[ADT] Remove APInt/APSInt toString() std::string variants

<string> is currently the highest impact header in a clang+llvm build:

https://commondatastorage.googleapis.com/chromium-browser-clang/llvm-include-analysis.html

One of the most common places this is being included is the APInt.h header, which needs it for an old toString() implementation that returns std::string - an inefficient method compared to the SmallString versions that it actually wraps.

This patch replaces these APInt/APSInt methods with a pair of llvm::toString() helpers inside StringExtras.h, adjusts users accordingly and removes the <string> from APInt.h - I was hoping that more of these users could be converted to use the SmallString methods, but it appears that most end up creating a std::string anyhow. I avoided trying to use the raw_ostream << operators as well as I didn't want to lose having the integer radix explicit in the code.

Differential Revision: https://reviews.llvm.org/D103888

include/llvm/ADT/APInt.h

@@ -20,7 +20,6 @@
 #include <cassert>
 #include <climits>
 #include <cstring>
-#include <string>
 
 namespace llvm {
 class FoldingSetNodeID;
@@ -1753,13 +1752,6 @@ public:
     toString(Str, Radix, true, false);
   }
 
-  /// Return the APInt as a std::string.
-  ///
-  /// Note that this is an inefficient method.  It is better to pass in a
-  /// SmallVector/SmallString to the methods above to avoid thrashing the heap
-  /// for the string.
-  std::string toString(unsigned Radix, bool Signed) const;
-
   /// \returns a byte-swapped representation of this APInt Value.
   APInt byteSwap() const;
 

include/llvm/ADT/APSInt.h

@@ -82,11 +82,6 @@ public:
   void toString(SmallVectorImpl<char> &Str, unsigned Radix = 10) const {
     APInt::toString(Str, Radix, isSigned());
   }
-  /// Converts an APInt to a std::string.  This is an inefficient
-  /// method; you should prefer passing in a SmallString instead.
-  std::string toString(unsigned Radix) const {
-    return APInt::toString(Radix, isSigned());
-  }
   using APInt::toString;
 
   /// Get the correctly-extended \c int64_t value.

include/llvm/ADT/StringExtras.h

@@ -13,6 +13,7 @@
 #ifndef LLVM_ADT_STRINGEXTRAS_H
 #define LLVM_ADT_STRINGEXTRAS_H
 
+#include "llvm/ADT/APSInt.h"
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/SmallString.h"
 #include "llvm/ADT/StringRef.h"
@@ -296,6 +297,17 @@ inline std::string itostr(int64_t X) {
     return utostr(static_cast<uint64_t>(X));
 }
 
+inline std::string toString(const APInt &I, unsigned Radix, bool Signed,
+                            bool formatAsCLiteral = false) {
+  SmallString<40> S;
+  I.toString(S, Radix, Signed, formatAsCLiteral);
+  return std::string(S.str());
+}
+
+inline std::string toString(const APSInt &I, unsigned Radix) {
+  return toString(I, Radix, I.isSigned());
+}
+
 /// StrInStrNoCase - Portable version of strcasestr.  Locates the first
 /// occurrence of string 's1' in string 's2', ignoring case.  Returns
 /// the offset of s2 in s1 or npos if s2 cannot be found.

lib/CodeGen/GlobalISel/GISelKnownBits.cpp

@@ -88,10 +88,10 @@ LLVM_ATTRIBUTE_UNUSED static void
 dumpResult(const MachineInstr &MI, const KnownBits &Known, unsigned Depth) {
   dbgs() << "[" << Depth << "] Compute known bits: " << MI << "[" << Depth
          << "] Computed for: " << MI << "[" << Depth << "] Known: 0x"
-         << (Known.Zero | Known.One).toString(16, false) << "\n"
+         << toString(Known.Zero | Known.One, 16, false) << "\n"
-         << "[" << Depth << "] Zero: 0x" << Known.Zero.toString(16, false)
+         << "[" << Depth << "] Zero: 0x" << toString(Known.Zero, 16, false)
          << "\n"
-         << "[" << Depth << "] One:  0x" << Known.One.toString(16, false)
+         << "[" << Depth << "] One:  0x" << toString(Known.One, 16, false)
          << "\n";
 }
 

