wrecparser.cpp

linux下开源浏览器WebKit的源码,市面上的很多商用浏览器都是移植自WebKit

        }
    }
    consume();

    // lazily catch reversed ranges ([z-a])in character classes
    if (constructor.isUpsideDown()) {
        setError(CharacterClassOutOfOrder);
        return false;
    }

    constructor.flush();
    CharacterClass charClass = constructor.charClass();
    return parseCharacterClassQuantifier(failures, charClass, invert);
}

bool Parser::parseNonCharacterEscape(JumpList& failures, const Escape& escape)
{
    switch (escape.type()) {
        case Escape::PatternCharacter:
            ASSERT_NOT_REACHED();
            return false;

        case Escape::CharacterClass:
            return parseCharacterClassQuantifier(failures, CharacterClassEscape::cast(escape).characterClass(), CharacterClassEscape::cast(escape).invert());

        case Escape::Backreference:
            return parseBackreferenceQuantifier(failures, BackreferenceEscape::cast(escape).subpatternId());

        case Escape::WordBoundaryAssertion:
            m_generator.generateAssertionWordBoundary(failures, WordBoundaryAssertionEscape::cast(escape).invert());
            return true;

        case Escape::Error:
            return false;
    }

    ASSERT_NOT_REACHED();
    return false;
}

Escape Parser::consumeEscape(bool inCharacterClass)
{
    switch (peek()) {
    case EndOfPattern:
        setError(EscapeUnterminated);
        return Escape(Escape::Error);

    // Assertions
    case 'b':
        consume();
        if (inCharacterClass)
            return PatternCharacterEscape('\b');
        return WordBoundaryAssertionEscape(false); // do not invert
    case 'B':
        consume();
        if (inCharacterClass)
            return PatternCharacterEscape('B');
        return WordBoundaryAssertionEscape(true); // invert

    // CharacterClassEscape
    case 'd':
        consume();
        return CharacterClassEscape(CharacterClass::digits(), false);
    case 's':
        consume();
        return CharacterClassEscape(CharacterClass::spaces(), false);
    case 'w':
        consume();
        return CharacterClassEscape(CharacterClass::wordchar(), false);
    case 'D':
        consume();
        return inCharacterClass
            ? CharacterClassEscape(CharacterClass::nondigits(), false)
            : CharacterClassEscape(CharacterClass::digits(), true);
    case 'S':
        consume();
        return inCharacterClass
            ? CharacterClassEscape(CharacterClass::nonspaces(), false)
            : CharacterClassEscape(CharacterClass::spaces(), true);
    case 'W':
        consume();
        return inCharacterClass
            ? CharacterClassEscape(CharacterClass::nonwordchar(), false)
            : CharacterClassEscape(CharacterClass::wordchar(), true);

    // DecimalEscape
    case '1':
    case '2':
    case '3':
    case '4':
    case '5':
    case '6':
    case '7':
    case '8':
    case '9': {
        if (peekDigit() > m_numSubpatterns || inCharacterClass) {
            // To match Firefox, we parse an invalid backreference in the range [1-7]
            // as an octal escape.
            return peekDigit() > 7 ? PatternCharacterEscape('\\') : PatternCharacterEscape(consumeOctal());
        }

        int value = 0;
        do {
            unsigned newValue = value * 10 + peekDigit();
            if (newValue > m_numSubpatterns)
                break;
            value = newValue;
            consume();
        } while (peekIsDigit());

        return BackreferenceEscape(value);
    }

    // Octal escape
    case '0':
        consume();
        return PatternCharacterEscape(consumeOctal());

    // ControlEscape
    case 'f':
        consume();
        return PatternCharacterEscape('\f');
    case 'n':
        consume();
        return PatternCharacterEscape('\n');
    case 'r':
        consume();
        return PatternCharacterEscape('\r');
    case 't':
        consume();
        return PatternCharacterEscape('\t');
    case 'v':
        consume();
        return PatternCharacterEscape('\v');

    // ControlLetter
    case 'c': {
        SavedState state(*this);
        consume();
        
        int control = consume();
        // To match Firefox, inside a character class, we also accept numbers
        // and '_' as control characters.
        if ((!inCharacterClass && !isASCIIAlpha(control)) || (!isASCIIAlphanumeric(control) && control != '_')) {
            state.restore();
            return PatternCharacterEscape('\\');
        }
        return PatternCharacterEscape(control & 31);
    }

    // HexEscape
    case 'x': {
        consume();

        SavedState state(*this);
        int x = consumeHex(2);
        if (x == -1) {
            state.restore();
            return PatternCharacterEscape('x');
        }
        return PatternCharacterEscape(x);
    }

    // UnicodeEscape
    case 'u': {
        consume();

        SavedState state(*this);
        int x = consumeHex(4);
        if (x == -1) {
            state.restore();
            return PatternCharacterEscape('u');
        }
        return PatternCharacterEscape(x);
    }

    // IdentityEscape
    default:
        return PatternCharacterEscape(consume());
    }
}

void Parser::parseAlternative(JumpList& failures)
{
    PatternCharacterSequence sequence(m_generator, failures);

    while (1) {
        switch (peek()) {
        case EndOfPattern:
        case '|':
        case ')':
            sequence.flush();
            return;

        case '*':
        case '+':
        case '?':
        case '{': {
            Quantifier q = consumeQuantifier();

            if (q.type == Quantifier::None) {
                sequence.append(consume());
                continue;
            }

            if (q.type == Quantifier::Error)
                return;

            if (!sequence.size()) {
                setError(QuantifierWithoutAtom);
                return;
            }

            sequence.flush(q);
            continue;
        }

        case '^':
            consume();

            sequence.flush();
            m_generator.generateAssertionBOL(failures);
            continue;

        case '$':
            consume();

            sequence.flush();
            m_generator.generateAssertionEOL(failures);
            continue;

        case '.':
            consume();

            sequence.flush();
            if (!parseCharacterClassQuantifier(failures, CharacterClass::newline(), true))
                return;
            continue;

        case '[':
            consume();

            sequence.flush();
            if (!parseCharacterClass(failures))
                return;
            continue;

        case '(':
            consume();

            sequence.flush();
            if (!parseParentheses(failures))
                return;
            continue;

        case '\\': {
            consume();

            Escape escape = consumeEscape(false);
            if (escape.type() == Escape::PatternCharacter) {
                sequence.append(PatternCharacterEscape::cast(escape).character());
                continue;
            }

            sequence.flush();
            if (!parseNonCharacterEscape(failures, escape))
                return;
            continue;
        }

        default:
            sequence.append(consume());
            continue;
        }
    }
}

/*
  TOS holds index.
*/
void Parser::parseDisjunction(JumpList& failures)
{
    parseAlternative(failures);
    if (peek() != '|')
        return;

    JumpList successes;
    do {
        consume();
        m_generator.terminateAlternative(successes, failures);
        parseAlternative(failures);
    } while (peek() == '|');

    m_generator.terminateDisjunction(successes);
}

Generator::ParenthesesType Parser::consumeParenthesesType()
{
    if (peek() != '?')
        return Generator::Capturing;
    consume();

    switch (consume()) {
    case ':':
        return Generator::NonCapturing;
    
    case '=':
        return Generator::Assertion;

    case '!':
        return Generator::InvertedAssertion;

    default:
        setError(ParenthesesTypeInvalid);
        return Generator::Error;
    }
}

} } // namespace JSC::WREC

#endif // ENABLE(WREC)