rangetoken.cpp

来自「IBM的解析xml的工具Xerces的源代码」· C++ 代码 · 共 674 行 · 第 1/2 页

        else if (rangeTok->fRanges[j] < fRanges[i]
                 || (rangeTok->fRanges[j] == fRanges[i]
                     && rangeTok->fRanges[j+1] < fRanges[i+1])) {

            for (int count = 0; count < 2; count++) {
                result[k++] = rangeTok->fRanges[j++];
            }
        }
        else {

            for (int count = 0; count < 2; count++) {

                result[k++] = fRanges[i++];
            }
        }
    }

    fMemoryManager->deallocate(fRanges);//delete [] fRanges;
    fElemCount += rangeTok->fElemCount;
    fRanges = result;
    fMaxCount = newMaxCount;
}

void RangeToken::subtractRanges(RangeToken* const tok) {

    if (fRanges == 0 || tok->fRanges == 0)
        return;

    if (tok->getTokenType() == T_NRANGE) {

        intersectRanges(tok);
        return;
    }

    fCaseIToken = 0;
    sortRanges();
    compactRanges();
    tok->sortRanges();
    tok->compactRanges();

    unsigned int newMax = (fElemCount + tok->fElemCount >= fMaxCount)
                             ? fMaxCount + tok->fMaxCount : fMaxCount;
    XMLInt32* result = (XMLInt32*) fMemoryManager->allocate
    (
        newMax * sizeof(XMLInt32)
    );//new XMLInt32[newMax];
    unsigned int newElemCount = 0;
    unsigned int srcCount = 0;
    unsigned int subCount = 0;

    while (srcCount < fElemCount && subCount < tok->fElemCount) {

        XMLInt32 srcBegin = fRanges[srcCount];
        XMLInt32 srcEnd = fRanges[srcCount + 1];
        XMLInt32 subBegin = tok->fRanges[subCount];
        XMLInt32 subEnd = tok->fRanges[subCount + 1];

        if (srcEnd < subBegin) { // no overlap

            result[newElemCount++] = fRanges[srcCount++];
            result[newElemCount++] = fRanges[srcCount++];
        }
        else if (srcEnd >= subBegin && srcBegin <= subEnd) {

            if (subBegin <= srcBegin && srcEnd <= subEnd) {
                srcCount += 2;
            }
            else if (subBegin <= srcBegin) {

                fRanges[srcCount] = subEnd + 1;
                subCount += 2;
            }
            else if (srcEnd <= subEnd) {

                result[newElemCount++] = srcBegin;
                result[newElemCount++] = subBegin - 1;
                srcCount += 2;
            }
            else {

                result[newElemCount++] = srcBegin;
                result[newElemCount++] = subBegin - 1;
                fRanges[srcCount] = subEnd + 1;
                subCount += 2;
            }
        }
        else if (subEnd < srcBegin) {
            subCount += 2;
        }
        else {
            fMemoryManager->deallocate(result);//delete [] result;
            ThrowXMLwithMemMgr(RuntimeException, XMLExcepts::Regex_SubtractRangesError, fMemoryManager);
        }
    } //end while

    while (srcCount < fElemCount) {

        result[newElemCount++] = fRanges[srcCount++];
        result[newElemCount++] = fRanges[srcCount++];
    }

    fMemoryManager->deallocate(fRanges);//delete [] fRanges;
    fRanges = result;
    fElemCount = newElemCount;
    fMaxCount = newMax;
}

/**
  * Ignore whether 'tok' is NRANGE or not.
  */
void RangeToken::intersectRanges(RangeToken* const tok) {

    if (fRanges == 0 || tok->fRanges == 0)
        return;

    fCaseIToken = 0;
    sortRanges();
    compactRanges();
    tok->sortRanges();
    tok->compactRanges();

    unsigned int newMax = (fElemCount + tok->fElemCount >= fMaxCount)
                             ? fMaxCount + tok->fMaxCount : fMaxCount;
    XMLInt32* result = (XMLInt32*) fMemoryManager->allocate
    (
        newMax * sizeof(XMLInt32)
    );//new XMLInt32[newMax];
    unsigned int newElemCount = 0;
    unsigned int srcCount = 0;
    unsigned int tokCount = 0;

    while (srcCount < fElemCount && tokCount < tok->fElemCount) {

        XMLInt32 srcBegin = fRanges[srcCount];
        XMLInt32 srcEnd = fRanges[srcCount + 1];
        XMLInt32 tokBegin = tok->fRanges[tokCount];
        XMLInt32 tokEnd = tok->fRanges[tokCount + 1];

        if (srcEnd < tokBegin) {
            srcCount += 2;
        }
        else if (srcEnd >= tokBegin && srcBegin <= tokEnd) {

            if (tokBegin <= srcBegin && srcEnd <= tokEnd) {

                result[newElemCount++] = srcBegin;
                result[newElemCount++] = srcEnd;
                srcCount += 2;
            }
            else if (tokBegin <= srcBegin) {

                result[newElemCount++] = srcBegin;
                result[newElemCount++] = tokEnd;
                tokCount += 2;

