bool PragmaCommentHandler::HandleComment(Preprocessor &PP, SourceRange Range) {
StringRef Text =
Lexer::getSourceText(CharSourceRange::getCharRange(Range),
PP.getSourceManager(), PP.getLangOpts());
size_t Pos = Text.find(IWYUPragma);
if (Pos == StringRef::npos)
return false;
StringRef RemappingFilePath = Text.substr(Pos + std::strlen(IWYUPragma));
Collector->addHeaderMapping(
PP.getSourceManager().getFilename(Range.getBegin()),
RemappingFilePath.trim("\"<>"));
return false;
}
bool HandleComment(Preprocessor &PP, SourceRange Range) override {
StringRef Text =
Lexer::getSourceText(CharSourceRange::getCharRange(Range),
PP.getSourceManager(), PP.getLangOpts());
if (!Text.consume_front(IWYUPragma))
return false;
// FIXME(ioeric): resolve the header and store actual file path. For now,
// we simply assume the written header is suitable to be #included.
Includes->addMapping(PP.getSourceManager().getFilename(Range.getBegin()),
isLiteralInclude(Text) ? Text.str()
: ("\"" + Text + "\"").str());
return false;
}
/// Returns true if the statement is expanded from a configuration macro.
static bool isExpandedFromConfigurationMacro(const Stmt *S,
Preprocessor &PP,
bool IgnoreYES_NO = false) {
// FIXME: This is not very precise. Here we just check to see if the
// value comes from a macro, but we can do much better. This is likely
// to be over conservative. This logic is factored into a separate function
// so that we can refine it later.
SourceLocation L = S->getBeginLoc();
if (L.isMacroID()) {
SourceManager &SM = PP.getSourceManager();
if (IgnoreYES_NO) {
// The Objective-C constant 'YES' and 'NO'
// are defined as macros. Do not treat them
// as configuration values.
SourceLocation TopL = getTopMostMacro(L, SM);
StringRef MacroName = PP.getImmediateMacroName(TopL);
if (MacroName == "YES" || MacroName == "NO")
return false;
} else if (!PP.getLangOpts().CPlusPlus) {
// Do not treat C 'false' and 'true' macros as configuration values.
SourceLocation TopL = getTopMostMacro(L, SM);
StringRef MacroName = PP.getImmediateMacroName(TopL);
if (MacroName == "false" || MacroName == "true")
return false;
}
return true;
}
return false;
}
/// Initializes an InclusionRewriter with a \p PP source and \p OS destination.
InclusionRewriter::InclusionRewriter(Preprocessor &PP, raw_ostream &OS,
bool ShowLineMarkers,
bool UseLineDirectives)
: PP(PP), SM(PP.getSourceManager()), OS(OS), MainEOL("\n"),
PredefinesBuffer(nullptr), ShowLineMarkers(ShowLineMarkers),
UseLineDirectives(UseLineDirectives),
LastInclusionLocation(SourceLocation()) {}
/// InclusionRewriterInInput - Implement -frewrite-includes mode.
void clang::RewriteIncludesInInput(Preprocessor &PP, raw_ostream *OS,
const PreprocessorOutputOptions &Opts) {
SourceManager &SM = PP.getSourceManager();
InclusionRewriter *Rewrite = new InclusionRewriter(PP, *OS,
Opts.ShowLineMarkers);
PP.addPPCallbacks(Rewrite);
PP.IgnorePragmas();
// First let the preprocessor process the entire file and call callbacks.
// Callbacks will record which #include's were actually performed.
PP.EnterMainSourceFile();
Token Tok;
// Only preprocessor directives matter here, so disable macro expansion
// everywhere else as an optimization.
// TODO: It would be even faster if the preprocessor could be switched
// to a mode where it would parse only preprocessor directives and comments,
// nothing else matters for parsing or processing.
PP.SetMacroExpansionOnlyInDirectives();
do {
PP.Lex(Tok);
} while (Tok.isNot(tok::eof));
Rewrite->setPredefinesBuffer(SM.getBuffer(PP.getPredefinesFileID()));
Rewrite->Process(PP.getPredefinesFileID(), SrcMgr::C_User);
Rewrite->Process(SM.getMainFileID(), SrcMgr::C_User);
OS->flush();
}
/// LexRawTokensFromMainFile - Lets all the raw tokens from the main file into
/// the specified vector.
static void LexRawTokensFromMainFile(Preprocessor &PP,
std::vector<Token> &RawTokens) {
SourceManager &SM = PP.getSourceManager();
// Create a lexer to lex all the tokens of the main file in raw mode. Even
// though it is in raw mode, it will not return comments.
const llvm::MemoryBuffer *FromFile = SM.getBuffer(SM.getMainFileID());
Lexer RawLex(SM.getMainFileID(), FromFile, SM, PP.getLangOpts());
// Switch on comment lexing because we really do want them.
RawLex.SetCommentRetentionState(true);
Token RawTok;
do {
RawLex.LexFromRawLexer(RawTok);
// If we have an identifier with no identifier info for our raw token, look
// up the indentifier info. This is important for equality comparison of
// identifier tokens.
if (RawTok.is(tok::raw_identifier))
PP.LookUpIdentifierInfo(RawTok);
RawTokens.push_back(RawTok);
} while (RawTok.isNot(tok::eof));
}
void clang::CacheTokens(Preprocessor &PP, llvm::raw_fd_ostream* OS) {
// Get the name of the main file.
const SourceManager &SrcMgr = PP.getSourceManager();
const FileEntry *MainFile = SrcMgr.getFileEntryForID(SrcMgr.getMainFileID());
SmallString<128> MainFilePath(MainFile->getName());
llvm::sys::fs::make_absolute(MainFilePath);
// Create the PTHWriter.
PTHWriter PW(*OS, PP);
// Install the 'stat' system call listener in the FileManager.
StatListener *StatCache = new StatListener(PW.getPM());
PP.getFileManager().addStatCache(StatCache, /*AtBeginning=*/true);
// Lex through the entire file. This will populate SourceManager with
// all of the header information.
Token Tok;
PP.EnterMainSourceFile();
do { PP.Lex(Tok); } while (Tok.isNot(tok::eof));
// Generate the PTH file.
PP.getFileManager().removeStatCache(StatCache);
PW.GeneratePTH(MainFilePath.str());
}
/// InitializePreprocessor - Initialize the preprocessor getting it and the
/// environment ready to process a single file. This returns true on error.
///
void clang::InitializePreprocessor(Preprocessor &PP,
const PreprocessorOptions &InitOpts,
const HeaderSearchOptions &HSOpts) {
std::vector<char> PredefineBuffer;
InitializeFileRemapping(PP.getDiagnostics(), PP.getSourceManager(),
PP.getFileManager(), InitOpts);
const char *LineDirective = "# 1 \"<built-in>\" 3\n";
PredefineBuffer.insert(PredefineBuffer.end(),
LineDirective, LineDirective+strlen(LineDirective));
// Install things like __POWERPC__, __GNUC__, etc into the macro table.
if (InitOpts.UsePredefines)
InitializePredefinedMacros(PP.getTargetInfo(), PP.getLangOptions(),
PredefineBuffer);
// Add on the predefines from the driver. Wrap in a #line directive to report
// that they come from the command line.
LineDirective = "# 1 \"<command line>\" 1\n";
PredefineBuffer.insert(PredefineBuffer.end(),
LineDirective, LineDirective+strlen(LineDirective));
// Process #define's and #undef's in the order they are given.
for (unsigned i = 0, e = InitOpts.Macros.size(); i != e; ++i) {
if (InitOpts.Macros[i].second) // isUndef
UndefineBuiltinMacro(PredefineBuffer, InitOpts.Macros[i].first.c_str());
else
DefineBuiltinMacro(PredefineBuffer, InitOpts.Macros[i].first.c_str(),
&PP.getDiagnostics());
}
// If -imacros are specified, include them now. These are processed before
// any -include directives.
for (unsigned i = 0, e = InitOpts.MacroIncludes.size(); i != e; ++i)
AddImplicitIncludeMacros(PredefineBuffer, InitOpts.MacroIncludes[i]);
// Process -include directives.
for (unsigned i = 0, e = InitOpts.Includes.size(); i != e; ++i) {
const std::string &Path = InitOpts.Includes[i];
if (Path == InitOpts.ImplicitPTHInclude)
AddImplicitIncludePTH(PredefineBuffer, PP, Path);
else
AddImplicitInclude(PredefineBuffer, Path);
}
// Exit the command line and go back to <built-in> (2 is LC_LEAVE).
