attrproj.c

数据挖掘经典的hierarchial clustering algorithm

/*------------------------------------------------------------------------------
 * function: AttrProj::CreateRecordList
 * Creates a record list (VectorArray) that can hold the projected
 * records specified in the attribute projection file.
 */
DevStatus
AttrProj::CreateRecordList(VectorArray *&vecArrayP)
{
        DO_DEBUG(printf("AttrProj::CreateRecordList()\n"));

        DevStatus       result = StatusOk;

        vecArrayP = new VectorArray(_projList.GetProjCount());

        int             projNum = 0;
        Projection * projP = _projList.GetFirstProj();
        while (projP != NULL)
        {
                vecArrayP->Init(projNum, projP->attrCount);
                projP = _projList.GetNextProj();
                projNum++;
        }

        return result;
}

/*------------------------------------------------------------------------------
 * function: AttrProj::ReadRec
 * Read a record from the data source, and project it onto the attribute
 * combinations corresponding to this object.
 */
DevStatus
AttrProj::ReadRec(RecId recId, VectorArray &vecArray)
{
        DO_DEBUG(printf("AttrProj::ReadRec(%d)\n", (int) recId));

        int                     dataSize;
        int                     numRecs;
        DevStatus       result = StatusOk;

        _tDataP->InitGetRecs(recId, recId);

        if (!_tDataP->GetRecs(_recBuf, _recBufSize, recId, numRecs, dataSize))
        {
                result = StatusFailed;
        }
        else
        {
                AttrList *      attrListP = _tDataP->GetAttrList();
                int             projNum = 0;
                Projection * projP = _projList.GetFirstProj();
                while (projP != NULL)
                {
                        Vector *        vectorP = vecArray.GetVector(projNum);
                        int             projAttrNum;
                        for (projAttrNum = 0; projAttrNum < projP->attrCount;
                                projAttrNum++)
                        {
                                int                     attrNum = projP->attrList[projAttrNum];
                                AttrInfo *      attrInfoP = attrListP->Get(attrNum);

                                vectorP->value[projAttrNum] = AttrToDouble(attrInfoP->type,
                                        _recBuf + attrInfoP->offset);
                        }

                        projP = _projList.GetNextProj();
                        projNum++;
                }
        }

        return result;
}

/*------------------------------------------------------------------------------
 * function: AttrProj::ReadWholeRec
 * Read an entire record (not its projections).
 */
DevStatus
AttrProj::ReadWholeRec(RecId recId, Vector &vector)
{
        DO_DEBUG(printf("AttrProj::ReadWholeRec()\n"));

        DevStatus       result = StatusOk;

        _tDataP->InitGetRecs(recId, recId);

        int                     dataSize;
        int                     numRecs;

        if (!_tDataP->GetRecs(_recBuf, _recBufSize, recId, numRecs, dataSize))
        {
                result = StatusFailed;
        }
        else
        {
                AttrList *      attrListP = _tDataP->GetAttrList();
                int                     attrNum = 0;

                attrListP->InitIterator();
                while (attrListP->More())
                {
                        AttrInfo *attrInfoP = attrListP->Next();

                        vector.value[attrNum] = AttrToDouble(attrInfoP->type,
                                _recBuf + attrInfoP->offset);
                        attrNum++;
                }
                attrListP->DoneIterator();
        }

        return result;
}

/*------------------------------------------------------------------------------
 * function: AttrProj::ParseProjection
 * Parse the attribute projection file and build up the corresponding
 * data structures.
 */
DevStatus
AttrProj::ParseProjection(char *attrProjFile)
{
        DO_DEBUG(printf("AttrProj::ParseProjection()\n"));

        DevStatus       result = StatusOk;

        FILE *          file = fopen(attrProjFile, "r");
        if (file == NULL)
        {
                fprintf(stderr, "Can't open attribute projection file\n");
                result = StatusFailed;
        }
        else
        {
                const int       bufSize = 4096;
                char            buf[bufSize];
                char            separators[] = " \t";

                /* Get each line in the attribute projection file. */
                while (fgets(buf, bufSize, file) != NULL)
                {
                        DOASSERT(buf[strlen(buf)-1] == '\n',
                                "Projection file line too long");

                        /* TEMPTEMP -- we should look for some kind of comment char. */

                        StripTrailingNewline(buf);
                        DO_DEBUG(printf("%s\n", buf));

                        Projection      projection;
                        char *          token = strtok(buf, separators);
                        if (token == NULL) continue;
                        projection.attrCount = atoi(token);
                        DO_DEBUG(printf("projection.attrCount = %d\n",
                                projection.attrCount));
                        projection.attrList = new int[projection.attrCount];

                        AttrList *      attrListP = _tDataP->GetAttrList();
                        int                     attrCount = attrListP->NumAttrs();
                        int                     projAttrNum = 0;

                        /* Find each attribute specified for this projection. */
                        while ((token = strtok(NULL, separators)) != NULL)
                        {
                                projection.attrList[projAttrNum] = illegalAttr;
                                DO_DEBUG(printf("  token = %s", token));

                                int                     attrNum;

                                /* Now find the attribute in the TData corresponding to
                                 * the name specified in the projection. */
                                for (attrNum = 0; attrNum < attrCount; attrNum++)
                                {
                                        AttrInfo *      attrInfoP = attrListP->Get(attrNum);

                                        if (!strcmp(token, attrInfoP->name))
                                        {
                                                DO_DEBUG(printf(" attrNum = %d\n", attrNum));
                                                projection.attrList[projAttrNum] = attrNum;
                                                break;
                                        }
                                }
                                DOASSERT(projection.attrList[projAttrNum] != illegalAttr,
                                        "Illegal attribute name in attribute projection file");
                                projAttrNum++;
                        }
                        DOASSERT(projAttrNum == projection.attrCount,
                                "Incorrect number of attributes in projection file");


                        _projList.AddProjection(projection);

                }

                fclose(file);
        }

        return result;
}

/*------------------------------------------------------------------------------
 * function: AttrToDouble
 * Convert an attribute value to a double.
 */
static double
AttrToDouble(AttrType type, char *valP)
{
        DO_DEBUG(printf("AttrToDouble()\n"));

        double          doubleVal;
        float           floatVal;
        int                     intVal;
        double          result = 0.0;

        switch (type)
        {
        case IntAttr:
                intVal = *(int *) valP;
                DO_DEBUG(printf("        %d\n", intVal));
                result = (double) intVal;
                break;

        case FloatAttr:
                floatVal = *(float *) valP;
                DO_DEBUG(printf("        %f\n", floatVal));
                result = (double) floatVal;
                break;

        case DoubleAttr:
                doubleVal = *(double *) valP;
                DO_DEBUG(printf("        %f\n", doubleVal));
                result = (double) doubleVal;
                break;

        case StringAttr:
                DOASSERT(false, "Can't deal with string attribute");
                break;

        case DateAttr:
                DOASSERT(false, "Can't deal with date attribute");
                break;

        default:
                DOASSERT(false, "Illegal attribute type");
                break;
        }

        return result;
}

/*============================================================================*/