xmlurl.cpp

来自「IBM的解析xml的工具Xerces的源代码」· C++ 代码 · 共 1,467 行 · 第 1/3 页

	//  parse the base URL string and conglomerate them.
	//
	if (isRelative() && baseURL)
	{
		if (*baseURL)
		{
			XMLURL basePart(baseURL, fMemoryManager);
			if (!conglomerateWithBase(basePart, false))
			{
				cleanup();
				ThrowXMLwithMemMgr(MalformedURLException, XMLExcepts::URL_RelativeBaseURL, fMemoryManager);
			}
		}
	}
}

// this version of setURL doesn't throw a malformedurl exception
// instead it returns false when it failed (or when it would of
// thrown a malformedurl exception)
bool XMLURL::setURL(const XMLCh* const    baseURL
                  , const XMLCh* const    relativeURL
                  , XMLURL& xmlURL)
{
    cleanup();

    // Parse our URL string
    if (parse(relativeURL, xmlURL))
    {
	    //  If its relative and the base is non-null and non-empty, then
	    //  parse the base URL string and conglomerate them.
	    //
	    if (isRelative() && baseURL && *baseURL)
	    {		   
	        XMLURL basePart(fMemoryManager);
            if (parse(baseURL, basePart)  && conglomerateWithBase(basePart, false))
            {    
		        return true;			                    
		    }
	    }
        else
            return true;
    }
    return false;
}

void XMLURL::setURL(const XMLURL&         baseURL
                  , const XMLCh* const    relativeURL)
{
    cleanup();

	// Parse our URL string
    parse(relativeURL);

    // If its relative, then conglomerate with the base URL
    if (isRelative())
		conglomerateWithBase(baseURL);
}


// ---------------------------------------------------------------------------
//  XMLURL: Miscellaneous methods
// ---------------------------------------------------------------------------
bool XMLURL::isRelative() const
{
    // If no protocol then relative
    if (fProtocol == Unknown)
        return true;

    // If no path, or the path is not absolute, then relative
    if (!fPath)
        return true;

    if (*fPath != chForwardSlash)
        return true;

    return false;
}


bool XMLURL::hasInvalidChar() const {
    return fHasInvalidChar;
}


BinInputStream* XMLURL::makeNewStream() const
{
    //
    //  If its a local host, then we short circuit it and use our own file
    //  stream support. Otherwise, we just let it fall through and let the
    //  installed network access object provide a stream.
    //
    if (fProtocol == XMLURL::File)
    {
        if (!fHost || !XMLString::compareIString(fHost, XMLUni::fgLocalHostString))
        {

            XMLCh* realPath = XMLString::replicate(fPath, fMemoryManager);
            ArrayJanitor<XMLCh> basePathName(realPath, fMemoryManager);

            //
            // Need to manually replace any character reference %xx first
            // HTTP protocol will be done automatically by the netaccessor
            //
            int end = XMLString::stringLen(realPath);
            int percentIndex = XMLString::indexOf(realPath, chPercent, 0, fMemoryManager);

            while (percentIndex != -1) {

                if (percentIndex+2 >= end ||
                    !isHexDigit(realPath[percentIndex+1]) ||
                    !isHexDigit(realPath[percentIndex+2]))
                {
                    XMLCh value1[4];
                    XMLString::moveChars(value1, &(realPath[percentIndex]), 3);
                    value1[3] = chNull;
                    ThrowXMLwithMemMgr2(MalformedURLException
                            , XMLExcepts::XMLNUM_URI_Component_Invalid_EscapeSequence
                            , realPath
                            , value1
                            , fMemoryManager);
                }

                unsigned int value = (xlatHexDigit(realPath[percentIndex+1]) * 16) + xlatHexDigit(realPath[percentIndex+2]);

                realPath[percentIndex] = XMLCh(value);

                int i =0;
                for (i = percentIndex + 1; i < end - 2 ; i++)
                    realPath[i] = realPath[i+2];
                realPath[i] = chNull;
                end = i;

                percentIndex = XMLString::indexOf(realPath, chPercent, percentIndex, fMemoryManager);
            }


            BinFileInputStream* retStrm = new (fMemoryManager) BinFileInputStream(realPath);
            if (!retStrm->getIsOpen())
            {
                delete retStrm;
                return 0;
            }
            return retStrm;
        }
    }

    //
    //  If we don't have have an installed net accessor object, then we
    //  have to just throw here.
    //
    if (!XMLPlatformUtils::fgNetAccessor)
        ThrowXMLwithMemMgr(MalformedURLException, XMLExcepts::URL_UnsupportedProto, fMemoryManager);

    // Else ask the net accessor to create the stream
    return XMLPlatformUtils::fgNetAccessor->makeNew(*this);
}

void XMLURL::makeRelativeTo(const XMLCh* const baseURLText)
{
    // If this one is not relative, don't bother
    if (!isRelative())
        return;

    XMLURL baseURL(baseURLText, fMemoryManager);
    conglomerateWithBase(baseURL);
}

void XMLURL::makeRelativeTo(const XMLURL& baseURL)
{
    // If this one is not relative, don't bother
    if (!isRelative())
        return;
    conglomerateWithBase(baseURL);
}




