Merge pull request #1952 from kuzudb/remove-query-normalizer

remove query normalizer

src/binder/CMakeLists.txt

@@ -10,8 +10,7 @@ add_library(kuzu_binder
         bound_statement_result.cpp
         bound_statement_visitor.cpp
         expression_binder.cpp
-        expression_visitor.cpp
-        query_normalizer.cpp)
+        expression_visitor.cpp)
 
 set(ALL_OBJECT_FILES
         ${ALL_OBJECT_FILES} $<TARGET_OBJECTS:kuzu_binder>

src/binder/bind/bind_projection_clause.cpp

@@ -1,6 +1,8 @@
 #include "binder/binder.h"
 #include "binder/expression/literal_expression.h"
 #include "binder/expression_visitor.h"
+#include "binder/query/return_with_clause/bound_return_clause.h"
+#include "binder/query/return_with_clause/bound_with_clause.h"
 #include "parser/expression/parsed_property_expression.h"
 
 using namespace kuzu::common;

src/binder/bind/bind_query.cpp

@@ -1,62 +1,73 @@
 #include "binder/binder.h"
+#include "binder/query/return_with_clause/bound_return_clause.h"
+#include "binder/query/return_with_clause/bound_with_clause.h"
+#include "parser/query/return_with_clause/return_clause.h"
+#include "parser/query/return_with_clause/with_clause.h"
 
 using namespace kuzu::parser;
 
 namespace kuzu {
 namespace binder {
 
 std::unique_ptr<BoundRegularQuery> Binder::bindQuery(const RegularQuery& regularQuery) {
-    std::vector<std::unique_ptr<BoundSingleQuery>> boundSingleQueries;
+    std::vector<std::unique_ptr<NormalizedSingleQuery>> normalizedSingleQueries;
     for (auto i = 0u; i < regularQuery.getNumSingleQueries(); i++) {
         // Don't clear scope within bindSingleQuery() yet because it is also used for subquery
         // binding.
         scope->clear();
-        boundSingleQueries.push_back(bindSingleQuery(*regularQuery.getSingleQuery(i)));
-    }
-    validateUnionColumnsOfTheSameType(boundSingleQueries);
-    assert(!boundSingleQueries.empty());
-    auto statementResult =
-        boundSingleQueries[0]->hasReturnClause() ?
-            boundSingleQueries[0]->getReturnClause()->getStatementResult()->copy() :
-            BoundStatementResult::createEmptyResult();
+        normalizedSingleQueries.push_back(bindSingleQuery(*regularQuery.getSingleQuery(i)));
+    }
+    validateUnionColumnsOfTheSameType(normalizedSingleQueries);
+    assert(!normalizedSingleQueries.empty());
     auto boundRegularQuery = std::make_unique<BoundRegularQuery>(
-        regularQuery.getIsUnionAll(), std::move(statementResult));
-    for (auto& boundSingleQuery : boundSingleQueries) {
-        auto normalizedSingleQuery = QueryNormalizer::normalizeQuery(*boundSingleQuery);
+        regularQuery.getIsUnionAll(), normalizedSingleQueries[0]->getStatementResult()->copy());
+    for (auto& normalizedSingleQuery : normalizedSingleQueries) {
         validateReadNotFollowUpdate(*normalizedSingleQuery);
         boundRegularQuery->addSingleQuery(std::move(normalizedSingleQuery));
     }
     validateIsAllUnionOrUnionAll(*boundRegularQuery);
     return boundRegularQuery;
 }
 
