kurlgoogle.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

            m_string = String::fromUTF8(m_utf8.data(), m_utf8.length());
        m_stringIsValid = true;
    }
    return m_string;
}

// KURL ------------------------------------------------------------------------

// Creates with NULL-terminated string input representing an absolute URL.
// WebCore generally calls this only with hardcoded strings, so the input is
// ASCII. We treat is as UTF-8 just in case.
KURL::KURL(const char *url)
{
    // FIXME The Mac code checks for beginning with a slash and converting to a
    // file: URL. We will want to add this as well once we can compile on a
    // system like that.
    m_url.init(KURL(), url, strlen(url), 0);

    // The one-argument constructors should never generate a NULL string.
    // This is a funny quirk of KURL.cpp (probably a bug) which we preserve.
    if (m_url.utf8String().isNull())
        m_url.setAscii(CString("", 0));
}

// Initializes with a string representing an absolute URL. No encoding
// information is specified. This generally happens when a KURL is converted
// to a string and then converted back. In this case, the URL is already
// canonical and in proper escaped form so needs no encoding. We treat it was
// UTF-8 just in case.
KURL::KURL(const String& url)
{
    if (!url.isNull())
        m_url.init(KURL(), url, 0);
    else {
        // WebCore expects us to preserve the nullness of strings when this
        // constructor is used. In all other cases, it expects a non-null
        // empty string, which is what init() will create.
        m_url.m_isValid = false;
        m_url.m_protocolInHTTPFamily = false;
    }
}

// Constructs a new URL given a base URL and a possibly relative input URL.
// This assumes UTF-8 encoding.
KURL::KURL(const KURL& base, const String& relative)
{
    m_url.init(base, relative, 0);
}

// Constructs a new URL given a base URL and a possibly relative input URL.
// Any query portion of the relative URL will be encoded in the given encoding.
KURL::KURL(const KURL& base,
           const String& relative,
           const TextEncoding& encoding)
{
    m_url.init(base, relative, &encoding.encodingForFormSubmission());
}

KURL::KURL(const CString& canonicalSpec,
           const url_parse::Parsed& parsed, bool isValid)
    : m_url(parsed, isValid)
{
    // We know the reference fragment is the only part that can be UTF-8, so
    // we know it's ASCII when there is no ref.
    if (parsed.ref.is_nonempty())
        m_url.setUtf8(canonicalSpec);
    else
        m_url.setAscii(canonicalSpec);
}

#if PLATFORM(CF)
KURL::KURL(CFURLRef)
{
    notImplemented();
    invalidate();
}

CFURLRef KURL::createCFURL() const
{
    notImplemented();
    return 0;
}
#endif

KURL KURL::copy() const
{
    KURL result = *this;
    m_url.copyTo(&result.m_url);
    return result;
}

bool KURL::isNull() const
{
    return m_url.utf8String().isNull();
}

bool KURL::isEmpty() const
{
    return !m_url.utf8String().length();
}

bool KURL::isValid() const
{
    return m_url.m_isValid;
}

bool KURL::protocolInHTTPFamily() const
{
    return m_url.m_protocolInHTTPFamily;
}

bool KURL::hasPath() const
{
    // Note that http://www.google.com/" has a path, the path is "/". This can
    // return false only for invalid or nonstandard URLs.
    return m_url.m_parsed.path.len >= 0;
}

// We handle "parameters" separated by a semicolon, while KURL.cpp does not,
// which can lead to different results in some cases.
String KURL::lastPathComponent() const
{
    // When the output ends in a slash, WebCore has different expectations than
    // the GoogleURL library. For "/foo/bar/" the library will return the empty
    // string, but WebCore wants "bar".
    url_parse::Component path = m_url.m_parsed.path;
    if (path.len > 0 && m_url.utf8String().data()[path.end() - 1] == '/')
        path.len--;

    url_parse::Component file;
    url_parse::ExtractFileName(m_url.utf8String().data(), path, &file);

    // Bug: https://bugs.webkit.org/show_bug.cgi?id=21015 this function returns
    // a null string when the path is empty, which we duplicate here.
    if (!file.is_nonempty())
        return String();
    return m_url.componentString(file);
}

String KURL::protocol() const
{
    return m_url.componentString(m_url.m_parsed.scheme);
}

String KURL::host() const
{
    // Note: KURL.cpp unescapes here.
    return m_url.componentString(m_url.m_parsed.host);
}

// Returns 0 when there is no port or it is invalid.
//
// We treat URL's with out-of-range port numbers as invalid URLs, and they will
// be rejected by the canonicalizer. KURL.cpp will allow them in parsing, but
// return 0 from this port() function, so we mirror that behavior here.
unsigned short KURL::port() const
{
    if (!m_url.m_isValid || m_url.m_parsed.port.len <= 0)
        return 0;
    int port = url_parse::ParsePort(m_url.utf8String().data(), m_url.m_parsed.port);
    if (port == url_parse::PORT_UNSPECIFIED)
        return 0;
    return static_cast<unsigned short>(port);
}

