create node table Tag (id INT64, name STRING, url STRING, PRIMARY KEY (id));
 create node table TagClass (id INT64, name STRING, url STRING, PRIMARY KEY (id));
-create rel table Comment_hasCreator_Person (FROM Comment TO Person, MANY_MANY);
-create rel table Comment_hasTag_Tag (FROM Comment TO Tag, MANY_MANY);
-create rel table Comment_isLocatedIn_Place (FROM Comment TO Place, MANY_ONE);
-create rel table Comment_replyOf_Comment (FROM Comment TO Comment, MANY_ONE);
-create rel table Comment_replyOf_Post (FROM Comment TO Post, MANY_ONE);
+create rel table Comment_hasCreator_Person (FROM `Comment` TO Person, MANY_MANY);
+create rel table Comment_hasTag_Tag (FROM `Comment` TO Tag, MANY_MANY);
+create rel table Comment_isLocatedIn_Place (FROM `Comment` TO Place, MANY_ONE);
+create rel table Comment_replyOf_Comment (FROM `Comment` TO `Comment`, MANY_ONE);
+create rel table Comment_replyOf_Post (FROM `Comment` TO Post, MANY_ONE);
 create rel table Forum_containerOf_Post (FROM Forum TO Post, ONE_MANY);
 create rel table Forum_hasMember_Person (FROM Forum TO Person, joinDate INT64, MANY_MANY);
 create rel table Forum_hasModerator_Person (FROM Forum TO Person, MANY_MANY);
@@ -19,7 +19,7 @@ create rel table Organisation_isLocatedIn_Place (FROM Organisation TO Place, MAN
 create rel table Person_hasInterest_Tag (FROM Person TO Tag, MANY_MANY);
 create rel table Person_isLocatedIn_Place (FROM Person TO Place, MANY_ONE);
 create rel table Person_knows_Person (FROM Person TO Person, creationDate INT64, MANY_MANY);
-create rel table Person_likes_Comment (FROM Person TO Comment, creationDate INT64, MANY_MANY);
+create rel table Person_likes_Comment (FROM Person TO `Comment`, creationDate INT64, MANY_MANY);
 create rel table Person_likes_Post (FROM Person TO Post, creationDate INT64, MANY_MANY);
 create rel table Person_studyAt_Organisation (FROM Person TO Organisation, classYear INT32, MANY_ONE);
 create rel table Person_workAt_Organisation (FROM Person TO Organisation, workFrom INT32, MANY_MANY);

dataset/sf-0.1/copy.cypher

@@ -1,5 +1,5 @@
 COPY City from "dataset/sf-0.1/City.csv" (HEADER=true, DELIM='|');
-COPY Comment from "dataset/sf-0.1/Comment.csv" (HEADER=true, DELIM='|');
+COPY `Comment` from "dataset/sf-0.1/Comment.csv" (HEADER=true, DELIM='|');
 COPY Company from "dataset/sf-0.1/Company.csv" (HEADER=true, DELIM='|');
 COPY Continent from "dataset/sf-0.1/Continent.csv" (HEADER=true, DELIM='|');
 COPY Country from "dataset/sf-0.1/Country.csv" (HEADER=true, DELIM='|');
@@ -33,4 +33,4 @@ COPY Post_hasTag_Tag from "dataset/sf-0.1/Post_hasTag_Tag.csv" (HEADER=true, DEL
 COPY Post_isLocatedIn_Country from "dataset/sf-0.1/Post_isLocatedIn_Country.csv" (HEADER=true, DELIM='|');
 COPY Tag_hasType_TagClass from "dataset/sf-0.1/Tag_hasType_TagClass.csv" (HEADER=true, DELIM='|');
 COPY TagClass_isSubclassOf_TagClass from "dataset/sf-0.1/TagClass_isSubclassOf_TagClass.csv" (HEADER=true, DELIM='|');
-COPY University_isLocatedIn_City from "dataset/sf-0.1/University_isLocatedIn_City.csv" (HEADER=true, DELIM='|');
+COPY University_isLocatedIn_City from "dataset/sf-0.1/University_isLocatedIn_City.csv" (HEADER=true, DELIM='|');

