Merge pull request #18292 from mhong/master_bytes_string

swift-ci · web-flow · commit 1623f424475f · 2018-08-06T09:05:58.000-07:00
diff --git a/include/swift/SIL/SILInstruction.h b/include/swift/SIL/SILInstruction.h
@@ -2891,6 +2891,7 @@ class StringLiteralInst final
 
 public:
   enum class Encoding {
+    Bytes,
     UTF8,
     UTF16,
     /// UTF-8 encoding of an Objective-C selector.
diff --git a/lib/IRGen/GenConstant.cpp b/lib/IRGen/GenConstant.cpp
@@ -54,6 +54,7 @@ llvm::Constant *irgen::emitConstantFP(IRGenModule &IGM, FloatLiteralInst *FLI) {
 llvm::Constant *irgen::emitAddrOfConstantString(IRGenModule &IGM,
                                                 StringLiteralInst *SLI) {
   switch (SLI->getEncoding()) {
+  case StringLiteralInst::Encoding::Bytes:
   case StringLiteralInst::Encoding::UTF8:
     return IGM.getAddrOfGlobalString(SLI->getValue());
 
diff --git a/lib/ParseSIL/ParseSIL.cpp b/lib/ParseSIL/ParseSIL.cpp
@@ -2354,6 +2354,8 @@ bool SILParser::parseSILInstruction(SILBuilder &B) {
       encoding = StringLiteralInst::Encoding::UTF16;
     } else if (P.Tok.getText() == "objc_selector") {
       encoding = StringLiteralInst::Encoding::ObjCSelector;
+    } else if (P.Tok.getText() == "bytes") {
+      encoding = StringLiteralInst::Encoding::Bytes;
     } else {
       P.diagnose(P.Tok, diag::sil_string_invalid_encoding, P.Tok.getText());
       return true;
@@ -2368,12 +2370,37 @@ bool SILParser::parseSILInstruction(SILBuilder &B) {
     // Drop the double quotes.
     StringRef rawString = P.Tok.getText().drop_front().drop_back();
 
-    // Ask the lexer to interpret the entire string as a literal segment.
-    SmallVector<char, 128> stringBuffer;
-    StringRef string = P.L->getEncodedStringSegment(rawString, stringBuffer);
     P.consumeToken(tok::string_literal);
     if (parseSILDebugLocation(InstLoc, B))
       return true;
+
+    // Ask the lexer to interpret the entire string as a literal segment.
+    SmallVector<char, 128> stringBuffer;
+
+    if (encoding == StringLiteralInst::Encoding::Bytes) {
+      // Decode hex bytes.
+      if (rawString.size() & 1) {
+        P.diagnose(P.Tok, diag::expected_tok_in_sil_instr,
+                   "even number of hex bytes");
+        return true;
+      }
+      while (!rawString.empty()) {
+        unsigned byte1 = llvm::hexDigitValue(rawString[0]);
+        unsigned byte2 = llvm::hexDigitValue(rawString[1]);
+        if (byte1 == -1U || byte2 == -1U) {
+          P.diagnose(P.Tok, diag::expected_tok_in_sil_instr,
+                     "hex bytes should contain 0-9, a-f, A-F only");
+          return true;
+        }
+        stringBuffer.push_back((unsigned char)(byte1 << 4) | byte2);
+        rawString = rawString.drop_front(2);
+      }
+
+      ResultVal = B.createStringLiteral(InstLoc, stringBuffer, encoding);
+      break;
+    }
+
+    StringRef string = P.L->getEncodedStringSegment(rawString, stringBuffer);
     ResultVal = B.createStringLiteral(InstLoc, string, encoding);
     break;
   }
diff --git a/lib/SIL/SILGlobalVariable.cpp b/lib/SIL/SILGlobalVariable.cpp
@@ -106,6 +106,7 @@ bool SILGlobalVariable::isValidStaticInitializerInst(const SILInstruction *I,
     }
     case SILInstructionKind::StringLiteralInst:
       switch (cast<StringLiteralInst>(I)->getEncoding()) {
+        case StringLiteralInst::Encoding::Bytes:
         case StringLiteralInst::Encoding::UTF8:
         case StringLiteralInst::Encoding::UTF16:
           return true;
diff --git a/lib/SIL/SILPrinter.cpp b/lib/SIL/SILPrinter.cpp
@@ -1184,6 +1184,7 @@ class SILPrinter : public SILInstructionVisitor<SILPrinter> {
   }
   static StringRef getStringEncodingName(StringLiteralInst::Encoding kind) {
     switch (kind) {
+    case StringLiteralInst::Encoding::Bytes: return "bytes ";
     case StringLiteralInst::Encoding::UTF8: return "utf8 ";
     case StringLiteralInst::Encoding::UTF16: return "utf16 ";
     case StringLiteralInst::Encoding::ObjCSelector: return "objc_selector ";
@@ -1192,8 +1193,17 @@ class SILPrinter : public SILInstructionVisitor<SILPrinter> {
   }
 
