| 1 |
//===--- LexerUtils.cpp - clang-tidy---------------------------------------===// |
| 2 |
// |
| 3 |
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
| 4 |
// See https://llvm.org/LICENSE.txt for license information. |
| 5 |
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
| 6 |
// |
| 7 |
//===----------------------------------------------------------------------===// |
| 8 |
|
| 9 |
#include "LexerUtils.h" |
| 10 |
#include "clang/AST/AST.h" |
| 11 |
#include "clang/Basic/SourceManager.h" |
| 12 |
#include <optional> |
| 13 |
|
| 14 |
namespace clang::tidy::utils::lexer { |
| 15 |
|
| 16 |
Token getPreviousToken(SourceLocation Location, const SourceManager &SM, |
| 17 |
const LangOptions &LangOpts, bool SkipComments) { |
| 18 |
Token Token; |
| 19 |
Token.setKind(tok::unknown); |
| 20 |
|
| 21 |
Location = Location.getLocWithOffset(-1); |
| 22 |
if (Location.isInvalid()) |
| 23 |
return Token; |
| 24 |
|
| 25 |
auto StartOfFile = SM.getLocForStartOfFile(SM.getFileID(Location)); |
| 26 |
while (Location != StartOfFile) { |
| 27 |
Location = Lexer::GetBeginningOfToken(Location, SM, LangOpts); |
| 28 |
if (!Lexer::getRawToken(Location, Token, SM, LangOpts) && |
| 29 |
(!SkipComments || !Token.is(tok::comment))) { |
| 30 |
break; |
| 31 |
} |
| 32 |
Location = Location.getLocWithOffset(-1); |
| 33 |
} |
| 34 |
return Token; |
| 35 |
} |
| 36 |
|
| 37 |
SourceLocation findPreviousTokenStart(SourceLocation Start, |
| 38 |
const SourceManager &SM, |
| 39 |
const LangOptions &LangOpts) { |
| 40 |
if (Start.isInvalid() || Start.isMacroID()) |
| 41 |
return SourceLocation(); |
| 42 |
|
| 43 |
SourceLocation BeforeStart = Start.getLocWithOffset(-1); |
| 44 |
if (BeforeStart.isInvalid() || BeforeStart.isMacroID()) |
| 45 |
return SourceLocation(); |
| 46 |
|
| 47 |
return Lexer::GetBeginningOfToken(BeforeStart, SM, LangOpts); |
| 48 |
} |
| 49 |
|
| 50 |
SourceLocation findPreviousTokenKind(SourceLocation Start, |
| 51 |
const SourceManager &SM, |
| 52 |
const LangOptions &LangOpts, |
| 53 |
tok::TokenKind TK) { |
| 54 |
if (Start.isInvalid() || Start.isMacroID()) |
| 55 |
return SourceLocation(); |
| 56 |
|
| 57 |
while (true) { |
| 58 |
SourceLocation L = findPreviousTokenStart(Start, SM, LangOpts); |
| 59 |
if (L.isInvalid() || L.isMacroID()) |
| 60 |
return SourceLocation(); |
| 61 |
|
| 62 |
Token T; |
| 63 |
if (Lexer::getRawToken(L, T, SM, LangOpts, /*IgnoreWhiteSpace=*/true)) |
| 64 |
return SourceLocation(); |
| 65 |
|
| 66 |
if (T.is(TK)) |
| 67 |
return T.getLocation(); |
| 68 |
|
| 69 |
Start = L; |
| 70 |
} |
| 71 |
} |
| 72 |
|
| 73 |
SourceLocation findNextTerminator(SourceLocation Start, const SourceManager &SM, |
| 74 |
const LangOptions &LangOpts) { |
| 75 |
return findNextAnyTokenKind(Start, SM, LangOpts, tok::comma, tok::semi); |
| 76 |
} |
| 77 |
|
