Merge pull request #1896 from kuzudb/parser-reorg

Reorg parser module

src/binder/bind/bind_create_macro.cpp

@@ -1,7 +1,7 @@
 #include "binder/binder.h"
 #include "binder/macro/bound_create_macro.h"
 #include "common/string_utils.h"
-#include "parser/macro/create_macro.h"
+#include "parser/create_macro.h"
 
 namespace kuzu {
 namespace binder {

src/binder/bind/bind_standalone_call.cpp

@@ -1,6 +1,6 @@
 #include "binder/binder.h"
 #include "binder/call/bound_standalone_call.h"
-#include "parser/call/standalone_call.h"
+#include "parser/standalone_call.h"
 
 namespace kuzu {
 namespace binder {

src/include/function/scalar_macro_function.h

@@ -2,7 +2,7 @@
 
 #include <unordered_map>
 
-#include "parser/macro/create_macro.h"
+#include "parser/create_macro.h"
 
 namespace kuzu {
 namespace function {

src/include/parser/copy.h

@@ -3,6 +3,7 @@
 #include <string>
 #include <unordered_map>
 
+#include "common/copier_config/copier_config.h"
 #include "parser/expression/parsed_expression.h"
 #include "parser/query/regular_query.h"
 #include "parser/statement.h"

src/include/parser/transformer.h

@@ -1,13 +1,24 @@
 #pragma once
 
 #include "cypher_parser.h"
-#include "parser/macro/create_macro.h"
-#include "parser/query/graph_pattern/pattern_element.h"
-#include "parser/query/regular_query.h"
+#include "expression/parsed_expression.h"
+#include "statement.h"
 
 namespace kuzu {
 namespace parser {
 
+class RegularQuery;
+class SingleQuery;
+class QueryPart;
+class UpdatingClause;
+class ReadingClause;
+class WithClause;
+class ReturnClause;
+class ProjectionBody;
+class PatternElement;
+class NodePattern;
+class PatternElementChain;
+class RelPattern;
 struct ParsedCaseAlternative;
 
 class Transformer {

src/parser/CMakeLists.txt

@@ -1,10 +1,11 @@
 add_subdirectory(antlr_parser)
 add_subdirectory(expression)
 add_subdirectory(query)
-add_subdirectory(macro)
+add_subdirectory(transform)
 
 add_library(kuzu_parser
         OBJECT
+        create_macro.cpp
         parser.cpp
         transformer.cpp
         parsed_expression_visitor.cpp)

src/parser/macro/create_macro.cpp → src/parser/create_macro.cpp

@@ -1,4 +1,4 @@
-#include "parser/macro/create_macro.h"
+#include "parser/create_macro.h"
 
 namespace kuzu {
 namespace parser {

src/parser/transform/CMakeLists.txt

@@ -0,0 +1,16 @@
+add_library(kuzu_parser_transform
+        OBJECT
+        transform_copy.cpp
+        transform_ddl.cpp
+        transform_expression.cpp
+        transform_graph_pattern.cpp
+        transform_macro.cpp
+        transform_projection.cpp
+        transform_query.cpp
+        transform_reading_clause.cpp
+        transform_standalone_call.cpp
+        transform_updating_clause.cpp)
+
+set(ALL_OBJECT_FILES
+        ${ALL_OBJECT_FILES} $<TARGET_OBJECTS:kuzu_parser_transform>
+        PARENT_SCOPE)

