Parse update queries (#1354)

This commit extends SPARQL parser s.t. it can parse `SPARQL UPDATE` queries and store them as part of the `ParsedQuery`. The updates are not yet processed, but (as of now) still lead to an error message that updates are not yet supported.
ad-freiburg · Jun 5, 2024 · a2ce6fa · a2ce6fa
1 parent f9c3132
commit a2ce6fa
Show file tree

Hide file tree

Showing 8 changed files with 668 additions and 246 deletions.
diff --git a/src/parser/ParsedQuery.h b/src/parser/ParsedQuery.h
@@ -18,6 +18,7 @@
 #include "parser/PropertyPath.h"
 #include "parser/SelectClause.h"
 #include "parser/TripleComponent.h"
+#include "parser/UpdateClause.h"
 #include "parser/data/GroupKey.h"
 #include "parser/data/LimitOffsetClause.h"
 #include "parser/data/OrderKey.h"
@@ -49,75 +50,8 @@ class SparqlPrefix {
   bool operator==(const SparqlPrefix&) const = default;
 };
 
-inline bool isVariable(const string& elem) { return elem.starts_with("?"); }
-inline bool isVariable(const TripleComponent& elem) {
-  return elem.isVariable();
-}
-
-inline bool isVariable(const PropertyPath& elem) {
-  return elem._operation == PropertyPath::Operation::IRI &&
-         isVariable(elem._iri);
-}
-
 std::ostream& operator<<(std::ostream& out, const PropertyPath& p);
 
-// Data container for parsed triples from the where clause.
-// It is templated on the predicate type, see the instantiations below.
-template <typename Predicate>
-class SparqlTripleBase {
- public:
-  using AdditionalScanColumns = std::vector<std::pair<ColumnIndex, Variable>>;
-  SparqlTripleBase(TripleComponent s, Predicate p, TripleComponent o,
-                   AdditionalScanColumns additionalScanColumns = {})
-      : s_(std::move(s)),
-        p_(std::move(p)),
-        o_(std::move(o)),
-        additionalScanColumns_(std::move(additionalScanColumns)) {}
-
-  bool operator==(const SparqlTripleBase& other) const = default;
-  TripleComponent s_;
-  Predicate p_;
-  TripleComponent o_;
-  // The additional columns (e.g. patterns) that are to be attached when
-  // performing an index scan using this triple.
-  // TODO<joka921> On this level we should not store `ColumnIndex`, but the
-  // special predicate IRIs that are to be attached here.
-  std::vector<std::pair<ColumnIndex, Variable>> additionalScanColumns_;
-};
-
-// A triple where the predicate is a `TripleComponent`, so a fixed entity or a
-// variable, but not a property path.
-class SparqlTripleSimple : public SparqlTripleBase<TripleComponent> {
-  using Base = SparqlTripleBase<TripleComponent>;
-  using Base::Base;
-};
-
-// A triple where the predicate is a `PropertyPath` (which technically still
-// might be a variable or fixed entity in the current implementation).
-class SparqlTriple : public SparqlTripleBase<PropertyPath> {
- public:
-  using Base = SparqlTripleBase<PropertyPath>;
-  using Base::Base;
-
-  // ___________________________________________________________________________
-  SparqlTriple(TripleComponent s, const std::string& p_iri, TripleComponent o)
-      : Base{std::move(s), PropertyPath::fromIri(p_iri), std::move(o)} {}
-
-  // ___________________________________________________________________________
-  [[nodiscard]] string asString() const;
-
-  // Convert to a simple triple. Fails with an exception if the predicate
-  // actually is a property path.
-  SparqlTripleSimple getSimple() const {
-    AD_CONTRACT_CHECK(p_.isIri());
-    TripleComponent p =
-        isVariable(p_._iri)
-            ? TripleComponent{Variable{p_._iri}}
-            : TripleComponent(TripleComponent::Iri::fromIriref(p_._iri));
-    return {s_, p, o_, additionalScanColumns_};
-  }
-};
-
 // Forward declaration
 namespace parsedQuery {
 struct GraphPatternOperation;
@@ -132,6 +66,8 @@ class ParsedQuery {
 
   using ConstructClause = parsedQuery::ConstructClause;
 
+  using UpdateClause = parsedQuery::UpdateClause;
+
   ParsedQuery() = default;
 
   GraphPattern _rootGraphPattern;
@@ -147,7 +83,8 @@ class ParsedQuery {
 
   // explicit default initialisation because the constructor
   // of SelectClause is private
-  std::variant<SelectClause, ConstructClause> _clause{SelectClause{}};
+  std::variant<SelectClause, ConstructClause, UpdateClause> _clause{
+      SelectClause{}};
 
   [[nodiscard]] bool hasSelectClause() const {
     return std::holds_alternative<SelectClause>(_clause);
@@ -157,6 +94,10 @@ class ParsedQuery {
     return std::holds_alternative<ConstructClause>(_clause);
   }
 
+  [[nodiscard]] bool hasUpdateClause() const {
+    return std::holds_alternative<UpdateClause>(_clause);
+  }
+
   [[nodiscard]] decltype(auto) selectClause() const {
     return std::get<SelectClause>(_clause);
   }
@@ -165,6 +106,10 @@ class ParsedQuery {
     return std::get<ConstructClause>(_clause);
   }
 
+  [[nodiscard]] decltype(auto) updateClause() const {
+    return std::get<UpdateClause>(_clause);
+  }
+
   [[nodiscard]] decltype(auto) selectClause() {
     return std::get<SelectClause>(_clause);
   }
@@ -173,6 +118,10 @@ class ParsedQuery {
     return std::get<ConstructClause>(_clause);
   }
 