LineDirective = "# 1 \"<built-in>\" 2\n";
PredefineBuffer.insert(PredefineBuffer.end(),
LineDirective, LineDirective+strlen(LineDirective));
// Null terminate PredefinedBuffer and add it.
PredefineBuffer.push_back(0);
PP.setPredefines(&PredefineBuffer[0]);
// Initialize the header search object.
ApplyHeaderSearchOptions(PP.getHeaderSearchInfo(), HSOpts,
PP.getLangOptions(),
PP.getTargetInfo().getTriple());
}
/// FindExpectedDiags - Lex the main source file to find all of the
// expected errors and warnings.
static void FindExpectedDiags(Preprocessor &PP, ExpectedData &ED) {
// Create a raw lexer to pull all the comments out of the main file. We don't
// want to look in #include'd headers for expected-error strings.
SourceManager &SM = PP.getSourceManager();
FileID FID = SM.getMainFileID();
if (SM.getMainFileID().isInvalid())
return;
// Create a lexer to lex all the tokens of the main file in raw mode.
const llvm::MemoryBuffer *FromFile = SM.getBuffer(FID);
Lexer RawLex(FID, FromFile, SM, PP.getLangOptions());
// Return comments as tokens, this is how we find expected diagnostics.
RawLex.SetCommentRetentionState(true);
Token Tok;
Tok.setKind(tok::comment);
while (Tok.isNot(tok::eof)) {
RawLex.Lex(Tok);
if (!Tok.is(tok::comment)) continue;
std::string Comment = PP.getSpelling(Tok);
if (Comment.empty()) continue;
// Find all expected errors/warnings/notes.
ParseDirective(&Comment[0], Comment.size(), ED, PP, Tok.getLocation());
};
}
/// Initializes an InclusionRewriter with a \p PP source and \p OS destination.
InclusionRewriter::InclusionRewriter(Preprocessor &PP, raw_ostream &OS,
bool ShowLineMarkers)
: PP(PP), SM(PP.getSourceManager()), OS(OS),
ShowLineMarkers(ShowLineMarkers),
LastInsertedFileChange(FileChanges.end()) {
// If we're in microsoft mode, use normal #line instead of line markers.
UseLineDirective = PP.getLangOpts().MicrosoftExt;
}
/// HandleComment - Hook into the preprocessor and extract comments containing
/// expected errors and warnings.
bool VerifyDiagnosticConsumer::HandleComment(Preprocessor &PP,
SourceRange Comment) {
SourceManager &SM = PP.getSourceManager();
// If this comment is for a different source manager, ignore it.
if (SrcManager && &SM != SrcManager)
return false;
SourceLocation CommentBegin = Comment.getBegin();
const char *CommentRaw = SM.getCharacterData(CommentBegin);
StringRef C(CommentRaw, SM.getCharacterData(Comment.getEnd()) - CommentRaw);
if (C.empty())
return false;
// Fold any "\<EOL>" sequences
size_t loc = C.find('\\');
if (loc == StringRef::npos) {
ParseDirective(C, &ED, SM, &PP, CommentBegin, Status, *Markers);
return false;
}
std::string C2;
C2.reserve(C.size());
for (size_t last = 0;; loc = C.find('\\', last)) {
if (loc == StringRef::npos || loc == C.size()) {
C2 += C.substr(last);
break;
}
C2 += C.substr(last, loc-last);
last = loc + 1;
if (C[last] == '\n' || C[last] == '\r') {
++last;
// Escape \r\n or \n\r, but not \n\n.
if (last < C.size())
if (C[last] == '\n' || C[last] == '\r')
if (C[last] != C[last-1])
++last;
} else {
// This was just a normal backslash.
C2 += '\\';
}
}
if (!C2.empty())
ParseDirective(C2, &ED, SM, &PP, CommentBegin, Status, *Markers);
return false;
}
/// InitializePreprocessor - Initialize the preprocessor getting it and the
/// environment ready to process a single file. This returns true on error.
///
void mlang::InitializePreprocessor(Preprocessor &PP,
const PreprocessorOptions &InitOpts,
const ImportSearchOptions &HSOpts,
const FrontendOptions &FEOpts) {
InitializeFileRemapping(PP.getDiagnostics(), PP.getSourceManager(),
PP.getFileManager(), InitOpts);
// Initialize the import search object.
// ApplyImportSearchOptions(PP.getImportSearchInfo(), HSOpts,
// PP.getLangOptions(),
// PP.getTargetInfo().getTriple());
}
static void emitPremigrationErrors(const CapturedDiagList &arcDiags,
const DiagnosticOptions &diagOpts,
Preprocessor &PP) {
TextDiagnosticPrinter printer(llvm::errs(), diagOpts);
IntrusiveRefCntPtr<DiagnosticIDs> DiagID(new DiagnosticIDs());
IntrusiveRefCntPtr<DiagnosticsEngine> Diags(
new DiagnosticsEngine(DiagID, &printer, /*ShouldOwnClient=*/false));
Diags->setSourceManager(&PP.getSourceManager());
printer.BeginSourceFile(PP.getLangOpts(), &PP);
arcDiags.reportDiagnostics(*Diags);
printer.EndSourceFile();
}
/// HandleComment - Hook into the preprocessor and extract comments containing
/// expected errors and warnings.
bool VerifyDiagnosticConsumer::HandleComment(Preprocessor &PP,
SourceRange Comment) {
SourceManager &SM = PP.getSourceManager();
SourceLocation CommentBegin = Comment.getBegin();
const char *CommentRaw = SM.getCharacterData(CommentBegin);
StringRef C(CommentRaw, SM.getCharacterData(Comment.getEnd()) - CommentRaw);
if (C.empty())
return false;
// Fold any "\<EOL>" sequences
size_t loc = C.find('\\');
if (loc == StringRef::npos) {
if (ParseDirective(C, ED, SM, CommentBegin, PP.getDiagnostics()))
if (const FileEntry *E = SM.getFileEntryForID(SM.getFileID(CommentBegin)))
FilesWithDirectives.insert(E);
return false;
}
std::string C2;
C2.reserve(C.size());
for (size_t last = 0;; loc = C.find('\\', last)) {
if (loc == StringRef::npos || loc == C.size()) {
C2 += C.substr(last);
break;
}
C2 += C.substr(last, loc-last);
last = loc + 1;
if (C[last] == '\n' || C[last] == '\r') {
++last;
// Escape \r\n or \n\r, but not \n\n.
if (last < C.size())
if (C[last] == '\n' || C[last] == '\r')
if (C[last] != C[last-1])
++last;
} else {
// This was just a normal backslash.
C2 += '\\';
}
}
if (!C2.empty())
if (ParseDirective(C2, ED, SM, CommentBegin, PP.getDiagnostics()))
if (const FileEntry *E = SM.getFileEntryForID(SM.getFileID(CommentBegin)))
FilesWithDirectives.insert(E);
return false;
}
/// FindExpectedDiags - Lex the main source file to find all of the
// expected errors and warnings.
static void FindExpectedDiags(Preprocessor &PP,
DiagList &ExpectedErrors,
DiagList &ExpectedWarnings,
DiagList &ExpectedNotes) {
// Create a raw lexer to pull all the comments out of the main file. We don't
// want to look in #include'd headers for expected-error strings.
FileID FID = PP.getSourceManager().getMainFileID();
// Create a lexer to lex all the tokens of the main file in raw mode.
Lexer RawLex(FID, PP.getSourceManager(), PP.getLangOptions());
// Return comments as tokens, this is how we find expected diagnostics.
RawLex.SetCommentRetentionState(true);
Token Tok;
Tok.setKind(tok::comment);
while (Tok.isNot(tok::eof)) {
RawLex.Lex(Tok);
if (!Tok.is(tok::comment)) continue;
std::string Comment = PP.getSpelling(Tok);
if (Comment.empty()) continue;
// Find all expected errors.
FindDiagnostics(&Comment[0], Comment.size(), ExpectedErrors, PP,
Tok.getLocation(), "expected-error");
// Find all expected warnings.
