[clangd] Improve Markup Rendering

clang-tools-extra/clangd/ClangdLSPServer.cpp

@@ -1262,11 +1262,9 @@ void ClangdLSPServer::onHover(const TextDocumentPositionParams &Params,
                       R.contents.kind = HoverContentFormat;
                       R.range = (*H)->SymRange;
                       switch (HoverContentFormat) {
-                      case MarkupKind::PlainText:
-                        R.contents.value = (*H)->present().asPlainText();
-                        return Reply(std::move(R));
                       case MarkupKind::Markdown:
-                        R.contents.value = (*H)->present().asMarkdown();
+                      case MarkupKind::PlainText:
+                        R.contents.value = (*H)->present(HoverContentFormat);
                         return Reply(std::move(R));
                       };
                       llvm_unreachable("unhandled MarkupKind");

clang-tools-extra/clangd/Config.h

@@ -177,6 +177,19 @@ struct Config {
     /// Controls highlighting modifiers that are disabled.
     std::vector<std::string> DisabledModifiers;
   } SemanticTokens;
+
+  enum class CommentFormatPolicy {
+    /// Treat comments as plain text.
+    PlainText,
+    /// Treat comments as Markdown.
+    Markdown,
+    /// Treat comments as doxygen.
+    Doxygen,
+  };
+
+  struct {
+    CommentFormatPolicy CommentFormat = CommentFormatPolicy::PlainText;
+  } Documentation;
 };
 
 } // namespace clangd

clang-tools-extra/clangd/ConfigCompile.cpp

@@ -198,6 +198,7 @@ struct FragmentCompiler {
     compile(std::move(F.InlayHints));
     compile(std::move(F.SemanticTokens));
     compile(std::move(F.Style));
+    compile(std::move(F.Documentation));
   }
 
   void compile(Fragment::IfBlock &&F) {
@@ -760,6 +761,22 @@ struct FragmentCompiler {
     }
   }
 
+  void compile(Fragment::DocumentationBlock &&F) {
+    if (F.CommentFormat) {
+      if (auto Val =
+              compileEnum<Config::CommentFormatPolicy>("CommentFormat",
+                                                       *F.CommentFormat)
+                  .map("Plaintext", Config::CommentFormatPolicy::PlainText)
+                  .map("Markdown",
+                       Config::CommentFormatPolicy::Markdown)
+                  .map("Doxygen", Config::CommentFormatPolicy::Doxygen)
+                  .value())
+        Out.Apply.push_back([Val](const Params &, Config &C) {
+          C.Documentation.CommentFormat = *Val;
+        });
+    }
+  }
+
   constexpr static llvm::SourceMgr::DiagKind Error = llvm::SourceMgr::DK_Error;
   constexpr static llvm::SourceMgr::DiagKind Warning =
       llvm::SourceMgr::DK_Warning;

clang-tools-extra/clangd/ConfigFragment.h

@@ -372,6 +372,17 @@ struct Fragment {
     std::vector<Located<std::string>> DisabledModifiers;
   };
   SemanticTokensBlock SemanticTokens;
+
+  /// Configures documentation style and behaviour.
+  struct DocumentationBlock {
+    /// Specifies the format of comments in the code.
+    /// Valid values are enum Config::CommentFormatPolicy values:
+    /// - Plaintext: Treat comments as plain text.
+    /// - Markdown: Treat comments as Markdown.
+    /// - Doxygen: Treat comments as doxygen.
+    std::optional<Located<std::string>> CommentFormat;
+  };
+  DocumentationBlock Documentation;
 };
 
 } // namespace config

clang-tools-extra/clangd/ConfigYAML.cpp

@@ -68,6 +68,7 @@ class Parser {
     Dict.handle("Hover", [&](Node &N) { parse(F.Hover, N); });
     Dict.handle("InlayHints", [&](Node &N) { parse(F.InlayHints, N); });
     Dict.handle("SemanticTokens", [&](Node &N) { parse(F.SemanticTokens, N); });
+    Dict.handle("Documentation", [&](Node &N) { parse(F.Documentation, N); });
     Dict.parse(N);
     return !(N.failed() || HadError);
   }
@@ -299,6 +300,15 @@ class Parser {
     Dict.parse(N);
   }
 
+  void parse(Fragment::DocumentationBlock &F, Node &N) {
+    DictParser Dict("Documentation", this);
+    Dict.handle("CommentFormat", [&](Node &N) {
+      if (auto Value = scalarValue(N, "CommentFormat"))
+        F.CommentFormat = *Value;
+    });
+    Dict.parse(N);
+  }
+
   // Helper for parsing mapping nodes (dictionaries).
   // We don't use YamlIO as we want to control over unknown keys.
   class DictParser {

clang-tools-extra/clangd/Hover.cpp

@@ -15,6 +15,7 @@
 #include "Headers.h"
 #include "IncludeCleaner.h"
 #include "ParsedAST.h"
+#include "Protocol.h"
 #include "Selection.h"
 #include "SourceCode.h"
 #include "clang-include-cleaner/Analysis.h"
@@ -960,42 +961,6 @@ std::optional<HoverInfo> getHoverContents(const Attr *A, ParsedAST &AST) {
   return HI;
 }
 
