/// Determine whether the given text matches a property name.
static bool textMatchesPropertyName(StringRef text,
const InheritedNameSet *allPropertyNames) {
if (!allPropertyNames) return false;
SmallString<16> localScratch;
auto name = camel_case::toLowercaseWord(text, localScratch);
// A property with exactly this name.
if (allPropertyNames->contains(name)) return true;
// From here on, we'll be working with scratch space.
if (name.data() != localScratch.data())
localScratch = name;
if (localScratch.back() == 'y') {
// If the last letter is a 'y', try 'ies'.
localScratch.pop_back();
localScratch += "ies";
if (allPropertyNames->contains(localScratch)) return true;
} else {
// Otherwise, add an 's' and try again.
localScratch += 's';
if (allPropertyNames->contains(localScratch)) return true;
// Alternatively, try to add 'es'.
localScratch.pop_back();
localScratch += "es";
if (allPropertyNames->contains(localScratch)) return true;
}
return false;
}
void ObjCMigrateASTConsumer::HandleTranslationUnit(ASTContext &Ctx) {
Rewriter rewriter(Ctx.getSourceManager(), Ctx.getLangOpts());
RewritesReceiver Rec(rewriter);
Editor->applyRewrites(Rec);
for (Rewriter::buffer_iterator
I = rewriter.buffer_begin(), E = rewriter.buffer_end(); I != E; ++I) {
FileID FID = I->first;
RewriteBuffer &buf = I->second;
const FileEntry *file = Ctx.getSourceManager().getFileEntryForID(FID);
assert(file);
SmallString<512> newText;
llvm::raw_svector_ostream vecOS(newText);
buf.write(vecOS);
vecOS.flush();
llvm::MemoryBuffer *memBuf = llvm::MemoryBuffer::getMemBufferCopy(
StringRef(newText.data(), newText.size()), file->getName());
SmallString<64> filePath(file->getName());
FileMgr.FixupRelativePath(filePath);
Remapper.remap(filePath.str(), memBuf);
}
if (IsOutputFile) {
Remapper.flushToFile(MigrateDir, Ctx.getDiagnostics());
} else {
Remapper.flushToDisk(MigrateDir, Ctx.getDiagnostics());
}
}
static void PrintHeaderInfo(raw_ostream *OutputFile, const char* Filename,
bool ShowDepth, unsigned CurrentIncludeDepth,
bool MSStyle) {
// Write to a temporary string to avoid unnecessary flushing on errs().
SmallString<512> Pathname(Filename);
if (!MSStyle)
Lexer::Stringify(Pathname);
SmallString<256> Msg;
if (MSStyle)
Msg += "Note: including file:";
if (ShowDepth) {
// The main source file is at depth 1, so skip one dot.
for (unsigned i = 1; i != CurrentIncludeDepth; ++i)
Msg += MSStyle ? ' ' : '.';
if (!MSStyle)
Msg += ' ';
}
Msg += Pathname;
Msg += '\n';
OutputFile->write(Msg.data(), Msg.size());
OutputFile->flush();
}
void HeaderIncludesCallback::FileChanged(SourceLocation Loc,
FileChangeReason Reason,
SrcMgr::CharacteristicKind NewFileType,
FileID PrevFID) {
// Unless we are exiting a #include, make sure to skip ahead to the line the
// #include directive was at.
PresumedLoc UserLoc = SM.getPresumedLoc(Loc);
if (UserLoc.isInvalid())
return;
// Adjust the current include depth.
if (Reason == PPCallbacks::EnterFile) {
++CurrentIncludeDepth;
} else if (Reason == PPCallbacks::ExitFile) {
if (CurrentIncludeDepth)
--CurrentIncludeDepth;
// We track when we are done with the predefines by watching for the first
// place where we drop back to a nesting depth of 1.
if (CurrentIncludeDepth == 1 && !HasProcessedPredefines)
HasProcessedPredefines = true;
return;
} else
return;
// Show the header if we are (a) past the predefines, or (b) showing all
// headers and in the predefines at a depth past the initial file and command
// line buffers.
bool ShowHeader = (HasProcessedPredefines ||
(ShowAllHeaders && CurrentIncludeDepth > 2));
// Dump the header include information we are past the predefines buffer or
// are showing all headers.
if (ShowHeader && Reason == PPCallbacks::EnterFile) {
// Write to a temporary string to avoid unnecessary flushing on errs().
SmallString<512> Filename(UserLoc.getFilename());
if (!MSStyle)
Lexer::Stringify(Filename);
SmallString<256> Msg;
if (MSStyle)
Msg += "Note: including file:";
if (ShowDepth) {
// The main source file is at depth 1, so skip one dot.
for (unsigned i = 1; i != CurrentIncludeDepth; ++i)
Msg += MSStyle ? ' ' : '.';
if (!MSStyle)
Msg += ' ';
}
Msg += Filename;
Msg += '\n';
OutputFile->write(Msg.data(), Msg.size());
OutputFile->flush();
}
}
void Value::setName(const Twine &NewName) {
// Fast path for common IRBuilder case of setName("") when there is no name.
if (NewName.isTriviallyEmpty() && !hasName())
return;
SmallString<256> NameData;
NewName.toVector(NameData);
const char *NameStr = NameData.data();
unsigned NameLen = NameData.size();
// Name isn't changing?
if (getName() == StringRef(NameStr, NameLen))
return;
assert(getType() != Type::getVoidTy(getContext()) &&
"Cannot assign a name to void values!");
// Get the symbol table to update for this object.