src/parser/transform/transform_copy.cpp

@@ -0,0 +1,53 @@
+#include "parser/copy.h"
+#include "parser/transformer.h"
+
+namespace kuzu {
+namespace parser {
+
+std::unique_ptr<Statement> Transformer::transformCopyTo(CypherParser::KU_CopyTOContext& ctx) {
+    std::string filePath = transformStringLiteral(*ctx.StringLiteral());
+    auto regularQuery = transformQuery(*ctx.oC_Query());
+    return std::make_unique<CopyTo>(std::move(filePath), std::move(regularQuery));
+}
+
+std::unique_ptr<Statement> Transformer::transformCopyFromCSV(
+    CypherParser::KU_CopyFromCSVContext& ctx) {
+    auto filePaths = transformFilePaths(ctx.kU_FilePaths()->StringLiteral());
+    auto tableName = transformSchemaName(*ctx.oC_SchemaName());
+    auto parsingOptions = ctx.kU_ParsingOptions() ?
+                              transformParsingOptions(*ctx.kU_ParsingOptions()) :
+                              std::unordered_map<std::string, std::unique_ptr<ParsedExpression>>();
+    return std::make_unique<CopyFrom>(std::move(filePaths), std::move(tableName),
+        std::move(parsingOptions), common::CopyDescription::FileType::UNKNOWN);
+}
+
+std::unique_ptr<Statement> Transformer::transformCopyFromNPY(
+    CypherParser::KU_CopyFromNPYContext& ctx) {
+    auto filePaths = transformFilePaths(ctx.StringLiteral());
+    auto tableName = transformSchemaName(*ctx.oC_SchemaName());
+    auto parsingOptions = std::unordered_map<std::string, std::unique_ptr<ParsedExpression>>();
+    return std::make_unique<CopyFrom>(std::move(filePaths), std::move(tableName),
+        std::move(parsingOptions), common::CopyDescription::FileType::NPY);
+}
+
+std::vector<std::string> Transformer::transformFilePaths(
+    std::vector<antlr4::tree::TerminalNode*> stringLiteral) {
+    std::vector<std::string> csvFiles;
+    for (auto& csvFile : stringLiteral) {
+        csvFiles.push_back(transformStringLiteral(*csvFile));
+    }
+    return csvFiles;
+}
+
+std::unordered_map<std::string, std::unique_ptr<ParsedExpression>>
+Transformer::transformParsingOptions(CypherParser::KU_ParsingOptionsContext& ctx) {
+    std::unordered_map<std::string, std::unique_ptr<ParsedExpression>> copyOptions;
+    for (auto loadOption : ctx.kU_ParsingOption()) {
+        auto optionName = transformSymbolicName(*loadOption->oC_SymbolicName());
+        copyOptions.emplace(optionName, transformLiteral(*loadOption->oC_Literal()));
+    }
+    return copyOptions;
+}
+
+} // namespace parser
+} // namespace kuzu