-bool isParagraphBreak(llvm::StringRef Rest) {
-  return Rest.ltrim(" \t").starts_with("\n");
-}
-
-bool punctuationIndicatesLineBreak(llvm::StringRef Line) {
-  constexpr llvm::StringLiteral Punctuation = R"txt(.:,;!?)txt";
-
-  Line = Line.rtrim();
-  return !Line.empty() && Punctuation.contains(Line.back());
-}
-
-bool isHardLineBreakIndicator(llvm::StringRef Rest) {
-  // '-'/'*' md list, '@'/'\' documentation command, '>' md blockquote,
-  // '#' headings, '`' code blocks
-  constexpr llvm::StringLiteral LinebreakIndicators = R"txt(-*@\>#`)txt";
-
-  Rest = Rest.ltrim(" \t");
-  if (Rest.empty())
-    return false;
-
-  if (LinebreakIndicators.contains(Rest.front()))
-    return true;
-
-  if (llvm::isDigit(Rest.front())) {
-    llvm::StringRef AfterDigit = Rest.drop_while(llvm::isDigit);
-    if (AfterDigit.starts_with(".") || AfterDigit.starts_with(")"))
-      return true;
-  }
-  return false;
-}
-
-bool isHardLineBreakAfter(llvm::StringRef Line, llvm::StringRef Rest) {
-  // Should we also consider whether Line is short?
-  return punctuationIndicatesLineBreak(Line) || isHardLineBreakIndicator(Rest);
-}
-
 void addLayoutInfo(const NamedDecl &ND, HoverInfo &HI) {
   if (ND.isInvalidDecl())
     return;
@@ -1571,6 +1536,25 @@ markup::Document HoverInfo::present() const {
   return Output;
 }
 
+std::string HoverInfo::present(MarkupKind Kind) const {
+  if (Kind == MarkupKind::Markdown) {
+    const Config &Cfg = Config::current();
+    if ((Cfg.Documentation.CommentFormat ==
+         Config::CommentFormatPolicy::Markdown) ||
+        (Cfg.Documentation.CommentFormat ==
+         Config::CommentFormatPolicy::Doxygen))
+      // If the user prefers Markdown, we use the present() method to generate
+      // the Markdown output.
+      return present().asMarkdown();
+    if (Cfg.Documentation.CommentFormat == Config::CommentFormatPolicy::PlainText)
+      // If the user prefers plain text, we use the present() method to generate
+      // the plain text output.
+      return present().asEscapedMarkdown();
+  }
+
+  return present().asPlainText();
+}
+
 // If the backtick at `Offset` starts a probable quoted range, return the range
 // (including the quotes).
 std::optional<llvm::StringRef> getBacktickQuoteRange(llvm::StringRef Line,
@@ -1601,51 +1585,32 @@ std::optional<llvm::StringRef> getBacktickQuoteRange(llvm::StringRef Line,
   return Line.slice(Offset, Next + 1);
 }
 
-void parseDocumentationLine(llvm::StringRef Line, markup::Paragraph &Out) {
+void parseDocumentationParagraph(llvm::StringRef Text, markup::Paragraph &Out) {
   // Probably this is appendText(Line), but scan for something interesting.
-  for (unsigned I = 0; I < Line.size(); ++I) {
-    switch (Line[I]) {
+  for (unsigned I = 0; I < Text.size(); ++I) {
+    switch (Text[I]) {
     case '`':
-      if (auto Range = getBacktickQuoteRange(Line, I)) {
-        Out.appendText(Line.substr(0, I));
+      if (auto Range = getBacktickQuoteRange(Text, I)) {
+        Out.appendText(Text.substr(0, I));
         Out.appendCode(Range->trim("`"), /*Preserve=*/true);
-        return parseDocumentationLine(Line.substr(I + Range->size()), Out);
+        return parseDocumentationParagraph(Text.substr(I + Range->size()), Out);
       }
       break;
     }
   }
-  Out.appendText(Line).appendSpace();
+  Out.appendText(Text);
 }
 
 void parseDocumentation(llvm::StringRef Input, markup::Document &Output) {
-  std::vector<llvm::StringRef> ParagraphLines;
-  auto FlushParagraph = [&] {
-    if (ParagraphLines.empty())
-      return;
-    auto &P = Output.addParagraph();
-    for (llvm::StringRef Line : ParagraphLines)
-      parseDocumentationLine(Line, P);
-    ParagraphLines.clear();
-  };
-
-  llvm::StringRef Line, Rest;
-  for (std::tie(Line, Rest) = Input.split('\n');
-       !(Line.empty() && Rest.empty());
-       std::tie(Line, Rest) = Rest.split('\n')) {
-
-    // After a linebreak remove spaces to avoid 4 space markdown code blocks.
-    // FIXME: make FlushParagraph handle this.
-    Line = Line.ltrim();
-    if (!Line.empty())
-      ParagraphLines.push_back(Line);
+  llvm::StringRef Paragraph, Rest;
+  for (std::tie(Paragraph, Rest) = Input.split("\n\n");
+       !(Paragraph.empty() && Rest.empty());
+       std::tie(Paragraph, Rest) = Rest.split("\n\n")) {
 
-    if (isParagraphBreak(Rest) || isHardLineBreakAfter(Line, Rest)) {
-      FlushParagraph();
-    }
+    if (!Paragraph.empty())
+      parseDocumentationParagraph(Paragraph, Output.addParagraph());
   }
-  FlushParagraph();
 }
-
 llvm::raw_ostream &operator<<(llvm::raw_ostream &OS,
                               const HoverInfo::PrintedType &T) {
   OS << T.Type;

clang-tools-extra/clangd/Hover.h

@@ -120,6 +120,8 @@ struct HoverInfo {
 
   /// Produce a user-readable information.
   markup::Document present() const;
+
+  std::string present(MarkupKind Kind) const;
 };
 
 inline bool operator==(const HoverInfo::PrintedType &LHS,