ValueSymbolTable *ST;
if (getSymTab(this, ST))
return; // Cannot set a name on this value (e.g. constant).
if (!ST) { // No symbol table to update? Just do the change.
if (NameLen == 0) {
// Free the name for this value.
Name->Destroy();
Name = 0;
return;
}
if (Name)
Name->Destroy();
// NOTE: Could optimize for the case the name is shrinking to not deallocate
// then reallocated.
// Create the new name.
Name = ValueName::Create(NameStr, NameStr+NameLen);
Name->setValue(this);
return;
}
// NOTE: Could optimize for the case the name is shrinking to not deallocate
// then reallocated.
if (hasName()) {
// Remove old name.
ST->removeValueName(Name);
Name->Destroy();
Name = 0;
if (NameLen == 0)
return;
}
// Name is changing to something new.
Name = ST->createValueName(StringRef(NameStr, NameLen), this);
}
void verifyEncoding(MCDwarfLineTableParams Params, int LineDelta, int AddrDelta,
ArrayRef<uint8_t> ExpectedEncoding) {
SmallString<16> Buffer;
raw_svector_ostream EncodingOS(Buffer);
MCDwarfLineAddr::Encode(getContext(), Params, LineDelta, AddrDelta,
EncodingOS);
ArrayRef<uint8_t> Encoding(reinterpret_cast<uint8_t *>(Buffer.data()),
Buffer.size());
EXPECT_EQ(ExpectedEncoding, Encoding);
}
void ObjCMigrateASTConsumer::HandleTranslationUnit(ASTContext &Ctx) {
TranslationUnitDecl *TU = Ctx.getTranslationUnitDecl();
if (MigrateProperty)
for (DeclContext::decl_iterator D = TU->decls_begin(), DEnd = TU->decls_end();
D != DEnd; ++D) {
if (ObjCInterfaceDecl *CDecl = dyn_cast<ObjCInterfaceDecl>(*D))
migrateObjCInterfaceDecl(Ctx, CDecl);
else if (ObjCProtocolDecl *PDecl = dyn_cast<ObjCProtocolDecl>(*D))
ObjCProtocolDecls.insert(PDecl);
else if (const ObjCImplementationDecl *ImpDecl =
dyn_cast<ObjCImplementationDecl>(*D))
migrateProtocolConformance(Ctx, ImpDecl);
else if (const EnumDecl *ED = dyn_cast<EnumDecl>(*D)) {
DeclContext::decl_iterator N = D;
++N;
if (N != DEnd)
if (const TypedefDecl *TD = dyn_cast<TypedefDecl>(*N))
migrateNSEnumDecl(Ctx, ED, TD);
}
// migrate methods which can have instancetype as their result type.
if (ObjCContainerDecl *CDecl = dyn_cast<ObjCContainerDecl>(*D))
migrateInstanceType(Ctx, CDecl);
}
Rewriter rewriter(Ctx.getSourceManager(), Ctx.getLangOpts());
RewritesReceiver Rec(rewriter);
Editor->applyRewrites(Rec);
for (Rewriter::buffer_iterator
I = rewriter.buffer_begin(), E = rewriter.buffer_end(); I != E; ++I) {
FileID FID = I->first;
RewriteBuffer &buf = I->second;
const FileEntry *file = Ctx.getSourceManager().getFileEntryForID(FID);
assert(file);
SmallString<512> newText;
llvm::raw_svector_ostream vecOS(newText);
buf.write(vecOS);
vecOS.flush();
llvm::MemoryBuffer *memBuf = llvm::MemoryBuffer::getMemBufferCopy(
StringRef(newText.data(), newText.size()), file->getName());
SmallString<64> filePath(file->getName());
FileMgr.FixupRelativePath(filePath);
Remapper.remap(filePath.str(), memBuf);
}
if (IsOutputFile) {
Remapper.flushToFile(MigrateDir, Ctx.getDiagnostics());
} else {
Remapper.flushToDisk(MigrateDir, Ctx.getDiagnostics());
}
}
bool MigrationProcess::applyTransform(TransformFn trans,
RewriteListener *listener) {
OwningPtr<CompilerInvocation> CInvok;
CInvok.reset(createInvocationForMigration(OrigCI));
CInvok->getDiagnosticOpts().IgnoreWarnings = true;
Remapper.applyMappings(CInvok->getPreprocessorOpts());
CapturedDiagList capturedDiags;
std::vector<SourceLocation> ARCMTMacroLocs;
assert(DiagClient);
IntrusiveRefCntPtr<DiagnosticIDs> DiagID(new DiagnosticIDs());
IntrusiveRefCntPtr<DiagnosticsEngine> Diags(
new DiagnosticsEngine(DiagID, DiagClient, /*ShouldOwnClient=*/false));
// Filter of all diagnostics.
CaptureDiagnosticConsumer errRec(*Diags, capturedDiags);
Diags->setClient(&errRec, /*ShouldOwnClient=*/false);
OwningPtr<ARCMTMacroTrackerAction> ASTAction;
ASTAction.reset(new ARCMTMacroTrackerAction(ARCMTMacroLocs));
OwningPtr<ASTUnit> Unit(
ASTUnit::LoadFromCompilerInvocationAction(CInvok.take(), Diags,
ASTAction.get()));
if (!Unit)
return true;
Unit->setOwnsRemappedFileBuffers(false); // FileRemapper manages that.
// Don't filter diagnostics anymore.