FindDiagnostics(&Comment[0], Comment.size(), ExpectedWarnings, PP,
Tok.getLocation(), "expected-warning");
// Find all expected notes.
FindDiagnostics(&Comment[0], Comment.size(), ExpectedNotes, PP,
Tok.getLocation(), "expected-note");
};
}
/// CheckResults - This compares the expected results to those that
/// were actually reported. It emits any discrepencies. Return "true" if there
/// were problems. Return "false" otherwise.
///
static bool CheckResults(Preprocessor &PP,
const DiagList &ExpectedErrors,
const DiagList &ExpectedWarnings,
const DiagList &ExpectedNotes) {
const DiagnosticClient *DiagClient = PP.getDiagnostics().getClient();
assert(DiagClient != 0 &&
"DiagChecker requires a valid TextDiagnosticBuffer");
const TextDiagnosticBuffer &Diags =
static_cast<const TextDiagnosticBuffer&>(*DiagClient);
SourceManager &SourceMgr = PP.getSourceManager();
// We want to capture the delta between what was expected and what was
// seen.
//
// Expected \ Seen - set expected but not seen
// Seen \ Expected - set seen but not expected
bool HadProblem = false;
// See if there are error mismatches.
HadProblem |= CompareDiagLists(SourceMgr,
ExpectedErrors.begin(), ExpectedErrors.end(),
Diags.err_begin(), Diags.err_end(),
"Errors expected but not seen:",
"Errors seen but not expected:");
// See if there are warning mismatches.
HadProblem |= CompareDiagLists(SourceMgr,
ExpectedWarnings.begin(),
ExpectedWarnings.end(),
Diags.warn_begin(), Diags.warn_end(),
"Warnings expected but not seen:",
"Warnings seen but not expected:");
// See if there are note mismatches.
HadProblem |= CompareDiagLists(SourceMgr,
ExpectedNotes.begin(),
ExpectedNotes.end(),
Diags.note_begin(), Diags.note_end(),
"Notes expected but not seen:",
"Notes seen but not expected:");
return HadProblem;
}
/// GetFirstChar - Get the first character of the token \arg Tok,
/// avoiding calls to getSpelling where possible.
static char GetFirstChar(Preprocessor &PP, const Token &Tok) {
if (IdentifierInfo *II = Tok.getIdentifierInfo()) {
// Avoid spelling identifiers, the most common form of token.
return II->getName()[0];
} else if (!Tok.needsCleaning()) {
if (Tok.isLiteral() && Tok.getLiteralData()) {
return *Tok.getLiteralData();
} else {
SourceManager &SM = PP.getSourceManager();
return *SM.getCharacterData(SM.getSpellingLoc(Tok.getLocation()));
}
} else if (Tok.getLength() < 256) {
char Buffer[256];
const char *TokPtr = Buffer;
PP.getSpelling(Tok, TokPtr);
return TokPtr[0];
} else {
return PP.getSpelling(Tok)[0];
}
}
void clang::DoRewriteTest(Preprocessor &PP, raw_ostream* OS) {
SourceManager &SM = PP.getSourceManager();
const LangOptions &LangOpts = PP.getLangOptions();
TokenRewriter Rewriter(SM.getMainFileID(), SM, LangOpts);
// Throw <i> </i> tags around comments.
for (TokenRewriter::token_iterator I = Rewriter.token_begin(),
E = Rewriter.token_end(); I != E; ++I) {
if (I->isNot(tok::comment)) continue;
Rewriter.AddTokenBefore(I, "<i>");
Rewriter.AddTokenAfter(I, "</i>");
}
// Print out the output.
for (TokenRewriter::token_iterator I = Rewriter.token_begin(),
E = Rewriter.token_end(); I != E; ++I)
*OS << PP.getSpelling(*I);
}
bool HandleComment(Preprocessor &PP, SourceRange Range) override {
StringRef Text =
Lexer::getSourceText(CharSourceRange::getCharRange(Range),
PP.getSourceManager(), PP.getLangOpts());
SmallVector<StringRef, 4> Matches;
if (!TodoMatch.match(Text, &Matches))
return false;
StringRef Username = Matches[1];
StringRef Comment = Matches[3];
if (!Username.empty())
return false;
std::string NewText = ("// TODO(" + Twine(User) + "): " + Comment).str();
Check.diag(Range.getBegin(), "missing username/bug in TODO")
<< FixItHint::CreateReplacement(CharSourceRange::getCharRange(Range),
NewText);
return false;
}
/// Finish - This does final analysis of the declspec, rejecting things like
/// "_Imaginary" (lacking an FP type). This returns a diagnostic to issue or
/// diag::NUM_DIAGNOSTICS if there is no error. After calling this method,
/// DeclSpec is guaranteed self-consistent, even if an error occurred.
void DeclSpec::Finish(Diagnostic &D, Preprocessor &PP) {
// Check the type specifier components first.
SourceManager &SrcMgr = PP.getSourceManager();
// signed/unsigned are only valid with int/char/wchar_t.
if (TypeSpecSign != TSS_unspecified) {
if (TypeSpecType == TST_unspecified)
TypeSpecType = TST_int; // unsigned -> unsigned int, signed -> signed int.
else if (TypeSpecType != TST_int &&
TypeSpecType != TST_char && TypeSpecType != TST_wchar) {
Diag(D, TSSLoc, SrcMgr, diag::err_invalid_sign_spec)
<< getSpecifierName((TST)TypeSpecType);
// signed double -> double.
TypeSpecSign = TSS_unspecified;
}
}
// Validate the width of the type.
switch (TypeSpecWidth) {
case TSW_unspecified: break;
case TSW_short: // short int
case TSW_longlong: // long long int
if (TypeSpecType == TST_unspecified)
TypeSpecType = TST_int; // short -> short int, long long -> long long int.
else if (TypeSpecType != TST_int) {
Diag(D, TSWLoc, SrcMgr,
TypeSpecWidth == TSW_short ? diag::err_invalid_short_spec
: diag::err_invalid_longlong_spec)
<< getSpecifierName((TST)TypeSpecType);
TypeSpecType = TST_int;
}
break;
case TSW_long: // long double, long int
if (TypeSpecType == TST_unspecified)
TypeSpecType = TST_int; // long -> long int.
else if (TypeSpecType != TST_int && TypeSpecType != TST_double) {
Diag(D, TSWLoc, SrcMgr, diag::err_invalid_long_spec)
<< getSpecifierName((TST)TypeSpecType);
TypeSpecType = TST_int;
}
break;
}
// TODO: if the implementation does not implement _Complex or _Imaginary,
// disallow their use. Need information about the backend.
if (TypeSpecComplex != TSC_unspecified) {
if (TypeSpecType == TST_unspecified) {
Diag(D, TSCLoc, SrcMgr, diag::ext_plain_complex)
<< CodeModificationHint::CreateInsertion(
PP.getLocForEndOfToken(getTypeSpecComplexLoc()),
" double");
TypeSpecType = TST_double; // _Complex -> _Complex double.
} else if (TypeSpecType == TST_int || TypeSpecType == TST_char) {
// Note that this intentionally doesn't include _Complex _Bool.
Diag(D, TSTLoc, SrcMgr, diag::ext_integer_complex);
} else if (TypeSpecType != TST_float && TypeSpecType != TST_double) {
Diag(D, TSCLoc, SrcMgr, diag::err_invalid_complex_spec)
<< getSpecifierName((TST)TypeSpecType);
TypeSpecComplex = TSC_unspecified;
}
}
// C++ [class.friend]p6:
// No storage-class-specifier shall appear in the decl-specifier-seq
// of a friend declaration.
if (isFriendSpecified() && getStorageClassSpec()) {
DeclSpec::SCS SC = getStorageClassSpec();
const char *SpecName = getSpecifierName(SC);
SourceLocation SCLoc = getStorageClassSpecLoc();
SourceLocation SCEndLoc = SCLoc.getFileLocWithOffset(strlen(SpecName));
Diag(D, SCLoc, SrcMgr, diag::err_friend_storage_spec)
<< SpecName
<< CodeModificationHint::CreateRemoval(SourceRange(SCLoc, SCEndLoc));
ClearStorageClassSpecs();
}
// Okay, now we can infer the real type.
