AkelPad

Code: Select all

// TODO: consider "(\n|abc)*" situations

// returns max lines that the cszRegExp matches
// -1 means unpredictable multi line match (e.g. "[\n]+")
int AnalyzeRegExp(const wchar_t* cszRegExp, int isDotMatchingNewLine)
{
    enum eStateFlags {
        reNormal           = 0,
        reEscapedJustFound = 0x0001, // just encountered '\'
        reEscaped          = 0x0002, // current char is escaped (after '\')
        reCharSet          = 0x0010, // inside [...]
        reNewLineJustFound = 0x0100, // just encountered '\n' or '\r' (or '.')
        reNewLineChar      = 0x0200, // previous char is a new line char
        reNewLineInCharSet = 0x0400, // a new line char inside [...]
        reDone             = 0x8000  // done
    };
    
    int nRegExpLines = 1;
    unsigned int state = reNormal;

    for (; state != reDone; ++cszRegExp)
    {
        const wchar_t ch = *cszRegExp;
        if (ch == 0)
        {
            state = reDone;
        }
        else if (ch == L'\\')
        {
            if (!(state & reEscaped))
                state |= reEscapedJustFound;
        }
        else if (ch == L'n' || ch == L'r')
        {
            if (state & reEscaped)
                state |= reNewLineJustFound;
        }
        else if (!(state & reEscaped))
        {
            if (ch == L'[')
            {
                state |= reCharSet;
            }
            else if (ch == L']')
            {
                if (state & reCharSet)
                {
                    state ^= reCharSet;
                    if (state & reNewLineInCharSet)
                    {
                        state ^= reNewLineInCharSet;
                        state |= reNewLineJustFound;
                    }
                }
            }
            else if (ch == L'+' || ch == L'*')
            {
                if (state & reNewLineChar)
                {
                    nRegExpLines = -1;
                    state = reDone;
                }
            }
            else if (ch == L'.')
            {
                if (isDotMatchingNewLine)
                    state |= reNewLineJustFound;
            }
        }

        if (state & reNewLineJustFound)
        {
            state ^= reNewLineJustFound;
            if (state & reCharSet)
            {
                state |= reNewLineInCharSet;
            }
            else
            {
                state |= reNewLineChar;
                ++nRegExpLines;
            }
        }
        else if (state & reNewLineChar)
            state ^= reNewLineChar;

        if (state & reEscapedJustFound)
        {
            state ^= reEscapedJustFound;
            state |= reEscaped;
        }
        else if (state & reEscaped)
            state ^= reEscaped;
    }

    return nRegExpLines;
}