| 78 |
std::optional<Token> |
| 79 |
findNextTokenSkippingComments(SourceLocation Start, const SourceManager &SM, |
| 80 |
const LangOptions &LangOpts) { |
| 81 |
std::optional<Token> CurrentToken; |
| 82 |
do { |
| 83 |
CurrentToken = Lexer::findNextToken(Start, SM, LangOpts); |
| 84 |
} while (CurrentToken && CurrentToken->is(tok::comment)); |
| 85 |
return CurrentToken; |
| 86 |
} |
| 87 |
|
| 88 |
bool rangeContainsExpansionsOrDirectives(SourceRange Range, |
| 89 |
const SourceManager &SM, |
| 90 |
const LangOptions &LangOpts) { |
| 91 |
assert(Range.isValid() && "Invalid Range for relexing provided"); |
| 92 |
SourceLocation Loc = Range.getBegin(); |
| 93 |
|
| 94 |
while (Loc < Range.getEnd()) { |
| 95 |
if (Loc.isMacroID()) |
| 96 |
return true; |
| 97 |
|
| 98 |
std::optional<Token> Tok = Lexer::findNextToken(Loc, SM, LangOpts); |
| 99 |
|
| 100 |
if (!Tok) |
| 101 |
return true; |
| 102 |
|
| 103 |
if (Tok->is(tok::hash)) |
| 104 |
return true; |
| 105 |
|
| 106 |
Loc = Lexer::getLocForEndOfToken(Loc, 0, SM, LangOpts).getLocWithOffset(1); |
| 107 |
} |
| 108 |
|
| 109 |
return false; |
| 110 |
} |
| 111 |
|
| 112 |
std::optional<Token> getQualifyingToken(tok::TokenKind TK, |
| 113 |
CharSourceRange Range, |
| 114 |
const ASTContext &Context, |
| 115 |
const SourceManager &SM) { |
| 116 |
assert((TK == tok::kw_const || TK == tok::kw_volatile || |
| 117 |
TK == tok::kw_restrict) && |
| 118 |
"TK is not a qualifier keyword"); |
| 119 |
std::pair<FileID, unsigned> LocInfo = SM.getDecomposedLoc(Range.getBegin()); |
| 120 |
StringRef File = SM.getBufferData(LocInfo.first); |
| 121 |
Lexer RawLexer(SM.getLocForStartOfFile(LocInfo.first), Context.getLangOpts(), |
| 122 |
File.begin(), File.data() + LocInfo.second, File.end()); |
| 123 |
std::optional<Token> LastMatchBeforeTemplate; |
| 124 |
std::optional<Token> LastMatchAfterTemplate; |
| 125 |
bool SawTemplate = false; |
| 126 |
Token Tok; |
| 127 |
while (!RawLexer.LexFromRawLexer(Tok) && |
| 128 |
Range.getEnd() != Tok.getLocation() && |
| 129 |
!SM.isBeforeInTranslationUnit(Range.getEnd(), Tok.getLocation())) { |
| 130 |
if (Tok.is(tok::raw_identifier)) { |
| 131 |
IdentifierInfo &Info = Context.Idents.get( |
| 132 |
StringRef(SM.getCharacterData(Tok.getLocation()), Tok.getLength())); |
| 133 |
Tok.setIdentifierInfo(&Info); |
| 134 |
Tok.setKind(Info.getTokenID()); |
| 135 |
} |
| 136 |
if (Tok.is(tok::less)) |
| 137 |
SawTemplate = true; |
| 138 |
else if (Tok.isOneOf(tok::greater, tok::greatergreater)) |
| 139 |
LastMatchAfterTemplate = std::nullopt; |
| 140 |
else if (Tok.is(TK)) { |
| 141 |
if (SawTemplate) |
| 142 |
LastMatchAfterTemplate = Tok; |
| 143 |
else |
| 144 |
LastMatchBeforeTemplate = Tok; |
| 145 |
} |
| 146 |
} |
| 147 |
return LastMatchAfterTemplate != std::nullopt ? LastMatchAfterTemplate |
| 148 |
