kurl.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

                parse(parseBuffer.data(), 0);
            } else {
                m_string = relative;
                invalidate();
            }
            return;
        }

        switch (str[0]) {
        case '\0':
            // the reference must be empty - the RFC says this is a
            // reference to the same document
            *this = base;
            break;
        case '#': {
            // must be fragment-only reference
            appendASCII(base.m_string.left(base.m_queryEnd), str, len, parseBuffer);
            parse(parseBuffer.data(), 0);
            break;
        }
        case '?': {
            // query-only reference, special case needed for non-URL results
            appendASCII(base.m_string.left(base.m_pathEnd), str, len, parseBuffer);
            parse(parseBuffer.data(), 0);
            break;
        }
        case '/':
            // must be net-path or absolute-path reference
            if (str[1] == '/') {
                // net-path
                appendASCII(base.m_string.left(base.m_schemeEnd + 1), str, len, parseBuffer);
                parse(parseBuffer.data(), 0);
            } else {
                // abs-path
                appendASCII(base.m_string.left(base.m_portEnd), str, len, parseBuffer);
                parse(parseBuffer.data(), 0);
            }
            break;
        default:
            {
                // must be relative-path reference

                // Base part plus relative part plus one possible slash added in between plus terminating \0 byte.
                parseBuffer.resize(base.m_pathEnd + 1 + len + 1);

                char* bufferPos = parseBuffer.data();

                // first copy everything before the path from the base
                unsigned baseLength = base.m_string.length();
                const UChar* baseCharacters = base.m_string.characters();
                CharBuffer baseStringBuffer(baseLength);
                copyASCII(baseCharacters, baseLength, baseStringBuffer.data());
                const char* baseString = baseStringBuffer.data();
                const char* baseStringStart = baseString;
                const char* pathStart = baseStringStart + base.m_portEnd;
                while (baseStringStart < pathStart)
                    *bufferPos++ = *baseStringStart++;
                char* bufferPathStart = bufferPos;

                // now copy the base path
                const char* baseStringEnd = baseString + base.m_pathEnd;

                // go back to the last slash
                while (baseStringEnd > baseStringStart && baseStringEnd[-1] != '/')
                    baseStringEnd--;

                if (baseStringEnd == baseStringStart) {
                    // no path in base, add a path separator if necessary
                    if (base.m_schemeEnd + 1 != base.m_pathEnd && *str && *str != '?' && *str != '#')
                        *bufferPos++ = '/';
                } else {
                    bufferPos += copyPathRemovingDots(bufferPos, baseStringStart, 0, baseStringEnd - baseStringStart);
                }

                const char* relStringStart = str;
                const char* relStringPos = relStringStart;

                while (*relStringPos && *relStringPos != '?' && *relStringPos != '#') {
                    if (relStringPos[0] == '.' && bufferPos[-1] == '/') {
                        if (isPathSegmentEndChar(relStringPos[1])) {
                            // skip over "." segment
                            relStringPos += 1;
                            if (relStringPos[0] == '/')
                                relStringPos++;
                            continue;
                        } else if (relStringPos[1] == '.' && isPathSegmentEndChar(relStringPos[2])) {
                            // skip over ".." segment and rewind the last segment
                            // the RFC leaves it up to the app to decide what to do with excess
                            // ".." segments - we choose to drop them since some web content
                            // relies on this.
                            relStringPos += 2;
                            if (relStringPos[0] == '/')
                                relStringPos++;
                            if (bufferPos > bufferPathStart + 1)
                                bufferPos--;
                            while (bufferPos > bufferPathStart + 1  && bufferPos[-1] != '/')
                                bufferPos--;
                            continue;
                        }
                    }

                    *bufferPos = *relStringPos;
                    relStringPos++;
                    bufferPos++;
                }

                // all done with the path work, now copy any remainder
                // of the relative reference; this will also add a null terminator
                strcpy(bufferPos, relStringPos);

                parse(parseBuffer.data(), 0);

                ASSERT(strlen(parseBuffer.data()) + 1 <= parseBuffer.size());
                break;
            }
        }
    }
}