-std::unique_ptr<BoundSingleQuery> Binder::bindSingleQuery(const SingleQuery& singleQuery) {
-    auto boundSingleQuery = std::make_unique<BoundSingleQuery>();
+std::unique_ptr<NormalizedSingleQuery> Binder::bindSingleQuery(const SingleQuery& singleQuery) {
+    auto normalizedSingleQuery = std::make_unique<NormalizedSingleQuery>();
     for (auto i = 0u; i < singleQuery.getNumQueryParts(); ++i) {
-        boundSingleQuery->addQueryPart(bindQueryPart(*singleQuery.getQueryPart(i)));
+        normalizedSingleQuery->appendQueryPart(bindQueryPart(*singleQuery.getQueryPart(i)));
     }
+    auto lastQueryPart = std::make_unique<NormalizedQueryPart>();
     for (auto i = 0u; i < singleQuery.getNumReadingClauses(); i++) {
-        boundSingleQuery->addReadingClause(bindReadingClause(*singleQuery.getReadingClause(i)));
+        lastQueryPart->addReadingClause(bindReadingClause(*singleQuery.getReadingClause(i)));
     }
     for (auto i = 0u; i < singleQuery.getNumUpdatingClauses(); ++i) {
-        boundSingleQuery->addUpdatingClause(bindUpdatingClause(*singleQuery.getUpdatingClause(i)));
+        lastQueryPart->addUpdatingClause(bindUpdatingClause(*singleQuery.getUpdatingClause(i)));
     }
+    std::unique_ptr<BoundStatementResult> statementResult;
     if (singleQuery.hasReturnClause()) {
-        boundSingleQuery->setReturnClause(bindReturnClause(*singleQuery.getReturnClause()));
+        auto boundReturnClause = bindReturnClause(*singleQuery.getReturnClause());
+        lastQueryPart->setProjectionBody(boundReturnClause->getProjectionBody()->copy());
+        statementResult = boundReturnClause->getStatementResult()->copy();
+    } else {
+        statementResult = BoundStatementResult::createEmptyResult();
     }
-    return boundSingleQuery;
+    normalizedSingleQuery->appendQueryPart(std::move(lastQueryPart));
+    normalizedSingleQuery->setStatementResult(std::move(statementResult));
+    return normalizedSingleQuery;
 }
 
-std::unique_ptr<BoundQueryPart> Binder::bindQueryPart(const QueryPart& queryPart) {
-    auto boundQueryPart = std::make_unique<BoundQueryPart>();
+std::unique_ptr<NormalizedQueryPart> Binder::bindQueryPart(const QueryPart& queryPart) {
+    auto normalizedQueryPart = std::make_unique<NormalizedQueryPart>();
     for (auto i = 0u; i < queryPart.getNumReadingClauses(); i++) {
-        boundQueryPart->addReadingClause(bindReadingClause(*queryPart.getReadingClause(i)));
+        normalizedQueryPart->addReadingClause(bindReadingClause(*queryPart.getReadingClause(i)));
     }
     for (auto i = 0u; i < queryPart.getNumUpdatingClauses(); ++i) {
-        boundQueryPart->addUpdatingClause(bindUpdatingClause(*queryPart.getUpdatingClause(i)));
+        normalizedQueryPart->addUpdatingClause(bindUpdatingClause(*queryPart.getUpdatingClause(i)));
+    }
+    auto boundWithClause = bindWithClause(*queryPart.getWithClause());
+    normalizedQueryPart->setProjectionBody(boundWithClause->getProjectionBody()->copy());
+    if (boundWithClause->hasWhereExpression()) {
+        normalizedQueryPart->setProjectionBodyPredicate(boundWithClause->getWhereExpression());
     }
-    boundQueryPart->setWithClause(bindWithClause(*queryPart.getWithClause()));
-    return boundQueryPart;
+    return normalizedQueryPart;
 }
 
 } // namespace binder

src/binder/bind/bind_reading_clause.cpp

@@ -1,6 +1,7 @@
 #include "binder/binder.h"
 #include "binder/expression/literal_expression.h"
 #include "binder/query/reading_clause/bound_in_query_call.h"
+#include "binder/query/reading_clause/bound_match_clause.h"
 #include "binder/query/reading_clause/bound_unwind_clause.h"
 #include "parser/query/reading_clause/in_query_call_clause.h"
 #include "parser/query/reading_clause/unwind_clause.h"

src/binder/binder.cpp

@@ -117,21 +117,21 @@ void Binder::validateOrderByFollowedBySkipOrLimitInWithClause(
 }
 