// ---------------------------------------------------------------------------
//  XMLURL: Private helper methods
// ---------------------------------------------------------------------------

//
//  This method will take the broken out parts of the URL and build up the
//  full text. We don't do this unless someone asks us to, since its often
//  never required.
//
void XMLURL::buildFullText()
{
    // Calculate the worst case size of the buffer required
    unsigned int bufSize = gMaxProtoLen + 1
                           + XMLString::stringLen(fFragment) + 1
                           + XMLString::stringLen(fHost) + 2
                           + XMLString::stringLen(fPassword) + 1
                           + XMLString::stringLen(fPath)
                           + XMLString::stringLen(fQuery) + 1
                           + XMLString::stringLen(fUser) + 1
                           + 32;

    // Clean up the existing buffer and allocate another
    fMemoryManager->deallocate(fURLText);//delete [] fURLText;
    fURLText = (XMLCh*) fMemoryManager->allocate((bufSize) * sizeof(XMLCh));//new XMLCh[bufSize];
    *fURLText = 0;

    XMLCh* outPtr = fURLText;
    if (fProtocol != Unknown)
    {
        XMLString::catString(fURLText, getProtocolName());
        outPtr += XMLString::stringLen(fURLText);
        *outPtr++ = chColon;
        *outPtr++ = chForwardSlash;
        *outPtr++ = chForwardSlash;
    }

    if (fUser)
    {
        XMLString::copyString(outPtr, fUser);
        outPtr += XMLString::stringLen(fUser);

        if (fPassword)
        {
            *outPtr++ = chColon;
            XMLString::copyString(outPtr, fPassword);
            outPtr += XMLString::stringLen(fPassword);
        }

        *outPtr++ = chAt;
    }

    if (fHost)
    {
        XMLString::copyString(outPtr, fHost);
        outPtr += XMLString::stringLen(fHost);

        //
        //  If the port is zero, then we don't put it in. Else we need
        //  to because it was explicitly provided.
        //
        if (fPortNum)
        {
            *outPtr++ = chColon;

            XMLCh tmpBuf[17];
            XMLString::binToText(fPortNum, tmpBuf, 16, 10, fMemoryManager);
            XMLString::copyString(outPtr, tmpBuf);
            outPtr += XMLString::stringLen(tmpBuf);
        }
    }

    if (fPath)
    {
        XMLString::copyString(outPtr, fPath);
        outPtr += XMLString::stringLen(fPath);
    }

    if (fQuery)
    {
        *outPtr++ = chQuestion;
        XMLString::copyString(outPtr, fQuery);
        outPtr += XMLString::stringLen(fQuery);
    }

    if (fFragment)
    {
        *outPtr++ = chPound;
        XMLString::copyString(outPtr, fFragment);
        outPtr += XMLString::stringLen(fFragment);
    }

    // Cap it off in case the last op was not a string copy
    *outPtr = 0;
}


//
//  Just a central place to handle cleanup, since its done from a number
//  of different spots.
//
void XMLURL::cleanup()
{
    fMemoryManager->deallocate(fFragment);//delete [] fFragment;
    fMemoryManager->deallocate(fHost);//delete [] fHost;
    fMemoryManager->deallocate(fPassword);//delete [] fPassword;
    fMemoryManager->deallocate(fPath);//delete [] fPath;
    fMemoryManager->deallocate(fQuery);//delete [] fQuery;
    fMemoryManager->deallocate(fUser);//delete [] fUser;
    fMemoryManager->deallocate(fURLText);//delete [] fURLText;

    fFragment = 0;
    fHost = 0;
    fPassword = 0;
    fPath = 0;
    fQuery = 0;
    fUser = 0;
    fURLText = 0;

    fProtocol = Unknown;
    fPortNum = 0;
    fHasInvalidChar = false;
}


//This function  has been modified to take a bool parameter and the
//functionality inside looks irrational but is only to make
//solaris 2.7 CC 5.0 optimized build happy.

bool XMLURL::conglomerateWithBase(const XMLURL& baseURL, bool useExceptions)
{
    // The base URL cannot be relative
    if (baseURL.isRelative())
    {
        if (useExceptions)
			ThrowXMLwithMemMgr(MalformedURLException, XMLExcepts::URL_RelativeBaseURL, fMemoryManager);
        else
            return false;
    }

    //
    //  Check a special case. If all we have is a fragment, then we want
    //  to just take the base host and path, plus our fragment.
    //
    if ((fProtocol == Unknown)
    &&  !fHost
    &&  !fPath
    &&  fFragment)
    {
        // Just in case, make sure we don't leak the user or password values
        fMemoryManager->deallocate(fUser);//delete [] fUser;
        fUser = 0;
        fMemoryManager->deallocate(fPassword);//delete [] fPassword;
        fPassword = 0;

        // Copy over the protocol and port number as is
        fProtocol = baseURL.fProtocol;
        fPortNum = baseURL.fPortNum;

