xmlstring.cpp

来自「IBM的解析xml的工具Xerces的源代码」· C++ 代码 · 共 1,907 行 · 第 1/4 页

{
    // Refer this one to the transcoding service
    XMLPlatformUtils::fgTransService->upperCase(toUpperCase);
}


void XMLString::lowerCase(XMLCh* const toLowerCase)
{
    // Refer this one to the transcoding service
    XMLPlatformUtils::fgTransService->lowerCase(toLowerCase);
}


void XMLString::subString(XMLCh* const targetStr, const XMLCh* const srcStr
                          , const int startIndex, const int endIndex
                          , MemoryManager* const manager)
{
    //if (startIndex < 0 || endIndex < 0)
    //    ThrowXMLwithMemMgr(ArrayIndexOutOfBoundsException, XMLExcepts::Str_NegativeIndex);

    if (targetStr == 0)
        ThrowXMLwithMemMgr(IllegalArgumentException, XMLExcepts::Str_ZeroSizedTargetBuf, manager);

    const int srcLen = stringLen(srcStr);
    const int copySize = endIndex - startIndex;

    // Make sure the start index is within the XMLString bounds
    if ( startIndex < 0 || startIndex > endIndex || endIndex > srcLen)
        ThrowXMLwithMemMgr(ArrayIndexOutOfBoundsException, XMLExcepts::Str_StartIndexPastEnd, manager);

    for (int i= startIndex; i < endIndex; i++) {
        targetStr[i-startIndex] = srcStr[i];
    }

    targetStr[copySize] = 0;
}

BaseRefVectorOf<XMLCh>* XMLString::tokenizeString(const XMLCh*      const   tokenizeSrc
                                               ,  MemoryManager*    const   manager)
{
    XMLCh* orgText = replicate(tokenizeSrc, manager);
    ArrayJanitor<XMLCh> janText(orgText, manager);
    XMLCh* tokenizeStr = orgText;

    RefArrayVectorOf<XMLCh>* tokenStack = new (manager) RefArrayVectorOf<XMLCh>(16, true, manager);

    unsigned int len = stringLen(tokenizeStr);
    unsigned int skip;
    unsigned int index = 0;

    while (index != len) {
        // find the first non-space character
        for (skip = index; skip < len; skip++)
        {
            if (!XMLPlatformUtils::fgTransService->isSpace(tokenizeStr[skip]))
                break;
        }
        index = skip;

        // find the delimiter (space character)
        for (; skip < len; skip++)
        {
            if (XMLPlatformUtils::fgTransService->isSpace(tokenizeStr[skip]))
                break;
        }

        // we reached the end of the string
        if (skip == index)
            break;

        // these tokens are adopted in the RefVector and will be deleted
        // when the vector is deleted by the caller
        XMLCh* token = (XMLCh*) manager->allocate
        (
            (skip+1-index) * sizeof(XMLCh)
        );//new XMLCh[skip+1-index];

        XMLString::subString(token, tokenizeStr, index, skip, manager);
        tokenStack->addElement(token);
        index = skip;
    }
    return tokenStack;
}

//
//  This method is called when we get a notation or enumeration type attribute
//  to validate. We have to confirm that the passed value to find is one of
//  the values in the passed list. The list is a space separated string of
//  values to match against.
//
bool XMLString::isInList(const XMLCh* const toFind, const XMLCh* const enumList)
{
    //
    //  We loop through the values in the list via this outer loop. We end
    //  when we hit the end of the enum list or get a match.
    //
    const XMLCh* listPtr = enumList;
    const unsigned int findLen = XMLString::stringLen(toFind);
    while (*listPtr)
    {
        unsigned int testInd;
        for (testInd = 0; testInd < findLen; testInd++)
        {
            //
            //  If they don't match, then reset and try again. Note that
            //  hitting the end of the current item will cause a mismatch
            //  because there can be no spaces in the toFind string.
            //
            if (listPtr[testInd] != toFind[testInd])
                break;
        }

        //
        //  If we went the distance, see if we matched. If we did, the current
        //  list character has to be null or space.
        //
        if (testInd == findLen)
        {
            if ((listPtr[testInd] == chSpace) || !listPtr[testInd])
                return true;
        }

        // Run the list pointer up to the next substring
        while ((*listPtr != chSpace) && *listPtr)
            listPtr++;

        // If we hit the end, then we failed
        if (!*listPtr)
            return false;