// TODO: return "auto function" and other bad things based on the real type.
// 'data definition has no type or storage class'?
}
/// RewriteMacrosInInput - Implement -rewrite-macros mode.
void clang::RewriteMacrosInInput(Preprocessor &PP, raw_ostream *OS) {
SourceManager &SM = PP.getSourceManager();
Rewriter Rewrite;
Rewrite.setSourceMgr(SM, PP.getLangOpts());
RewriteBuffer &RB = Rewrite.getEditBuffer(SM.getMainFileID());
std::vector<Token> RawTokens;
LexRawTokensFromMainFile(PP, RawTokens);
unsigned CurRawTok = 0;
Token RawTok = GetNextRawTok(RawTokens, CurRawTok, false);
// Get the first preprocessing token.
PP.EnterMainSourceFile();
Token PPTok;
PP.Lex(PPTok);
// Preprocess the input file in parallel with raw lexing the main file. Ignore
// all tokens that are preprocessed from a file other than the main file (e.g.
// a header). If we see tokens that are in the preprocessed file but not the
// lexed file, we have a macro expansion. If we see tokens in the lexed file
// that aren't in the preprocessed view, we have macros that expand to no
// tokens, or macro arguments etc.
while (RawTok.isNot(tok::eof) || PPTok.isNot(tok::eof)) {
SourceLocation PPLoc = SM.getExpansionLoc(PPTok.getLocation());
// If PPTok is from a different source file, ignore it.
if (!SM.isFromMainFile(PPLoc)) {
PP.Lex(PPTok);
continue;
}
// If the raw file hits a preprocessor directive, they will be extra tokens
// in the raw file that don't exist in the preprocsesed file. However, we
// choose to preserve them in the output file and otherwise handle them
// specially.
if (RawTok.is(tok::hash) && RawTok.isAtStartOfLine()) {
// If this is a #warning directive or #pragma mark (GNU extensions),
// comment the line out.
if (RawTokens[CurRawTok].is(tok::identifier)) {
const IdentifierInfo *II = RawTokens[CurRawTok].getIdentifierInfo();
if (II->getName() == "warning") {
// Comment out #warning.
RB.InsertTextAfter(SM.getFileOffset(RawTok.getLocation()), "//");
} else if (II->getName() == "pragma" &&
RawTokens[CurRawTok+1].is(tok::identifier) &&
(RawTokens[CurRawTok+1].getIdentifierInfo()->getName() ==
"mark")) {
// Comment out #pragma mark.
RB.InsertTextAfter(SM.getFileOffset(RawTok.getLocation()), "//");
}
}
// Otherwise, if this is a #include or some other directive, just leave it
// in the file by skipping over the line.
RawTok = GetNextRawTok(RawTokens, CurRawTok, false);
while (!RawTok.isAtStartOfLine() && RawTok.isNot(tok::eof))
RawTok = GetNextRawTok(RawTokens, CurRawTok, false);
continue;
}
// Okay, both tokens are from the same file. Get their offsets from the
// start of the file.
unsigned PPOffs = SM.getFileOffset(PPLoc);
unsigned RawOffs = SM.getFileOffset(RawTok.getLocation());
// If the offsets are the same and the token kind is the same, ignore them.
if (PPOffs == RawOffs && isSameToken(RawTok, PPTok)) {
RawTok = GetNextRawTok(RawTokens, CurRawTok, false);
PP.Lex(PPTok);
continue;
}
// If the PP token is farther along than the raw token, something was
// deleted. Comment out the raw token.
if (RawOffs <= PPOffs) {
// Comment out a whole run of tokens instead of bracketing each one with
// comments. Add a leading space if RawTok didn't have one.
bool HasSpace = RawTok.hasLeadingSpace();
RB.InsertTextAfter(RawOffs, &" /*"[HasSpace]);
unsigned EndPos;
do {
EndPos = RawOffs+RawTok.getLength();
RawTok = GetNextRawTok(RawTokens, CurRawTok, true);
RawOffs = SM.getFileOffset(RawTok.getLocation());
if (RawTok.is(tok::comment)) {
// Skip past the comment.
RawTok = GetNextRawTok(RawTokens, CurRawTok, false);
break;
}
} while (RawOffs <= PPOffs && !RawTok.isAtStartOfLine() &&
(PPOffs != RawOffs || !isSameToken(RawTok, PPTok)));
RB.InsertTextBefore(EndPos, "*/");
continue;
}
// Otherwise, there was a replacement an expansion. Insert the new token
// in the output buffer. Insert the whole run of new tokens at once to get
// them in the right order.
unsigned InsertPos = PPOffs;
std::string Expansion;
while (PPOffs < RawOffs) {
Expansion += ' ' + PP.getSpelling(PPTok);
PP.Lex(PPTok);
PPLoc = SM.getExpansionLoc(PPTok.getLocation());
PPOffs = SM.getFileOffset(PPLoc);
}
Expansion += ' ';
RB.InsertTextBefore(InsertPos, Expansion);
}
// Get the buffer corresponding to MainFileID. If we haven't changed it, then
// we are done.
if (const RewriteBuffer *RewriteBuf =
Rewrite.getRewriteBufferFor(SM.getMainFileID())) {
//printf("Changed:\n");
*OS << std::string(RewriteBuf->begin(), RewriteBuf->end());
} else {
fprintf(stderr, "No changes\n");
}
OS->flush();
}
/// HighlightMacros - This uses the macro table state from the end of the
/// file, to re-expand macros and insert (into the HTML) information about the
/// macro expansions. This won't be perfectly perfect, but it will be
/// reasonably close.
void html::HighlightMacros(Rewriter &R, FileID FID, Preprocessor& PP) {
// Re-lex the raw token stream into a token buffer.
const SourceManager &SM = PP.getSourceManager();
std::vector<Token> TokenStream;
Lexer L(FID, SM, PP.getLangOptions());
// Lex all the tokens in raw mode, to avoid entering #includes or expanding
// macros.
while (1) {
Token Tok;
L.LexFromRawLexer(Tok);
// If this is a # at the start of a line, discard it from the token stream.
// We don't want the re-preprocess step to see #defines, #includes or other
// preprocessor directives.
if (Tok.is(tok::hash) && Tok.isAtStartOfLine())
continue;
// If this is a ## token, change its kind to unknown so that repreprocessing
// it will not produce an error.
if (Tok.is(tok::hashhash))
Tok.setKind(tok::unknown);
// If this raw token is an identifier, the raw lexer won't have looked up
// the corresponding identifier info for it. Do this now so that it will be
// macro expanded when we re-preprocess it.
if (Tok.is(tok::identifier)) {
// Change the kind of this identifier to the appropriate token kind, e.g.
// turning "for" into a keyword.
Tok.setKind(PP.LookUpIdentifierInfo(Tok)->getTokenID());
}
TokenStream.push_back(Tok);
if (Tok.is(tok::eof)) break;
}
// Temporarily change the diagnostics object so that we ignore any generated
// diagnostics from this pass.
IgnoringDiagClient TmpDC;
Diagnostic TmpDiags(&TmpDC);
Diagnostic *OldDiags = &PP.getDiagnostics();
PP.setDiagnostics(TmpDiags);
// Inform the preprocessor that we don't want comments.
PP.SetCommentRetentionState(false, false);
// Enter the tokens we just lexed. This will cause them to be macro expanded
// but won't enter sub-files (because we removed #'s).
PP.EnterTokenStream(&TokenStream[0], TokenStream.size(), false, false);
TokenConcatenation ConcatInfo(PP);
// Lex all the tokens.
Token Tok;
PP.Lex(Tok);
while (Tok.isNot(tok::eof)) {
// Ignore non-macro tokens.
if (!Tok.getLocation().isMacroID()) {
PP.Lex(Tok);
continue;
}
// Okay, we have the first token of a macro expansion: highlight the
// instantiation by inserting a start tag before the macro instantiation and
// end tag after it.
std::pair<SourceLocation, SourceLocation> LLoc =
SM.getInstantiationRange(Tok.getLocation());
// Ignore tokens whose instantiation location was not the main file.
if (SM.getFileID(LLoc.first) != FID) {
PP.Lex(Tok);
continue;
}
assert(SM.getFileID(LLoc.second) == FID &&
"Start and end of expansion must be in the same ultimate file!");
std::string Expansion = PP.getSpelling(Tok);
unsigned LineLen = Expansion.size();
Token PrevTok = Tok;
// Okay, eat this token, getting the next one.
