/// \brief Lex the specified source file to determine whether it contains
/// any expected-* directives. As a Lexer is used rather than a full-blown
/// Preprocessor, directives inside skipped #if blocks will still be found.
///
/// \return true if any directives were found.
static bool findDirectives(SourceManager &SM, FileID FID,
const LangOptions &LangOpts) {
// Create a raw lexer to pull all the comments out of FID.
if (FID.isInvalid())
return false;
// 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, LangOpts);
// Return comments as tokens, this is how we find expected diagnostics.
RawLex.SetCommentRetentionState(true);
Token Tok;
Tok.setKind(tok::comment);
VerifyDiagnosticConsumer::DirectiveStatus Status =
VerifyDiagnosticConsumer::HasNoDirectives;
while (Tok.isNot(tok::eof)) {
RawLex.Lex(Tok);
if (!Tok.is(tok::comment)) continue;
std::string Comment = RawLex.getSpelling(Tok, SM, LangOpts);
if (Comment.empty()) continue;
// Find first directive.
if (ParseDirective(Comment, 0, SM, Tok.getLocation(),
SM.getDiagnostics(), Status))
return true;
}
return false;
}
/// ParseDirective - Go through the comment and see if it indicates expected
/// diagnostics. If so, then put them in the appropriate directive list.
///
/// Returns true if any valid directives were found.
static bool ParseDirective(StringRef S, ExpectedData *ED, SourceManager &SM,
Preprocessor *PP, SourceLocation Pos,
VerifyDiagnosticConsumer::DirectiveStatus &Status,
VerifyDiagnosticConsumer::MarkerTracker &Markers) {
DiagnosticsEngine &Diags = PP ? PP->getDiagnostics() : SM.getDiagnostics();
// First, scan the comment looking for markers.
for (ParseHelper PH(S); !PH.Done();) {
if (!PH.Search("#", true))
break;
PH.C = PH.P;
if (!PH.NextMarker()) {
PH.Next("#");
PH.Advance();
continue;
}
PH.Advance();
Markers.addMarker(PH.Match(), Pos);
}
// A single comment may contain multiple directives.
bool FoundDirective = false;
for (ParseHelper PH(S); !PH.Done();) {
// Search for the initial directive token.
// If one prefix, save time by searching only for its directives.
// Otherwise, search for any potential directive token and check it later.
const auto &Prefixes = Diags.getDiagnosticOptions().VerifyPrefixes;
if (!(Prefixes.size() == 1 ? PH.Search(*Prefixes.begin(), true, true)
: PH.Search("", true, true)))
break;
StringRef DToken = PH.Match();
PH.Advance();
// Default directive kind.
UnattachedDirective D;
const char *KindStr = "string";
// Parse the initial directive token in reverse so we can easily determine
// its exact actual prefix. If we were to parse it from the front instead,
// it would be harder to determine where the prefix ends because there
// might be multiple matching -verify prefixes because some might prefix
// others.
// Regex in initial directive token: -re
if (DToken.endswith("-re")) {
D.RegexKind = true;
KindStr = "regex";
DToken = DToken.substr(0, DToken.size()-3);
}
// Type in initial directive token: -{error|warning|note|no-diagnostics}
bool NoDiag = false;
StringRef DType;
if (DToken.endswith(DType="-error"))
D.DL = ED ? &ED->Errors : nullptr;
else if (DToken.endswith(DType="-warning"))
D.DL = ED ? &ED->Warnings : nullptr;
else if (DToken.endswith(DType="-remark"))
D.DL = ED ? &ED->Remarks : nullptr;
else if (DToken.endswith(DType="-note"))
D.DL = ED ? &ED->Notes : nullptr;
else if (DToken.endswith(DType="-no-diagnostics")) {
NoDiag = true;
if (D.RegexKind)
continue;
}
else
continue;
DToken = DToken.substr(0, DToken.size()-DType.size());
// What's left in DToken is the actual prefix. That might not be a -verify
// prefix even if there is only one -verify prefix (for example, the full
// DToken is foo-bar-warning, but foo is the only -verify prefix).
