libSyntax: parse codeblock syntax node. (#12771)

This commit teaches parser to generate code block syntax node. As a support for this, 
SyntaxParsingContext can be created by a single syntax kind, indicating the whole context 
should be parsed into a node of that given syntax. Another change is to bridge created syntax 
node with the given context kind. For instance, if a statement context results into an expression 
node, the expression node will be bridged to a statement by wrapping it with a ExpressionStmt 
node.

Author: nkcsgexi

include/swift/Syntax/SyntaxParsingContext.h

@@ -13,7 +13,8 @@
 #ifndef SWIFT_SYNTAX_PARSING_CONTEXT_H
 #define SWIFT_SYNTAX_PARSING_CONTEXT_H
 
- #include "swift/Syntax/Syntax.h"
+#include "swift/Basic/SourceLoc.h"
+#include "swift/Syntax/Syntax.h"
 
 namespace swift {
   class SourceLoc;
@@ -25,19 +26,28 @@ namespace syntax {
   struct RawSyntax;
   enum class SyntaxKind;
 
+enum class SyntaxContextKind: uint8_t{
+  Expr,
+  Decl,
+  Stmt,
+};
+
 /// The handler for parser to generate libSyntax entities.
 struct RawSyntaxInfo {
-  /// Start location of this syntax node.
-  SourceLoc StartLoc;
-
-  /// The number of tokens belong to the syntax node.
-  unsigned TokCount;
+  /// Start and end location of this syntax node.
+  SourceRange SyntaxRange;
 
   /// The raw node.
   RC<RawSyntax> RawNode;
+  RawSyntaxInfo(RC<RawSyntax> RawNode): RawNode(RawNode) {}
   RawSyntaxInfo(SourceLoc StartLoc, RC<RawSyntax> RawNode):
-    RawSyntaxInfo(StartLoc, 1, RawNode) {}
-  RawSyntaxInfo(SourceLoc StartLoc, unsigned TokCount, RC<RawSyntax> RawNode);
+    SyntaxRange(StartLoc), RawNode(RawNode) {}
+  RawSyntaxInfo(SourceRange SyntaxRange, RC<RawSyntax> RawNode):
+    SyntaxRange(SyntaxRange), RawNode(RawNode) {}
+
+  bool isImplicit() const { return SyntaxRange.isInvalid(); }
+  SourceLoc getStartLoc() const { return SyntaxRange.Start; }
+  SourceLoc getEndLoc() const { return SyntaxRange.End; }
 
   template <typename SyntaxNode>
   SyntaxNode makeSyntax() const { return make<SyntaxNode>(RawNode); }
@@ -46,6 +56,7 @@ struct RawSyntaxInfo {
   RC<RawSyntaxNode> getRaw() const {
     return RC<RawSyntaxNode>(cast<RawSyntaxNode>(RawNode));
   }
+  void brigeWithContext(SyntaxContextKind Kind);
 };
 
 enum class SyntaxParsingContextKind: uint8_t {
@@ -96,22 +107,28 @@ class SyntaxParsingContextRoot: public SyntaxParsingContext {
   };
 };
 
-enum class SyntaxContextKind: uint8_t{
-  Expr,
-  Decl,
-};
-
 // The base class for contexts that are created from a parent context.
 // The stack instance will set the context holder when the context
 // is firstly created and reset the context holder to the parent when
 // it's destructed.
 class SyntaxParsingContextChild: public SyntaxParsingContext {
   SyntaxParsingContext *Parent;
   SyntaxParsingContext *&ContextHolder;
-  const SyntaxContextKind Kind;
+  Optional<SyntaxContextKind> Kind;
+  Optional<SyntaxKind> KnownSyntax;
+  void makeNodeWhole(SyntaxKind Kind);
+  SyntaxParsingContextChild(SyntaxParsingContext *&ContextHolder,
+                            Optional<SyntaxContextKind> Kind,
+                            Optional<SyntaxKind> KnownSyntax);
 public:
   SyntaxParsingContextChild(SyntaxParsingContext *&ContextHolder,
-                            SyntaxContextKind Kind);
+    SyntaxContextKind Kind): SyntaxParsingContextChild(ContextHolder,
+                                                       Kind, None) {}
+
+  SyntaxParsingContextChild(SyntaxParsingContext *&ContextHolder,
+    SyntaxKind KnownSyntax): SyntaxParsingContextChild(ContextHolder,
+                             None, KnownSyntax) {};
+
   ~SyntaxParsingContextChild();
   void makeNode(SyntaxKind Kind) override;
   void addTokenSyntax(SourceLoc Loc) override;

lib/Parse/ParseStmt.cpp

@@ -22,12 +22,15 @@
 #include "swift/Parse/Lexer.h"
 #include "swift/Parse/CodeCompletionCallbacks.h"
 #include "swift/Subsystems.h"
+#include "swift/Syntax/TokenSyntax.h"
+#include "swift/Syntax/SyntaxParsingContext.h"
 #include "llvm/ADT/PointerUnion.h"
 #include "llvm/ADT/Twine.h"
 #include "llvm/Support/Compiler.h"
 #include "llvm/Support/SaveAndRestore.h"
 
 using namespace swift;
+using namespace swift::syntax;
 