Diags->setClient(DiagClient, /*ShouldOwnClient=*/false);
ASTContext &Ctx = Unit->getASTContext();
if (Diags->hasFatalErrorOccurred()) {
Diags->Reset();
DiagClient->BeginSourceFile(Ctx.getLangOpts(), &Unit->getPreprocessor());
capturedDiags.reportDiagnostics(*Diags);
DiagClient->EndSourceFile();
return true;
}
// After parsing of source files ended, we want to reuse the
// diagnostics objects to emit further diagnostics.
// We call BeginSourceFile because DiagnosticConsumer requires that
// diagnostics with source range information are emitted only in between
// BeginSourceFile() and EndSourceFile().
DiagClient->BeginSourceFile(Ctx.getLangOpts(), &Unit->getPreprocessor());
Rewriter rewriter(Ctx.getSourceManager(), Ctx.getLangOpts());
TransformActions TA(*Diags, capturedDiags, Ctx, Unit->getPreprocessor());
MigrationPass pass(Ctx, OrigCI.getLangOpts()->getGC(),
Unit->getSema(), TA, ARCMTMacroLocs);
trans(pass);
{
RewritesApplicator applicator(rewriter, Ctx, listener);
TA.applyRewrites(applicator);
}
DiagClient->EndSourceFile();
if (DiagClient->getNumErrors())
return true;
for (Rewriter::buffer_iterator
I = rewriter.buffer_begin(), E = rewriter.buffer_end(); I != E; ++I) {
FileID FID = I->first;
RewriteBuffer &buf = I->second;
const FileEntry *file = Ctx.getSourceManager().getFileEntryForID(FID);
assert(file);
std::string newFname = file->getName();
newFname += "-trans";
SmallString<512> newText;
llvm::raw_svector_ostream vecOS(newText);
buf.write(vecOS);
vecOS.flush();
llvm::MemoryBuffer *memBuf = llvm::MemoryBuffer::getMemBufferCopy(
StringRef(newText.data(), newText.size()), newFname);
SmallString<64> filePath(file->getName());
Unit->getFileManager().FixupRelativePath(filePath);
Remapper.remap(filePath.str(), memBuf);
}
return false;
}
static StringRef omitNeedlessWords(StringRef name,
OmissionTypeName typeName,
NameRole role,
const InheritedNameSet *allPropertyNames,
StringScratchSpace &scratch) {
// If we have no name or no type name, there is nothing to do.
if (name.empty() || typeName.empty()) return name;
// Get the camel-case words in the name and type name.
auto nameWords = camel_case::getWords(name);
auto typeWords = camel_case::getWords(typeName.Name);
// Match the last words in the type name to the last words in the
// name.
auto nameWordRevIter = nameWords.rbegin(),
nameWordRevIterBegin = nameWordRevIter,
firstMatchingNameWordRevIter = nameWordRevIter,
nameWordRevIterEnd = nameWords.rend();
auto typeWordRevIter = typeWords.rbegin(),
typeWordRevIterEnd = typeWords.rend();
bool anyMatches = false;
auto matched = [&] {
if (anyMatches) return;
anyMatches = true;
firstMatchingNameWordRevIter = nameWordRevIter;
};
while (nameWordRevIter != nameWordRevIterEnd &&
typeWordRevIter != typeWordRevIterEnd) {
// If the names match, continue.
auto nameWord = *nameWordRevIter;
if (matchNameWordToTypeWord(nameWord, *typeWordRevIter)) {
matched();
++nameWordRevIter;
++typeWordRevIter;
continue;
}
// Special case: "Indexes" and "Indices" in the name match
// "IndexSet" in the type.
if ((matchNameWordToTypeWord(nameWord, "Indexes") ||
matchNameWordToTypeWord(nameWord, "Indices")) &&
*typeWordRevIter == "Set") {
auto nextTypeWordRevIter = typeWordRevIter;
++nextTypeWordRevIter;
if (nextTypeWordRevIter != typeWordRevIterEnd &&
matchNameWordToTypeWord("Index", *nextTypeWordRevIter)) {
matched();
++nameWordRevIter;
typeWordRevIter = nextTypeWordRevIter;
++typeWordRevIter;
continue;
}
}
// Special case: "Index" in the name matches "Int" or "Integer" in the type.
if (matchNameWordToTypeWord(nameWord, "Index") &&
(matchNameWordToTypeWord("Int", *typeWordRevIter) ||
matchNameWordToTypeWord("Integer", *typeWordRevIter))) {
matched();
++nameWordRevIter;
++typeWordRevIter;
continue;
}
// Special case: if the word in the name ends in 's', and we have
// a collection element type, see if this is a plural.
if (!typeName.CollectionElement.empty() && nameWord.size() > 2 &&
nameWord.back() == 's' && role != NameRole::BaseNameSelf) {
// Check <element name>s.
auto shortenedNameWord
= name.substr(0, nameWordRevIter.base().getPosition()-1);
auto newShortenedNameWord
= omitNeedlessWords(shortenedNameWord, typeName.CollectionElement,
NameRole::Partial, allPropertyNames, scratch);
if (shortenedNameWord != newShortenedNameWord) {
matched();
unsigned targetSize = newShortenedNameWord.size();
while (nameWordRevIter.base().getPosition() > targetSize)
++nameWordRevIter;
continue;
}
}
// If this is a skippable suffix, skip it and keep looking.