if (!std::binary_search(Prefixes.begin(), Prefixes.end(), DToken))
continue;
if (NoDiag) {
if (Status == VerifyDiagnosticConsumer::HasOtherExpectedDirectives)
Diags.Report(Pos, diag::err_verify_invalid_no_diags)
<< /*IsExpectedNoDiagnostics=*/true;
else
Status = VerifyDiagnosticConsumer::HasExpectedNoDiagnostics;
continue;
}
if (Status == VerifyDiagnosticConsumer::HasExpectedNoDiagnostics) {
Diags.Report(Pos, diag::err_verify_invalid_no_diags)
<< /*IsExpectedNoDiagnostics=*/false;
continue;
}
Status = VerifyDiagnosticConsumer::HasOtherExpectedDirectives;
// If a directive has been found but we're not interested
// in storing the directive information, return now.
if (!D.DL)
return true;
// Next optional token: @
SourceLocation ExpectedLoc;
StringRef Marker;
bool MatchAnyLine = false;
if (!PH.Next("@")) {
ExpectedLoc = Pos;
} else {
PH.Advance();
unsigned Line = 0;
bool FoundPlus = PH.Next("+");
if (FoundPlus || PH.Next("-")) {
// Relative to current line.
PH.Advance();
bool Invalid = false;
unsigned ExpectedLine = SM.getSpellingLineNumber(Pos, &Invalid);
if (!Invalid && PH.Next(Line) && (FoundPlus || Line < ExpectedLine)) {
if (FoundPlus) ExpectedLine += Line;
else ExpectedLine -= Line;
ExpectedLoc = SM.translateLineCol(SM.getFileID(Pos), ExpectedLine, 1);
}
} else if (PH.Next(Line)) {
// Absolute line number.
if (Line > 0)
ExpectedLoc = SM.translateLineCol(SM.getFileID(Pos), Line, 1);
} else if (PH.NextMarker()) {
Marker = PH.Match();
} else if (PP && PH.Search(":")) {
// Specific source file.
StringRef Filename(PH.C, PH.P-PH.C);
PH.Advance();
// Lookup file via Preprocessor, like a #include.
const DirectoryLookup *CurDir;
const FileEntry *FE =
PP->LookupFile(Pos, Filename, false, nullptr, nullptr, CurDir,
nullptr, nullptr, nullptr, nullptr, nullptr);
if (!FE) {
Diags.Report(Pos.getLocWithOffset(PH.C-PH.Begin),
diag::err_verify_missing_file) << Filename << KindStr;
continue;
}
if (SM.translateFile(FE).isInvalid())
SM.createFileID(FE, Pos, SrcMgr::C_User);
if (PH.Next(Line) && Line > 0)
ExpectedLoc = SM.translateFileLineCol(FE, Line, 1);
else if (PH.Next("*")) {
MatchAnyLine = true;
ExpectedLoc = SM.translateFileLineCol(FE, 1, 1);
}
} else if (PH.Next("*")) {
MatchAnyLine = true;
ExpectedLoc = SourceLocation();
}
if (ExpectedLoc.isInvalid() && !MatchAnyLine && Marker.empty()) {
Diags.Report(Pos.getLocWithOffset(PH.C-PH.Begin),
diag::err_verify_missing_line) << KindStr;
continue;
}
PH.Advance();
}
// Skip optional whitespace.
PH.SkipWhitespace();
// Next optional token: positive integer or a '+'.
if (PH.Next(D.Min)) {
PH.Advance();
// A positive integer can be followed by a '+' meaning min
// or more, or by a '-' meaning a range from min to max.
if (PH.Next("+")) {
D.Max = Directive::MaxCount;
PH.Advance();
} else if (PH.Next("-")) {
PH.Advance();
if (!PH.Next(D.Max) || D.Max < D.Min) {
Diags.Report(Pos.getLocWithOffset(PH.C-PH.Begin),
diag::err_verify_invalid_range) << KindStr;
continue;
}
PH.Advance();
} else {
D.Max = D.Min;
}
} else if (PH.Next("+")) {
// '+' on its own means "1 or more".
D.Max = Directive::MaxCount;
PH.Advance();
}
// Skip optional whitespace.
