Merge pull request #26768 from davidungar/A-8-21-parserPR

[Parser, NameLookup, ASTScope] Parser changes for lazy ASTScopes

Author: David Ungar

include/swift/AST/Expr.h

@@ -926,11 +926,13 @@ class TapExpr : public Expr {
   void setBody(BraceStmt * b) { Body = b; }
 
   SourceLoc getLoc() const { return SubExpr ? SubExpr->getLoc() : SourceLoc(); }
-
-  SourceRange getSourceRange() const {
-    return SubExpr ? SubExpr->getSourceRange() : SourceRange();
+  
+  SourceLoc getStartLoc() const {
+    return SubExpr ? SubExpr->getStartLoc() : SourceLoc();
   }
 
+  SourceLoc getEndLoc() const;
+
   static bool classof(const Expr *E) {
     return E->getKind() == ExprKind::Tap;
   }
@@ -1027,6 +1029,8 @@ class InterpolatedStringLiteralExpr : public LiteralExpr {
     // token, so the range should be (Start == End).
     return Loc;
   }
+  
+  /// Could also be computed by relexing.
   SourceLoc getTrailingQuoteLoc() const {
     return TrailingQuoteLoc;
   }

include/swift/Parse/Parser.h

@@ -571,7 +571,7 @@ class Parser {
 
   /// Retrieve the location just past the end of the previous
   /// source location.
-  SourceLoc getEndOfPreviousLoc();
+  SourceLoc getEndOfPreviousLoc() const;
 
   /// If the current token is the specified kind, consume it and
   /// return true.  Otherwise, return false without consuming it.
@@ -812,6 +812,19 @@ class Parser {
   bool parseMatchingToken(tok K, SourceLoc &TokLoc, Diag<> ErrorDiag,
                           SourceLoc OtherLoc);
 
+  /// Returns the proper location for a missing right brace, parenthesis, etc.
+  SourceLoc getLocForMissingMatchingToken() const;
+
+  /// When encountering an error or a missing matching token (e.g. '}'), return
+  /// the location to use for it. This value should be at the last token in
+  /// the ASTNode being parsed so that it nests within any enclosing nodes, and,
+  /// for ASTScope lookups, it does not preceed any identifiers to be looked up.
+  /// However, the latter case does not hold when  parsing an interpolated
+  /// string literal because there may be identifiers to be looked up in the
+  /// literal and their locations will not precede the location of a missing
+  /// close brace.
+  SourceLoc getErrorOrMissingLoc() const;
+
   /// Parse a comma separated list of some elements.
   ParserStatus parseList(tok RightK, SourceLoc LeftLoc, SourceLoc &RightLoc,
                          bool AllowSepAfterLast, Diag<> ErrorDiag,
@@ -1068,6 +1081,9 @@ class Parser {
 
   ParserResult<TypeRepr> parseDeclResultType(Diag<> MessageID);
 
+  /// Get the location for a type error.
+  SourceLoc getTypeErrorLoc() const;
+
   //===--------------------------------------------------------------------===//
   // Type Parsing
 

lib/AST/ASTVerifier.cpp

@@ -3646,7 +3646,9 @@ class Verifier : public ASTWalker {
         if (D->isImplicit())
           return;
         // FIXME: This is not working well for decl parents.
-        return;
+        // But it must work when using lazy ASTScopes for IterableDeclContexts
+        if (!isa<IterableDeclContext>(D))
+          return;
       } else if (Stmt *S = Parent.getAsStmt()) {
         Enclosing = S->getSourceRange();
         if (S->isImplicit())

lib/AST/Expr.cpp

@@ -2227,6 +2227,20 @@ VarDecl * TapExpr::getVar() const {
   return dyn_cast<VarDecl>(Body->getElement(0).dyn_cast<Decl *>());
 }
 
+SourceLoc TapExpr::getEndLoc() const {
+  // Include the body in the range, assuming the body follows the SubExpr.
+  // Also, be (perhaps overly) defensive about null pointers & invalid
+  // locations.
+  if (auto *const b = getBody()) {
+    const auto be = b->getEndLoc();
+    if (be.isValid())
+      return be;
+  }
+  if (auto *const se = getSubExpr())
+    return se->getEndLoc();
+  return SourceLoc();
+}
+
 // See swift/Basic/Statistic.h for declaration: this enables tracing Exprs, is
 // defined here to avoid too much layering violation / circular linkage
 // dependency.

lib/Parse/ParseExpr.cpp

@@ -1992,6 +1992,9 @@ ParserResult<Expr> Parser::parseExprStringLiteral() {
   llvm::SaveAndRestore<Token> SavedTok(Tok);
   llvm::SaveAndRestore<ParsedTrivia> SavedLeadingTrivia(LeadingTrivia);
   llvm::SaveAndRestore<ParsedTrivia> SavedTrailingTrivia(TrailingTrivia);
+  // For errors, we need the real PreviousLoc, i.e. the start of the
+  // whole InterpolatedStringLiteral.
+  llvm::SaveAndRestore<SourceLoc> SavedPreviousLoc(PreviousLoc);
 