 void Binder::validateUnionColumnsOfTheSameType(
-    const std::vector<std::unique_ptr<BoundSingleQuery>>& boundSingleQueries) {
-    if (boundSingleQueries.size() <= 1) {
+    const std::vector<std::unique_ptr<NormalizedSingleQuery>>& normalizedSingleQueries) {
+    if (normalizedSingleQueries.size() <= 1) {
         return;
     }
-    auto expressionsToProject = boundSingleQueries[0]->getExpressionsToCollect();
-    for (auto i = 1u; i < boundSingleQueries.size(); i++) {
-        auto expressionsToProjectToCheck = boundSingleQueries[i]->getExpressionsToCollect();
-        if (expressionsToProject.size() != expressionsToProjectToCheck.size()) {
+    auto columns = normalizedSingleQueries[0]->getStatementResult()->getColumns();
+    for (auto i = 1u; i < normalizedSingleQueries.size(); i++) {
+        auto otherColumns = normalizedSingleQueries[i]->getStatementResult()->getColumns();
+        if (columns.size() != otherColumns.size()) {
             throw BinderException("The number of columns to union/union all must be the same.");
         }
         // Check whether the dataTypes in union expressions are exactly the same in each single
         // query.
-        for (auto j = 0u; j < expressionsToProject.size(); j++) {
-            ExpressionBinder::validateExpectedDataType(*expressionsToProjectToCheck[j],
-                expressionsToProject[j]->dataType.getLogicalTypeID());
+        for (auto j = 0u; j < columns.size(); j++) {
+            ExpressionBinder::validateExpectedDataType(
+                *otherColumns[j], columns[j]->dataType.getLogicalTypeID());
         }
     }
 }

src/include/binder/binder.h

@@ -5,7 +5,6 @@
 #include "expression_binder.h"
 #include "parser/copy.h"
 #include "parser/query/regular_query.h"
-#include "query_normalizer.h"
 
 namespace kuzu {
 namespace main {
@@ -17,6 +16,8 @@ namespace binder {
 class BoundCreateInfo;
 class BoundSetPropertyInfo;
 class BoundDeleteInfo;
+class BoundWithClause;
+class BoundReturnClause;
 
 // BinderScope keeps track of expressions in scope and their aliases. We maintain the order of
 // expressions in
@@ -110,8 +111,8 @@ class Binder {
 
     /*** bind query ***/
     std::unique_ptr<BoundRegularQuery> bindQuery(const parser::RegularQuery& regularQuery);
-    std::unique_ptr<BoundSingleQuery> bindSingleQuery(const parser::SingleQuery& singleQuery);
-    std::unique_ptr<BoundQueryPart> bindQueryPart(const parser::QueryPart& queryPart);
+    std::unique_ptr<NormalizedSingleQuery> bindSingleQuery(const parser::SingleQuery& singleQuery);
+    std::unique_ptr<NormalizedQueryPart> bindQueryPart(const parser::QueryPart& queryPart);
 
     /*** bind call ***/
     std::unique_ptr<BoundStatement> bindStandaloneCall(const parser::Statement& statement);
@@ -216,7 +217,7 @@ class Binder {
         const BoundProjectionBody& boundProjectionBody);
 
     static void validateUnionColumnsOfTheSameType(
-        const std::vector<std::unique_ptr<BoundSingleQuery>>& boundSingleQueries);
+        const std::vector<std::unique_ptr<NormalizedSingleQuery>>& normalizedSingleQueries);
 
     static void validateIsAllUnionOrUnionAll(const BoundRegularQuery& regularQuery);
 

src/include/binder/query/normalized_single_query.h

@@ -8,17 +8,20 @@ namespace binder {
 
 class NormalizedSingleQuery {
 public:
-    NormalizedSingleQuery() = default;
-    ~NormalizedSingleQuery() = default;
-
     inline void appendQueryPart(std::unique_ptr<NormalizedQueryPart> queryPart) {
         queryParts.push_back(std::move(queryPart));
     }
     inline uint32_t getNumQueryParts() const { return queryParts.size(); }
     inline NormalizedQueryPart* getQueryPart(uint32_t idx) const { return queryParts[idx].get(); }
 
+    inline void setStatementResult(std::unique_ptr<BoundStatementResult> result) {
+        statementResult = std::move(result);
+    }
+    inline BoundStatementResult* getStatementResult() const { return statementResult.get(); }
+
 private:
     std::vector<std::unique_ptr<NormalizedQueryPart>> queryParts;
+    std::unique_ptr<BoundStatementResult> statementResult;
 };
 
 } // namespace binder