PH.SkipWhitespace();
// Next token: {{
if (!PH.Next("{{")) {
Diags.Report(Pos.getLocWithOffset(PH.C-PH.Begin),
diag::err_verify_missing_start) << KindStr;
continue;
}
PH.Advance();
const char* const ContentBegin = PH.C; // mark content begin
// Search for token: }}
if (!PH.SearchClosingBrace("{{", "}}")) {
Diags.Report(Pos.getLocWithOffset(PH.C-PH.Begin),
diag::err_verify_missing_end) << KindStr;
continue;
}
const char* const ContentEnd = PH.P; // mark content end
PH.Advance();
D.DirectivePos = Pos;
D.ContentBegin = Pos.getLocWithOffset(ContentBegin - PH.Begin);
// Build directive text; convert \n to newlines.
StringRef NewlineStr = "\\n";
StringRef Content(ContentBegin, ContentEnd-ContentBegin);
size_t CPos = 0;
size_t FPos;
while ((FPos = Content.find(NewlineStr, CPos)) != StringRef::npos) {
D.Text += Content.substr(CPos, FPos-CPos);
D.Text += '\n';
CPos = FPos + NewlineStr.size();
}
if (D.Text.empty())
D.Text.assign(ContentBegin, ContentEnd);
// Check that regex directives contain at least one regex.
if (D.RegexKind && D.Text.find("{{") == StringRef::npos) {
Diags.Report(D.ContentBegin, diag::err_verify_missing_regex) << D.Text;
return false;
}
if (Marker.empty())
attachDirective(Diags, D, ExpectedLoc, MatchAnyLine);
else
Markers.addDirective(Marker, D);
FoundDirective = true;
}
return FoundDirective;
}
/// ParseDirective - Go through the comment and see if it indicates expected
/// diagnostics. If so, then put them in the appropriate directive list.
///
/// Returns true if any valid directives were found.
static bool ParseDirective(StringRef S, ExpectedData *ED, SourceManager &SM,
Preprocessor *PP, SourceLocation Pos,
VerifyDiagnosticConsumer::DirectiveStatus &Status) {
DiagnosticsEngine &Diags = PP ? PP->getDiagnostics() : SM.getDiagnostics();
// A single comment may contain multiple directives.
bool FoundDirective = false;
for (ParseHelper PH(S); !PH.Done();) {
// Search for token: expected
if (!PH.Search("expected", true))
break;
PH.Advance();
// Next token: -
if (!PH.Next("-"))
continue;
PH.Advance();
// Next token: { error | warning | note }
DirectiveList *DL = nullptr;
if (PH.Next("error"))
DL = ED ? &ED->Errors : nullptr;
else if (PH.Next("warning"))
DL = ED ? &ED->Warnings : nullptr;
else if (PH.Next("remark"))
DL = ED ? &ED->Remarks : nullptr;
else if (PH.Next("note"))
DL = ED ? &ED->Notes : nullptr;
else if (PH.Next("no-diagnostics")) {
if (Status == VerifyDiagnosticConsumer::HasOtherExpectedDirectives)
Diags.Report(Pos, diag::err_verify_invalid_no_diags)
<< /*IsExpectedNoDiagnostics=*/true;
else
Status = VerifyDiagnosticConsumer::HasExpectedNoDiagnostics;
continue;
} else
continue;
PH.Advance();
if (Status == VerifyDiagnosticConsumer::HasExpectedNoDiagnostics) {
Diags.Report(Pos, diag::err_verify_invalid_no_diags)
<< /*IsExpectedNoDiagnostics=*/false;
continue;
}
Status = VerifyDiagnosticConsumer::HasOtherExpectedDirectives;
// If a directive has been found but we're not interested
// in storing the directive information, return now.
if (!DL)
return true;
// Default directive kind.
bool RegexKind = false;
const char* KindStr = "string";
// Next optional token: -
if (PH.Next("-re")) {
PH.Advance();
RegexKind = true;
KindStr = "regex";
}
// Next optional token: @
SourceLocation ExpectedLoc;
bool MatchAnyLine = false;
if (!PH.Next("@")) {
ExpectedLoc = Pos;
} else {
PH.Advance();
unsigned Line = 0;
bool FoundPlus = PH.Next("+");
if (FoundPlus || PH.Next("-")) {
// Relative to current line.
