[libSyntax] Store the token's text in the SyntaxArena

include/swift/Syntax/RawSyntax.h

@@ -150,7 +150,7 @@ typedef unsigned SyntaxNodeId;
 ///
 /// This is implementation detail - do not expose it in public API.
 class RawSyntax final
-    : private llvm::TrailingObjects<RawSyntax, RC<RawSyntax>, OwnedString> {
+    : private llvm::TrailingObjects<RawSyntax, RC<RawSyntax>> {
   friend TrailingObjects;
 
   /// The ID that shall be used for the next node that is created and does not
@@ -160,15 +160,17 @@ class RawSyntax final
   /// An ID of this node that is stable across incremental parses
   SyntaxNodeId NodeId;
 
+  mutable std::atomic<int> RefCount;
+
   /// If this node was allocated using a \c SyntaxArena's bump allocator, a
   /// reference to the arena to keep the underlying memory buffer of this node
   /// alive. If this is a \c nullptr, the node owns its own memory buffer.
   RC<SyntaxArena> Arena;
 
   union {
     struct {
-      // FIXME: Reduce TextLength to 30 bits so that common fits in 4 bytes?
-      /// Number of bytes this node takes up spelled out in the source code
+      /// Number of bytes this node takes up spelled out in the source code.
+      /// Always 0 if the node is missing.
       unsigned TextLength : 32;
       /// Whether this piece of syntax was actually present in the source.
       unsigned Presence : 1;
@@ -196,18 +198,19 @@ class RawSyntax final
                     "Only 64 bits reserved for standard syntax bits");
       uint64_t : bitmax(NumRawSyntaxBits, 64); // align to 16 bits
       /// The kind of token this "token" node represents.
+      const char *LeadingTrivia;
+      const char *TokenText;
+      const char *TrailingTrivia;
+      unsigned LeadingTriviaLength : 32;
+      unsigned TokenLength : 32;
+      unsigned TrailingTriviaLength : 32;
       unsigned TokenKind : 16;
-      StringRef LeadingTrivia;
-      StringRef TrailingTrivia;
     } Token;
   } Bits;
 
   size_t numTrailingObjects(OverloadToken<RC<RawSyntax>>) const {
     return isToken() ? 0 : Bits.Layout.NumChildren;
   }
-  size_t numTrailingObjects(OverloadToken<OwnedString>) const {
-    return isToken() ? 1 : 0;
-  }
 
   /// Constructor for creating layout nodes.
   /// If the node has been allocated inside the bump allocator of a
@@ -223,7 +226,7 @@ class RawSyntax final
   /// underlying storage.
   /// If \p NodeId is \c None, the next free NodeId is used, if it is passed,
   /// the caller needs to assure that the NodeId has not been used yet.
-  RawSyntax(tok TokKind, OwnedString Text, size_t TextLength,
+  RawSyntax(tok TokKind, StringRef Text, size_t TextLength,
             StringRef LeadingTrivia, StringRef TrailingTrivia,
             SourcePresence Presence, const RC<SyntaxArena> &Arena,
             llvm::Optional<SyntaxNodeId> NodeId);
@@ -240,8 +243,6 @@ class RawSyntax final
     return TextLength;
   }
 
-  mutable std::atomic<int> RefCount;
-
 public:
   ~RawSyntax();
 
@@ -286,15 +287,15 @@ class RawSyntax final
   }
 
   /// Make a raw "token" syntax node.
-  static RC<RawSyntax> make(tok TokKind, OwnedString Text, size_t TextLength,
+  static RC<RawSyntax> make(tok TokKind, StringRef Text, size_t TextLength,
                             StringRef LeadingTrivia, StringRef TrailingTrivia,
                             SourcePresence Presence,
                             const RC<SyntaxArena> &Arena = SyntaxArena::make(),
                             llvm::Optional<SyntaxNodeId> NodeId = llvm::None);
 
   /// Make a raw "token" syntax node that was allocated in \p Arena.
   static RC<RawSyntax>
-  makeAndCalcLength(tok TokKind, OwnedString Text, StringRef LeadingTrivia,
+  makeAndCalcLength(tok TokKind, StringRef Text, StringRef LeadingTrivia,
                     StringRef TrailingTrivia, SourcePresence Presence,
                     const RC<SyntaxArena> &Arena = SyntaxArena::make(),
                     llvm::Optional<SyntaxNodeId> NodeId = llvm::None) {
@@ -316,7 +317,7 @@ class RawSyntax final
 
