C++ SourceManager::extractText方法代码示例

/// \brief Tokenizes a string literal, taking into account string interpolation.
static void getStringPartTokens(const Token &Tok, const LangOptions &LangOpts,
                                const SourceManager &SM,
                                int BufID, std::vector<Token> &Toks) {
  assert(Tok.is(tok::string_literal));
  bool IsMultiline = Tok.IsMultilineString();
  unsigned QuoteLen = IsMultiline ? 3 : 1;
  SmallVector<Lexer::StringSegment, 4> Segments;
  Lexer::getStringLiteralSegments(Tok, Segments, /*Diags=*/nullptr);
  for (unsigned i = 0, e = Segments.size(); i != e; ++i) {
    Lexer::StringSegment &Seg = Segments[i];
    bool isFirst = i == 0;
    bool isLast = i == e-1;
    if (Seg.Kind == Lexer::StringSegment::Literal) {
      SourceLoc Loc = Seg.Loc;
      unsigned Len = Seg.Length;
      if (isFirst) {
        // Include the quote.
        Loc = Loc.getAdvancedLoc(-QuoteLen);
        Len += QuoteLen;
      }
      if (isLast) {
        // Include the quote.
        Len += QuoteLen;
      }

      StringRef Text = SM.extractText({ Loc, Len });
      Token NewTok;
      NewTok.setToken(tok::string_literal, Text, IsMultiline);
      Toks.push_back(NewTok);

    } else {
      assert(Seg.Kind == Lexer::StringSegment::Expr &&
             "new enumerator was introduced ?");
      unsigned Offset = SM.getLocOffsetInBuffer(Seg.Loc, BufID);
      unsigned EndOffset = Offset + Seg.Length;

      if (isFirst) {
        // Add a token for the quote character.
        StringRef Text = SM.extractText({ Seg.Loc.getAdvancedLoc(-2), 1 });
        Token NewTok;
        NewTok.setToken(tok::string_literal, Text);
        Toks.push_back(NewTok);
      }

      std::vector<Token> NewTokens = swift::tokenize(LangOpts, SM, BufID,
                                                     Offset, EndOffset,
                                                     /*KeepComments=*/true);
      Toks.insert(Toks.end(), NewTokens.begin(), NewTokens.end());

      if (isLast) {
        // Add a token for the quote character.
        StringRef Text = SM.extractText({ Seg.Loc.getAdvancedLoc(Seg.Length),
                                          1 });
        Token NewTok;
        NewTok.setToken(tok::string_literal, Text);
        Toks.push_back(NewTok);
      }
    }
  }
}

SingleRawComment::SingleRawComment(CharSourceRange Range,
                                   const SourceManager &SourceMgr)
    : Range(Range), RawText(SourceMgr.extractText(Range)),
      Kind(static_cast<unsigned>(getCommentKind(RawText))) {
  auto StartLineAndColumn = SourceMgr.getLineAndColumn(Range.getStart());
  StartLine = StartLineAndColumn.first;
  StartColumn = StartLineAndColumn.second;
  EndLine = SourceMgr.getLineNumber(Range.getEnd());
}

/// \brief Extract a character immediately before \p Loc. If \p Loc is the
/// start of the buffer, return '\f'.
static char extractCharBefore(SourceManager &SM, SourceLoc Loc) {
  // We have to be careful not to go off the front of the buffer.
  auto bufferID = SM.findBufferContainingLoc(Loc);
  auto bufferRange = SM.getRangeForBuffer(bufferID);
  if (bufferRange.getStart() == Loc)
    return '\f';
  auto chars = SM.extractText({Loc.getAdvancedLoc(-1), 1}, bufferID);
  assert(!chars.empty() && "Couldn't extractText with valid range");
  return chars[0];
}

static SourceLoc findEndOfLine(SourceManager &SM, SourceLoc loc,
                               unsigned bufferID) {
  CharSourceRange entireBuffer = SM.getRangeForBuffer(bufferID);
  CharSourceRange rangeFromLoc{SM, loc, entireBuffer.getEnd()};
  StringRef textFromLoc = SM.extractText(rangeFromLoc);
  size_t newlineOffset = textFromLoc.find_first_of({"\r\n\0", 3});
  if (newlineOffset == StringRef::npos)
    return entireBuffer.getEnd();
  return loc.getAdvancedLoc(newlineOffset);
}

SourceCompleteResult
ide::isSourceInputComplete(std::unique_ptr<llvm::MemoryBuffer> MemBuf) {
  SourceManager SM;
  auto BufferID = SM.addNewSourceBuffer(std::move(MemBuf));
  ParserUnit Parse(SM, BufferID);
  Parser &P = Parse.getParser();

  bool Done;
  do {
    P.parseTopLevel();
    Done = P.Tok.is(tok::eof);
  } while (!Done);

  SourceCompleteResult SCR;
  SCR.IsComplete = !P.isInputIncomplete();

  // Use the same code that was in the REPL code to track the indent level
  // for now. In the future we should get this from the Parser if possible.

  CharSourceRange entireRange = SM.getRangeForBuffer(BufferID);
  StringRef Buffer = SM.extractText(entireRange);
  const char *SourceStart = Buffer.data();
  const char *SourceEnd = Buffer.data() + Buffer.size();
  const char *LineStart = SourceStart;
  const char *LineSourceStart = nullptr;
  uint32_t LineIndent = 0;
  struct IndentInfo {
    StringRef Prefix;
    uint32_t Indent;
    IndentInfo(const char *s, size_t n, uint32_t i) :
      Prefix(s, n),
      Indent(i) {}
  };
  SmallVector<IndentInfo, 4> IndentInfos;
  for (const char *p = SourceStart; p<SourceEnd; ++p) {
    switch (*p) {
    case '\r':
    case '\n':
      LineIndent = 0;
      LineSourceStart = nullptr;
      LineStart = p + 1;
      break;

    case '"':
      p = skipStringInCode (p, SourceEnd);
      break;

    case '{':
    case '(':
    case '[':
      ++LineIndent;
      if (LineSourceStart == nullptr)
        IndentInfos.push_back(IndentInfo(LineStart,
                                         p - LineStart,
                                         LineIndent));
      else
        IndentInfos.push_back(IndentInfo(LineStart,
                                         LineSourceStart - LineStart,
                                         LineIndent));
      break;

    case '}':
    case ')':
    case ']':
      if (LineIndent > 0)
        --LineIndent;
      if (!IndentInfos.empty())
        IndentInfos.pop_back();
      break;

    default:
      if (LineSourceStart == nullptr && !isspace(*p))
        LineSourceStart = p;
      break;
    }
    if (*p == '\0')
      break;
  }
  if (!IndentInfos.empty()) {
    SCR.IndentPrefix = IndentInfos.back().Prefix.str();
    // Trim off anything that follows a non-space character
    const size_t pos = SCR.IndentPrefix.find_first_not_of(" \t");
    if (pos != std::string::npos)
        SCR.IndentPrefix.erase(pos);
    SCR.IndentLevel = IndentInfos.back().Indent;
  }
  return SCR;
}

/// \brief Extract a character at \p Loc. If \p Loc is the end of the buffer,
/// return '\f'.
static char extractCharAfter(SourceManager &SM, SourceLoc Loc) {
  auto chars = SM.extractText({Loc, 1});
  return chars.empty() ? '\f' : chars[0];
}

/// Extract the text for the given expression.
static StringRef extractExprText(const Expr *E, SourceManager &SM) {
  const auto CSR = Lexer::getCharSourceRangeFromSourceRange(SM,
      E->getSourceRange());
  return SM.extractText(CSR);
}