teak-llvm/clang/lib/Rewrite/TokenRewriter.cpp
Chandler Carruth 2946cd7010 Update the file headers across all of the LLVM projects in the monorepo
to reflect the new license.

We understand that people may be surprised that we're moving the header
entirely to discuss the new license. We checked this carefully with the
Foundation's lawyer and we believe this is the correct approach.

Essentially, all code in the project is now made available by the LLVM
project under our new license, so you will see that the license headers
include that license only. Some of our contributors have contributed
code under our old license, and accordingly, we have retained a copy of
our old license notice in the top-level files in each project and
repository.

llvm-svn: 351636
2019-01-19 08:50:56 +00:00

100 lines
3.2 KiB
C++

//===- TokenRewriter.cpp - Token-based code rewriting interface -----------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
//
// This file implements the TokenRewriter class, which is used for code
// transformations.
//
//===----------------------------------------------------------------------===//
#include "clang/Rewrite/Core/TokenRewriter.h"
#include "clang/Basic/SourceManager.h"
#include "clang/Lex/Lexer.h"
#include "clang/Lex/ScratchBuffer.h"
#include "clang/Lex/Token.h"
#include <cassert>
#include <cstring>
#include <map>
#include <utility>
using namespace clang;
TokenRewriter::TokenRewriter(FileID FID, SourceManager &SM,
const LangOptions &LangOpts) {
ScratchBuf.reset(new ScratchBuffer(SM));
// 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 all comments and whitespace as tokens.
RawLex.SetKeepWhitespaceMode(true);
// Lex the file, populating our datastructures.
Token RawTok;
RawLex.LexFromRawLexer(RawTok);
while (RawTok.isNot(tok::eof)) {
#if 0
if (Tok.is(tok::raw_identifier)) {
// Look up the identifier info for the token. This should use
// IdentifierTable directly instead of PP.
PP.LookUpIdentifierInfo(Tok);
}
#endif
AddToken(RawTok, TokenList.end());
RawLex.LexFromRawLexer(RawTok);
}
}
TokenRewriter::~TokenRewriter() = default;
/// RemapIterator - Convert from token_iterator (a const iterator) to
/// TokenRefTy (a non-const iterator).
TokenRewriter::TokenRefTy TokenRewriter::RemapIterator(token_iterator I) {
if (I == token_end()) return TokenList.end();
// FIXME: This is horrible, we should use our own list or something to avoid
// this.
std::map<SourceLocation, TokenRefTy>::iterator MapIt =
TokenAtLoc.find(I->getLocation());
assert(MapIt != TokenAtLoc.end() && "iterator not in rewriter?");
return MapIt->second;
}
/// AddToken - Add the specified token into the Rewriter before the other
/// position.
TokenRewriter::TokenRefTy
TokenRewriter::AddToken(const Token &T, TokenRefTy Where) {
Where = TokenList.insert(Where, T);
bool InsertSuccess = TokenAtLoc.insert(std::make_pair(T.getLocation(),
Where)).second;
assert(InsertSuccess && "Token location already in rewriter!");
(void)InsertSuccess;
return Where;
}
TokenRewriter::token_iterator
TokenRewriter::AddTokenBefore(token_iterator I, const char *Val) {
unsigned Len = strlen(Val);
// Plop the string into the scratch buffer, then create a token for this
// string.
Token Tok;
Tok.startToken();
const char *Spelling;
Tok.setLocation(ScratchBuf->getToken(Val, Len, Spelling));
Tok.setLength(Len);
// TODO: Form a whole lexer around this and relex the token! For now, just
// set kind to tok::unknown.
Tok.setKind(tok::unknown);
return AddToken(Tok, RemapIterator(I));
}