+  [[nodiscard]] decltype(auto) updateClause() {
+    return std::get<UpdateClause>(_clause);
+  }
+
   // Add a variable, that was found in the query body.
   void registerVariableVisibleInQueryBody(const Variable& variable);
 

diff --git a/src/parser/SparqlTriple.h b/src/parser/SparqlTriple.h
@@ -0,0 +1,91 @@
+//  Copyright 2024, University of Freiburg,
+//  Chair of Algorithms and Data Structures.
+//  Authors: Björn Buchhold ([email protected])
+//           Johannes Kalmbach <[email protected]>
+
+#pragma once
+
+#include <vector>
+
+#include "PropertyPath.h"
+#include "TripleComponent.h"
+#include "global/Id.h"
+#include "parser/data/Variable.h"
+
+inline bool isVariable(const string& elem) { return elem.starts_with("?"); }
+inline bool isVariable(const TripleComponent& elem) {
+  return elem.isVariable();
+}
+
+inline bool isVariable(const PropertyPath& elem) {
+  return elem._operation == PropertyPath::Operation::IRI &&
+         isVariable(elem._iri);
+}
+
+// Data container for parsed triples from the where clause.
+// It is templated on the predicate type, see the instantiations below.
+template <typename Predicate>
+class SparqlTripleBase {
+ public:
+  using AdditionalScanColumns = std::vector<std::pair<ColumnIndex, Variable>>;
+  SparqlTripleBase(TripleComponent s, Predicate p, TripleComponent o,
+                   AdditionalScanColumns additionalScanColumns = {})
+      : s_(std::move(s)),
+        p_(std::move(p)),
+        o_(std::move(o)),
+        additionalScanColumns_(std::move(additionalScanColumns)) {}
+
+  bool operator==(const SparqlTripleBase& other) const = default;
+  TripleComponent s_;
+  Predicate p_;
+  TripleComponent o_;
+  // The additional columns (e.g. patterns) that are to be attached when
+  // performing an index scan using this triple.
+  // TODO<joka921> On this level we should not store `ColumnIndex`, but the
+  // special predicate IRIs that are to be attached here.
+  std::vector<std::pair<ColumnIndex, Variable>> additionalScanColumns_;
+};
+
+// A triple where the predicate is a `TripleComponent`, so a fixed entity or a
+// variable, but not a property path.
+class SparqlTripleSimple : public SparqlTripleBase<TripleComponent> {
+  using Base = SparqlTripleBase<TripleComponent>;
+  using Base::Base;
+};
+
+// A triple where the predicate is a `PropertyPath` (which technically still
+// might be a variable or fixed entity in the current implementation).
+class SparqlTriple : public SparqlTripleBase<PropertyPath> {
+ public:
+  using Base = SparqlTripleBase<PropertyPath>;
+  using Base::Base;
+
+  // ___________________________________________________________________________
+  SparqlTriple(TripleComponent s, const std::string& iri, TripleComponent o)
+      : Base{std::move(s), PropertyPath::fromIri(iri), std::move(o)} {}
+
+  // ___________________________________________________________________________
+  [[nodiscard]] string asString() const;
+
+  // Convert to a simple triple. Fails with an exception if the predicate
+  // actually is a property path.
+  SparqlTripleSimple getSimple() const {
+    AD_CONTRACT_CHECK(p_.isIri());
+    TripleComponent p =
+        isVariable(p_._iri)
+            ? TripleComponent{Variable{p_._iri}}
+            : TripleComponent(TripleComponent::Iri::fromIriref(p_._iri));
+    return {s_, p, o_, additionalScanColumns_};
+  }
+
+  // Constructs SparqlTriple from a simple triple. Fails with an exception if
+  // the predicate is neither a variable nor an iri.
+  static SparqlTriple fromSimple(const SparqlTripleSimple& triple) {
+    AD_CONTRACT_CHECK(triple.p_.isVariable() || triple.p_.isIri());
+    PropertyPath p = triple.p_.isVariable()
+                         ? PropertyPath::fromVariable(triple.p_.getVariable())
+                         : PropertyPath::fromIri(
+                               triple.p_.getIri().toStringRepresentation());
+    return {triple.s_, p, triple.o_};
+  }
+};
diff --git a/src/parser/UpdateClause.h b/src/parser/UpdateClause.h
@@ -0,0 +1,21 @@
+//  Copyright 2024, University of Freiburg,
+//  Chair of Algorithms and Data Structures.
+//  Author: Julian Mundhahs ([email protected])
+
+#pragma once
+
+#include "parser/SelectClause.h"
+#include "parser/SparqlTriple.h"
+#include "parser/data/Types.h"
+
+namespace parsedQuery {
+struct UpdateClause : ClauseBase {
+  std::vector<SparqlTripleSimple> toInsert_;
+  std::vector<SparqlTripleSimple> toDelete_;
+
+  UpdateClause() = default;
+  UpdateClause(std::vector<SparqlTripleSimple> toInsert,
+               std::vector<SparqlTripleSimple> toDelete)
+      : toInsert_{std::move(toInsert)}, toDelete_{std::move(toDelete)} {}
+};
+}  // namespace parsedQuery
diff --git a/src/parser/data/GraphRef.h b/src/parser/data/GraphRef.h
@@ -0,0 +1,17 @@
+// Copyright 2024, University of Freiburg,
+//                 Chair of Algorithms and Data Structures.
+// Author: Julian Mundhahs ([email protected])
+
+#pragma once
+
+#include <variant>
+
+#include "parser/Iri.h"
+
+using GraphRef = TripleComponent::Iri;
+struct DEFAULT {};
+struct NAMED {};
+struct ALL {};
+
+using GraphRefAll = std::variant<GraphRef, DEFAULT, NAMED, ALL>;
+using GraphOrDefault = std::variant<GraphRef, DEFAULT>;