dataset/sf-0.1/schema.cypher

@@ -1,5 +1,5 @@
 create node table City (id int64, PRIMARY KEY (id));
-create node table Comment (id int64, PRIMARY KEY (id));
+create node table `Comment` (id int64, PRIMARY KEY (id));
 create node table Company (id int64, PRIMARY KEY (id));
 create node table Continent (id int64, PRIMARY KEY (id));
 create node table Country (id int64, PRIMARY KEY (id));
@@ -10,11 +10,11 @@ create node table Tag (id int64, PRIMARY KEY (id));
 create node table TagClass (id int64, PRIMARY KEY (id));
 create node table University (id int64, PRIMARY KEY (id));
 create rel table City_isPartOf_Country (FROM City TO Country, MANY_ONE);
-create rel table Comment_hasCreator_Person (FROM Comment TO Person, MANY_ONE);
-create rel table Comment_hasTag_Tag (FROM Comment TO Tag, MANY_MANY);
-create rel table Comment_isLocatedIn_Country (FROM Comment TO Country, MANY_ONE);
-create rel table Comment_replyOf_Comment (FROM Comment TO Comment, MANY_ONE);
-create rel table Comment_replyOf_Post (FROM Comment TO Post, MANY_ONE);
+create rel table Comment_hasCreator_Person (FROM `Comment` TO Person, MANY_ONE);
+create rel table Comment_hasTag_Tag (FROM `Comment` TO Tag, MANY_MANY);
+create rel table Comment_isLocatedIn_Country (FROM `Comment` TO Country, MANY_ONE);
+create rel table Comment_replyOf_Comment (FROM `Comment` TO `Comment`, MANY_ONE);
+create rel table Comment_replyOf_Post (FROM `Comment` TO Post, MANY_ONE);
 create rel table Company_isLocatedIn_Country (FROM Company TO Country, MANY_ONE);
 create rel table Country_isPartOf_Continent (FROM Country TO Continent, MANY_ONE);
 create rel table Forum_containerOf_Post (FROM Forum TO Post, MANY_MANY);
@@ -24,7 +24,7 @@ create rel table Forum_hasTag_Tag (FROM Forum TO Tag, MANY_MANY);
 create rel table Person_hasInterest_Tag (FROM Person TO Tag, MANY_MANY);
 create rel table Person_isLocatedIn_City (FROM Person TO City, MANY_ONE);
 create rel table Person_knows_Person (FROM Person TO Person, MANY_MANY);
-create rel table Person_likes_Comment (FROM Person TO Comment, MANY_MANY);
+create rel table Person_likes_Comment (FROM Person TO `Comment`, MANY_MANY);
 create rel table Person_likes_Post (FROM Person TO Post, MANY_MANY);
 create rel table Person_studyAt_University (FROM Person TO University, MANY_ONE);
 create rel table Person_workAt_Company (FROM Person TO Company, MANY_MANY);

src/antlr4/Cypher.g4

@@ -30,6 +30,11 @@ kU_StandaloneCall
 
 CALL : ( 'C' | 'c' ) ( 'A' | 'a' ) ( 'L' | 'l' ) ( 'L' | 'l' ) ;
 
+kU_CommentOn
+    : COMMENT SP ON SP TABLE SP oC_SchemaName SP IS SP StringLiteral ;
+
+COMMENT : ( 'C' | 'c' ) ( 'O' | 'o' ) ( 'M' | 'm' ) ( 'M' | 'm' ) ( 'E' | 'e' ) ( 'N' | 'n' ) ( 'T' | 't' ) ;
+
 kU_CreateMacro
     : CREATE SP MACRO SP oC_FunctionName SP? '(' SP? kU_PositionalArgs? SP? kU_DefaultArg? ( SP? ',' SP? kU_DefaultArg )* SP? ')' SP AS SP oC_Expression ;
 
