[llvm-cxxfilt] Add --quote option to quote demangled function names

This is useful when looking at LLVM/MLIR assembly produced from C++ sources.
For example
 cir.call @_ZN3aie4tileILi1ELi4EE7programIZ4mainE3$_0EEvOT_(%2, %7) :
will be translated to
cir.call @"void aie::tile<1, 4>::program<main::$_0>(main::$_0&&)"(%2, %7) :
which can be parsed as valid MLIR by the right mlir-lsp-server.

If a symbol is already quoted, do not quote it more.

Author: keryell

llvm/docs/CommandGuide/llvm-cxxfilt.rst

@@ -52,6 +52,11 @@ OPTIONS
 
   Do not demangle function parameters or return types.
 
+.. option:: --quote, -q
+
+  Add `"` `"` around demangled function symbols, typically to keep LLVM/MLIR
+  assembly files with `@<symbols>` valid. Do not quote already quoted symbols.
+
 .. option:: --no-strip-underscore, -n
 
   Do not strip a leading underscore. This is the default for all platforms

llvm/test/tools/llvm-cxxfilt/quote.test

@@ -0,0 +1,12 @@
+// Show that llvm-cxxfilt --quote can emit quoted demangled symbols but not
+// without double-quoting in the case they are already quoted
+
+RUN: llvm-cxxfilt --quote < %s | FileCheck %s
+      cir.call @_ZN3aie4tileILi1ELi4EE7programIZ4mainE3$_0EEvOT_(%2, %7) : (!cir.ptr<!ty_aie3A3Atile3C12C_43E>, !cir.ptr<!ty_anon2E0_>) -> () loc(#loc74)
+  cir.func lambda internal private  @_ZZ4mainENK3$_1clEv(%arg0: !cir.ptr<!ty_anon2E1_> loc("example.cpp":31:26)) extra(#fn_attr) {
+module @"example.cpp" attributes {cir.global_annotations = #cir<global_annotations [["_ZN3aie6deviceILNS_3$_0E42EE4tileILi1ELi4EEENS_4tileIXT_EXT0_EEEv", #cir.annotation<name = "aie.device.tile", args = [1 : i32, 4 : i32, 42 : i8, 42 : i8]>]]}
+
+CHECK: cir.call @"void aie::tile<1, 4>::program<main::$_0>(main::$_0&&)"(%2, %7) 
+CHECK-NEXT: cir.func lambda internal private  @"main::$_1::operator()() const"(%arg0: !cir.ptr<!ty_anon2E1_>
+// \todo Is there a simpler way to escape these [[ leading to "error: invalid variable name" otherwise?
+CHECK-NEXT: module @"example.cpp" attributes {cir.global_annotations = #cir<global_annotations {{[[]}}["aie::tile<1, 4> aie::device<(aie::$_0)42>::tile<1, 4>()", #cir.annotation<name = "aie.device.tile", args = [1 : i32, 4 : i32, 42 : i8, 42 : i8]>]]}

llvm/tools/llvm-cxxfilt/Opts.td

@@ -15,6 +15,7 @@ multiclass Eq<string name, string help> {
 }
 
 def help : FF<"help", "Display this help">;
+def quote : FF<"quote", "Quote demangled function names with \" \" if not already quoted. Useful for symbols appearing after an @ in MLIR/LLVM assembly">;
 defm strip_underscore : BB<"strip-underscore", "Strip the leading underscore", "Don't strip the leading underscore">;
 def types : FF<"types", "Attempt to demangle types as well as function names">;
 def no_params : FF<"no-params", "Skip function parameters and return types">;
@@ -27,4 +28,5 @@ def : F<"_", "Alias for --strip-underscore">, Alias<strip_underscore>;
 def : F<"h", "Alias for --help">, Alias<help>;
 def : F<"n", "Alias for --no-strip-underscore">, Alias<no_strip_underscore>;
 def : F<"p", "Alias for --no-params">, Alias<no_params>;
+def : F<"q", "Alias for --quote">, Alias<quote>;
 def : F<"t", "Alias for --types">, Alias<types>;

llvm/tools/llvm-cxxfilt/llvm-cxxfilt.cpp

@@ -20,6 +20,7 @@
 #include "llvm/TargetParser/Triple.h"
 #include <cstdlib>
 #include <iostream>
+#include <string>
 
 using namespace llvm;
 
@@ -54,6 +55,7 @@ class CxxfiltOptTable : public opt::GenericOptTable {
 } // namespace
 
 static bool ParseParams;
+static bool Quote;
 static bool StripUnderscore;
 static bool Types;
 
@@ -64,7 +66,15 @@ static void error(const Twine &Message) {
   exit(1);
 }
 
-static std::string demangle(const std::string &Mangled) {
+// Quote Undecorated with "" if asked for and not already followed by a '"'
+static std::string optionalQuote(std::string Undecorated,
+                                 StringRef Delimiters) {
+  if (Quote && (Delimiters.empty() || Delimiters[0] != '"'))
+    return '"' + Undecorated + '"';
+  return Undecorated;
+}
+
+static std::string demangle(const std::string &Mangled, StringRef Delimiters) {
   using llvm::itanium_demangle::starts_with;
   std::string_view DecoratedStr = Mangled;
   bool CanHaveLeadingDot = true;
@@ -76,7 +86,7 @@ static std::string demangle(const std::string &Mangled) {
   std::string Result;
   if (nonMicrosoftDemangle(DecoratedStr, Result, CanHaveLeadingDot,
                            ParseParams))
-    return Result;
+    return optionalQuote(Result, Delimiters);
 
   std::string Prefix;
   char *Undecorated = nullptr;
@@ -89,7 +99,8 @@ static std::string demangle(const std::string &Mangled) {
     Undecorated = itaniumDemangle(DecoratedStr.substr(6), ParseParams);
   }
 
-  Result = Undecorated ? Prefix + Undecorated : Mangled;
+  Result =
+      Undecorated ? Prefix + optionalQuote(Undecorated, Delimiters) : Mangled;
   free(Undecorated);
   return Result;
 }
@@ -137,9 +148,10 @@ static void demangleLine(llvm::raw_ostream &OS, StringRef Mangled, bool Split) {
     SmallVector<std::pair<StringRef, StringRef>, 16> Words;
     SplitStringDelims(Mangled, Words, IsLegalItaniumChar);
     for (const auto &Word : Words)
-      Result += ::demangle(std::string(Word.first)) + Word.second.str();
+      Result +=
+          ::demangle(std::string(Word.first), Word.second) + Word.second.str();
   } else
-    Result = ::demangle(std::string(Mangled));
+    Result = ::demangle(std::string(Mangled), "");
   OS << Result << '\n';
   OS.flush();
 }
@@ -172,6 +184,8 @@ int llvm_cxxfilt_main(int argc, char **argv, const llvm::ToolContext &) {
 
   Types = Args.hasArg(OPT_types);
 
+  Quote = Args.hasArg(OPT_quote);
+
   std::vector<std::string> Decorated = Args.getAllArgValues(OPT_INPUT);
   if (Decorated.empty())
     for (std::string Mangled; std::getline(std::cin, Mangled);)