Merge pull request #36920 from bnbarham/cleanup-rawcomment

[Gardening] Do not store start/end line in SingleRawComment

Author: bnbarham

include/swift/AST/RawComment.h

@@ -34,12 +34,10 @@ struct SingleRawComment {
   StringRef RawText;
 
   unsigned Kind : 8;
-  unsigned StartColumn : 16;
-  unsigned StartLine;
-  unsigned EndLine;
+  unsigned ColumnIndent : 16;
 
   SingleRawComment(CharSourceRange Range, const SourceManager &SourceMgr);
-  SingleRawComment(StringRef RawText, unsigned StartColumn);
+  SingleRawComment(StringRef RawText, unsigned ColumnIndent);
 
   SingleRawComment(const SingleRawComment &) = default;
   SingleRawComment &operator=(const SingleRawComment &) = default;

lib/AST/ASTPrinter.cpp

@@ -641,7 +641,7 @@ class PrintAST : public ASTVisitor<PrintAST> {
       Lines.clear();
 
       StringRef RawText = SRC.RawText.rtrim("\n\r");
-      unsigned WhitespaceToTrim = SRC.StartColumn - 1;
+      unsigned WhitespaceToTrim = SRC.ColumnIndent - 1;
       trimLeadingWhitespaceFromLines(RawText, WhitespaceToTrim, Lines);
 
       for (auto Line : Lines) {

lib/AST/Module.cpp

@@ -879,10 +879,10 @@ SourceFile::getBasicLocsForDecl(const Decl *D) const {
   Result.SourceFilePath = SM.getDisplayNameForLoc(D->getLoc());
 
   for (const auto &SRC : D->getRawComment(/*SerializedOK*/false).Comments) {
-    Result.DocRanges.push_back(std::make_pair(
-      SourcePosition { SRC.StartLine, SRC.StartColumn },
-      SRC.Range.getByteLength())
-    );
+    auto LineAndCol = SM.getLineAndColumnInBuffer(SRC.Range.getStart());
+    Result.DocRanges.push_back(
+        std::make_pair(SourcePosition{LineAndCol.first, LineAndCol.second},
+                       SRC.Range.getByteLength()));
   }
 
   auto setLineColumn = [&SM](SourcePosition &Home, SourceLoc Loc) {

lib/AST/RawComment.cpp

@@ -61,71 +61,57 @@ SingleRawComment::SingleRawComment(CharSourceRange Range,
                                    const SourceManager &SourceMgr)
     : Range(Range), RawText(SourceMgr.extractText(Range)),
       Kind(static_cast<unsigned>(getCommentKind(RawText))) {
-  auto StartLineAndColumn =
-      SourceMgr.getLineAndColumnInBuffer(Range.getStart());
-  StartLine = StartLineAndColumn.first;
-  StartColumn = StartLineAndColumn.second;
-  EndLine = SourceMgr.getLineAndColumnInBuffer(Range.getEnd()).first;
+  ColumnIndent = SourceMgr.getLineAndColumnInBuffer(Range.getStart()).second;
 }
 
-SingleRawComment::SingleRawComment(StringRef RawText, unsigned StartColumn)
+SingleRawComment::SingleRawComment(StringRef RawText, unsigned ColumnIndent)
     : RawText(RawText), Kind(static_cast<unsigned>(getCommentKind(RawText))),
-      StartColumn(StartColumn), StartLine(0), EndLine(0) {}
-
-static void addCommentToList(SmallVectorImpl<SingleRawComment> &Comments,
-                             const SingleRawComment &SRC) {
-  // TODO: consider producing warnings when we decide not to merge comments.
-
-  if (SRC.isOrdinary()) {
-    // Skip gyb comments that are line number markers.
-    if (SRC.RawText.startswith("// ###"))
-      return;
-
-    Comments.clear();
-    return;
-  }
-
-  // If this is the first documentation comment, save it (because there isn't
-  // anything to merge it with).
-  if (Comments.empty()) {
-    Comments.push_back(SRC);
-    return;
-  }
-
-  auto &Last = Comments.back();
-
-  // Merge comments if they are on same or consecutive lines.
-  if (Last.EndLine + 1 < SRC.StartLine) {
-    Comments.clear();
-    return;
-  }
-
-  Comments.push_back(SRC);
-}
+      ColumnIndent(ColumnIndent) {}
 
 static RawComment toRawComment(ASTContext &Context, CharSourceRange Range) {
   if (Range.isInvalid())
     return RawComment();
 