@@ -175,6 +180,7 @@ oC_Statement
         | kU_CopyTO
         | kU_StandaloneCall
         | kU_CreateMacro
+        | kU_CommentOn
         | kU_Transaction ;
 
 kU_Transaction

src/binder/bind/CMakeLists.txt

@@ -1,16 +1,17 @@
 add_library(
         kuzu_binder_bind
         OBJECT
-        bind_standalone_call.cpp
-        bind_transaction.cpp
-        bind_create_macro.cpp
+        bind_comment_on.cpp
         bind_copy.cpp
+        bind_create_macro.cpp
         bind_ddl.cpp
         bind_explain.cpp
         bind_graph_pattern.cpp
         bind_projection_clause.cpp
         bind_query.cpp
         bind_reading_clause.cpp
+        bind_standalone_call.cpp
+        bind_transaction.cpp
         bind_updating_clause.cpp)
 
 set(ALL_OBJECT_FILES

src/binder/bind/bind_comment_on.cpp

@@ -0,0 +1,21 @@
+#include "binder/binder.h"
+#include "binder/bound_comment_on.h"
+#include "parser/comment_on.h"
+
+namespace kuzu {
+namespace binder {
+
+std::unique_ptr<BoundStatement> Binder::bindCommentOn(const parser::Statement& statement) {
+    auto& commentOnStatement = reinterpret_cast<const parser::CommentOn&>(statement);
+    auto tableName = commentOnStatement.getTable();
+    auto comment = commentOnStatement.getComment();
+
+    validateTableExist(catalog, tableName);
+    auto catalogContent = catalog.getReadOnlyVersion();
+    auto tableID = catalogContent->getTableID(tableName);
+
+    return std::make_unique<BoundCommentOn>(tableID, tableName, comment);
+}
+
+} // namespace binder
+} // namespace kuzu

src/binder/binder.cpp

@@ -45,6 +45,9 @@ std::unique_ptr<BoundStatement> Binder::bind(const Statement& statement) {
     case StatementType::STANDALONE_CALL: {
         boundStatement = bindStandaloneCall(statement);
     } break;
+    case StatementType::COMMENT_ON: {
+        boundStatement = bindCommentOn(statement);
+    } break;
     case StatementType::EXPLAIN: {
         boundStatement = bindExplain(statement);
     } break;

src/binder/bound_statement_visitor.cpp

@@ -37,6 +37,9 @@ void BoundStatementVisitor::visit(const kuzu::binder::BoundStatement& statement)
     case StatementType::STANDALONE_CALL: {
         visitStandaloneCall(statement);
     } break;
+    case StatementType::COMMENT_ON: {
+        visitCommentOn(statement);
+    } break;
     case StatementType::EXPLAIN: {
         visitExplain(statement);
     } break;

src/catalog/catalog.cpp

@@ -147,5 +147,10 @@ void Catalog::addScalarMacroFunction(
     catalogContentForWriteTrx->addScalarMacroFunction(std::move(name), std::move(macro));
 }
 
+void Catalog::setTableComment(table_id_t tableID, const std::string& comment) {
+    initCatalogContentForWriteTrxIfNecessary();
+    catalogContentForWriteTrx->getTableSchema(tableID)->setComment(comment);
+}
+
 } // namespace catalog
 } // namespace kuzu

src/catalog/catalog_content.cpp

@@ -145,6 +145,14 @@ Property* CatalogContent::getRelProperty(
     throw CatalogException("Cannot find rel property " + propertyName + ".");
 }
 
+std::vector<TableSchema*> CatalogContent::getTableSchemas() const {
+    std::vector<TableSchema*> allTableSchemas;
+    for (auto&& [_, schema] : tableSchemas) {
+        allTableSchemas.push_back(schema.get());
+    }
+    return allTableSchemas;
+}
+
 void CatalogContent::dropTableSchema(table_id_t tableID) {
     auto tableName = getTableName(tableID);
     tableNameToIDMap.erase(tableName);

src/catalog/table_schema.cpp

@@ -73,6 +73,7 @@ void TableSchema::serialize(FileInfo* fileInfo, uint64_t& offset) {
     SerDeser::serializeValue(tableID, fileInfo, offset);
     SerDeser::serializeValue(tableType, fileInfo, offset);
     SerDeser::serializeVectorOfPtrs(properties, fileInfo, offset);
+    SerDeser::serializeValue(comment, fileInfo, offset);
     SerDeser::serializeValue(nextPropertyID, fileInfo, offset);
     serializeInternal(fileInfo, offset);
 }
@@ -82,11 +83,13 @@ std::unique_ptr<TableSchema> TableSchema::deserialize(FileInfo* fileInfo, uint64
     table_id_t tableID;
     TableType tableType;
     std::vector<std::unique_ptr<Property>> properties;
+    std::string comment;
     property_id_t nextPropertyID;
     SerDeser::deserializeValue(tableName, fileInfo, offset);
     SerDeser::deserializeValue(tableID, fileInfo, offset);
     SerDeser::deserializeValue(tableType, fileInfo, offset);
     SerDeser::deserializeVectorOfPtrs(properties, fileInfo, offset);
+    SerDeser::deserializeValue(comment, fileInfo, offset);
     SerDeser::deserializeValue(nextPropertyID, fileInfo, offset);
     std::unique_ptr<TableSchema> result;
     switch (tableType) {
@@ -110,6 +113,7 @@ std::unique_ptr<TableSchema> TableSchema::deserialize(FileInfo* fileInfo, uint64
     result->tableID = tableID;
     result->tableType = tableType;
     result->properties = std::move(properties);
+    result->comment = std::move(comment);
     result->nextPropertyID = nextPropertyID;
     return result;
 }