   void visitStringLiteralInst(StringLiteralInst *SLI) {
-    *this << getStringEncodingName(SLI->getEncoding())
-          << QuotedString(SLI->getValue());
+    *this << getStringEncodingName(SLI->getEncoding());
+
+    if (SLI->getEncoding() != StringLiteralInst::Encoding::Bytes) {
+      // FIXME: this isn't correct: this doesn't properly handle translating
+      // UTF16 into UTF8, and the SIL parser always parses as UTF8.
+      *this << QuotedString(SLI->getValue());
+      return;
+    }
+
+    // "Bytes" are always output in a hexadecimal form.
+    *this << '"' << llvm::toHex(SLI->getValue()) << '"';
   }
 
   static StringRef
diff --git a/lib/SILGen/SILGenApply.cpp b/lib/SILGen/SILGenApply.cpp
@@ -1522,8 +1522,9 @@ static RValue emitStringLiteral(SILGenFunction &SGF, Expr *E, StringRef Str,
     TypeElts = TypeEltsArray;
     break;
 
+  case StringLiteralInst::Encoding::Bytes:
   case StringLiteralInst::Encoding::ObjCSelector:
-    llvm_unreachable("Objective-C selectors cannot be formed here");
+    llvm_unreachable("these cannot be formed here");
   }
 
   CanType ty =
diff --git a/lib/SILOptimizer/Utils/SpecializationMangler.cpp b/lib/SILOptimizer/Utils/SpecializationMangler.cpp
@@ -236,6 +236,7 @@ FunctionSignatureSpecializationMangler::mangleConstantProp(LiteralInst *LI) {
 
     ArgOpBuffer << 's';
     switch (SLI->getEncoding()) {
+      case StringLiteralInst::Encoding::Bytes: ArgOpBuffer << 'B'; break;
       case StringLiteralInst::Encoding::UTF8: ArgOpBuffer << 'b'; break;
       case StringLiteralInst::Encoding::UTF16: ArgOpBuffer << 'w'; break;
       case StringLiteralInst::Encoding::ObjCSelector: ArgOpBuffer << 'c'; break;
diff --git a/lib/Serialization/DeserializeSIL.cpp b/lib/Serialization/DeserializeSIL.cpp
@@ -50,6 +50,7 @@ STATISTIC(NumDeserializedFunc, "Number of deserialized SIL functions");
 static Optional<StringLiteralInst::Encoding>
 fromStableStringEncoding(unsigned value) {
   switch (value) {
+  case SIL_BYTES: return StringLiteralInst::Encoding::Bytes;
   case SIL_UTF8: return StringLiteralInst::Encoding::UTF8;
   case SIL_UTF16: return StringLiteralInst::Encoding::UTF16;
   case SIL_OBJC_SELECTOR: return StringLiteralInst::Encoding::ObjCSelector;
diff --git a/lib/Serialization/SILFormat.h b/lib/Serialization/SILFormat.h
@@ -33,6 +33,7 @@ enum SILStringEncoding : uint8_t {
   SIL_UTF8,
   SIL_UTF16,
   SIL_OBJC_SELECTOR,
+  SIL_BYTES
 };
 
 enum SILLinkageEncoding : uint8_t {
diff --git a/lib/Serialization/SerializeSIL.cpp b/lib/Serialization/SerializeSIL.cpp
@@ -44,6 +44,7 @@ using llvm::BCBlockRAII;
 