PP.Lex(Tok);
// Skip all the rest of the tokens that are part of this macro
// instantiation. It would be really nice to pop up a window with all the
// spelling of the tokens or something.
while (!Tok.is(tok::eof) &&
SM.getInstantiationLoc(Tok.getLocation()) == LLoc.first) {
// Insert a newline if the macro expansion is getting large.
if (LineLen > 60) {
Expansion += "<br>";
LineLen = 0;
}
LineLen -= Expansion.size();
// If the tokens were already space separated, or if they must be to avoid
// them being implicitly pasted, add a space between them.
if (Tok.hasLeadingSpace() ||
ConcatInfo.AvoidConcat(PrevTok, Tok))
Expansion += ' ';
// Escape any special characters in the token text.
Expansion += EscapeText(PP.getSpelling(Tok));
LineLen += Expansion.size();
PrevTok = Tok;
PP.Lex(Tok);
}
// Insert the expansion as the end tag, so that multi-line macros all get
// highlighted.
Expansion = "<span class='expansion'>" + Expansion + "</span></span>";
HighlightRange(R, LLoc.first, LLoc.second,
"<span class='macro'>", Expansion.c_str());
}
// Restore diagnostics object back to its own thing.
PP.setDiagnostics(*OldDiags);
}
/// HighlightMacros - This uses the macro table state from the end of the
/// file, to re-expand macros and insert (into the HTML) information about the
/// macro expansions. This won't be perfectly perfect, but it will be
/// reasonably close.
void html::HighlightMacros(Rewriter &R, FileID FID, const Preprocessor& PP) {
// Re-lex the raw token stream into a token buffer.
const SourceManager &SM = PP.getSourceManager();
std::vector<Token> TokenStream;
const llvm::MemoryBuffer *FromFile = SM.getBuffer(FID);
Lexer L(FID, FromFile, SM, PP.getLangOpts());
// Lex all the tokens in raw mode, to avoid entering #includes or expanding
// macros.
while (1) {
Token Tok;
L.LexFromRawLexer(Tok);
// If this is a # at the start of a line, discard it from the token stream.
// We don't want the re-preprocess step to see #defines, #includes or other
// preprocessor directives.
if (Tok.is(tok::hash) && Tok.isAtStartOfLine())
continue;
// If this is a ## token, change its kind to unknown so that repreprocessing
// it will not produce an error.
if (Tok.is(tok::hashhash))
Tok.setKind(tok::unknown);
// If this raw token is an identifier, the raw lexer won't have looked up
// the corresponding identifier info for it. Do this now so that it will be
// macro expanded when we re-preprocess it.
if (Tok.is(tok::raw_identifier))
PP.LookUpIdentifierInfo(Tok);
TokenStream.push_back(Tok);
if (Tok.is(tok::eof)) break;
}
// Temporarily change the diagnostics object so that we ignore any generated
// diagnostics from this pass.
DiagnosticsEngine TmpDiags(PP.getDiagnostics().getDiagnosticIDs(),
&PP.getDiagnostics().getDiagnosticOptions(),
new IgnoringDiagConsumer);
// FIXME: This is a huge hack; we reuse the input preprocessor because we want
// its state, but we aren't actually changing it (we hope). This should really
// construct a copy of the preprocessor.
Preprocessor &TmpPP = const_cast<Preprocessor&>(PP);
DiagnosticsEngine *OldDiags = &TmpPP.getDiagnostics();
TmpPP.setDiagnostics(TmpDiags);
// Inform the preprocessor that we don't want comments.
TmpPP.SetCommentRetentionState(false, false);
// We don't want pragmas either. Although we filtered out #pragma, removing
// _Pragma and __pragma is much harder.
bool PragmasPreviouslyEnabled = TmpPP.getPragmasEnabled();
TmpPP.setPragmasEnabled(false);
// Enter the tokens we just lexed. This will cause them to be macro expanded
// but won't enter sub-files (because we removed #'s).
TmpPP.EnterTokenStream(TokenStream, false);
TokenConcatenation ConcatInfo(TmpPP);
// Lex all the tokens.
Token Tok;
TmpPP.Lex(Tok);
while (Tok.isNot(tok::eof)) {
// Ignore non-macro tokens.
if (!Tok.getLocation().isMacroID()) {
TmpPP.Lex(Tok);
continue;
}
// Okay, we have the first token of a macro expansion: highlight the
// expansion by inserting a start tag before the macro expansion and
// end tag after it.
std::pair<SourceLocation, SourceLocation> LLoc =
SM.getExpansionRange(Tok.getLocation());
// Ignore tokens whose instantiation location was not the main file.
if (SM.getFileID(LLoc.first) != FID) {
TmpPP.Lex(Tok);
continue;
}
assert(SM.getFileID(LLoc.second) == FID &&
"Start and end of expansion must be in the same ultimate file!");
std::string Expansion = EscapeText(TmpPP.getSpelling(Tok));
unsigned LineLen = Expansion.size();
Token PrevPrevTok;
Token PrevTok = Tok;
// Okay, eat this token, getting the next one.
TmpPP.Lex(Tok);
// Skip all the rest of the tokens that are part of this macro
// instantiation. It would be really nice to pop up a window with all the
// spelling of the tokens or something.
while (!Tok.is(tok::eof) &&
SM.getExpansionLoc(Tok.getLocation()) == LLoc.first) {
// Insert a newline if the macro expansion is getting large.
if (LineLen > 60) {
Expansion += "<br>";
LineLen = 0;
}
LineLen -= Expansion.size();
// If the tokens were already space separated, or if they must be to avoid
// them being implicitly pasted, add a space between them.
if (Tok.hasLeadingSpace() ||
ConcatInfo.AvoidConcat(PrevPrevTok, PrevTok, Tok))
Expansion += ' ';
// Escape any special characters in the token text.
Expansion += EscapeText(TmpPP.getSpelling(Tok));
LineLen += Expansion.size();
PrevPrevTok = PrevTok;
PrevTok = Tok;
TmpPP.Lex(Tok);
}
// Insert the expansion as the end tag, so that multi-line macros all get
// highlighted.
Expansion = "<span class='expansion'>" + Expansion + "</span></span>";
HighlightRange(R, LLoc.first, LLoc.second,
"<span class='macro'>", Expansion.c_str());
}
// Restore the preprocessor's old state.
TmpPP.setDiagnostics(*OldDiags);
TmpPP.setPragmasEnabled(PragmasPreviouslyEnabled);
}
void DependencyCollector::attachToPreprocessor(Preprocessor &PP) {
PP.addPPCallbacks(
llvm::make_unique<DepCollectorPPCallbacks>(*this, PP.getSourceManager()));
}
/// Finish - This does final analysis of the declspec, rejecting things like
/// "_Imaginary" (lacking an FP type). This returns a diagnostic to issue or
/// diag::NUM_DIAGNOSTICS if there is no error. After calling this method,
/// DeclSpec is guaranteed self-consistent, even if an error occurred.
void DeclSpec::Finish(Diagnostic &D, Preprocessor &PP) {
// Before possibly changing their values, save specs as written.
SaveWrittenBuiltinSpecs();
SaveStorageSpecifierAsWritten();
// Check the type specifier components first.