src/function/built_in_table_functions.cpp

@@ -12,6 +12,7 @@ void BuiltInTableFunctions::registerTableFunctions() {
     tableFunctions.insert({TABLE_INFO_FUNC_NAME, TableInfoFunction::getDefinitions()});
     tableFunctions.insert({DB_VERSION_FUNC_NAME, DBVersionFunction::getDefinitions()});
     tableFunctions.insert({CURRENT_SETTING_FUNC_NAME, CurrentSettingFunction::getDefinitions()});
+    tableFunctions.insert({SHOW_TABLES_FUNC_NAME, ShowTablesFunction::getDefinitions()});
 }
 
 TableFunctionDefinition* BuiltInTableFunctions::mathTableFunction(const std::string& name) {

src/function/table_functions.cpp

@@ -97,5 +97,51 @@ std::unique_ptr<TableFuncBindData> CurrentSettingFunction::bindFunc(main::Client
         std::move(returnTypes), std::move(returnColumnNames), 1 /* one row result */);
 }
 
+void ShowTablesFunction::tableFunc(std::pair<offset_t, offset_t> morsel,
+    function::TableFuncBindData* bindData, std::vector<ValueVector*> outputVectors) {
+    auto tables = reinterpret_cast<function::ShowTablesBindData*>(bindData)->tables;
+    auto numTablesToOutput = morsel.second - morsel.first;
+    auto outVectorPos = 0;
+    for (auto i = 0u; i < numTablesToOutput; i++) {
+        auto tableSchema = tables[morsel.first + i];
+        outputVectors[0]->setValue(outVectorPos, tableSchema->tableName);
+
+        std::string typeString;
+        switch (tableSchema->tableType) {
+        case TableType::NODE:
+            typeString = "NODE";
+            break;
+        case TableType::REL:
+            typeString = "REL";
+            break;
+        case TableType::RDF:
+            typeString = "RDF";
+            break;
+        };
+        outputVectors[1]->setValue(outVectorPos, typeString);
+        outputVectors[2]->setValue(outVectorPos, tableSchema->comment);
+
+        outVectorPos++;
+    }
+    for (auto& outputVector : outputVectors) {
+        outputVector->state->selVector->selectedSize = outVectorPos;
+    }
+}
+
+std::unique_ptr<TableFuncBindData> ShowTablesFunction::bindFunc(main::ClientContext* context,
+    kuzu::function::TableFuncBindInput input, catalog::CatalogContent* catalog) {
+    std::vector<std::string> returnColumnNames;
+    std::vector<LogicalType> returnTypes;
+    returnColumnNames.emplace_back("TableName");
+    returnTypes.emplace_back(LogicalTypeID::STRING);
+    returnColumnNames.emplace_back("TableType");
+    returnTypes.emplace_back(LogicalTypeID::STRING);
+    returnColumnNames.emplace_back("TableComment");
+    returnTypes.emplace_back(LogicalTypeID::STRING);
+
+    return std::make_unique<ShowTablesBindData>(catalog->getTableSchemas(), std::move(returnTypes),
+        std::move(returnColumnNames), catalog->getTableCount());
+}
+
 } // namespace function
 } // namespace kuzu

src/include/binder/binder.h

@@ -138,6 +138,9 @@ class Binder {
     /*** bind transaction ***/
     std::unique_ptr<BoundStatement> bindTransaction(const parser::Statement& statement);
 