        // Else move past the space and try again
        listPtr++;
    }

    // We never found it
    return false;
}

//
// a string is whitespace:replaced, is having no
//    #xD  Carriage Return
//    #xA  Line Feed
//    #x9  TAB
//
bool XMLString::isWSReplaced(const XMLCh* const toCheck)
{
    // If no string, then its a OK
    if (( !toCheck ) || ( !*toCheck ))
        return true;

    const XMLCh* startPtr = toCheck;
    while ( *startPtr )
    {
        if ( ( *startPtr == chCR) ||
             ( *startPtr == chLF) ||
             ( *startPtr == chHTab))
        return false;

        startPtr++;
    }

    return true;
}

//
//   to replace characters listed below to #x20
//    #xD  Carriage Return
//    #xA  Line Feed
//    #x9  TAB
//
void XMLString::replaceWS(XMLCh* const toConvert
                          , MemoryManager* const  manager)
{
    int strLen = XMLString::stringLen(toConvert);
    if (strLen == 0)
        return;

    XMLCh* retBuf = (XMLCh*) manager->allocate
    (
        (strLen+1) * sizeof(XMLCh)
    );//new XMLCh[strLen+1];
    XMLCh* retPtr = &(retBuf[0]);
    XMLCh* startPtr = toConvert;

    while ( *startPtr )
    {
        if ( ( *startPtr == chCR) ||
             ( *startPtr == chLF) ||
             ( *startPtr == chHTab))
            *retPtr = chSpace;
        else
            *retPtr = *startPtr;

        retPtr++;
        startPtr++;
    }

    retBuf[strLen] = chNull;

    XMLString::moveChars(toConvert, retBuf, strLen);
    manager->deallocate(retBuf);//delete[] retBuf;
    return;
}

//
// a string is whitespace:collapsed, is whitespace::replaced
// and no
//    leading space (#x20)
//    trailing space
//    no contiguous sequences of spaces
//
bool XMLString::isWSCollapsed(const XMLCh* const toCheck)
{
    if (( !toCheck ) || ( !*toCheck ))
        return true;

    // shall be whitespace::replaced first
    if ( !isWSReplaced(toCheck) )
        return false;

    // no leading or trailing space
    if ((*toCheck == chSpace) ||
        (toCheck[XMLString::stringLen(toCheck)-1] == chSpace))
        return false;

    const XMLCh* startPtr = toCheck;
    XMLCh theChar;
    bool  inSpace = false;
    while ( (theChar = *startPtr) != 0 )
    {
        if ( theChar == chSpace)
        {
            if (inSpace)
                return false;
            else
                inSpace = true;
        }
        else
            inSpace = false;

        startPtr++;

    }

    return true;
}

//
// no leading and/or trailing spaces
// no continuous sequences of spaces
//
void XMLString::collapseWS(XMLCh* const toConvert
                           , MemoryManager* const  manager)
{
    // If no string, then its a failure
    if (( !toConvert ) || ( !*toConvert ))
        return;

    // replace whitespace first
    replaceWS(toConvert, manager);

    // remove leading spaces
    const XMLCh* startPtr = toConvert;
    while ( *startPtr == chSpace )
        startPtr++;

    if (!*startPtr)
        return;

    // remove trailing spaces
    const XMLCh* endPtr = toConvert + stringLen(toConvert);
    while (*(endPtr - 1) == chSpace)
        endPtr--;

    //
    //  Work through what remains and chop continuous spaces
    //
    XMLCh* retBuf = (XMLCh*) manager->allocate
    (
        (endPtr - startPtr + 1) * sizeof(XMLCh)
    );//new XMLCh[endPtr - startPtr + 1];
    XMLCh* retPtr = &(retBuf[0]);
    bool  inSpace = false;
    while (startPtr < endPtr)
    {
        if ( *startPtr == chSpace)
        {
            if (inSpace)
            {
                //discard it;
            }
            else
            {
                inSpace = true;
                *retPtr = chSpace;  //copy the first chSpace
                retPtr++;
            }
        }
        else
        {
            inSpace = false;
            *retPtr = *startPtr;
            retPtr++;
        }

        startPtr++;
    }

    *retPtr = chNull;
    XMLString::moveChars(toConvert, retBuf, stringLen(retBuf)+1); //copy the last chNull as well
    manager->deallocate(retBuf);//delete[] retBuf;
    return;
}