SourceManager &SrcMgr = PP.getSourceManager();
// Validate and finalize AltiVec vector declspec.
if (TypeAltiVecVector) {
if (TypeAltiVecBool) {
// Sign specifiers are not allowed with vector bool. (PIM 2.1)
if (TypeSpecSign != TSS_unspecified) {
Diag(D, TSSLoc, SrcMgr, diag::err_invalid_vector_bool_decl_spec)
<< getSpecifierName((TSS)TypeSpecSign);
}
// Only char/int are valid with vector bool. (PIM 2.1)
if (((TypeSpecType != TST_unspecified) && (TypeSpecType != TST_char) &&
(TypeSpecType != TST_int)) || TypeAltiVecPixel) {
Diag(D, TSTLoc, SrcMgr, diag::err_invalid_vector_bool_decl_spec)
<< (TypeAltiVecPixel ? "__pixel" :
getSpecifierName((TST)TypeSpecType));
}
// Only 'short' is valid with vector bool. (PIM 2.1)
if ((TypeSpecWidth != TSW_unspecified) && (TypeSpecWidth != TSW_short))
Diag(D, TSWLoc, SrcMgr, diag::err_invalid_vector_bool_decl_spec)
<< getSpecifierName((TSW)TypeSpecWidth);
// Elements of vector bool are interpreted as unsigned. (PIM 2.1)
if ((TypeSpecType == TST_char) || (TypeSpecType == TST_int) ||
(TypeSpecWidth != TSW_unspecified))
TypeSpecSign = TSS_unsigned;
}
if (TypeAltiVecPixel) {
//TODO: perform validation
TypeSpecType = TST_int;
TypeSpecSign = TSS_unsigned;
TypeSpecWidth = TSW_short;
TypeSpecOwned = false;
}
}
// signed/unsigned are only valid with int/char/wchar_t.
if (TypeSpecSign != TSS_unspecified) {
if (TypeSpecType == TST_unspecified)
TypeSpecType = TST_int; // unsigned -> unsigned int, signed -> signed int.
else if (TypeSpecType != TST_int &&
TypeSpecType != TST_char && TypeSpecType != TST_wchar) {
Diag(D, TSSLoc, SrcMgr, diag::err_invalid_sign_spec)
<< getSpecifierName((TST)TypeSpecType);
// signed double -> double.
TypeSpecSign = TSS_unspecified;
}
}
// Validate the width of the type.
switch (TypeSpecWidth) {
case TSW_unspecified: break;
case TSW_short: // short int
case TSW_longlong: // long long int
if (TypeSpecType == TST_unspecified)
TypeSpecType = TST_int; // short -> short int, long long -> long long int.
else if (TypeSpecType != TST_int) {
Diag(D, TSWLoc, SrcMgr,
TypeSpecWidth == TSW_short ? diag::err_invalid_short_spec
: diag::err_invalid_longlong_spec)
<< getSpecifierName((TST)TypeSpecType);
TypeSpecType = TST_int;
TypeSpecOwned = false;
}
break;
case TSW_long: // long double, long int
if (TypeSpecType == TST_unspecified)
TypeSpecType = TST_int; // long -> long int.
else if (TypeSpecType != TST_int && TypeSpecType != TST_double) {
Diag(D, TSWLoc, SrcMgr, diag::err_invalid_long_spec)
<< getSpecifierName((TST)TypeSpecType);
TypeSpecType = TST_int;
TypeSpecOwned = false;
}
break;
}
// TODO: if the implementation does not implement _Complex or _Imaginary,
// disallow their use. Need information about the backend.
if (TypeSpecComplex != TSC_unspecified) {
if (TypeSpecType == TST_unspecified) {
Diag(D, TSCLoc, SrcMgr, diag::ext_plain_complex)
<< FixItHint::CreateInsertion(
PP.getLocForEndOfToken(getTypeSpecComplexLoc()),
" double");
TypeSpecType = TST_double; // _Complex -> _Complex double.
} else if (TypeSpecType == TST_int || TypeSpecType == TST_char) {
// Note that this intentionally doesn't include _Complex _Bool.
Diag(D, TSTLoc, SrcMgr, diag::ext_integer_complex);
} else if (TypeSpecType != TST_float && TypeSpecType != TST_double) {
Diag(D, TSCLoc, SrcMgr, diag::err_invalid_complex_spec)
<< getSpecifierName((TST)TypeSpecType);
TypeSpecComplex = TSC_unspecified;
}
}
// C++ [class.friend]p6:
// No storage-class-specifier shall appear in the decl-specifier-seq
// of a friend declaration.
if (isFriendSpecified() && getStorageClassSpec()) {
DeclSpec::SCS SC = getStorageClassSpec();
const char *SpecName = getSpecifierName(SC);
SourceLocation SCLoc = getStorageClassSpecLoc();
SourceLocation SCEndLoc = SCLoc.getFileLocWithOffset(strlen(SpecName));
Diag(D, SCLoc, SrcMgr, diag::err_friend_storage_spec)
<< SpecName
<< FixItHint::CreateRemoval(SourceRange(SCLoc, SCEndLoc));
ClearStorageClassSpecs();
}
assert(!TypeSpecOwned || isDeclRep((TST) TypeSpecType));
// Okay, now we can infer the real type.
// TODO: return "auto function" and other bad things based on the real type.
// 'data definition has no type or storage class'?
}
void DependencyCollector::attachToPreprocessor(Preprocessor &PP) {
PP.addPPCallbacks(new DepCollectorPPCallbacks(*this, PP.getSourceManager()));
}
/// InitializePreprocessor - Initialize the preprocessor getting it and the
/// environment ready to process a single file. This returns true on error.
///
void clang::InitializePreprocessor(Preprocessor &PP,
const PreprocessorOptions &InitOpts,
const HeaderSearchOptions &HSOpts,
const FrontendOptions &FEOpts) {
const LangOptions &LangOpts = PP.getLangOpts();
std::string PredefineBuffer;
PredefineBuffer.reserve(4080);
llvm::raw_string_ostream Predefines(PredefineBuffer);
MacroBuilder Builder(Predefines);
InitializeFileRemapping(PP.getDiagnostics(), PP.getSourceManager(),
PP.getFileManager(), InitOpts);
// Emit line markers for various builtin sections of the file. We don't do
// this in asm preprocessor mode, because "# 4" is not a line marker directive
// in this mode.
if (!PP.getLangOpts().AsmPreprocessor)
Builder.append("# 1 \"<built-in>\" 3");
// Install things like __POWERPC__, __GNUC__, etc into the macro table.
if (InitOpts.UsePredefines) {
InitializePredefinedMacros(PP.getTargetInfo(), LangOpts, FEOpts, Builder);
// Install definitions to make Objective-C++ ARC work well with various
// C++ Standard Library implementations.
if (LangOpts.ObjC1 && LangOpts.CPlusPlus && LangOpts.ObjCAutoRefCount) {
switch (InitOpts.ObjCXXARCStandardLibrary) {
case ARCXX_nolib:
case ARCXX_libcxx:
break;
case ARCXX_libstdcxx:
AddObjCXXARCLibstdcxxDefines(LangOpts, Builder);
break;
}
}
}
// Even with predefines off, some macros are still predefined.
// These should all be defined in the preprocessor according to the
// current language configuration.
InitializeStandardPredefinedMacros(PP.getTargetInfo(), PP.getLangOpts(),
FEOpts, Builder);
// Add on the predefines from the driver. Wrap in a #line directive to report
// that they come from the command line.
if (!PP.getLangOpts().AsmPreprocessor)
Builder.append("# 1 \"<command line>\" 1");
// Process #define's and #undef's in the order they are given.
for (unsigned i = 0, e = InitOpts.Macros.size(); i != e; ++i) {
if (InitOpts.Macros[i].second) // isUndef
Builder.undefineMacro(InitOpts.Macros[i].first);
else
DefineBuiltinMacro(Builder, InitOpts.Macros[i].first,
PP.getDiagnostics());
}
// If -imacros are specified, include them now. These are processed before
// any -include directives.
for (unsigned i = 0, e = InitOpts.MacroIncludes.size(); i != e; ++i)
AddImplicitIncludeMacros(Builder, InitOpts.MacroIncludes[i],
PP.getFileManager());
// Process -include-pch/-include-pth directives.
if (!InitOpts.ImplicitPCHInclude.empty())
AddImplicitIncludePCH(Builder, PP, InitOpts.ImplicitPCHInclude);
if (!InitOpts.ImplicitPTHInclude.empty())
AddImplicitIncludePTH(Builder, PP, InitOpts.ImplicitPTHInclude);
// Process -include directives.
for (unsigned i = 0, e = InitOpts.Includes.size(); i != e; ++i) {
const std::string &Path = InitOpts.Includes[i];
AddImplicitInclude(Builder, Path, PP.getFileManager());
}
// Exit the command line and go back to <built-in> (2 is LC_LEAVE).
if (!PP.getLangOpts().AsmPreprocessor)
Builder.append("# 1 \"<built-in>\" 2");
// Instruct the preprocessor to skip the preamble.