src/parser/transform/transform_ddl.cpp

@@ -0,0 +1,118 @@
+#include "parser/ddl/add_property.h"
+#include "parser/ddl/create_node_clause.h"
+#include "parser/ddl/create_rel_clause.h"
+#include "parser/ddl/drop_property.h"
+#include "parser/ddl/drop_table.h"
+#include "parser/ddl/rename_property.h"
+#include "parser/ddl/rename_table.h"
+#include "parser/expression/parsed_literal_expression.h"
+#include "parser/transformer.h"
+
+namespace kuzu {
+namespace parser {
+
+std::unique_ptr<Statement> Transformer::transformDDL(CypherParser::KU_DDLContext& ctx) {
+    if (ctx.kU_CreateNode()) {
+        return transformCreateNodeClause(*ctx.kU_CreateNode());
+    } else if (root.oC_Statement()->kU_DDL()->kU_CreateRel()) {
+        return transformCreateRelClause(*root.oC_Statement()->kU_DDL()->kU_CreateRel());
+    } else if (root.oC_Statement()->kU_DDL()->kU_DropTable()) {
+        return transformDropTable(*root.oC_Statement()->kU_DDL()->kU_DropTable());
+    } else {
+        return transformAlterTable(*root.oC_Statement()->kU_DDL()->kU_AlterTable());
+    }
+}
+
+std::unique_ptr<Statement> Transformer::transformAlterTable(
+    CypherParser::KU_AlterTableContext& ctx) {
+    if (ctx.kU_AlterOptions()->kU_AddProperty()) {
+        return transformAddProperty(ctx);
+    } else if (ctx.kU_AlterOptions()->kU_DropProperty()) {
+        return transformDropProperty(ctx);
+    } else if (ctx.kU_AlterOptions()->kU_RenameTable()) {
+        return transformRenameTable(ctx);
+    } else {
+        return transformRenameProperty(ctx);
+    }
+}
+
+std::unique_ptr<Statement> Transformer::transformCreateNodeClause(
+    CypherParser::KU_CreateNodeContext& ctx) {
+    auto schemaName = transformSchemaName(*ctx.oC_SchemaName());
+    auto propertyDefinitions = transformPropertyDefinitions(*ctx.kU_PropertyDefinitions());
+    auto pkColName =
+        ctx.kU_CreateNodeConstraint() ? transformPrimaryKey(*ctx.kU_CreateNodeConstraint()) : "";
+    return std::make_unique<CreateNodeTableClause>(
+        std::move(schemaName), std::move(propertyDefinitions), pkColName);
+}
+
+std::unique_ptr<Statement> Transformer::transformCreateRelClause(
+    CypherParser::KU_CreateRelContext& ctx) {
+    auto schemaName = transformSchemaName(*ctx.oC_SchemaName(0));
+    auto propertyDefinitions = ctx.kU_PropertyDefinitions() ?
+                                   transformPropertyDefinitions(*ctx.kU_PropertyDefinitions()) :
+                                   std::vector<std::pair<std::string, std::string>>();
+    auto relMultiplicity =
+        ctx.oC_SymbolicName() ? transformSymbolicName(*ctx.oC_SymbolicName()) : "MANY_MANY";
+    return make_unique<CreateRelClause>(std::move(schemaName), std::move(propertyDefinitions),
+        relMultiplicity, transformSchemaName(*ctx.oC_SchemaName(1)),
+        transformSchemaName(*ctx.oC_SchemaName(2)));
+}
+
+std::unique_ptr<Statement> Transformer::transformDropTable(CypherParser::KU_DropTableContext& ctx) {
+    return std::make_unique<DropTable>(transformSchemaName(*ctx.oC_SchemaName()));
+}
+
+std::unique_ptr<Statement> Transformer::transformRenameTable(
+    CypherParser::KU_AlterTableContext& ctx) {
+    return std::make_unique<RenameTable>(transformSchemaName(*ctx.oC_SchemaName()),
+        transformSchemaName(*ctx.kU_AlterOptions()->kU_RenameTable()->oC_SchemaName()));
+}
+
+std::unique_ptr<Statement> Transformer::transformAddProperty(
+    CypherParser::KU_AlterTableContext& ctx) {
+    return std::make_unique<AddProperty>(transformSchemaName(*ctx.oC_SchemaName()),
+        transformPropertyKeyName(*ctx.kU_AlterOptions()->kU_AddProperty()->oC_PropertyKeyName()),
+        transformDataType(*ctx.kU_AlterOptions()->kU_AddProperty()->kU_DataType()),
+        ctx.kU_AlterOptions()->kU_AddProperty()->oC_Expression() ?
+            transformExpression(*ctx.kU_AlterOptions()->kU_AddProperty()->oC_Expression()) :
+            std::make_unique<ParsedLiteralExpression>(
+                std::make_unique<common::Value>(common::Value::createNullValue()), "NULL"));
+}
+
+std::unique_ptr<Statement> Transformer::transformDropProperty(
+    CypherParser::KU_AlterTableContext& ctx) {
+    return std::make_unique<DropProperty>(transformSchemaName(*ctx.oC_SchemaName()),
+        transformPropertyKeyName(*ctx.kU_AlterOptions()->kU_DropProperty()->oC_PropertyKeyName()));
+}
+
+std::unique_ptr<Statement> Transformer::transformRenameProperty(
+    CypherParser::KU_AlterTableContext& ctx) {
+    return std::make_unique<RenameProperty>(transformSchemaName(*ctx.oC_SchemaName()),
+        transformPropertyKeyName(
+            *ctx.kU_AlterOptions()->kU_RenameProperty()->oC_PropertyKeyName()[0]),
+        transformPropertyKeyName(
+            *ctx.kU_AlterOptions()->kU_RenameProperty()->oC_PropertyKeyName()[1]));
+}
+
+std::vector<std::pair<std::string, std::string>> Transformer::transformPropertyDefinitions(
+    CypherParser::KU_PropertyDefinitionsContext& ctx) {
+    std::vector<std::pair<std::string, std::string>> propertyNameDataTypes;
+    for (auto property : ctx.kU_PropertyDefinition()) {
+        propertyNameDataTypes.emplace_back(
+            transformPropertyKeyName(*property->oC_PropertyKeyName()),
+            transformDataType(*property->kU_DataType()));
+    }
+    return propertyNameDataTypes;
+}
+
+std::string Transformer::transformDataType(CypherParser::KU_DataTypeContext& ctx) {
+    return ctx.getText();
+}
+
+std::string Transformer::transformPrimaryKey(CypherParser::KU_CreateNodeConstraintContext& ctx) {
+    return transformPropertyKeyName(*ctx.oC_PropertyKeyName());
+}
+
+} // namespace parser
+} // namespace kuzu