        // Replicate the base fields that are provided
        fHost = XMLString::replicate(baseURL.fHost, fMemoryManager);
        fUser = XMLString::replicate(baseURL.fUser, fMemoryManager);
        fPassword = XMLString::replicate(baseURL.fPassword, fMemoryManager);
        fPath = XMLString::replicate(baseURL.fPath, fMemoryManager);
        return true;
    }

    //
    //  All we have to do is run up through our fields and, for each one
    //  that we don't have, use the based URL's. Once we hit one field
    //  that we have, we stop.
    //
    if (fProtocol != Unknown)
        return true;
    fProtocol = baseURL.fProtocol;

    //
    //  If the protocol is not file, and we either already have our own
    //  host, or the base does not have one, then we are done.
    //
    if (fProtocol != File)
    {
        if (fHost || !baseURL.fHost)
            return true;
    }

    // Replicate all of the hosty stuff if the base has one
    if (baseURL.fHost)
    {
        // Just in case, make sure we don't leak a user or password field
        fMemoryManager->deallocate(fUser);//delete [] fUser;
        fUser = 0;
        fMemoryManager->deallocate(fPassword);//delete [] fPassword;
        fPassword = 0;
        fMemoryManager->deallocate(fHost);//delete [] fHost;
        fHost = 0;

        fHost = XMLString::replicate(baseURL.fHost, fMemoryManager);
        fUser = XMLString::replicate(baseURL.fUser, fMemoryManager);
        fPassword = XMLString::replicate(baseURL.fPassword, fMemoryManager);

        fPortNum = baseURL.fPortNum;
    }

    // If we have a path and its absolute, then we are done
    const bool hadPath = (fPath != 0);
    if (hadPath)
    {
        if (*fPath == chForwardSlash)
            return true;
    }

    // Its a relative path, so weave them together.
    if (baseURL.fPath) {
        XMLCh* temp = XMLPlatformUtils::weavePaths(baseURL.fPath, fPath ,fMemoryManager);
        fMemoryManager->deallocate(fPath);//delete [] fPath;
        fPath = temp;
    }

    // If we had any original path, then we are done
    if (hadPath)
        return true;

    // We had no original path, so go on to deal with the query/fragment parts
    if (fQuery || !baseURL.fQuery)
        return true;
    fQuery = XMLString::replicate(baseURL.fQuery, fMemoryManager);

    if (fFragment || !baseURL.fFragment)
        return true;
    fFragment = XMLString::replicate(baseURL.fFragment, fMemoryManager);
	return true;
}


void XMLURL::parse(const XMLCh* const urlText)
{
    // Simplify things by checking for the psycho scenarios first
    if (!*urlText)
        ThrowXMLwithMemMgr(MalformedURLException, XMLExcepts::URL_NoProtocolPresent, fMemoryManager);

    // Before we start, check if this urlText contains valid uri characters
    if (!XMLUri::isURIString(urlText))
        fHasInvalidChar = true;
    else
        fHasInvalidChar = false;

    //
    //  The first thing we will do is to check for a file name, so that
    //  we don't waste time thinking its a URL. If its in the form x:\
    //  or x:/ and x is an ASCII letter, then assume that's the deal.
    //
    if (((*urlText >= chLatin_A) && (*urlText <= chLatin_Z))
    ||  ((*urlText >= chLatin_a) && (*urlText <= chLatin_z)))
    {
        if (*(urlText + 1) == chColon)
        {
            if ((*(urlText + 2) == chForwardSlash)
            ||  (*(urlText + 2) == chBackSlash))
            {
                ThrowXMLwithMemMgr(MalformedURLException, XMLExcepts::URL_NoProtocolPresent, fMemoryManager);
            }
        }
    }

    // Get a copy of the URL that we can modify
    XMLCh* srcCpy = XMLString::replicate(urlText, fMemoryManager);
    ArrayJanitor<XMLCh> janSrcCopy(srcCpy, fMemoryManager);

    //
    //  Get a pointer now that we can run up thrown the source as we parse
    //  bits and pieces out of it.
    //
    XMLCh* srcPtr = srcCpy;

    // Run up past any spaces
    while (*srcPtr)
    {
        if (!XMLPlatformUtils::fgTransService->isSpace(*srcPtr))
            break;
        srcPtr++;
    }

    // Make sure it wasn't all space
    if (!*srcPtr)
        ThrowXMLwithMemMgr(MalformedURLException, XMLExcepts::URL_NoProtocolPresent, fMemoryManager);

    //
    //  Ok, the next thing we have to do is to find either a / or : character.
    //  If the : is first, we assume we have a protocol. If the / is first,
    //  then we skip to the host processing.
    //
    XMLCh* ptr1 = XMLString::findAny(srcPtr, gListOne);
    XMLCh* ptr2;

    // If we found a protocol, then deal with it
    if (ptr1)
    {
        if (*ptr1 == chColon)
        {
            // Cap the string at the colon
            *ptr1 = 0;

            // And try to find it in our list of protocols
            fProtocol = lookupByName(srcPtr);

            if (fProtocol == Unknown)
            {
                ThrowXMLwithMemMgr1
                (
                    MalformedURLException
                    , XMLExcepts::URL_UnsupportedProto1