PP.setSkipMainFilePreamble(InitOpts.PrecompiledPreambleBytes.first,
InitOpts.PrecompiledPreambleBytes.second);
// Copy PredefinedBuffer into the Preprocessor.
PP.setPredefines(Predefines.str());
// Initialize the header search object.
ApplyHeaderSearchOptions(PP.getHeaderSearchInfo(), HSOpts,
PP.getLangOpts(),
PP.getTargetInfo().getTriple());
}
void DependencyCollector::attachToPreprocessor(Preprocessor &PP) {
PP.addPPCallbacks(
llvm::make_unique<DepCollectorPPCallbacks>(*this, PP.getSourceManager()));
PP.getHeaderSearchInfo().getModuleMap().addModuleMapCallbacks(
llvm::make_unique<DepCollectorMMCallbacks>(*this));
}
/// Finish - This does final analysis of the declspec, rejecting things like
/// "_Imaginary" (lacking an FP type). This returns a diagnostic to issue or
/// diag::NUM_DIAGNOSTICS if there is no error. After calling this method,
/// DeclSpec is guaranteed self-consistent, even if an error occurred.
void DeclSpec::Finish(DiagnosticsEngine &D, Preprocessor &PP, const PrintingPolicy &Policy) {
// Before possibly changing their values, save specs as written.
SaveWrittenBuiltinSpecs();
// Check the type specifier components first.
// If decltype(auto) is used, no other type specifiers are permitted.
if (TypeSpecType == TST_decltype_auto &&
(TypeSpecWidth != TSW_unspecified ||
TypeSpecComplex != TSC_unspecified ||
TypeSpecSign != TSS_unspecified ||
TypeAltiVecVector || TypeAltiVecPixel || TypeAltiVecBool ||
TypeQualifiers)) {
const unsigned NumLocs = 8;
SourceLocation ExtraLocs[NumLocs] = {
TSWLoc, TSCLoc, TSSLoc, AltiVecLoc,
TQ_constLoc, TQ_restrictLoc, TQ_volatileLoc, TQ_atomicLoc
};
FixItHint Hints[NumLocs];
SourceLocation FirstLoc;
for (unsigned I = 0; I != NumLocs; ++I) {
if (!ExtraLocs[I].isInvalid()) {
if (FirstLoc.isInvalid() ||
PP.getSourceManager().isBeforeInTranslationUnit(ExtraLocs[I],
FirstLoc))
FirstLoc = ExtraLocs[I];
Hints[I] = FixItHint::CreateRemoval(ExtraLocs[I]);
}
}
TypeSpecWidth = TSW_unspecified;
TypeSpecComplex = TSC_unspecified;
TypeSpecSign = TSS_unspecified;
TypeAltiVecVector = TypeAltiVecPixel = TypeAltiVecBool = false;
TypeQualifiers = 0;
Diag(D, TSTLoc, diag::err_decltype_auto_cannot_be_combined)
<< Hints[0] << Hints[1] << Hints[2] << Hints[3]
<< Hints[4] << Hints[5] << Hints[6] << Hints[7];
}
// Validate and finalize AltiVec vector declspec.
if (TypeAltiVecVector) {
if (TypeAltiVecBool) {
// Sign specifiers are not allowed with vector bool. (PIM 2.1)
if (TypeSpecSign != TSS_unspecified) {
Diag(D, TSSLoc, diag::err_invalid_vector_bool_decl_spec)
<< getSpecifierName((TSS)TypeSpecSign);
}
// Only char/int are valid with vector bool. (PIM 2.1)
if (((TypeSpecType != TST_unspecified) && (TypeSpecType != TST_char) &&
(TypeSpecType != TST_int)) || TypeAltiVecPixel) {
Diag(D, TSTLoc, diag::err_invalid_vector_bool_decl_spec)
<< (TypeAltiVecPixel ? "__pixel" :
getSpecifierName((TST)TypeSpecType, Policy));
}
// Only 'short' and 'long long' are valid with vector bool. (PIM 2.1)
if ((TypeSpecWidth != TSW_unspecified) && (TypeSpecWidth != TSW_short) &&
(TypeSpecWidth != TSW_longlong))
Diag(D, TSWLoc, diag::err_invalid_vector_bool_decl_spec)
<< getSpecifierName((TSW)TypeSpecWidth);
// vector bool long long requires VSX support.
if ((TypeSpecWidth == TSW_longlong) &&
(!PP.getTargetInfo().hasFeature("vsx")) &&
(!PP.getTargetInfo().hasFeature("power8-vector")))
Diag(D, TSTLoc, diag::err_invalid_vector_long_long_decl_spec);
// Elements of vector bool are interpreted as unsigned. (PIM 2.1)
if ((TypeSpecType == TST_char) || (TypeSpecType == TST_int) ||
(TypeSpecWidth != TSW_unspecified))
TypeSpecSign = TSS_unsigned;
} else if (TypeSpecType == TST_double) {
// vector long double and vector long long double are never allowed.
// vector double is OK for Power7 and later.
if (TypeSpecWidth == TSW_long || TypeSpecWidth == TSW_longlong)
Diag(D, TSWLoc, diag::err_invalid_vector_long_double_decl_spec);
else if (!PP.getTargetInfo().hasFeature("vsx"))
Diag(D, TSTLoc, diag::err_invalid_vector_double_decl_spec);
} else if (TypeSpecWidth == TSW_long) {
Diag(D, TSWLoc, diag::warn_vector_long_decl_spec_combination)
<< getSpecifierName((TST)TypeSpecType, Policy);
}
if (TypeAltiVecPixel) {
//TODO: perform validation
TypeSpecType = TST_int;
TypeSpecSign = TSS_unsigned;
TypeSpecWidth = TSW_short;
TypeSpecOwned = false;
}
}
// signed/unsigned are only valid with int/char/wchar_t.
if (TypeSpecSign != TSS_unspecified) {
if (TypeSpecType == TST_unspecified)
TypeSpecType = TST_int; // unsigned -> unsigned int, signed -> signed int.
else if (TypeSpecType != TST_int && TypeSpecType != TST_int128 &&
TypeSpecType != TST_char && TypeSpecType != TST_wchar) {
Diag(D, TSSLoc, diag::err_invalid_sign_spec)
<< getSpecifierName((TST)TypeSpecType, Policy);
// signed double -> double.
TypeSpecSign = TSS_unspecified;
}
}
// Validate the width of the type.
switch (TypeSpecWidth) {
case TSW_unspecified: break;
case TSW_short: // short int
case TSW_longlong: // long long int
if (TypeSpecType == TST_unspecified)
TypeSpecType = TST_int; // short -> short int, long long -> long long int.
else if (TypeSpecType != TST_int) {
Diag(D, TSWLoc,
TypeSpecWidth == TSW_short ? diag::err_invalid_short_spec
: diag::err_invalid_longlong_spec)
<< getSpecifierName((TST)TypeSpecType, Policy);
TypeSpecType = TST_int;
TypeSpecOwned = false;
}
break;
case TSW_long: // long double, long int
if (TypeSpecType == TST_unspecified)
TypeSpecType = TST_int; // long -> long int.
else if (TypeSpecType != TST_int && TypeSpecType != TST_double) {
Diag(D, TSWLoc, diag::err_invalid_long_spec)
<< getSpecifierName((TST)TypeSpecType, Policy);
TypeSpecType = TST_int;
TypeSpecOwned = false;
}
break;
}
// TODO: if the implementation does not implement _Complex or _Imaginary,
// disallow their use. Need information about the backend.
if (TypeSpecComplex != TSC_unspecified) {
if (TypeSpecType == TST_unspecified) {
Diag(D, TSCLoc, diag::ext_plain_complex)
<< FixItHint::CreateInsertion(
PP.getLocForEndOfToken(getTypeSpecComplexLoc()),
" double");
TypeSpecType = TST_double; // _Complex -> _Complex double.