 static unsigned toStableStringEncoding(StringLiteralInst::Encoding encoding) {
   switch (encoding) {
+  case StringLiteralInst::Encoding::Bytes: return SIL_BYTES;
   case StringLiteralInst::Encoding::UTF8: return SIL_UTF8;
   case StringLiteralInst::Encoding::UTF16: return SIL_UTF16;
   case StringLiteralInst::Encoding::ObjCSelector: return SIL_OBJC_SELECTOR;
diff --git a/test/SIL/Parser/string_literal.sil b/test/SIL/Parser/string_literal.sil
@@ -8,6 +8,8 @@ bb0:
   %2 = const_string_literal utf8 "\u{0B}"
 // CHECK: const_string_literal utf16 "\u{0B}"
   %3 = const_string_literal utf16 "\u{0B}"
+// CHECK: string_literal bytes "ABCD"
+  %4 = string_literal bytes "ABCD"
   %6 = tuple ()                                   // user: %7
   return %6 : $()
 }
diff --git a/test/SIL/Serialization/literals.sil b/test/SIL/Serialization/literals.sil
@@ -14,6 +14,8 @@ bb0:
   %2 = const_string_literal utf8 "\u{0B}"
 // CHECK: const_string_literal utf16 "\u{0B}"
   %3 = const_string_literal utf16 "\u{0B}"
+// CHECK: string_literal bytes "ABCD"
+  %4 = string_literal bytes "ABCD"
   %6 = tuple ()
   return %6 : $()
 }

Original file line number	Diff line number	Diff line change
`@@ -106,6 +106,7 @@ bool SILGlobalVariable::isValidStaticInitializerInst(const SILInstruction *I,`
`106`	`106`	`}`
`107`	`107`	`case SILInstructionKind::StringLiteralInst:`
`108`	`108`	`switch (cast<StringLiteralInst>(I)->getEncoding()) {`
	`109`	`+ case StringLiteralInst::Encoding::Bytes:`
`109`	`110`	`case StringLiteralInst::Encoding::UTF8:`
`110`	`111`	`case StringLiteralInst::Encoding::UTF16:`
`111`	`112`	`return true;`
Original file line number	Diff line number	Diff line change
`@@ -1522,8 +1522,9 @@ static RValue emitStringLiteral(SILGenFunction &SGF, Expr *E, StringRef Str,`
`1522`	`1522`	`TypeElts = TypeEltsArray;`
`1523`	`1523`	`break;`
`1524`	`1524`
	`1525`	`+ case StringLiteralInst::Encoding::Bytes:`
`1525`	`1526`	`case StringLiteralInst::Encoding::ObjCSelector:`
`1526`		`- llvm_unreachable("Objective-C selectors cannot be formed here");`
	`1527`	`+ llvm_unreachable("these cannot be formed here");`
`1527`	`1528`	`}`
`1528`	`1529`
`1529`	`1530`	`CanType ty =`
Original file line number	Diff line number	Diff line change
`@@ -8,6 +8,8 @@ bb0:`
`8`	`8`	`%2 = const_string_literal utf8 "\u{0B}"`
`9`	`9`	`// CHECK: const_string_literal utf16 "\u{0B}"`
`10`	`10`	`%3 = const_string_literal utf16 "\u{0B}"`
	`11`	`+// CHECK: string_literal bytes "ABCD"`
	`12`	`+ %4 = string_literal bytes "ABCD"`
`11`	`13`	`%6 = tuple () // user: %7`
`12`	`14`	`return %6 : $()`
`13`	`15`	`}`
Original file line number	Diff line number	Diff line change
`@@ -14,6 +14,8 @@ bb0:`
`14`	`14`	`%2 = const_string_literal utf8 "\u{0B}"`
`15`	`15`	`// CHECK: const_string_literal utf16 "\u{0B}"`
`16`	`16`	`%3 = const_string_literal utf16 "\u{0B}"`
	`17`	`+// CHECK: string_literal bytes "ABCD"`
	`18`	`+ %4 = string_literal bytes "ABCD"`
`17`	`19`	`%6 = tuple ()`
`18`	`20`	`return %6 : $()`
`19`	`21`	`}`