PH.Advance();
bool Invalid = false;
unsigned ExpectedLine = SM.getSpellingLineNumber(Pos, &Invalid);
if (!Invalid && PH.Next(Line) && (FoundPlus || Line < ExpectedLine)) {
if (FoundPlus) ExpectedLine += Line;
else ExpectedLine -= Line;
ExpectedLoc = SM.translateLineCol(SM.getFileID(Pos), ExpectedLine, 1);
}
} else if (PH.Next(Line)) {
// Absolute line number.
if (Line > 0)
ExpectedLoc = SM.translateLineCol(SM.getFileID(Pos), Line, 1);
} else if (PP && PH.Search(":")) {
// Specific source file.
StringRef Filename(PH.C, PH.P-PH.C);
PH.Advance();
// Lookup file via Preprocessor, like a #include.
const DirectoryLookup *CurDir;
const FileEntry *FE =
PP->LookupFile(Pos, Filename, false, nullptr, nullptr, CurDir,
nullptr, nullptr, nullptr);
if (!FE) {
Diags.Report(Pos.getLocWithOffset(PH.C-PH.Begin),
diag::err_verify_missing_file) << Filename << KindStr;
continue;
}
if (SM.translateFile(FE).isInvalid())
SM.createFileID(FE, Pos, SrcMgr::C_User);
if (PH.Next(Line) && Line > 0)
ExpectedLoc = SM.translateFileLineCol(FE, Line, 1);
else if (PH.Next("*")) {
MatchAnyLine = true;
ExpectedLoc = SM.translateFileLineCol(FE, 1, 1);
}
}
if (ExpectedLoc.isInvalid()) {
Diags.Report(Pos.getLocWithOffset(PH.C-PH.Begin),
diag::err_verify_missing_line) << KindStr;
continue;
}
PH.Advance();
}
// Skip optional whitespace.
PH.SkipWhitespace();
// Next optional token: positive integer or a '+'.
unsigned Min = 1;
unsigned Max = 1;
if (PH.Next(Min)) {
PH.Advance();
// A positive integer can be followed by a '+' meaning min
// or more, or by a '-' meaning a range from min to max.
if (PH.Next("+")) {
Max = Directive::MaxCount;
PH.Advance();
} else if (PH.Next("-")) {
PH.Advance();
if (!PH.Next(Max) || Max < Min) {
Diags.Report(Pos.getLocWithOffset(PH.C-PH.Begin),
diag::err_verify_invalid_range) << KindStr;
continue;
}
PH.Advance();
} else {
Max = Min;
}
} else if (PH.Next("+")) {
// '+' on its own means "1 or more".
Max = Directive::MaxCount;
PH.Advance();
}
// Skip optional whitespace.
PH.SkipWhitespace();
// Next token: {{
if (!PH.Next("{{")) {
Diags.Report(Pos.getLocWithOffset(PH.C-PH.Begin),
diag::err_verify_missing_start) << KindStr;
continue;
}
PH.Advance();
const char* const ContentBegin = PH.C; // mark content begin
// Search for token: }}
if (!PH.SearchClosingBrace("{{", "}}")) {
Diags.Report(Pos.getLocWithOffset(PH.C-PH.Begin),
diag::err_verify_missing_end) << KindStr;
continue;
}
const char* const ContentEnd = PH.P; // mark content end
PH.Advance();
// Build directive text; convert \n to newlines.
std::string Text;
StringRef NewlineStr = "\\n";
StringRef Content(ContentBegin, ContentEnd-ContentBegin);
size_t CPos = 0;
size_t FPos;
while ((FPos = Content.find(NewlineStr, CPos)) != StringRef::npos) {
Text += Content.substr(CPos, FPos-CPos);
Text += '\n';
CPos = FPos + NewlineStr.size();
}
if (Text.empty())
Text.assign(ContentBegin, ContentEnd);
// Check that regex directives contain at least one regex.
if (RegexKind && Text.find("{{") == StringRef::npos) {
Diags.Report(Pos.getLocWithOffset(ContentBegin-PH.Begin),
diag::err_verify_missing_regex) << Text;
return false;
}
// Construct new directive.
std::unique_ptr<Directive> D = Directive::create(
RegexKind, Pos, ExpectedLoc, MatchAnyLine, Text, Min, Max);
std::string Error;
if (D->isValid(Error)) {
DL->push_back(std::move(D));
FoundDirective = true;
} else {
Diags.Report(Pos.getLocWithOffset(ContentBegin-PH.Begin),
diag::err_verify_invalid_content)
<< KindStr << Error;
}
}
return FoundDirective;
}