   // We're not in a place where an interpolation would be valid.
   if (!CurLocalContext) {
@@ -3426,7 +3429,8 @@ ParserResult<Expr> Parser::parseExprCollection() {
 
   if (Status.isError()) {
     // If we've already got errors, don't emit missing RightK diagnostics.
-    RSquareLoc = Tok.is(tok::r_square) ? consumeToken() : PreviousLoc;
+    RSquareLoc = Tok.is(tok::r_square) ? consumeToken()
+                                       : getLocForMissingMatchingToken();
   } else if (parseMatchingToken(tok::r_square, RSquareLoc,
                                 diag::expected_rsquare_array_expr,
                                 LSquareLoc)) {

lib/Parse/ParseType.cpp

@@ -520,7 +520,8 @@ ParserResult<TypeRepr> Parser::parseDeclResultType(Diag<> MessageID) {
     auto diag = diagnose(Tok, diag::extra_rbracket);
     diag.fixItInsert(result.get()->getStartLoc(), getTokenText(tok::l_square));
     consumeToken();
-    return makeParserErrorResult(new (Context) ErrorTypeRepr(Tok.getLoc()));
+    return makeParserErrorResult(new (Context)
+                                     ErrorTypeRepr(getTypeErrorLoc()));
   } else if (!result.isParseError() && Tok.is(tok::colon)) {
     auto colonTok = consumeToken();
     auto secondType = parseType(diag::expected_dictionary_value_type);
@@ -534,11 +535,17 @@ ParserResult<TypeRepr> Parser::parseDeclResultType(Diag<> MessageID) {
         diag.fixItInsertAfter(secondType.get()->getEndLoc(), getTokenText(tok::r_square));
       }
     }
-    return makeParserErrorResult(new (Context) ErrorTypeRepr(Tok.getLoc()));
+    return makeParserErrorResult(new (Context)
+                                     ErrorTypeRepr(getTypeErrorLoc()));
   }
   return result;
 }
 
+SourceLoc Parser::getTypeErrorLoc() const {
+  // Use the same location as a missing close brace, etc.
+  return getErrorOrMissingLoc();
+}
+
 ParserStatus Parser::parseGenericArguments(SmallVectorImpl<TypeRepr *> &Args,
                                            SourceLoc &LAngleLoc,
                                            SourceLoc &RAngleLoc) {

lib/Parse/Parser.cpp

@@ -596,7 +596,7 @@ ParsedTokenSyntax Parser::consumeTokenSyntax() {
   return ParsedToken;
 }
 
-SourceLoc Parser::getEndOfPreviousLoc() {
+SourceLoc Parser::getEndOfPreviousLoc() const {
   return Lexer::getLocForEndOfToken(SourceMgr, PreviousLoc);
 }
 
@@ -968,8 +968,6 @@ bool Parser::parseToken(tok K, SourceLoc &TokLoc, const Diagnostic &D) {
   return true;
 }
 
-/// Parse the specified expected token and return its location on success.  On failure, emit the specified
-/// error diagnostic,  a note at the specified note location, and return the location of the previous token.
 bool Parser::parseMatchingToken(tok K, SourceLoc &TokLoc, Diag<> ErrorDiag,
                                 SourceLoc OtherLoc) {
   Diag<> OtherNote;
@@ -982,13 +980,34 @@ bool Parser::parseMatchingToken(tok K, SourceLoc &TokLoc, Diag<> ErrorDiag,
   if (parseToken(K, TokLoc, ErrorDiag)) {
     diagnose(OtherLoc, OtherNote);
 
-    TokLoc = PreviousLoc;
+    TokLoc = getLocForMissingMatchingToken();
     return true;
   }
 
   return false;
 }
 
+SourceLoc Parser::getLocForMissingMatchingToken() const {
+  // At present, use the same location whether it's an error or whether
+  // the matching token is missing.
+  // Both cases supply a location for something the user didn't type.
+  return getErrorOrMissingLoc();
+}
+
+SourceLoc Parser::getErrorOrMissingLoc() const {
+  // The next token might start a new enclosing construct,
+  // and SourceLoc's are always at the start of a token (for example, for
+  // fixits, so use the previous token's SourceLoc and allow a subnode to end
+  // right at the same place as its supernode.
+
+  // The tricky case is when the previous token is an InterpolatedStringLiteral.
+  // Then, there will be names in scope whose SourceLoc is *after* the
+  // the location of a missing close brace.
+  // ASTScope tree creation will have to cope.
+
+  return PreviousLoc;
+}
+
 static SyntaxKind getListElementKind(SyntaxKind ListKind) {
   switch (ListKind) {
   case SyntaxKind::FunctionCallArgumentList:
@@ -1094,7 +1113,8 @@ Parser::parseList(tok RightK, SourceLoc LeftLoc, SourceLoc &RightLoc,
 
   if (Status.isError()) {
     // If we've already got errors, don't emit missing RightK diagnostics.
-    RightLoc = Tok.is(RightK) ? consumeToken() : PreviousLoc;
+    RightLoc =
+        Tok.is(RightK) ? consumeToken() : getLocForMissingMatchingToken();
   } else if (parseMatchingToken(RightK, RightLoc, ErrorDiag, LeftLoc)) {
     Status.setIsParseError();
   }