// Returns the empty string if there is no password.
String KURL::pass() const
{
    // Bug: https://bugs.webkit.org/show_bug.cgi?id=21015 this function returns
    // a null string when the password is empty, which we duplicate here.
    if (!m_url.m_parsed.password.is_nonempty())
        return String();

    // Note: KURL.cpp unescapes here.
    return m_url.componentString(m_url.m_parsed.password);
}

// Returns the empty string if there is no username.
String KURL::user() const
{
    // Note: KURL.cpp unescapes here.
    return m_url.componentString(m_url.m_parsed.username);
}

String KURL::ref() const
{
    // Empty but present refs ("foo.com/bar#") should result in the empty
    // string, which m_url.componentString will produce. Nonexistant refs should be
    // the NULL string.
    if (!m_url.m_parsed.ref.is_valid())
        return String();

    // Note: KURL.cpp unescapes here.
    return m_url.componentString(m_url.m_parsed.ref);
}

bool KURL::hasRef() const
{
    // Note: KURL.cpp unescapes here.
    // FIXME determine if KURL.cpp agrees about an empty ref
    return m_url.m_parsed.ref.len >= 0;
}

String KURL::query() const
{
    if (m_url.m_parsed.query.len >= 0) {
        // KURL's query() includes the question mark, even though the reference
        // doesn't. Move the query component backwards one to account for it
        // (our library doesn't count the question mark).
        url_parse::Component queryComp = m_url.m_parsed.query;
        queryComp.begin--;
        queryComp.len++;
        return m_url.componentString(queryComp);
    }

    // Bug: https://bugs.webkit.org/show_bug.cgi?id=21015 this function returns
    // an empty string when the query is empty rather than a null (not sure
    // which is right).
    return String("", 0);
}

String KURL::path() const
{
    // Note: KURL.cpp unescapes here.
    return m_url.componentString(m_url.m_parsed.path);
}

void KURL::setProtocol(const String& protocol)
{
    KURLGooglePrivate::Replacements replacements;
    replacements.SetScheme(CharactersOrEmpty(protocol),
                           url_parse::Component(0, protocol.length()));
    m_url.replaceComponents(replacements);
}

void KURL::setHost(const String& host)
{
    KURLGooglePrivate::Replacements replacements;
    replacements.SetHost(CharactersOrEmpty(host),
                         url_parse::Component(0, host.length()));
    m_url.replaceComponents(replacements);
}

// This function is used only in the JSC build.
void KURL::setHostAndPort(const String& s)
{
    String newhost = s.left(s.find(":"));
    String newport = s.substring(s.find(":") + 1);

    KURLGooglePrivate::Replacements replacements;
    // Host can't be removed, so we always set.
    replacements.SetHost(CharactersOrEmpty(newhost),
                         url_parse::Component(0, newhost.length()));

    if (newport.isEmpty())  // Port may be removed, so we support clearing.
        replacements.ClearPort();
    else
        replacements.SetPort(CharactersOrEmpty(newport), url_parse::Component(0, newport.length()));
    m_url.replaceComponents(replacements);
}

void KURL::setPort(unsigned short i)
{
    KURLGooglePrivate::Replacements replacements;
    String portStr;
    if (i) {
        portStr = String::number(static_cast<int>(i));
        replacements.SetPort(
            reinterpret_cast<const url_parse::UTF16Char*>(portStr.characters()),
            url_parse::Component(0, portStr.length()));

    } else {
        // Clear any existing port when it is set to 0.
        replacements.ClearPort();
    }
    m_url.replaceComponents(replacements);
}

void KURL::setUser(const String& user)
{
    // This function is commonly called to clear the username, which we
    // normally don't have, so we optimize this case.
    if (user.isEmpty() && !m_url.m_parsed.username.is_valid())
        return;

    // The canonicalizer will clear any usernames that are empty, so we
    // don't have to explicitly call ClearUsername() here.
    KURLGooglePrivate::Replacements replacements;
    replacements.SetUsername(CharactersOrEmpty(user),
                             url_parse::Component(0, user.length()));
    m_url.replaceComponents(replacements);
}

void KURL::setPass(const String& pass)
{
    // This function is commonly called to clear the password, which we
    // normally don't have, so we optimize this case.
    if (pass.isEmpty() && !m_url.m_parsed.password.is_valid())
        return;

    // The canonicalizer will clear any passwords that are empty, so we
    // don't have to explicitly call ClearUsername() here.
    KURLGooglePrivate::Replacements replacements;
    replacements.SetPassword(CharactersOrEmpty(pass),
                             url_parse::Component(0, pass.length()));
    m_url.replaceComponents(replacements);
}

void KURL::setRef(const String& ref)
{
    // This function is commonly called to clear the ref, which we
    // normally don't have, so we optimize this case.
    if (ref.isNull() && !m_url.m_parsed.ref.is_valid())
        return;

    KURLGooglePrivate::Replacements replacements;