if (nameWordRevIter == nameWordRevIterBegin) {
if (auto withoutSuffix = skipTypeSuffix(typeName.Name)) {
typeName.Name = *withoutSuffix;
typeWords = camel_case::getWords(typeName.Name);
typeWordRevIter = typeWords.rbegin();
typeWordRevIterEnd = typeWords.rend();
continue;
}
}
// If we're matching the base name of a method against the type of
// 'Self', and we haven't matched anything yet, skip over words in
// the name.
if (role == NameRole::BaseNameSelf && !anyMatches) {
++nameWordRevIter;
continue;
}
break;
}
StringRef origName = name;
// If we matched anything above, update the name appropriately.
if (anyMatches) {
// Handle complete name matches.
if (nameWordRevIter == nameWordRevIterEnd) {
// If we're doing a partial match, return the empty string.
if (role == NameRole::Partial) return "";
// Leave the name alone.
return name;
}
// Don't strip just "Error".
if (nameWordRevIter != nameWordRevIterBegin) {
auto nameWordPrev = std::prev(nameWordRevIter);
if (nameWordPrev == nameWordRevIterBegin && *nameWordPrev == "Error")
return name;
}
switch (role) {
case NameRole::Property:
// Always strip off type information.
name = name.substr(0, nameWordRevIter.base().getPosition());
break;
case NameRole::BaseNameSelf:
switch (getPartOfSpeech(*nameWordRevIter)) {
case PartOfSpeech::Verb: {
// Splice together the parts before and after the matched
// type. For example, if we matched "ViewController" in
// "dismissViewControllerAnimated", stitch together
// "dismissAnimated".
SmallString<16> newName =
name.substr(0, nameWordRevIter.base().getPosition());
newName
+= name.substr(firstMatchingNameWordRevIter.base().getPosition());
name = scratch.copyString(newName);
break;
}
case PartOfSpeech::Preposition:
case PartOfSpeech::Gerund:
case PartOfSpeech::Unknown:
return name;
}
break;
case NameRole::BaseName:
case NameRole::FirstParameter:
case NameRole::Partial:
case NameRole::SubsequentParameter:
// Classify the part of speech of the word before the type
// information we would strip off.
switch (getPartOfSpeech(*nameWordRevIter)) {
case PartOfSpeech::Preposition:
if (role == NameRole::BaseName) {
// Strip off the part of the name that is redundant with
// type information, so long as there's something preceding the
// preposition.
if (std::next(nameWordRevIter) != nameWordRevIterEnd)
name = name.substr(0, nameWordRevIter.base().getPosition());
break;
}
SWIFT_FALLTHROUGH;
case PartOfSpeech::Verb:
case PartOfSpeech::Gerund:
// Don't prune redundant type information from the base name if
// there is a corresponding property (either singular or plural).
if (allPropertyNames && role == NameRole::BaseName) {
SmallString<16> localScratch;
auto removedText = name.substr(nameWordRevIter.base().getPosition());
auto removedName = camel_case::toLowercaseWord(removedText,
localScratch);
// A property with exactly this name.
if (allPropertyNames->contains(removedName)) return name;
// From here on, we'll be working with scratch space.
if (removedName.data() != localScratch.data())
localScratch = removedName;
if (localScratch.back() == 'y') {
// If the last letter is a 'y', try 'ies'.
localScratch.pop_back();
localScratch += "ies";
if (allPropertyNames->contains(localScratch)) return name;
} else {
// Otherwise, add an 's' and try again.
localScratch += 's';
if (allPropertyNames->contains(localScratch)) return name;
// Alternatively, try to add 'es'.
localScratch.pop_back();
localScratch += "es";
if (allPropertyNames->contains(localScratch)) return name;
}
}
// Strip off the part of the name that is redundant with
// type information.
name = name.substr(0, nameWordRevIter.base().getPosition());
break;
case PartOfSpeech::Unknown:
// Assume it's a noun or adjective; don't strip anything.
break;
}
break;
}
}
// If we ended up with a vacuous name like "get" or "set", do nothing.
if (isVacuousName(name))
return origName;
switch (role) {
case NameRole::BaseName:
case NameRole::BaseNameSelf:
case NameRole::Property:
// If we ended up with a keyword for a property name or base name,
// do nothing.
if (isKeyword(name))
return origName;
break;
case NameRole::SubsequentParameter:
case NameRole::FirstParameter:
case NameRole::Partial:
break;
}
// We're done.
return name;
}
C2::ExprResult C2Sema::ActOnNumericConstant(const Token& Tok) {
#ifdef SEMA_DEBUG
std::cerr << COL_SEMA << "SEMA: numeric constant" << ANSI_NORMAL"\n";
#endif
// Fast path for a single digit (which is quite common). A single digit
// cannot have a trigraph, escaped newline, radix prefix, or suffix.
if (Tok.getLength() == 1) {
const char Val = PP.getSpellingOfSingleCharacterNumericConstant(Tok);
return ActOnIntegerConstant(Tok.getLocation(), Val-'0');
}
SmallString<128> SpellingBuffer;
// NumericLiteralParser wants to overread by one character. Add padding to
// the buffer in case the token is copied to the buffer. If getSpelling()
// returns a StringRef to the memory buffer, it should have a null char at
// the EOF, so it is also safe.
