Exemplo n.º 1
0
/**
 * Initialize Table
 * ================================
 *
 * The following reads in the PEG in the passed filename and converts the
 * grammar specification into a set of tokens for traversal when parsing other files
 * according to said grammar.
 */
void Parser::initializeTable(Scanner& input)
{
    // These regexes are used to read in a custom PEG grammar.
    // Refer to /grammars/arithmetic.peg for a more thorough explanation
    // on the grammar. Note all other terms can be manipulated just
    // by reading in the remainder of a line or reading in words.
    Regex arrowOperator = Regex::fromPool("pparser-arrow", "\\->");
    Regex markedWord = Regex::fromPool("pparser-marked-word", "\\A+'?");

    // On any given line, the first two terminals should be the nonterminal
    // being defined and the arrow operator or we've encountered a comment.
    while(input.peek() != EOF) {

        if(input.peek() == PPARSER_COMMENT) {
            input.readLine();
        } else {

            // First read in nonterminal and find start if possible
            std::string nonterminal = input.next(markedWord);
            if(nonterminal.back() == PPARSER_START) {
                nonterminal.pop_back();
                if(start.empty()) {
                    start = nonterminal;
                } else {
                    throw InvalidGrammar("Multiple starting nonterminals", input.getCurrentState());
                }
            }

            // Can now read in the arrow operator
            input.next(arrowOperator);

            // Rest of line is dedicated to definition
            table[nonterminal] = std::make_shared<Choices>(input);
        }
    }

    // Must have at least one starting nonterminal
    if(start.empty()) {
        throw InvalidGrammar("No starting nonterminal specified");
    }

}
Exemplo n.º 2
0
Inst ExtManager::parseExtInstMnemo(Scanner& scanner, Brigantine& bw, int* vx) const
{
    assert(scanner.peek().kind() == EExtInstName);

    // Parse mnemo prefix adding suffices until there is an extension which can handle it.
    // This prefix has the form <vendor>_<extension>

    string prefix = scanner.scan().text();
    while (!isMnemoPrefix(prefix) && scanner.peek().kind() == EExtInstSuff) 
    {
        prefix += scanner.scan().text();
    }

    // Parse remaining part of instruction mnemo (typically in the form <opcode>_<suff>)

    if (const Extension* e = getByPrefix(prefix)) return e->parseInstMnemo(prefix, scanner, bw, vx);

    // Enabled extensions failed to parse this mnemo.
    // Search for a disabled extension which might have handled it
    const char* s = 0;
    for (unsigned i = 0; i < size(); ++i)
    {
        if (!isEnabled[i] && (s = extension[i]->matchInstMnemo(prefix)) != 0)
        {
            scanner.syntaxError(string("Instruction \"") + s + "\" cannot be used (extension \"" + extension[i]->getName() + "\" is not enabled)");
            return Inst();
        }
    }

    // Extensions (including disabled) cannot handle this mnemo

    if (!hasEnabled()) scanner.scan(); // show offending token
    scanner.syntaxError("Undefined instruction");

    return Inst();
}