KURL KURL::copy() const
{
    KURL result = *this;
    result.m_string = result.m_string.copy();
    return result;
}

bool KURL::hasPath() const
{
    return m_pathEnd != m_portEnd;
}

String KURL::lastPathComponent() const
{
    if (!hasPath())
        return String();

    int end = m_pathEnd - 1;
    if (m_string[end] == '/')
        --end;

    int start = m_string.reverseFind('/', end);
    if (start < m_portEnd)
        return String();
    ++start;

    return m_string.substring(start, end - start + 1);
}

String KURL::protocol() const
{
    return m_string.left(m_schemeEnd);
}

String KURL::host() const
{
    int start = hostStart();
    return decodeURLEscapeSequences(m_string.substring(start, m_hostEnd - start));
}

unsigned short KURL::port() const
{
    if (m_hostEnd == m_portEnd)
        return 0;

    int number = m_string.substring(m_hostEnd + 1, m_portEnd - m_hostEnd - 1).toInt();
    if (number < 0 || number > 0xFFFF)
        return 0;
    return number;
}

String KURL::pass() const
{
    if (m_passwordEnd == m_userEnd)
        return String();

    return decodeURLEscapeSequences(m_string.substring(m_userEnd + 1, m_passwordEnd - m_userEnd - 1)); 
}

String KURL::user() const
{
    return decodeURLEscapeSequences(m_string.substring(m_userStart, m_userEnd - m_userStart));
}

String KURL::ref() const
{
    if (m_fragmentEnd == m_queryEnd)
        return String();

    return m_string.substring(m_queryEnd + 1, m_fragmentEnd - (m_queryEnd + 1));
}

bool KURL::hasRef() const
{
    return m_fragmentEnd != m_queryEnd;
}

#ifdef NDEBUG

static inline void assertProtocolIsGood(const char*)
{
}

#else

static void assertProtocolIsGood(const char* protocol)
{
    const char* p = protocol;
    while (*p) {
        ASSERT(*p > ' ' && *p < 0x7F && !(*p >= 'A' && *p <= 'Z'));
        ++p;
    }
}

#endif

bool KURL::protocolIs(const char* protocol) const
{
    // Do the comparison without making a new string object.
    assertProtocolIsGood(protocol);
    if (!m_isValid)
        return false;
    for (int i = 0; i < m_schemeEnd; ++i) {
        if (!protocol[i] || toASCIILower(m_string[i]) != protocol[i])
            return false;
    }
    return !protocol[m_schemeEnd]; // We should have consumed all characters in the argument.
}

String KURL::query() const
{
    if (m_queryEnd == m_pathEnd)
        return String();

    return m_string.substring(m_pathEnd + 1, m_queryEnd - (m_pathEnd + 1)); 
}

String KURL::path() const
{
    return decodeURLEscapeSequences(m_string.substring(m_portEnd, m_pathEnd - m_portEnd)); 
}

void KURL::setProtocol(const String& s)
{
    // FIXME: Non-ASCII characters must be encoded and escaped to match parse() expectations,
    // and to avoid changing more than just the protocol.

    if (!m_isValid) {
        parse(s + ":" + m_string);
        return;
    }

    parse(s + m_string.substring(m_schemeEnd));
}

void KURL::setHost(const String& s)
{
    if (!m_isValid)
        return;

    // FIXME: Non-ASCII characters must be encoded and escaped to match parse() expectations,
    // and to avoid changing more than just the host.

    bool slashSlashNeeded = m_userStart == m_schemeEnd + 1;

    parse(m_string.left(hostStart()) + (slashSlashNeeded ? "//" : "") + s + m_string.substring(m_hostEnd));
}

void KURL::setPort(unsigned short i)
{
    if (!m_isValid)
        return;

    if (i) {
        bool colonNeeded = m_portEnd == m_hostEnd;
        int portStart = (colonNeeded ? m_hostEnd : m_hostEnd + 1);

        parse(m_string.left(portStart) + (colonNeeded ? ":" : "") + String::number(i) + m_string.substring(m_portEnd));
    } else
        parse(m_string.left(m_hostEnd) + m_string.substring(m_portEnd));
}

void KURL::setHostAndPort(const String& hostAndPort)
{
    if (!m_isValid)
        return;