SpellingBuffer.resize(Tok.getLength() + 1);
// Get the spelling of the token, which eliminates trigraphs, etc.
bool Invalid = false;
StringRef TokSpelling = PP.getSpelling(Tok, SpellingBuffer, &Invalid);
if (Invalid)
return ExprError();
NumericLiteralParser Literal(TokSpelling, Tok.getLocation(), PP);
if (Literal.hadError)
return ExprError();
if (Literal.hasUDSuffix()) {
assert(0 && "HUH?");
}
Expr* Res;
if (Literal.isFloatingLiteral()) {
// clang::Sema::BuildFloatingLiteral()
// TEMP Hardcoded
const llvm::fltSemantics& Format = llvm::APFloat::IEEEsingle;
APFloat Val(Format);
APFloat::opStatus result = Literal.GetFloatValue(Val);
// Overflow is always an error, but underflow is only an error if
// we underflowed to zero (APFloat reports denormals as underflow).
if ((result & APFloat::opOverflow) ||
((result & APFloat::opUnderflow) && Val.isZero())) {
assert(0 && "TODO");
#if 0
unsigned diagnostic;
SmallString<20> buffer;
if (result & APFloat::opOverflow) {
diagnostic = diag::warn_float_overflow;
APFloat::getLargest(Format).toString(buffer);
} else {
diagnostic = diag::warn_float_underflow;
APFloat::getSmallest(Format).toString(buffer);
}
Diag(Tok.getLocation(), diagnostic)
<< Ty
<< StringRef(buffer.data(), buffer.size());
#endif
}
//bool isExact = (result == APFloat::opOK);
//return FloatingLiteral::Create(S.Context, Val, isExact, Ty, Loc);
Res = new FloatingLiteral(Tok.getLocation(), Val);
} else if (!Literal.isIntegerLiteral()) {
return ExprError();
} else {
QualType ty;
const unsigned MaxWidth = 64; // for now limit to 64 bits
llvm::APInt ResultVal(MaxWidth, 0);
if (Literal.GetIntegerValue(ResultVal)) {
Diag(Tok.getLocation(), diag::err_integer_literal_too_large) << 1;
} else {
#if 0
// Octal, Hexadecimal, and integers with a U suffix are allowed to
// be an unsigned.
bool AllowUnsigned = Literal.isUnsigned || Literal.getRadix() != 10;
// Check from smallest to largest, picking the smallest type we can.
unsigned Width = 0;
if (!Literal.isLong && !Literal.isLongLong) {
// Are int/unsigned possibilities?
unsigned IntSize = 64;
// Does it fit in a unsigned?
if (ResultVal.isIntN(IntSize)) {
// Does it fit in a signed int?
#if 0
if (!Literal.isUnsigned && ResultVal[IntSize-1] == 0)
Ty = Context.IntTy;
else if (AllowUnsigned)
Ty = Context.UnsignedIntTy;
#endif
Width = IntSize;
}
}
// Check long long if needed.
if (Width == 0) {
if (ResultVal.isIntN(64)) {
#if 0
if (!Literal.isUnsigned && (ResultVal[LongLongSize-1] == 0 ||
(getLangOpts().MicrosoftExt && Literal.isLongLong)))
Ty = Context.LongLongTy;
else if (AllowUnsigned)
Ty = Context.UnsignedLongLongTy;
#endif
Width = 64;
}
}
if (Width == 0) {
fprintf(stderr, "TOO LARGE\n");
assert(0 && "TODO");
}
// set correct width
if (ResultVal.getBitWidth() != Width) {
ResultVal = ResultVal.trunc(Width);
}
#endif
}
Res = new IntegerLiteral(Tok.getLocation(), ResultVal, Literal.getRadix());
}
return ExprResult(Res);
}
/// StringifyArgument - Implement C99 6.10.3.2p2, converting a sequence of
/// tokens into the literal string token that should be produced by the C #
/// preprocessor operator. If Charify is true, then it should be turned into
/// a character literal for the Microsoft charize (#@) extension.
///
Token MacroArgs::StringifyArgument(const Token *ArgToks,
Preprocessor &PP, bool Charify,
SourceLocation ExpansionLocStart,
SourceLocation ExpansionLocEnd) {
Token Tok;
Tok.startToken();
Tok.setKind(Charify ? tok::char_constant : tok::string_literal);
const Token *ArgTokStart = ArgToks;
// Stringify all the tokens.
SmallString<128> Result;
Result += "\"";
bool isFirst = true;
for (; ArgToks->isNot(tok::eof); ++ArgToks) {
const Token &Tok = *ArgToks;
if (!isFirst && (Tok.hasLeadingSpace() || Tok.isAtStartOfLine()))
Result += ' ';
isFirst = false;
// If this is a string or character constant, escape the token as specified
// by 6.10.3.2p2.
if (tok::isStringLiteral(Tok.getKind()) || // "foo", u8R"x(foo)x"_bar, etc.
Tok.is(tok::char_constant) || // 'x'
Tok.is(tok::wide_char_constant) || // L'x'.
Tok.is(tok::utf8_char_constant) || // u8'x'.
Tok.is(tok::utf16_char_constant) || // u'x'.