+    /*** bind comment on ***/
+    std::unique_ptr<BoundStatement> bindCommentOn(const parser::Statement& statement);
+
     /*** bind explain ***/
     std::unique_ptr<BoundStatement> bindExplain(const parser::Statement& statement);
 

src/include/binder/bound_comment_on.h

@@ -0,0 +1,26 @@
+#pragma once
+
+#include "bound_statement_result.h"
+#include "common/statement_type.h"
+
+namespace kuzu {
+namespace binder {
+
+class BoundCommentOn : public BoundStatement {
+public:
+    BoundCommentOn(common::table_id_t tableID, std::string tableName, std::string comment)
+        : BoundStatement{common::StatementType::COMMENT_ON,
+              BoundStatementResult::createSingleStringColumnResult()},
+          tableID(tableID), tableName(tableName), comment(comment) {}
+
+    inline common::table_id_t getTableID() const { return tableID; }
+    inline std::string getTableName() const { return tableName; }
+    inline std::string getComment() const { return comment; }
+
+private:
+    common::table_id_t tableID;
+    std::string tableName, comment;
+};
+
+} // namespace binder
+} // namespace kuzu

src/include/binder/bound_statement_visitor.h

@@ -26,6 +26,7 @@ class BoundStatementVisitor {
     virtual void visitRenameProperty(const BoundStatement& statement) {}
     virtual void visitCopy(const BoundStatement& statement) {}
     virtual void visitStandaloneCall(const BoundStatement& statement) {}
+    virtual void visitCommentOn(const BoundStatement& statement) {}
     virtual void visitExplain(const BoundStatement& statement);
     virtual void visitCreateMacro(const BoundStatement& statement) {}
     virtual void visitTransaction(const BoundStatement& statement) {}

src/include/binder/visitor/statement_read_write_analyzer.h

@@ -12,14 +12,15 @@ class StatementReadWriteAnalyzer : public BoundStatementVisitor {
     bool isReadOnly(const BoundStatement& statement);
 
 private:
-    void visitCreateTable(const BoundStatement& statement) override { readOnly = false; }
-    void visitDropTable(const BoundStatement& statement) override { readOnly = false; }
-    void visitRenameTable(const BoundStatement& statement) override { readOnly = false; }
     void visitAddProperty(const BoundStatement& statement) override { readOnly = false; }
-    void visitDropProperty(const BoundStatement& statement) override { readOnly = false; }
-    void visitRenameProperty(const BoundStatement& statement) override { readOnly = false; }
+    void visitCommentOn(const BoundStatement& statement) override { readOnly = false; }
     void visitCopy(const BoundStatement& statement) override { readOnly = false; }
     void visitCreateMacro(const BoundStatement& statement) override { readOnly = false; }
+    void visitCreateTable(const BoundStatement& statement) override { readOnly = false; }
+    void visitDropProperty(const BoundStatement& statement) override { readOnly = false; }
+    void visitDropTable(const BoundStatement& statement) override { readOnly = false; }
+    void visitRenameProperty(const BoundStatement& statement) override { readOnly = false; }
+    void visitRenameTable(const BoundStatement& statement) override { readOnly = false; }
     void visitQueryPart(const NormalizedQueryPart& queryPart) final;
 
 private:

src/include/catalog/catalog.h

@@ -74,6 +74,8 @@ class Catalog {
     void addScalarMacroFunction(
         std::string name, std::unique_ptr<function::ScalarMacroFunction> macro);
 
+    void setTableComment(common::table_id_t tableID, const std::string& comment);
+
     // TODO(Ziyi): pass transaction pointer here.
     inline function::ScalarMacroFunction* getScalarMacroFunction(const std::string& name) const {
         return catalogContentForReadOnlyTrx->macros.at(name).get();

src/include/catalog/catalog_content.h

@@ -69,6 +69,7 @@ class CatalogContent {
         assert(tableSchemas.contains(tableID));
         return tableSchemas.at(tableID)->getProperties();
     }
+    inline size_t getTableCount() const { return tableSchemas.size(); }
     inline std::vector<common::table_id_t> getNodeTableIDs() const {
         return getTableIDs(common::TableType::NODE);
     }
@@ -82,6 +83,8 @@ class CatalogContent {
         return getTableSchemas(common::TableType::REL);
     }
 
+    std::vector<TableSchema*> getTableSchemas() const;
+
     inline bool containMacro(const std::string& macroName) const {
         return macros.contains(macroName);
     }