lib/CodeGen/TargetLoweringObjectFileImpl.cpp

@@ -1930,7 +1930,7 @@ const MCExpr *TargetLoweringObjectFileCOFF::lowerRelativeReference(
 
 static std::string APIntToHexString(const APInt &AI) {
   unsigned Width = (AI.getBitWidth() / 8) * 2;
-  std::string HexString = AI.toString(16, /*Signed=*/false);
+  std::string HexString = toString(AI, 16, /*Signed=*/false);
   llvm::transform(HexString, HexString.begin(), tolower);
   unsigned Size = HexString.size();
   assert(Width >= Size && "hex string is too large!");

lib/Support/APInt.cpp

@@ -2281,14 +2281,6 @@ void APInt::toString(SmallVectorImpl<char> &Str, unsigned Radix,
   std::reverse(Str.begin()+StartDig, Str.end());
 }
 
-/// Returns the APInt as a std::string. Note that this is an inefficient method.
-/// It is better to pass in a SmallVector/SmallString to the methods above.
-std::string APInt::toString(unsigned Radix = 10, bool Signed = true) const {
-  SmallString<40> S;
-  toString(S, Radix, Signed, /* formatAsCLiteral = */false);
-  return std::string(S.str());
-}
-
 #if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
 LLVM_DUMP_METHOD void APInt::dump() const {
   SmallString<40> S, U;

lib/Target/X86/X86MCInstLower.cpp

@@ -2167,7 +2167,7 @@ static void addConstantComments(const MachineInstr *MI,
       const MachineOperand &DstOp = MI->getOperand(0);
       CS << X86ATTInstPrinter::getRegisterName(DstOp.getReg()) << " = ";
       if (auto *CF = dyn_cast<ConstantFP>(C)) {
-        CS << "0x" << CF->getValueAPF().bitcastToAPInt().toString(16, false);
+        CS << "0x" << toString(CF->getValueAPF().bitcastToAPInt(), 16, false);
         OutStreamer.AddComment(CS.str());
       }
     }

lib/Transforms/IPO/OpenMPOpt.cpp

@@ -585,7 +585,7 @@ struct OpenMPOpt {
           return ORA << "OpenMP ICV " << ore::NV("OpenMPICV", ICVInfo.Name)
                      << " Value: "
                      << (ICVInfo.InitValue
-                             ? ICVInfo.InitValue->getValue().toString(10, true)
+                             ? toString(ICVInfo.InitValue->getValue(), 10, true)
                              : "IMPLEMENTATION_DEFINED");
         };
 

tools/llvm-exegesis/lib/BenchmarkResult.cpp

@@ -255,7 +255,7 @@ template <> struct ScalarTraits<exegesis::RegisterValue> {
                      raw_ostream &Out) {
     YamlContext &Context = getTypedContext(Ctx);
     Out << Context.getRegName(RV.Register) << "=0x"
-        << RV.Value.toString(kRadix, kSigned);
+        << toString(RV.Value, kRadix, kSigned);
   }
 
   static StringRef input(StringRef String, void *Ctx,

tools/llvm-pdbutil/MinimalSymbolDumper.cpp

@@ -559,7 +559,7 @@ Error MinimalSymbolDumper::visitKnownRecord(CVSymbol &CVR,
   P.format(" `{0}`", Constant.Name);
   AutoIndent Indent(P, 7);
   P.formatLine("type = {0}, value = {1}", typeIndex(Constant.Type),
-               Constant.Value.toString(10));
+               toString(Constant.Value, 10));
   return Error::success();
 }
 

tools/llvm-pdbutil/MinimalTypeDumper.cpp

@@ -557,7 +557,7 @@ Error MinimalTypeDumpVisitor::visitKnownMember(CVMemberRecord &CVR,
 Error MinimalTypeDumpVisitor::visitKnownMember(CVMemberRecord &CVR,
                                                EnumeratorRecord &Enum) {
   P.format(" [{0} = {1}]", Enum.Name,
-           Enum.Value.toString(10, Enum.Value.isSigned()));
+           toString(Enum.Value, 10, Enum.Value.isSigned()));
   return Error::success();
 }
 