Tok.is(tok::utf32_char_constant)) { // U'x'.
bool Invalid = false;
std::string TokStr = PP.getSpelling(Tok, &Invalid);
if (!Invalid) {
std::string Str = Lexer::Stringify(TokStr);
Result.append(Str.begin(), Str.end());
}
} else if (Tok.is(tok::code_completion)) {
PP.CodeCompleteNaturalLanguage();
} else {
// Otherwise, just append the token. Do some gymnastics to get the token
// in place and avoid copies where possible.
unsigned CurStrLen = Result.size();
Result.resize(CurStrLen+Tok.getLength());
const char *BufPtr = Result.data() + CurStrLen;
bool Invalid = false;
unsigned ActualTokLen = PP.getSpelling(Tok, BufPtr, &Invalid);
if (!Invalid) {
// If getSpelling returned a pointer to an already uniqued version of
// the string instead of filling in BufPtr, memcpy it onto our string.
if (ActualTokLen && BufPtr != &Result[CurStrLen])
memcpy(&Result[CurStrLen], BufPtr, ActualTokLen);
// If the token was dirty, the spelling may be shorter than the token.
if (ActualTokLen != Tok.getLength())
Result.resize(CurStrLen+ActualTokLen);
}
}
}
// If the last character of the string is a \, and if it isn't escaped, this
// is an invalid string literal, diagnose it as specified in C99.
if (Result.back() == '\\') {
// Count the number of consecutive \ characters. If even, then they are
// just escaped backslashes, otherwise it's an error.
unsigned FirstNonSlash = Result.size()-2;
// Guaranteed to find the starting " if nothing else.
while (Result[FirstNonSlash] == '\\')
--FirstNonSlash;
if ((Result.size()-1-FirstNonSlash) & 1) {
// Diagnose errors for things like: #define F(X) #X / F(\)
PP.Diag(ArgToks[-1], diag::pp_invalid_string_literal);
Result.pop_back(); // remove one of the \'s.
}
}
Result += '"';
// If this is the charify operation and the result is not a legal character
// constant, diagnose it.
if (Charify) {
// First step, turn double quotes into single quotes:
Result[0] = '\'';
Result[Result.size()-1] = '\'';
// Check for bogus character.
bool isBad = false;
if (Result.size() == 3)
isBad = Result[1] == '\''; // ''' is not legal. '\' already fixed above.
else
isBad = (Result.size() != 4 || Result[1] != '\\'); // Not '\x'
if (isBad) {
PP.Diag(ArgTokStart[0], diag::err_invalid_character_to_charify);
Result = "' '"; // Use something arbitrary, but legal.
}
}
PP.CreateString(Result, Tok,
ExpansionLocStart, ExpansionLocEnd);
return Tok;
}
/// PasteTokens - Tok is the LHS of a ## operator, and CurToken is the ##
/// operator. Read the ## and RHS, and paste the LHS/RHS together. If there
/// are more ## after it, chomp them iteratively. Return the result as Tok.
/// If this returns true, the caller should immediately return the token.
bool TokenLexer::PasteTokens(Token &Tok) {
// MSVC: If previous token was pasted, this must be a recovery from an invalid
// paste operation. Ignore spaces before this token to mimic MSVC output.
// Required for generating valid UUID strings in some MS headers.
if (PP.getLangOpts().MicrosoftExt && (CurToken >= 2) &&
Tokens[CurToken - 2].is(tok::hashhash))
Tok.clearFlag(Token::LeadingSpace);
SmallString<128> Buffer;
const char *ResultTokStrPtr = nullptr;
SourceLocation StartLoc = Tok.getLocation();
SourceLocation PasteOpLoc;
do {
// Consume the ## operator if any.
PasteOpLoc = Tokens[CurToken].getLocation();
if (Tokens[CurToken].is(tok::hashhash))
++CurToken;
assert(!isAtEnd() && "No token on the RHS of a paste operator!");
// Get the RHS token.
const Token &RHS = Tokens[CurToken];
// Allocate space for the result token. This is guaranteed to be enough for
// the two tokens.
Buffer.resize(Tok.getLength() + RHS.getLength());
// Get the spelling of the LHS token in Buffer.
const char *BufPtr = &Buffer[0];
bool Invalid = false;
unsigned LHSLen = PP.getSpelling(Tok, BufPtr, &Invalid);
if (BufPtr != &Buffer[0]) // Really, we want the chars in Buffer!
memcpy(&Buffer[0], BufPtr, LHSLen);
if (Invalid)
return true;
BufPtr = Buffer.data() + LHSLen;
unsigned RHSLen = PP.getSpelling(RHS, BufPtr, &Invalid);
if (Invalid)
return true;
if (RHSLen && BufPtr != &Buffer[LHSLen])
// Really, we want the chars in Buffer!
memcpy(&Buffer[LHSLen], BufPtr, RHSLen);
// Trim excess space.
Buffer.resize(LHSLen+RHSLen);
// Plop the pasted result (including the trailing newline and null) into a
// scratch buffer where we can lex it.
Token ResultTokTmp;
ResultTokTmp.startToken();
// Claim that the tmp token is a string_literal so that we can get the
// character pointer back from CreateString in getLiteralData().
ResultTokTmp.setKind(tok::string_literal);
PP.CreateString(Buffer, ResultTokTmp);
SourceLocation ResultTokLoc = ResultTokTmp.getLocation();
ResultTokStrPtr = ResultTokTmp.getLiteralData();
// Lex the resultant pasted token into Result.
Token Result;
if (Tok.isAnyIdentifier() && RHS.isAnyIdentifier()) {
// Common paste case: identifier+identifier = identifier. Avoid creating
// a lexer and other overhead.
PP.IncrementPasteCounter(true);
Result.startToken();
Result.setKind(tok::raw_identifier);
Result.setRawIdentifierData(ResultTokStrPtr);
Result.setLocation(ResultTokLoc);
Result.setLength(LHSLen+RHSLen);
} else {
PP.IncrementPasteCounter(false);
assert(ResultTokLoc.isFileID() &&
"Should be a raw location into scratch buffer");
SourceManager &SourceMgr = PP.getSourceManager();
FileID LocFileID = SourceMgr.getFileID(ResultTokLoc);
bool Invalid = false;
const char *ScratchBufStart
= SourceMgr.getBufferData(LocFileID, &Invalid).data();
if (Invalid)
return false;
// Make a lexer to lex this string from. Lex just this one token.
