view release on metacpan or  search on metacpan

libsass/util.cpp  view on Meta::CPAN

      } else if (
        i == ' ' ||
        i == '\r' ||
        i == '\n' ||
        i == '	'
      ) {
        // only add one space
        if (!ws) text += ' ';
        ws = true;
      } else {
        ws = false;
        text += i;
      }
    }
    if (esc) text += '\\';
    return text;
  }

  // find best quote_mark by detecting if the string contains any single
  // or double quotes. When a single quote is found, we not we want a double
  // quote as quote_mark. Otherwise we check if the string cotains any double
  // quotes, which will trigger the use of single quotes as best quote_mark.
  char detect_best_quotemark(const char* s, char qm)
  {
    // ensure valid fallback quote_mark
    char quote_mark = qm && qm != '*' ? qm : '"';
    while (*s) {
      // force double quotes as soon
      // as one single quote is found
      if (*s == '\'') { return '"'; }
      // a single does not force quote_mark
      // maybe we see a double quote later
      else if (*s == '"') { quote_mark = '\''; }
      ++ s;
    }
    return quote_mark;
  }

  string unquote(const string& s, char* qd)
  {

    // not enough room for quotes
    // no possibility to unquote
    if (s.length() < 2) return s;

    char q;
    bool skipped = false;

    // this is no guarantee that the unquoting will work
    // what about whitespace before/after the quote_mark?
    if      (*s.begin() == '"'  && *s.rbegin() == '"')  q = '"';
    else if (*s.begin() == '\'' && *s.rbegin() == '\'') q = '\'';
    else                                                return s;

    string unq;
    unq.reserve(s.length()-2);

    for (size_t i = 1, L = s.length() - 1; i < L; ++i) {

      // implement the same strange ruby sass behavior
      // an escape sequence can also mean a unicode char
      if (s[i] == '\\' && !skipped) {
        // remember
        skipped = true;

        // skip it
        // ++ i;

        // if (i == L) break;

        // escape length
        size_t len = 1;

        // parse as many sequence chars as possible
        // ToDo: Check if ruby aborts after possible max
        while (i + len < L && s[i + len] && isxdigit(s[i + len])) ++ len;

        // hex string?
        if (len > 1) {

          // convert the extracted hex string to code point value
          // ToDo: Maybe we could do this without creating a substring
          uint32_t cp = strtol(s.substr (i + 1, len - 1).c_str(), nullptr, 16);

          // assert invalid code points
          if (cp == 0) cp = 0xFFFD;
          // replace bell character
          // if (cp == 10) cp = 32;

          // use a very simple approach to convert via utf8 lib
          // maybe there is a more elegant way; maybe we shoud
          // convert the whole output from string to a stream!?
          // allocate memory for utf8 char and convert to utf8
          unsigned char u[5] = {0,0,0,0,0}; utf8::append(cp, u);
          for(size_t m = 0; u[m] && m < 5; m++) unq.push_back(u[m]);

          // skip some more chars?
          i += len - 1; skipped = false;

        }


      }
      // check for unexpected delimiter
      // be strict and throw error back
      // else if (!skipped && q == s[i]) {
      //   // don't be that strict
      //   return s;
      //   // this basically always means an internal error and not users fault
      //   error("Unescaped delimiter in string to unquote found. [" + s + "]", ParserState("[UNQUOTE]"));
      // }
      else {
        skipped = false;
        unq.push_back(s[i]);
      }

    }
    if (skipped) { return s; }
    if (qd) *qd = q;
    return unq;