 /// isStartOfStmt - Return true if the current token starts a statement.
 ///
@@ -223,7 +226,7 @@ void Parser::consumeTopLevelDecl(ParserPosition BeginParserPosition,
 ParserStatus Parser::parseBraceItems(SmallVectorImpl<ASTNode> &Entries,
                                      BraceItemListKind Kind,
                                      BraceItemListKind ConditionalBlockKind) {
-  
+  SyntaxParsingContextChild StmtListContext(SyntaxContext, SyntaxKind::StmtList);
   bool IsTopLevel = (Kind == BraceItemListKind::TopLevelCode) ||
                     (Kind == BraceItemListKind::TopLevelLibrary);
   bool isActiveConditionalBlock =
@@ -259,6 +262,7 @@ ParserStatus Parser::parseBraceItems(SmallVectorImpl<ASTNode> &Entries,
          Tok.isNot(tok::kw_sil_default_witness_table) &&
          (isConditionalBlock ||
           !isTerminatorForBraceItemListKind(Kind, Entries))) {
+    SyntaxParsingContextChild StmtContext(SyntaxContext, SyntaxContextKind::Stmt);
     if (Kind == BraceItemListKind::TopLevelLibrary &&
         skipExtraTopLevelRBraces())
       continue;
@@ -489,7 +493,7 @@ static ParserResult<Stmt> recoverFromInvalidCase(Parser &P) {
 }
 
 ParserResult<Stmt> Parser::parseStmt() {
-
+  SyntaxParsingContextChild LocalContext(SyntaxContext, SyntaxContextKind::Stmt);
   // Note that we're parsing a statement.
   StructureMarkerRAII ParsingStmt(*this, Tok.getLoc(),
                                   StructureMarkerKind::Statement);
@@ -586,15 +590,19 @@ ParserResult<BraceStmt> Parser::parseBraceItemList(Diag<> ID) {
     if (Tok.isNot(tok::l_brace))
       return nullptr;
   }
+  SyntaxParsingContextChild LocalContext(SyntaxContext, SyntaxKind::CodeBlock);
   SourceLoc LBLoc = consumeToken(tok::l_brace);
+  LocalContext.addTokenSyntax(LBLoc);
 
   SmallVector<ASTNode, 16> Entries;
   SourceLoc RBLoc;
 
   ParserStatus Status = parseBraceItems(Entries, BraceItemListKind::Brace,
                                         BraceItemListKind::Brace);
-  parseMatchingToken(tok::r_brace, RBLoc,
-                     diag::expected_rbrace_in_brace_stmt, LBLoc);
+  if (!parseMatchingToken(tok::r_brace, RBLoc,
+                          diag::expected_rbrace_in_brace_stmt, LBLoc)) {
+    LocalContext.addTokenSyntax(RBLoc);
+  }
 
   return makeParserResult(Status,
                           BraceStmt::create(Context, LBLoc, Entries, RBLoc));

lib/Parse/Parser.cpp

@@ -68,7 +68,7 @@ class ParseDelayedFunctionBodies : public ASTWalker {
     SourceManager &SourceMgr = SF.getASTContext().SourceMgr;
     unsigned BufferID = SourceMgr.findBufferContainingLoc(AFD->getLoc());
     Parser TheParser(BufferID, SF, nullptr, &ParserState);
-
+    TheParser.SyntaxContext->disable();
     std::unique_ptr<CodeCompletionCallbacks> CodeCompletion;
     if (CodeCompletionFactory) {
       CodeCompletion.reset(
@@ -102,6 +102,9 @@ static void parseDelayedDecl(
     SourceMgr.findBufferContainingLoc(ParserState.getDelayedDeclLoc());
   Parser TheParser(BufferID, SF, nullptr, &ParserState);
 
+  // Disable libSyntax creation in the delayed parsing.
+  TheParser.SyntaxContext->disable();
+
   std::unique_ptr<CodeCompletionCallbacks> CodeCompletion;
   if (CodeCompletionFactory) {
     CodeCompletion.reset(

lib/Syntax/RawSyntax.cpp

@@ -29,6 +29,7 @@ static bool isTrivialSyntaxKind(SyntaxKind Kind) {
   case SyntaxKind::SourceFile:
   case SyntaxKind::TopLevelCodeDecl:
   case SyntaxKind::ExpressionStmt:
+  case SyntaxKind::DeclarationStmt:
     return true;
   default:
     return false;