//
// remove whitespace
//
void XMLString::removeWS(XMLCh* const toConvert
                         , MemoryManager* const manager)
{
    // If no string, then its a failure
    if (( !toConvert ) || ( !*toConvert ))
        return;

    XMLCh* retBuf = (XMLCh*) manager->allocate
    (
        (XMLString::stringLen(toConvert) + 1) * sizeof(XMLCh)
    );//new XMLCh[ XMLString::stringLen(toConvert) + 1];
    XMLCh* retPtr = &(retBuf[0]);
    XMLCh* startPtr = toConvert;

    while (*startPtr)
    {
        if ( ( *startPtr != chCR)    &&
             ( *startPtr != chLF)    &&
             ( *startPtr != chHTab)  &&
             ( *startPtr != chSpace)  )
        {
            *retPtr++ = *startPtr;
        }

        startPtr++;

    }

    *retPtr = chNull;
    XMLString::moveChars(toConvert, retBuf, stringLen(retBuf)+1); //copy the last chNull as well
    manager->deallocate(retBuf);//delete[] retBuf;
    return;
}

void XMLString::removeChar(const XMLCh*     const srcString
                         , const XMLCh&           toRemove
                         ,       XMLBuffer&       dstBuffer)
{
    const XMLCh* pszSrc = srcString;
    XMLCh c;

    dstBuffer.reset();

    while (c=*pszSrc++) 
    {
        if (c != toRemove) 
            dstBuffer.append(c);

    }
}

/**
 * Fixes a platform dependent absolute path filename to standard URI form.
 * 1. Windows: fix 'x:' to 'file:///x:' and convert any backslash to forward slash
 * 2. UNIX: fix '/blah/blahblah' to 'file:///blah/blahblah'
 */
void XMLString::fixURI(const XMLCh* const str, XMLCh* const target)
{
    if (!str || !*str)
        return;

    int colonIdx = XMLString::indexOf(str, chColon);

    // If starts with a '/' we assume
    // this is an absolute (UNIX) file path and prefix it with file://
    if (colonIdx == -1 && XMLString::indexOf(str, chForwardSlash) == 0) {
        unsigned index = 0;
        target[index++] = chLatin_f;
        target[index++] = chLatin_i;
        target[index++] = chLatin_l;
        target[index++] = chLatin_e;
        target[index++] = chColon;
        target[index++] = chForwardSlash;
        target[index++] = chForwardSlash;

        // copy the string
        const XMLCh* inPtr = str;
        while (*inPtr)
            target[index++] = *inPtr++;

        target[index] = chNull;
    }
    else if (colonIdx == 1 && XMLString::isAlpha(*str)) {
        // If starts with a driver letter 'x:' we assume
        // this is an absolute (Windows) file path and prefix it with file:///
        unsigned index = 0;
        target[index++] = chLatin_f;
        target[index++] = chLatin_i;
        target[index++] = chLatin_l;
        target[index++] = chLatin_e;
        target[index++] = chColon;
        target[index++] = chForwardSlash;
        target[index++] = chForwardSlash;
        target[index++] = chForwardSlash;

        // copy the string and fix any backward slash
        const XMLCh* inPtr = str;
        while (*inPtr) {
            if (*inPtr == chYenSign ||
                *inPtr == chWonSign ||
                *inPtr == chBackSlash)
                target[index++] = chForwardSlash;
            else
                target[index++] = *inPtr;
            inPtr++;
        }

        // cap it with null
        target[index] = chNull;
    }
    else {
        // not specific case, so just copy the string over
        copyString(target, str);
    }
}

void XMLString::release(char** buf)
{
    delete [] *buf;
    *buf = 0;
}

void XMLString::release(XMLCh** buf)
{
    delete [] *buf;
    *buf = 0;
}

void XMLString::release(XMLByte** buf)
{
    delete [] *buf;
    *buf = 0;
}

// ---------------------------------------------------------------------------
//  XMLString: Private static methods
// ---------------------------------------------------------------------------
void XMLString::initString(XMLLCPTranscoder* const defToUse,
                           MemoryManager* const manager)
{
    // Store away the default transcoder that we are to use
    gTranscoder = defToUse;

    // Store memory manager
    fgMemoryManager = manager;
}

void XMLString::termString()
{
    // Just clean up our local code page transcoder
    delete gTranscoder;
    gTranscoder = 0;

    // reset memory manager
    fgMemoryManager = 0;
}

XERCES_CPP_NAMESPACE_END