unittests/ADT/APIntTest.cpp

@@ -1461,7 +1461,10 @@ TEST(APIntTest, mul_clear) {
   APInt ValC(65, 0);
   ValC = ValA * ValB;
   ValA *= ValB;
-  EXPECT_EQ(ValA.toString(10, false), ValC.toString(10, false));
+  SmallString<16> StrA, StrC;
+  ValA.toString(StrA, 10, false);
+  ValC.toString(StrC, 10, false);
+  EXPECT_EQ(std::string(StrA), std::string(StrC));
 }
 
 TEST(APIntTest, Rotate) {

unittests/ADT/StringExtrasTest.cpp

@@ -229,3 +229,48 @@ TEST(StringExtrasTest, ListSeparator) {
   S = LS2;
   EXPECT_EQ(S, " ");
 }
+
+TEST(StringExtrasTest, toStringAPInt) {
+  bool isSigned;
+
+  EXPECT_EQ(toString(APInt(8, 0), 2, true, true), "0b0");
+  EXPECT_EQ(toString(APInt(8, 0), 8, true, true), "00");
+  EXPECT_EQ(toString(APInt(8, 0), 10, true, true), "0");
+  EXPECT_EQ(toString(APInt(8, 0), 16, true, true), "0x0");
+  EXPECT_EQ(toString(APInt(8, 0), 36, true, false), "0");
+
+  isSigned = false;
+  EXPECT_EQ(toString(APInt(8, 255, isSigned), 2, isSigned, true), "0b11111111");
+  EXPECT_EQ(toString(APInt(8, 255, isSigned), 8, isSigned, true), "0377");
+  EXPECT_EQ(toString(APInt(8, 255, isSigned), 10, isSigned, true), "255");
+  EXPECT_EQ(toString(APInt(8, 255, isSigned), 16, isSigned, true), "0xFF");
+  EXPECT_EQ(toString(APInt(8, 255, isSigned), 36, isSigned, false), "73");
+
+  isSigned = true;
+  EXPECT_EQ(toString(APInt(8, 255, isSigned), 2, isSigned, true), "-0b1");
+  EXPECT_EQ(toString(APInt(8, 255, isSigned), 8, isSigned, true), "-01");
+  EXPECT_EQ(toString(APInt(8, 255, isSigned), 10, isSigned, true), "-1");
+  EXPECT_EQ(toString(APInt(8, 255, isSigned), 16, isSigned, true), "-0x1");
+  EXPECT_EQ(toString(APInt(8, 255, isSigned), 36, isSigned, false), "-1");
+}
+
+TEST(StringExtrasTest, toStringAPSInt) {
+  bool isUnsigned;
+
+  EXPECT_EQ(toString(APSInt(APInt(8, 0), false), 2), "0");
+  EXPECT_EQ(toString(APSInt(APInt(8, 0), false), 8), "0");
+  EXPECT_EQ(toString(APSInt(APInt(8, 0), false), 10), "0");
+  EXPECT_EQ(toString(APSInt(APInt(8, 0), false), 16), "0");
+
+  isUnsigned = true;
+  EXPECT_EQ(toString(APSInt(APInt(8, 255), isUnsigned), 2), "11111111");
+  EXPECT_EQ(toString(APSInt(APInt(8, 255), isUnsigned), 8), "377");
+  EXPECT_EQ(toString(APSInt(APInt(8, 255), isUnsigned), 10), "255");
+  EXPECT_EQ(toString(APSInt(APInt(8, 255), isUnsigned), 16), "FF");
+
+  isUnsigned = false;
+  EXPECT_EQ(toString(APSInt(APInt(8, 255), isUnsigned), 2), "-1");
+  EXPECT_EQ(toString(APSInt(APInt(8, 255), isUnsigned), 8), "-1");
+  EXPECT_EQ(toString(APSInt(APInt(8, 255), isUnsigned), 10), "-1");
+  EXPECT_EQ(toString(APSInt(APInt(8, 255), isUnsigned), 16), "-1");
+}