-  auto &SourceMgr = Context.SourceMgr;
-  unsigned BufferID = SourceMgr.findBufferContainingLoc(Range.getStart());
-  unsigned Offset = SourceMgr.getLocOffsetInBuffer(Range.getStart(), BufferID);
-  unsigned EndOffset = SourceMgr.getLocOffsetInBuffer(Range.getEnd(), BufferID);
+  auto &SM = Context.SourceMgr;
+  unsigned BufferID = SM.findBufferContainingLoc(Range.getStart());
+  unsigned Offset = SM.getLocOffsetInBuffer(Range.getStart(), BufferID);
+  unsigned EndOffset = SM.getLocOffsetInBuffer(Range.getEnd(), BufferID);
   LangOptions FakeLangOpts;
-  Lexer L(FakeLangOpts, SourceMgr, BufferID, nullptr, LexerMode::Swift,
-          HashbangMode::Disallowed,
-          CommentRetentionMode::ReturnAsTokens,
-          TriviaRetentionMode::WithoutTrivia,
-          Offset, EndOffset);
+  Lexer L(FakeLangOpts, SM, BufferID, nullptr, LexerMode::Swift,
+          HashbangMode::Disallowed, CommentRetentionMode::ReturnAsTokens,
+          TriviaRetentionMode::WithoutTrivia, Offset, EndOffset);
+
   SmallVector<SingleRawComment, 16> Comments;
   Token Tok;
+  unsigned LastEnd = 0;
   while (true) {
     L.lex(Tok);
     if (Tok.is(tok::eof))
       break;
     assert(Tok.is(tok::comment));
-    addCommentToList(Comments, SingleRawComment(Tok.getRange(), SourceMgr));
+
+    auto SRC = SingleRawComment(Tok.getRange(), SM);
+    if (SRC.isOrdinary()) {
+      // Skip gyb comments that are line number markers.
+      if (!SRC.RawText.startswith("// ###")) {
+        Comments.clear();
+        LastEnd = 0;
+      }
+      continue;
+    }
+
+    // Merge comments if they are on same or consecutive lines.
+    unsigned Start =
+        SM.getLineAndColumnInBuffer(Tok.getRange().getStart()).first;
+    if (LastEnd > 0 && LastEnd + 1 < Start) {
+      Comments.clear();
+      LastEnd = 0;
+    } else {
+      Comments.push_back(SRC);
+      LastEnd = SM.getLineAndColumnInBuffer(Tok.getRange().getEnd()).first;
+    }
   }
+
   RawComment Result;
   Result.Comments = Context.AllocateCopy(Comments);
   return Result;
@@ -159,25 +145,24 @@ RawComment Decl::getRawComment(bool SerializedOK) const {
   switch (Unit->getKind()) {
   case FileUnitKind::SerializedAST: {
     if (SerializedOK) {
-      if (const auto *CachedLocs = getSerializedLocs()) {
-        if (!CachedLocs->DocRanges.empty()) {
-          SmallVector<SingleRawComment, 4> SRCs;
-          for (const auto &Range : CachedLocs->DocRanges) {
-            if (Range.isValid()) {
-              SRCs.push_back({ Range, Context.SourceMgr });
-            } else {
-              // if we've run into an invalid range, don't bother trying to load
-              // any of the other comments
-              SRCs.clear();
-              break;
-            }
+      auto *CachedLocs = getSerializedLocs();
+      if (!CachedLocs->DocRanges.empty()) {
+        SmallVector<SingleRawComment, 4> SRCs;
+        for (const auto &Range : CachedLocs->DocRanges) {
+          if (Range.isValid()) {
+            SRCs.push_back({Range, Context.SourceMgr});
+          } else {
+            // if we've run into an invalid range, don't bother trying to load
+            // any of the other comments
+            SRCs.clear();
+            break;
           }
-          auto RC = RawComment(Context.AllocateCopy(llvm::makeArrayRef(SRCs)));
+        }
 
-          if (!RC.isEmpty()) {
-            Context.setRawComment(this, RC, true);
-            return RC;
-          }
+        if (!SRCs.empty()) {
+          auto RC = RawComment(Context.AllocateCopy(llvm::makeArrayRef(SRCs)));
+          Context.setRawComment(this, RC, true);
+          return RC;
         }
       }
     }

lib/Markup/LineList.cpp

@@ -118,7 +118,7 @@ LineList MarkupContext::getLineList(swift::RawComment RC) {
         unsigned NewlineBytes = swift::measureNewline(Cleaned);
         Cleaned = Cleaned.drop_front(NewlineBytes);
         CleanedStartLoc = CleanedStartLoc.getAdvancedLocOrInvalid(NewlineBytes);
-        HasASCIIArt = measureASCIIArt(Cleaned, C.StartColumn - 1) != 0;
+        HasASCIIArt = measureASCIIArt(Cleaned, C.ColumnIndent - 1) != 0;
       }
 
       while (!Cleaned.empty()) {
@@ -129,7 +129,7 @@ LineList MarkupContext::getLineList(swift::RawComment RC) {
         // Skip over ASCII art, if present.
         if (HasASCIIArt)
           if (unsigned ASCIIArtBytes =
-              measureASCIIArt(Cleaned, C.StartColumn - 1)) {
+                  measureASCIIArt(Cleaned, C.ColumnIndent - 1)) {
             Cleaned = Cleaned.drop_front(ASCIIArtBytes);
             CleanedStartLoc =
             CleanedStartLoc.getAdvancedLocOrInvalid(ASCIIArtBytes);

lib/Serialization/SerializeDoc.cpp

@@ -247,7 +247,7 @@ class DeclCommentTableInfo {
     out << data.Brief;
     writer.write<uint32_t>(data.Raw.Comments.size());
     for (auto C : data.Raw.Comments) {
-      writer.write<uint32_t>(C.StartColumn);
+      writer.write<uint32_t>(C.ColumnIndent);
       writer.write<uint32_t>(C.RawText.size());
       out << C.RawText;
     }