Method Lex

library/cpp/regex/pire/pire/re_lexer.cpp:223–286

        bool Accepts(wchar32 c) const { return c == '[' || c == (Control | '[') || c == (Control | ']'); }

        Term Lex()
        {
            static const char* controls = "^[]-\\";
            static const char* controls2 = "*+{}()$?.&~";
            wchar32 ch = CorrectChar(GetChar(), controls);
            if (ch == '[' || ch == ']')
                return Term::Character(ch);

            Term::CharacterRange cs;
            ch = CorrectChar(GetChar(), controls);
            if (ch == (Control | '^')) {
                cs.second = true;
                ch = CorrectChar(GetChar(), controls);
            }

            bool firstUnicode;
            wchar32 unicodeSymbol = 0;

            for (; ch != End && ch != (Control | ']'); ch = CorrectChar(GetChar(), controls)) {
                if (ch == (Control | 'x')) {
                    UngetChar(ch);
                    firstUnicode = true;
                    unicodeSymbol = ReadUnicodeCharacter();
                } else {
                    firstUnicode = false;
                }

                if (((ch & ControlMask) != Control || firstUnicode) && CorrectChar(PeekChar(), controls) == (Control | '-')) {
                    GetChar();
                    wchar32 current = GetChar();

                    bool secondUnicode = (current == (Control | 'x'));

                    wchar32 begin = (firstUnicode) ? unicodeSymbol : ch;
                    wchar32 end;
                    if (secondUnicode) {
                        UngetChar(current);
                        end = ReadUnicodeCharacter();
                    } else {
                        end = CorrectChar(current, controls);
                        if ((end & ControlMask) == Control)
                            Error("Wrong character range");
                    }

                    for (ch = begin; ch <= end; ++ch) {
                        cs.first.insert(Term::String(1, ch));
                    }
                } else if (ch == (Control | '-')) {
                    cs.first.insert(Term::String(1, '-'));
                }
                else if ((ch & ControlMask) == Control && (strchr(controls2, ch & ~ControlMask) || strchr(controls, ch & ~ControlMask))) {
                    cs.first.insert(Term::String(1, ch & ~ControlMask));
                }
                else if ((ch & ControlMask) != Control || !strchr(controls, ch & ~ControlMask)) {
                    cs.first.insert(Term::String(1, (firstUnicode) ? unicodeSymbol : ch));
                } else {
                    Error("Wrong character in range");
                }