instancereader.java

id3的java实现

        
        //the following algorithm is based on identifying "newly removed"
        //attributes.  A newly removed attribute is one which has a valid
        //(i.e. >= 0) entry in the assim map, but a false in the projMap.
        //For each newly removed attribute, set it to unmapped in the
        //assim map and increment displacement.
        //Also, throughout the loop decrement the assimilation values of
        //any valid attribtes by displacement.  This will correct for
        //the newly removed attributes.
        for(int i=0; i< projMap.length;i++) {
            if(assimMap[i] >= 0) {
                if(projMap[i] == false) {
                    //newly removed
                    displacement++;
                    assimMap[i] = unmapped;
                }
                else {
                    //decrement by displacement to correct for other removed attrs
                    assimMap[i] -= displacement;
                    //ASSERT(assimMap[i] >= 0);
                }
            }
        }
    }
    
    private void update_for_overflows() {
        build_proj_maps(projMap, listProjMap);
        instList.update_for_overflows(listProjMap);
        update_assim_map(projMap);
    }
    
    private int warn_projected_columns() {
        //LOG(5, "Checking schema againts names-file schema);
        //ASSERT(fileSchema);
        
        int attrNum = 0;
        int origAttrNum = 0;
        int numIgnored = 0;
        
        Schema schema = get_schema();
        while(origAttrNum <  fileSchema.num_attr()) {
            if(attrNum < schema.num_attr())
                ;//LOG(5, "Schema: attr "+attrNum+" ("
            //+ schema.attr_name(attrNum) +").");
            else
                ;//LOG(5, "Schema: no more attrs.");
            
            //LOG(5, "Original: attr " + origAttrNum +" ("
            //   + fileSchema.attrInfos[origAttrNum].name() + ").");
            
            //If the left out column is a weight or label, don't do anything
            //Only leave out the weight if ignoreWeightColumn is set.
            if(fileSchema.get_label_column() == origAttrNum ||
            (fileSchema.get_weight_column() == origAttrNum &&
            fileSchema.get_ignore_weight_column()))
                ; //ignore this case
            
            //Check names.  If they match, we're ok and we increment
            // out index.  If not, print a warning
            else if(attrNum < schema.num_attr() &&
            schema.attr_name(attrNum) ==
            fileSchema.attrInfos[origAttrNum].name() )
                attrNum++;
            else{
                //Only nominals should get left out.  If it's not nominal,
                // then it's an error
                AttrInfo ai = fileSchema.attrInfos[origAttrNum];
                if(!ai.can_cast_to_nominal())
                    Error.err("InstanceList::diff_original"
                    +"_schema: non-nominal attribute "+ai.name()+" was "
                    +"projected out -->fatal_error");
                NominalAttrInfo nai = ai.cast_to_nominal();
                if(nai.num_values() >= attrValueLimit)
                    System.out.println("Warning-->attribute "+ai.name()
                    +" ignored: exceeded "+attrValueLimit+ " values.");
                else if(nai.num_values() == 0)
                    System.out.println("Warning-->attribute "+ai.name()
                    +" this attribtue has no values except unknowns.");
                else if(nai.get_ignore())
                    System.out.println("Warning-->attribute "+ai.name()
                    +"ignored: ignore flag set.");
                else
                    Error.err("InstanceReader::warn_on...:"
                    +"attribute "+ai.name() +" was poject out for an "
                    +"invalid reason -->fatal_error");
                
                numIgnored++;
            }
            
            origAttrNum++;
        }
        
        if(numIgnored > 0)
            ;//LOG(1, "\nWarning: some attributes are ignored.  The default"
        //+" limit of +"attrValueLimit+" may be increased\n"
        //+" by changing the MAX_ATTR_VALS paramter");
        
        //if no attributes remain, post a warning
        if(schema.num_attr()==0)
            Error.err("InstanceReader::warn_projected_columns"
            +": all attributes have been ignored!  This example is probably"
            +" no useful for learnign -->fatal_error");
        