                if (tokCount < tok->fElemCount)
                    fRanges[srcCount] = tokEnd + 1;
                else
                    srcCount += 2;
            }
            else if (srcEnd <= tokEnd) {

                result[newElemCount++] = tokBegin;
                result[newElemCount++] = srcEnd;
                srcCount += 2;
            }
            else {

                result[newElemCount++] = tokBegin;
                result[newElemCount++] = tokEnd;
                tokCount += 2;

                if (tokCount < tok->fElemCount)
                    fRanges[srcCount] = tokEnd + 1;
                else
                    srcCount += 2;
            }
        }
        else if (tokEnd < srcBegin) {
            tokCount += 2;

            if (tokCount >= tok->fElemCount)
                srcCount += 2;
        }
        else {

            fMemoryManager->deallocate(result);//delete [] result;
            ThrowXMLwithMemMgr(RuntimeException, XMLExcepts::Regex_IntersectRangesError, fMemoryManager);
        }
    } //end while

    fMemoryManager->deallocate(fRanges);//delete [] fRanges;
    fRanges = result;
    fElemCount = newElemCount;
    fMaxCount = newMax;
}

/**
  * for RANGE: Creates complement.
  * for NRANGE: Creates the same meaning RANGE.
  */
Token* RangeToken::complementRanges(RangeToken* const tok,
                                    TokenFactory* const tokFactory,
                                    MemoryManager* const manager) {

    if (tok->getTokenType() != T_RANGE && tok->getTokenType() != T_NRANGE)
        ThrowXMLwithMemMgr(IllegalArgumentException, XMLExcepts::Regex_ComplementRangesInvalidArg, manager);

    tok->sortRanges();
    tok->compactRanges();

    XMLInt32 lastElem = tok->fRanges[tok->fElemCount - 1];
    RangeToken* rangeTok = tokFactory->createRange();

    if (tok->fRanges[0] > 0) {
        rangeTok->addRange(0, tok->fRanges[0] - 1);
    }

    for (unsigned int i= 1; i< tok->fElemCount - 2; i += 2) {
        rangeTok->addRange(tok->fRanges[i] + 1, tok->fRanges[i+1] - 1);
    }

    if (lastElem != UTF16_MAX) {
        rangeTok->addRange(lastElem + 1, UTF16_MAX);
    }

    rangeTok->fCompacted = true;

    return rangeTok;
}


// ---------------------------------------------------------------------------
//  RangeToken: Match methods
// ---------------------------------------------------------------------------
bool RangeToken::match(const XMLInt32 ch) {

    if (fMap == 0)
        createMap();

    bool ret;

    if (getTokenType() == T_RANGE) {

        if (ch < MAPSIZE)
            return ((fMap[ch/32] & (1<<(ch&0x1f))) != 0);

        ret = false;

        for (unsigned int i= fNonMapIndex; i< fElemCount; i +=2) {

            if (fRanges[i] <= ch && ch <= fRanges[i+1])
                return true;
        }
    }
    else {

        if (ch < MAPSIZE)
            return ((fMap[ch/32] & (1<<(ch&0x1f))) == 0);

        ret = true;

        for (unsigned int i= fNonMapIndex; i< fElemCount; i += 2) {

            if (fRanges[i] <= ch && ch <= fRanges[i+1])
                return false;
        }
    }

    return ret;
}

// ---------------------------------------------------------------------------
//  RangeToken: Private helpers methods
// ---------------------------------------------------------------------------
void RangeToken::expand(const unsigned int length) {

    unsigned int newMax = fElemCount + length;

    // Avoid too many reallocations by expanding by a percentage
    unsigned int minNewMax = (unsigned int)((double)fElemCount * 1.25);
    if (newMax < minNewMax)
        newMax = minNewMax;

    XMLInt32* newList = (XMLInt32*) fMemoryManager->allocate
    (
        newMax * sizeof(XMLInt32)
    );//new XMLInt32[newMax];
    for (unsigned int index = 0; index < fElemCount; index++)
        newList[index] = fRanges[index];

    fMemoryManager->deallocate(fRanges);//delete [] fRanges;
    fRanges = newList;
    fMaxCount = newMax;
}

void RangeToken::createMap() {

    int asize = MAPSIZE/32;
    fMap = (int*) fMemoryManager->allocate(asize * sizeof(int));//new int[asize];
    fNonMapIndex = fElemCount;

    for (int i = 0; i < asize; i++) {
        fMap[i] = 0;
    }

    for (unsigned int j= 0; j < fElemCount; j += 2) {

        XMLInt32 begin = fRanges[j];
        XMLInt32 end = fRanges[j+1];

        if (begin < MAPSIZE) {

            for (int k = begin; k <= end && k < MAPSIZE; k++) {
                fMap[k/32] |= 1<<(k&0x1F);
            }
        }
        else {

            fNonMapIndex = j;
            break;
        }

        if (end >= MAPSIZE) {

            fNonMapIndex = j;
            break;
        }
    }
}

XERCES_CPP_NAMESPACE_END

/**
  * End of file RangeToken.cpp
  */