   /// Make a missing raw "token" syntax node.
   static RC<RawSyntax>
-  missing(tok TokKind, OwnedString Text,
+  missing(tok TokKind, StringRef Text,
           const RC<SyntaxArena> &Arena = SyntaxArena::make()) {
     return make(TokKind, Text, /*TextLength=*/0, {}, {},
                 SourcePresence::Missing, Arena);
@@ -390,27 +391,24 @@ class RawSyntax final
     return static_cast<tok>(Bits.Token.TokenKind);
   }
 
-  /// Return the text of the token as an \c OwnedString. Keeping a reference to
-  /// this string will keep it alive even if the syntax node gets freed.
-  OwnedString getOwnedTokenText() const {
-    assert(isToken());
-    return *getTrailingObjects<OwnedString>();
-  }
-
   /// Return the text of the token as a reference. The referenced buffer may
   /// disappear when the syntax node gets freed.
-  StringRef getTokenText() const { return getOwnedTokenText().str(); }
+  StringRef getTokenText() const {
+    assert(isToken());
+    return StringRef(Bits.Token.TokenText, Bits.Token.TokenLength);
+  }
 
   /// Return the unparsed leading trivia of the token.
   StringRef getLeadingTrivia() const {
     assert(isToken());
-    return Bits.Token.LeadingTrivia;
+    return StringRef(Bits.Token.LeadingTrivia, Bits.Token.LeadingTriviaLength);
   }
 
   /// Return the unparsed trailing trivia of the token.
   StringRef getTrailingTrivia() const {
     assert(isToken());
-    return Bits.Token.TrailingTrivia;
+    return StringRef(Bits.Token.TrailingTrivia,
+                     Bits.Token.TrailingTriviaLength);
   }
 
   /// Return pieces that make up the leading trivia of the token.
@@ -434,17 +432,15 @@ class RawSyntax final
   /// Return a new token like this one, but with the given leading
   /// trivia instead.
   RC<RawSyntax> withLeadingTrivia(StringRef NewLeadingTrivia) const {
-    return makeAndCalcLength(getTokenKind(), getOwnedTokenText(),
-                             NewLeadingTrivia, getTrailingTrivia(),
-                             getPresence());
+    return makeAndCalcLength(getTokenKind(), getTokenText(), NewLeadingTrivia,
+                             getTrailingTrivia(), getPresence());
   }
 
   /// Return a new token like this one, but with the given trailing
   /// trivia instead.
   RC<RawSyntax> withTrailingTrivia(StringRef NewTrailingTrivia) const {
-    return makeAndCalcLength(getTokenKind(), getOwnedTokenText(),
-                             getLeadingTrivia(), NewTrailingTrivia,
-                             getPresence());
+    return makeAndCalcLength(getTokenKind(), getTokenText(), getLeadingTrivia(),
+                             NewTrailingTrivia, getPresence());
   }
 
   /// @}
@@ -503,7 +499,7 @@ class RawSyntax final
   /// Dump this piece of syntax recursively.
   void dump(llvm::raw_ostream &OS, unsigned Indent = 0) const;
 
-  static void Profile(llvm::FoldingSetNodeID &ID, tok TokKind, OwnedString Text,
+  static void Profile(llvm::FoldingSetNodeID &ID, tok TokKind, StringRef Text,
                       StringRef LeadingTrivia, StringRef TrailingTrivia);
 };
 

include/swift/Syntax/Serialization/SyntaxDeserialization.h

@@ -163,8 +163,8 @@ template <> struct MappingTraits<swift::RC<swift::RawSyntax>> {
       in.mapRequired("id", nodeIdString);
       unsigned nodeId = std::atoi(nodeIdString.data());
       value = swift::RawSyntax::makeAndCalcLength(
-          tokenKind, swift::OwnedString::makeRefCounted(text), leadingTrivia,
-          trailingTrivia, presence, swift::SyntaxArena::make(), nodeId);
+          tokenKind, text, leadingTrivia, trailingTrivia, presence,
+          swift::SyntaxArena::make(), nodeId);
     } else {
       swift::SyntaxKind kind;
       in.mapRequired("kind", kind);