    // FIXME: Non-ASCII characters must be encoded and escaped to match parse() expectations,
    // and to avoid changing more than just host and port.

    bool slashSlashNeeded = m_userStart == m_schemeEnd + 1;

    parse(m_string.left(hostStart()) + (slashSlashNeeded ? "//" : "") + hostAndPort + m_string.substring(m_portEnd));
}

void KURL::setUser(const String& user)
{
    if (!m_isValid)
        return;

    // FIXME: Non-ASCII characters must be encoded and escaped to match parse() expectations,
    // and to avoid changing more than just the user login.
    String u;
    int end = m_userEnd;
    if (!user.isEmpty()) {
        u = user;
        if (m_userStart == m_schemeEnd + 1)
            u = "//" + u;
        // Add '@' if we didn't have one before.
        if (end == m_hostEnd || (end == m_passwordEnd && m_string[end] != '@'))
            u.append('@');
    } else {
        // Remove '@' if we now have neither user nor password.
        if (m_userEnd == m_passwordEnd && end != m_hostEnd && m_string[end] == '@')
            end += 1;
    }
    parse(m_string.left(m_userStart) + u + m_string.substring(end));
}

void KURL::setPass(const String& password)
{
    if (!m_isValid)
        return;

    // FIXME: Non-ASCII characters must be encoded and escaped to match parse() expectations,
    // and to avoid changing more than just the user password.
    String p;
    int end = m_passwordEnd;
    if (!password.isEmpty()) {
        p = ":" + password + "@";
        if (m_userEnd == m_schemeEnd + 1)
            p = "//" + p;
        // Eat the existing '@' since we are going to add our own.
        if (end != m_hostEnd && m_string[end] == '@')
            end += 1;
    } else {
        // Remove '@' if we now have neither user nor password.
        if (m_userStart == m_userEnd && end != m_hostEnd && m_string[end] == '@')
            end += 1;
    }
    parse(m_string.left(m_userEnd) + p + m_string.substring(end));
}

void KURL::setRef(const String& s)
{
    if (!m_isValid)
        return;

    // FIXME: Non-ASCII characters must be encoded and escaped to match parse() expectations.
    parse(m_string.left(m_queryEnd) + (s.isNull() ? "" : "#" + s));
}

void KURL::removeRef()
{
    if (!m_isValid)
        return;
    parse(m_string.left(m_queryEnd));
}
    
void KURL::setQuery(const String& query)
{
    if (!m_isValid)
        return;

    // FIXME: '#' and non-ASCII characters must be encoded and escaped.
    // Usually, the query is encoded using document encoding, not UTF-8, but we don't have
    // access to the document in this function.
    if ((query.isEmpty() || query[0] != '?') && !query.isNull())
        parse(m_string.left(m_pathEnd) + "?" + query + m_string.substring(m_queryEnd));
    else
        parse(m_string.left(m_pathEnd) + query + m_string.substring(m_queryEnd));

}

void KURL::setPath(const String& s)
{
    if (!m_isValid)
        return;

    // FIXME: encodeWithURLEscapeSequences does not correctly escape '#' and '?', so fragment and query parts
    // may be inadvertently affected.
    parse(m_string.left(m_portEnd) + encodeWithURLEscapeSequences(s) + m_string.substring(m_pathEnd));
}

String KURL::prettyURL() const
{
    if (!m_isValid)
        return m_string;

    Vector<UChar> result;

    append(result, protocol());
    result.append(':');

    Vector<UChar> authority;

    if (m_hostEnd != m_passwordEnd) {
        if (m_userEnd != m_userStart) {
            append(authority, user());
            authority.append('@');
        }
        append(authority, host());
        if (port() != 0) {
            authority.append(':');
            append(authority, String::number(port()));
        }
    }

    if (!authority.isEmpty()) {
        result.append('/');
        result.append('/');
        result.append(authority);