: LastMatchBeforeTemplate; |
| 149 |
} |
| 150 |
|
| 151 |
static bool breakAndReturnEnd(const Stmt &S) { |
| 152 |
return isa<CompoundStmt, DeclStmt, NullStmt>(S); |
| 153 |
} |
| 154 |
|
| 155 |
static bool breakAndReturnEndPlus1Token(const Stmt &S) { |
| 156 |
return isa<Expr, DoStmt, ReturnStmt, BreakStmt, ContinueStmt, GotoStmt, SEHLeaveStmt>(S); |
| 157 |
} |
| 158 |
|
| 159 |
// Given a Stmt which does not include it's semicolon this method returns the |
| 160 |
// SourceLocation of the semicolon. |
| 161 |
static SourceLocation getSemicolonAfterStmtEndLoc(const SourceLocation &EndLoc, |
| 162 |
const SourceManager &SM, |
| 163 |
const LangOptions &LangOpts) { |
| 164 |
|
| 165 |
if (EndLoc.isMacroID()) { |
| 166 |
// Assuming EndLoc points to a function call foo within macro F. |
| 167 |
// This method is supposed to return location of the semicolon within |
| 168 |
// those macro arguments: |
| 169 |
// F ( foo() ; ) |
| 170 |
// ^ EndLoc ^ SpellingLoc ^ next token of SpellingLoc |
| 171 |
const SourceLocation SpellingLoc = SM.getSpellingLoc(EndLoc); |
| 172 |
std::optional<Token> NextTok = |
| 173 |
findNextTokenSkippingComments(SpellingLoc, SM, LangOpts); |
| 174 |
|
| 175 |
// Was the next token found successfully? |
| 176 |
// All macro issues are simply resolved by ensuring it's a semicolon. |
| 177 |
if (NextTok && NextTok->is(tok::TokenKind::semi)) { |
| 178 |
// Ideally this would return `F` with spelling location `;` (NextTok) |
| 179 |
// following the example above. For now simply return NextTok location. |
| 180 |
return NextTok->getLocation(); |
| 181 |
} |
| 182 |
|
| 183 |
// Fallthrough to 'normal handling'. |
| 184 |
// F ( foo() ) ; |
| 185 |
// ^ EndLoc ^ SpellingLoc ) ^ next token of EndLoc |
| 186 |
} |
| 187 |
|
| 188 |
std::optional<Token> NextTok = |
| 189 |
findNextTokenSkippingComments(EndLoc, SM, LangOpts); |
| 190 |
|
| 191 |
// Testing for semicolon again avoids some issues with macros. |
| 192 |
if (NextTok && NextTok->is(tok::TokenKind::semi)) |
| 193 |
return NextTok->getLocation(); |
| 194 |
|
| 195 |
return SourceLocation(); |
| 196 |
} |
| 197 |
|
| 198 |
SourceLocation getUnifiedEndLoc(const Stmt &S, const SourceManager &SM, |
| 199 |
const LangOptions &LangOpts) { |
| 200 |
|
| 201 |
const Stmt *LastChild = &S; |
| 202 |
while (!LastChild->children().empty() && !breakAndReturnEnd(*LastChild) && |
| 203 |
!breakAndReturnEndPlus1Token(*LastChild)) { |
| 204 |
for (const Stmt *Child : LastChild->children()) |
| 205 |
LastChild = Child; |
| 206 |
} |
| 207 |
|
| 208 |
if (!breakAndReturnEnd(*LastChild) && |
| 209 |
breakAndReturnEndPlus1Token(*LastChild)) |
| 210 |
return getSemicolonAfterStmtEndLoc(S.getEndLoc(), SM, LangOpts); |
| 211 |
|
| 212 |
return S.getEndLoc(); |
| 213 |
} |
| 214 |
|
| 215 |
} // namespace clang::tidy::utils::lexer |