// Make a lexer object so that we lex and expand the paste result.
Lexer TL(SourceMgr.getLocForStartOfFile(LocFileID),
PP.getLangOpts(), ScratchBufStart,
ResultTokStrPtr, ResultTokStrPtr+LHSLen+RHSLen);
// Lex a token in raw mode. This way it won't look up identifiers
// automatically, lexing off the end will return an eof token, and
// warnings are disabled. This returns true if the result token is the
// entire buffer.
bool isInvalid = !TL.LexFromRawLexer(Result);
// If we got an EOF token, we didn't form even ONE token. For example, we
// did "/ ## /" to get "//".
isInvalid |= Result.is(tok::eof);
// If pasting the two tokens didn't form a full new token, this is an
// error. This occurs with "x ## +" and other stuff. Return with Tok
// unmodified and with RHS as the next token to lex.
if (isInvalid) {
// Explicitly convert the token location to have proper expansion
// information so that the user knows where it came from.
SourceManager &SM = PP.getSourceManager();
SourceLocation Loc =
SM.createExpansionLoc(PasteOpLoc, ExpandLocStart, ExpandLocEnd, 2);
// Test for the Microsoft extension of /##/ turning into // here on the
// error path.
if (PP.getLangOpts().MicrosoftExt && Tok.is(tok::slash) &&
RHS.is(tok::slash)) {
HandleMicrosoftCommentPaste(Tok, Loc);
return true;
}
// Do not emit the error when preprocessing assembler code.
if (!PP.getLangOpts().AsmPreprocessor) {
// If we're in microsoft extensions mode, downgrade this from a hard
// error to an extension that defaults to an error. This allows
// disabling it.
PP.Diag(Loc, PP.getLangOpts().MicrosoftExt ? diag::ext_pp_bad_paste_ms
: diag::err_pp_bad_paste)
<< Buffer;
}
// An error has occurred so exit loop.
break;
}
// Turn ## into 'unknown' to avoid # ## # from looking like a paste
// operator.
if (Result.is(tok::hashhash))
Result.setKind(tok::unknown);
}
// Transfer properties of the LHS over the Result.
Result.setFlagValue(Token::StartOfLine , Tok.isAtStartOfLine());
Result.setFlagValue(Token::LeadingSpace, Tok.hasLeadingSpace());
// Finally, replace LHS with the result, consume the RHS, and iterate.
++CurToken;
Tok = Result;
} while (!isAtEnd() && Tokens[CurToken].is(tok::hashhash));
SourceLocation EndLoc = Tokens[CurToken - 1].getLocation();
// The token's current location indicate where the token was lexed from. We
// need this information to compute the spelling of the token, but any
// diagnostics for the expanded token should appear as if the token was
// expanded from the full ## expression. Pull this information together into
// a new SourceLocation that captures all of this.
SourceManager &SM = PP.getSourceManager();
if (StartLoc.isFileID())
StartLoc = getExpansionLocForMacroDefLoc(StartLoc);
if (EndLoc.isFileID())
EndLoc = getExpansionLocForMacroDefLoc(EndLoc);
FileID MacroFID = SM.getFileID(MacroExpansionStart);
while (SM.getFileID(StartLoc) != MacroFID)
StartLoc = SM.getImmediateExpansionRange(StartLoc).first;
while (SM.getFileID(EndLoc) != MacroFID)
EndLoc = SM.getImmediateExpansionRange(EndLoc).second;
Tok.setLocation(SM.createExpansionLoc(Tok.getLocation(), StartLoc, EndLoc,
Tok.getLength()));
// Now that we got the result token, it will be subject to expansion. Since
// token pasting re-lexes the result token in raw mode, identifier information
// isn't looked up. As such, if the result is an identifier, look up id info.
if (Tok.is(tok::raw_identifier)) {
// Look up the identifier info for the token. We disabled identifier lookup
// by saying we're skipping contents, so we need to do this manually.
PP.LookUpIdentifierInfo(Tok);
}
return false;
}
int main(int ac, char **av)
{
SourceMgr SrcMgr;
//LLVMInitializeX86TargetInfo();
llvm::InitializeAllTargetInfos();
//LLVMInitializeX86AsmParser();
llvm::InitializeAllTargetMCs();
//LLVMInitializeX86TargetMC();
llvm::InitializeAllAsmParsers();
//LLVMInitializeX86AsmParser();
llvm::InitializeAllDisassemblers();
//LLVMInitializeX86Disassembler();
// arg0:
// llvm::Target encapsulating the "x86_64-apple-darwin14.5.0" information
// see /lib/Support/Triple.cpp for the details
//spec = llvm::sys::getDefaultTargetTriple();
//std::string machSpec = "x86_64-apple-windows"; // will produce a COFF
//std::string machSpec = "x86_64-apple-darwin14.5.0"; // will produce a Mach-O
std::string machSpec = "arm-none-none-eabi"; //
//std::string machSpec = "x86_64-apple-darwin";
//std::string machSpec = "x86_64-thumb-linux-gnu";
//std::string machSpec = "x86_64-unknown-linux-gnu";
printf("machine spec: %s\n", machSpec.c_str());
machSpec = Triple::normalize(machSpec);
printf("machine spec (normalized): %s\n", machSpec.c_str());
Triple TheTriple(machSpec);
// Get the target specific parser.