} else if (TypeSpecType == TST_int || TypeSpecType == TST_char) {
// Note that this intentionally doesn't include _Complex _Bool.
if (!PP.getLangOpts().CPlusPlus)
Diag(D, TSTLoc, diag::ext_integer_complex);
} else if (TypeSpecType != TST_float && TypeSpecType != TST_double) {
Diag(D, TSCLoc, diag::err_invalid_complex_spec)
<< getSpecifierName((TST)TypeSpecType, Policy);
TypeSpecComplex = TSC_unspecified;
}
}
// C11 6.7.1/3, C++11 [dcl.stc]p1, GNU TLS: __thread, thread_local and
// _Thread_local can only appear with the 'static' and 'extern' storage class
// specifiers. We also allow __private_extern__ as an extension.
if (ThreadStorageClassSpec != TSCS_unspecified) {
switch (StorageClassSpec) {
case SCS_unspecified:
case SCS_extern:
case SCS_private_extern:
case SCS_static:
break;
default:
if (PP.getSourceManager().isBeforeInTranslationUnit(
getThreadStorageClassSpecLoc(), getStorageClassSpecLoc()))
Diag(D, getStorageClassSpecLoc(),
diag::err_invalid_decl_spec_combination)
<< DeclSpec::getSpecifierName(getThreadStorageClassSpec())
<< SourceRange(getThreadStorageClassSpecLoc());
else
Diag(D, getThreadStorageClassSpecLoc(),
diag::err_invalid_decl_spec_combination)
<< DeclSpec::getSpecifierName(getStorageClassSpec())
<< SourceRange(getStorageClassSpecLoc());
// Discard the thread storage class specifier to recover.
ThreadStorageClassSpec = TSCS_unspecified;
ThreadStorageClassSpecLoc = SourceLocation();
}
}
// If no type specifier was provided and we're parsing a language where
// the type specifier is not optional, but we got 'auto' as a storage
// class specifier, then assume this is an attempt to use C++0x's 'auto'
// type specifier.
if (PP.getLangOpts().CPlusPlus &&
TypeSpecType == TST_unspecified && StorageClassSpec == SCS_auto) {
TypeSpecType = TST_auto;
StorageClassSpec = SCS_unspecified;
TSTLoc = TSTNameLoc = StorageClassSpecLoc;
StorageClassSpecLoc = SourceLocation();
}
// Diagnose if we've recovered from an ill-formed 'auto' storage class
// specifier in a pre-C++11 dialect of C++.
if (!PP.getLangOpts().CPlusPlus11 && TypeSpecType == TST_auto)
Diag(D, TSTLoc, diag::ext_auto_type_specifier);
if (PP.getLangOpts().CPlusPlus && !PP.getLangOpts().CPlusPlus11 &&
StorageClassSpec == SCS_auto)
Diag(D, StorageClassSpecLoc, diag::warn_auto_storage_class)
<< FixItHint::CreateRemoval(StorageClassSpecLoc);
if (TypeSpecType == TST_char16 || TypeSpecType == TST_char32)
Diag(D, TSTLoc, diag::warn_cxx98_compat_unicode_type)
<< (TypeSpecType == TST_char16 ? "char16_t" : "char32_t");
if (Constexpr_specified)
Diag(D, ConstexprLoc, diag::warn_cxx98_compat_constexpr);
// C++ [class.friend]p6:
// No storage-class-specifier shall appear in the decl-specifier-seq
// of a friend declaration.
if (isFriendSpecified() &&
(getStorageClassSpec() || getThreadStorageClassSpec())) {
SmallString<32> SpecName;
SourceLocation SCLoc;
FixItHint StorageHint, ThreadHint;
if (DeclSpec::SCS SC = getStorageClassSpec()) {
SpecName = getSpecifierName(SC);
SCLoc = getStorageClassSpecLoc();
StorageHint = FixItHint::CreateRemoval(SCLoc);
}
if (DeclSpec::TSCS TSC = getThreadStorageClassSpec()) {
if (!SpecName.empty()) SpecName += " ";
SpecName += getSpecifierName(TSC);
SCLoc = getThreadStorageClassSpecLoc();
ThreadHint = FixItHint::CreateRemoval(SCLoc);
}
Diag(D, SCLoc, diag::err_friend_decl_spec)
<< SpecName << StorageHint << ThreadHint;
ClearStorageClassSpecs();
}
// C++11 [dcl.fct.spec]p5:
// The virtual specifier shall be used only in the initial
// declaration of a non-static class member function;
// C++11 [dcl.fct.spec]p6:
// The explicit specifier shall be used only in the declaration of
// a constructor or conversion function within its class
// definition;
if (isFriendSpecified() && (isVirtualSpecified() || isExplicitSpecified())) {
StringRef Keyword;
SourceLocation SCLoc;
if (isVirtualSpecified()) {
Keyword = "virtual";
SCLoc = getVirtualSpecLoc();
} else {
Keyword = "explicit";
SCLoc = getExplicitSpecLoc();
}
FixItHint Hint = FixItHint::CreateRemoval(SCLoc);
Diag(D, SCLoc, diag::err_friend_decl_spec)
<< Keyword << Hint;
FS_virtual_specified = FS_explicit_specified = false;
FS_virtualLoc = FS_explicitLoc = SourceLocation();
}
assert(!TypeSpecOwned || isDeclRep((TST) TypeSpecType));
// Okay, now we can infer the real type.
// TODO: return "auto function" and other bad things based on the real type.
// 'data definition has no type or storage class'?
}
/// SyntaxHighlight - Relex the specified FileID and annotate the HTML with
/// information about keywords, macro expansions etc. This uses the macro
/// table state from the end of the file, so it won't be perfectly perfect,
/// but it will be reasonably close.
void html::SyntaxHighlight(Rewriter &R, FileID FID, const Preprocessor &PP) {
RewriteBuffer &RB = R.getEditBuffer(FID);
const SourceManager &SM = PP.getSourceManager();
const llvm::MemoryBuffer *FromFile = SM.getBuffer(FID);
Lexer L(FID, FromFile, SM, PP.getLangOpts());
const char *BufferStart = L.getBuffer().data();
// Inform the preprocessor that we want to retain comments as tokens, so we
// can highlight them.
L.SetCommentRetentionState(true);
// Lex all the tokens in raw mode, to avoid entering #includes or expanding
// macros.
Token Tok;
L.LexFromRawLexer(Tok);
while (Tok.isNot(tok::eof)) {
// Since we are lexing unexpanded tokens, all tokens are from the main
// FileID.
unsigned TokOffs = SM.getFileOffset(Tok.getLocation());
unsigned TokLen = Tok.getLength();
switch (Tok.getKind()) {
default: break;
case tok::identifier:
llvm_unreachable("tok::identifier in raw lexing mode!");
case tok::raw_identifier: {
// Fill in Result.IdentifierInfo and update the token kind,
// looking up the identifier in the identifier table.
PP.LookUpIdentifierInfo(Tok);
// If this is a pp-identifier, for a keyword, highlight it as such.
if (Tok.isNot(tok::identifier))
HighlightRange(RB, TokOffs, TokOffs+TokLen, BufferStart,
"<span class='keyword'>", "</span>");
break;
}
case tok::comment:
HighlightRange(RB, TokOffs, TokOffs+TokLen, BufferStart,
"<span class='comment'>", "</span>");
break;
case tok::utf8_string_literal:
// Chop off the u part of u8 prefix
++TokOffs;
--TokLen;
// FALL THROUGH to chop the 8
case tok::wide_string_literal:
case tok::utf16_string_literal:
case tok::utf32_string_literal:
// Chop off the L, u, U or 8 prefix
++TokOffs;
--TokLen;
// FALL THROUGH.
case tok::string_literal:
// FIXME: Exclude the optional ud-suffix from the highlighted range.
HighlightRange(RB, TokOffs, TokOffs+TokLen, BufferStart,
"<span class='string_literal'>", "</span>");
break;
case tok::hash: {
// If this is a preprocessor directive, all tokens to end of line are too.
if (!Tok.isAtStartOfLine())
break;
// Eat all of the tokens until we get to the next one at the start of
// line.
unsigned TokEnd = TokOffs+TokLen;
L.LexFromRawLexer(Tok);
while (!Tok.isAtStartOfLine() && Tok.isNot(tok::eof)) {
TokEnd = SM.getFileOffset(Tok.getLocation())+Tok.getLength();
L.LexFromRawLexer(Tok);
}
// Find end of line. This is a hack.
HighlightRange(RB, TokOffs, TokEnd, BufferStart,
"<span class='directive'>", "</span>");
// Don't skip the next token.
continue;
}
}
L.LexFromRawLexer(Tok);
}
}