include/swift/Syntax/SyntaxFactory.h.gyb

@@ -46,7 +46,7 @@ class SyntaxArena;
 struct SyntaxFactory {
   /// Make any kind of token.
   static TokenSyntax makeToken(tok Kind,
-      OwnedString Text, StringRef LeadingTrivia, StringRef TrailingTrivia,
+      StringRef Text, StringRef LeadingTrivia, StringRef TrailingTrivia,
       SourcePresence Presence,
       const RC<SyntaxArena> &Arena = SyntaxArena::make()
   );
@@ -108,7 +108,7 @@ struct SyntaxFactory {
             StringRef TrailingTrivia,
       const RC<SyntaxArena> &Arena = SyntaxArena::make());
 %   else:
-  static TokenSyntax make${token.name}(OwnedString Text,
+  static TokenSyntax make${token.name}(StringRef Text,
             StringRef LeadingTrivia, StringRef TrailingTrivia,
       const RC<SyntaxArena> &Arena = SyntaxArena::make());
 %   end
@@ -139,7 +139,7 @@ struct SyntaxFactory {
 
   /// Creates a TypeIdentifierSyntax with the provided name and leading/trailing
   /// trivia.
-  static TypeSyntax makeTypeIdentifier(OwnedString TypeName,
+  static TypeSyntax makeTypeIdentifier(StringRef TypeName,
       StringRef LeadingTrivia = {}, StringRef TrailingTrivia = {},
       const RC<SyntaxArena> &Arena = SyntaxArena::make()
   );

include/swift/Syntax/TokenSyntax.h

@@ -35,7 +35,7 @@ class TokenSyntax final : public Syntax {
 public:
   TokenSyntax(const SyntaxData Data) : Syntax(Data) {}
 
-  static TokenSyntax missingToken(const tok Kind, OwnedString Text) {
+  static TokenSyntax missingToken(const tok Kind, StringRef Text) {
     return makeRoot<TokenSyntax>(RawSyntax::missing(Kind, Text));
   }
 

lib/Parse/Parser.cpp

@@ -336,12 +336,11 @@ swift::tokenizeWithTrivia(const LangOptions &LangOpts, const SourceManager &SM,
       /*SplitTokens=*/ArrayRef<Token>(),
       [&](const Token &Tok, StringRef LeadingTrivia, StringRef TrailingTrivia) {
         CharSourceRange TokRange = Tok.getRange();
-        auto Text = OwnedString::makeRefCounted(Tok.getRawText());
         size_t TextLength = LeadingTrivia.size() + TokRange.getByteLength() +
                             TrailingTrivia.size();
-        auto ThisToken =
-            RawSyntax::make(Tok.getKind(), Text, TextLength, LeadingTrivia,
-                            TrailingTrivia, SourcePresence::Present);
+        auto ThisToken = RawSyntax::make(
+            Tok.getKind(), Tok.getRawText(), TextLength, LeadingTrivia,
+            TrailingTrivia, SourcePresence::Present);
 
         auto ThisTokenPos =
             RunningPos.advancedBy(ThisToken->getLeadingTriviaLength());

lib/Syntax/RawSyntax.cpp

@@ -111,8 +111,8 @@ RawSyntax::RawSyntax(SyntaxKind Kind, ArrayRef<RC<RawSyntax>> Layout,
                      size_t TextLength, SourcePresence Presence,
                      const RC<SyntaxArena> &Arena,
                      llvm::Optional<unsigned> NodeId)
-    : Arena(Arena), Bits({{unsigned(TextLength), unsigned(Presence), false}}),
-      RefCount(0) {
+    : RefCount(0), Arena(Arena),
+      Bits({{unsigned(TextLength), unsigned(Presence), false}}) {
   assert(Arena && "RawSyntax nodes must always be allocated in an arena");
   assert(Kind != SyntaxKind::Token &&
          "'token' syntax node must be constructed with dedicated constructor");
@@ -139,36 +139,41 @@ RawSyntax::RawSyntax(SyntaxKind Kind, ArrayRef<RC<RawSyntax>> Layout,
                           getTrailingObjects<RC<RawSyntax>>());
 }
 
-RawSyntax::RawSyntax(tok TokKind, OwnedString Text, size_t TextLength,
+RawSyntax::RawSyntax(tok TokKind, StringRef Text, size_t TextLength,
                      StringRef LeadingTrivia, StringRef TrailingTrivia,
                      SourcePresence Presence, const RC<SyntaxArena> &Arena,
                      llvm::Optional<unsigned> NodeId)
-    : Arena(Arena), Bits({{unsigned(TextLength), unsigned(Presence), true}}),
-      RefCount(0) {
+    : RefCount(0), Arena(Arena),
+      Bits({{unsigned(TextLength), unsigned(Presence), true}}) {
   assert(Arena && "RawSyntax nodes must always be allocated in an arena");
   copyToArenaIfNecessary(LeadingTrivia, Arena);
+  copyToArenaIfNecessary(Text, Arena);
   copyToArenaIfNecessary(TrailingTrivia, Arena);
 