std::string Error;
const Target *TheTarget = TargetRegistry::lookupTarget(/*arch*/"", TheTriple, Error);
if (!TheTarget) {
errs() << Error;
return -1;
}
machSpec = TheTriple.getTriple();
printf("machine spec (returned): %s\n", machSpec.c_str());
printf("Target.getName(): %s\n", TheTarget->getName());
printf("Target.getShortDescription(): %s\n", TheTarget->getShortDescription());
/* from the target we get almost everything */
std::unique_ptr<MCRegisterInfo> MRI(TheTarget->createMCRegInfo(machSpec));
std::unique_ptr<MCAsmInfo> MAI(TheTarget->createMCAsmInfo(*MRI, machSpec));
std::unique_ptr<MCInstrInfo> MCII(TheTarget->createMCInstrInfo()); /* describes target instruction set */
MCSubtargetInfo *STI = TheTarget->createMCSubtargetInfo(machSpec, "", ""); /* subtarget instr set */
MCAsmBackend *MAB = TheTarget->createMCAsmBackend(*MRI, machSpec, /* specific CPU */ "");
// arg0:
// llvm::SourceMgr (Support/SourceMgr.h) that holds assembler source
// has vector of llvm::SrcBuffer encaps (Support/MemoryBuffer.h) and vector of include dirs
//std::string asmSrc = ".org 0x100, 0xAA\nfoo:\nxor %eax, %ebx\npush %rbp\njmp foo\nrdtsc\n";
//std::string asmSrc = ".text\n" "ldr pc, data_foo\n" "\n" "data_foo:\n" " .int 0x8\n" "\n" "loop:\n" "b loop\n";
//std::string asmSrc = ".text\n" "mov r2, r1\n";
std::string asmSrc = ".text\n" "ldr pc, data_foo\n" "data_foo:\n" ".int 0x8\n" "loop:\n" "b loop\n";
std::unique_ptr<MemoryBuffer> memBuf = MemoryBuffer::getMemBuffer(asmSrc);
SrcMgr.AddNewSourceBuffer(std::move(memBuf), SMLoc());
// arg1: the machine code context
MCObjectFileInfo MOFI;
MCContext Ctx(MAI.get(), MRI.get(), &MOFI, &SrcMgr);
MOFI.InitMCObjectFileInfo(TheTriple, Reloc::Default, CodeModel::Default, Ctx);
// this is the assembler interface
// -methods per .s statements (emit bytes, handle directive, etc.)
// -remembers current section
// -implementations that write a .s, or .o in various formats
//
// 1. the output stream ... a formatted_raw_ostream wraps a raw_ostream to provide
// tracking of line and column position for padding and shit
//
// but raw_ostream is abstract and is implemented by raw_fd_ostream, raw_string_ostream, etc.
/* output stream:
raw_svector_ostream is a raw_pwrite_stream is a raw_ostream
since a SmallString is SmallVector (svector) we can use this and
retrieve bytes later with its .data() method */
SmallString<1024> smallString;
raw_svector_ostream rso(smallString);
/* code emitter needs 1) instruction set info 2) register info */
MCCodeEmitter *CE = TheTarget->createMCCodeEmitter(*MCII, *MRI, Ctx);
MCStreamer *as = TheTarget->createMCObjectStreamer(
TheTriple, /* Triple */
Ctx, /* the MCContext */
*MAB, /* the AsmBackend, (fixups, relaxation, objs and elfs) */
rso, /* output stream raw_pwrite_stream */
CE, /* code emitter */
*STI, /* subtarget info */
true, /* relax all fixups */
true, /* incremental linker compatible */
false /* DWARFMustBeAtTheEnd */
);
std::string abi = "none";
MCTargetOptions toptions;
toptions.MCUseDwarfDirectory = false;
toptions.ABIName = abi;
printf("trying to assemble, let's go..\n");
AssembleInput(TheTarget, SrcMgr, Ctx, *as, *MAI, *STI,
*MCII, toptions);
printf("done with AssembleInput()\n");
/* dump to file for debugging */
FILE *fp;
fp = fopen("out.bin", "wb");
fwrite(smallString.data(), 1, smallString.size(), fp);
fclose(fp);
//int n = smallString.size();
int codeOffset=0, codeSize = 0;
char *data = smallString.data();
if(*(uint32_t *)data == 0xFEEDFACF) {
unsigned int idx = 0;
idx += 0x20; /* skip mach_header_64 to first command */
idx += 0x48; /* advance into segment_command_64 to first section */
idx += 0x28; /* advance into section_64 to size */
uint64_t scn_size = *(uint64_t *)(data + idx);
idx += 0x8; /* advance into section_64 to offset */
uint64_t scn_offset = *(uint64_t *)(data + idx);
codeOffset = scn_offset;
codeSize = scn_size;
}
else if(0==memcmp(data, "\x7F" "ELF\x01\x01\x01\x00", 8)) {
/* assume four sections: NULL, .strtab, .text, .symtab */
uint32_t e_shoff = *(uint32_t *)(data + 0x20);
uint32_t sh_offset = *(uint32_t *)(data + e_shoff + 2*0x28 + 0x10); /* second shdr */
uint32_t sh_size = *(uint32_t *)(data + e_shoff + 2*0x28 + 0x14); /* second shdr */
codeOffset = sh_offset;
codeSize = sh_size;
}
else {
printf("ERROR: couldn't identify type of output file\n");
}
dump_bytes((unsigned char *)data + codeOffset, codeSize, 0);
return 0;
}