+  if (Presence == SourcePresence::Missing) {
+    assert(TextLength == 0);
+  } else {
+    assert(TextLength ==
+           LeadingTrivia.size() + Text.size() + TrailingTrivia.size());
+  }
+
   if (NodeId.hasValue()) {
     this->NodeId = NodeId.getValue();
     NextFreeNodeId = std::max(this->NodeId + 1, NextFreeNodeId);
   } else {
     this->NodeId = NextFreeNodeId++;
   }
+  Bits.Token.LeadingTrivia = LeadingTrivia.data();
+  Bits.Token.TokenText = Text.data();
+  Bits.Token.TrailingTrivia = TrailingTrivia.data();
+  Bits.Token.LeadingTriviaLength = LeadingTrivia.size();
+  Bits.Token.TokenLength = Text.size();
+  Bits.Token.TrailingTriviaLength = TrailingTrivia.size();
   Bits.Token.TokenKind = unsigned(TokKind);
-  // FIXME: Copy the backing storage of the string into the arena
-  Bits.Token.LeadingTrivia = LeadingTrivia;
-  Bits.Token.TrailingTrivia = TrailingTrivia;
-
-  // Initialize token text.
-  ::new (static_cast<void *>(getTrailingObjects<OwnedString>()))
-      OwnedString(Text);
 }
 
 RawSyntax::~RawSyntax() {
-  if (isToken()) {
-    getTrailingObjects<OwnedString>()->~OwnedString();
-  } else {
+  if (!isToken()) {
     for (auto &child : getLayout())
       child.~RC<RawSyntax>();
   }
@@ -179,19 +184,19 @@ RC<RawSyntax> RawSyntax::make(SyntaxKind Kind, ArrayRef<RC<RawSyntax>> Layout,
                               const RC<SyntaxArena> &Arena,
                               llvm::Optional<unsigned> NodeId) {
   assert(Arena && "RawSyntax nodes must always be allocated in an arena");
-  auto size = totalSizeToAlloc<RC<RawSyntax>, OwnedString>(Layout.size(), 0);
+  auto size = totalSizeToAlloc<RC<RawSyntax>>(Layout.size());
   void *data = Arena->Allocate(size, alignof(RawSyntax));
   return RC<RawSyntax>(
       new (data) RawSyntax(Kind, Layout, TextLength, Presence, Arena, NodeId));
 }
 
-RC<RawSyntax> RawSyntax::make(tok TokKind, OwnedString Text, size_t TextLength,
+RC<RawSyntax> RawSyntax::make(tok TokKind, StringRef Text, size_t TextLength,
                               StringRef LeadingTrivia, StringRef TrailingTrivia,
                               SourcePresence Presence,
                               const RC<SyntaxArena> &Arena,
                               llvm::Optional<unsigned> NodeId) {
   assert(Arena && "RawSyntax nodes must always be allocated in an arena");
-  auto size = totalSizeToAlloc<RC<RawSyntax>, OwnedString>(0, 1);
+  auto size = totalSizeToAlloc<RC<RawSyntax>>(0);
   void *data = Arena->Allocate(size, alignof(RawSyntax));
   return RC<RawSyntax>(new (data)
                            RawSyntax(TokKind, Text, TextLength, LeadingTrivia,
@@ -306,9 +311,8 @@ void RawSyntax::dump(llvm::raw_ostream &OS, unsigned Indent) const {
   OS << ')';
 }
 
-void RawSyntax::Profile(llvm::FoldingSetNodeID &ID, tok TokKind,
-                        OwnedString Text, StringRef LeadingTrivia,
-                        StringRef TrailingTrivia) {
+void RawSyntax::Profile(llvm::FoldingSetNodeID &ID, tok TokKind, StringRef Text,
+                        StringRef LeadingTrivia, StringRef TrailingTrivia) {
   ID.AddInteger(unsigned(TokKind));
   ID.AddInteger(LeadingTrivia.size());
   ID.AddInteger(TrailingTrivia.size());
@@ -320,7 +324,7 @@ void RawSyntax::Profile(llvm::FoldingSetNodeID &ID, tok TokKind,
 #include "swift/Syntax/TokenKinds.def"
     break;
   default:
-    ID.AddString(Text.str());
+    ID.AddString(Text);
     break;
   }
   ID.AddString(LeadingTrivia);