        //Make sure we've accounted for all attributes in this schema
        //ASSERT(origAttrNum == fileSchema.num_attr());
        //ASSERT(attrNum == schema.num_attr());
        return numIgnored;
        
    }
    
    /** Checks if the Instance being read is labelled.
     * @return TRUE if the Instance is labelled, FALSE otherwise.
     */
    public boolean is_labelled(){return get_schema().is_labelled();}
    
    private void prepare_to_set(int attrNum, boolean isNominal, boolean isReal) {
        //check range on the incoming attrNum
        if(attrNum < 0 || attrNum > fileSchema.num_attr())
            Error.err("InstanceReader::prepare_to_set: "
            +"attribute number "+attrNum+" is out of range -->fatal_error");
        
        if(isNominal && isReal)
            Error.err("InstanceReader::prepare_to_set: "
            +"isNominal and isReal may not both be set -->fatal_error");
        
        if(isNominal && !fileSchema.attrInfos[attrNum].can_cast_to_nominal())
            Error.err("InstanceReader::prepare_to_set: "
            +"attempted to call a nominal setting function on a non-nominal"
            +" attribute -->fatal_error");
        
        if(isReal && !fileSchema.attrInfos[attrNum].can_cast_to_real())
            Error.err("InstanceReader::prepare_to_set: "
            +"attempted to call a real setting function on a non-real"
            +" attribute -->fatal_error");
        
        if(setAttr[attrNum])
            Error.err("InstanceReader::prepare_to_set: "
            +"attribute "+attrNum+" was already set -->fatal_Error");
        
        setAttr[attrNum] = true;
        anySet = true;
    }
    
    /** Returns the Schema being used to read data.
     * @return The Schema containing details about the file beinig read.
     */
    public Schema get_schema() {
        check_has_list();
        return instList.get_schema();
    }
    
    private void check_has_list() {
        if(!has_list())
            Error.err("InstanceReader::check_has_list: "
            +"this reader has had its list released -->fatal_error");
    }
    
    /** Checks if this InstanceReader has an InstanceList to store Instances in.
     * @return TRUE if there is an InstanceList present, FALSE otherwise.
     */
    public boolean has_list() {
        return instList != null;
    }
    
    /** Explicitly sets a nominal value. The attribute's type must support nominal.
     *
     * @param attrNum The number of the attribute containing the nominal value.
     * @param attrVal The value to be set as a nominal value.
     */
    public void set_nominal(int attrNum, String attrVal) {
        prepare_to_set(attrNum, true, false);
        //map attribute number if needed. If it doesn't map, return
        int mapNum = assimMap[attrNum];
        
        //map to label if requested. Do nothing if unmapped.  You cannot
        //map a nominal to the weight so abort if this is requested.
        if(mapNum==unmapped) {
            return;
        }
        else if(mapNum==mapToLabel){
            //ASSERT(fileSchema.get_label_column() != unmapped);
            if(vals[attrNum]==null)vals[attrNum]=new AttrValue();
            get_schema().label_info().cast_to_nominal().set_nominal_string(vals[attrNum], attrVal,makeUnknowns);
            
        }
        else {
            //set the value in vals from the list's attribute info.  If
            //makeUnknowns is set, we need to check if the value exists.
            //if not, don't add it but rather set an unknown value.
            if(vals[attrNum]==null)vals[attrNum]=new AttrValue();
            get_schema().attr_info(mapNum).cast_to_nominal().set_nominal_string(vals[attrNum],attrVal, makeUnknowns);
            
        }
        
        //force an update on the attrInfo in fileSchema if the nominal
        //belongs to a non-fixed attribute.
        //Only do this if makeUnknowns is not set
        if(!makeUnknowns &&
        !fileSchema.attrInfos[attrNum].cast_to_nominal().is_fixed()){
            AttrValue dummyAttrVal = new AttrValue();
            fileSchema.attrInfos[attrNum].set_nominal_string(dummyAttrVal, attrVal,false);
        }
        
    }
    
    /** Explicitly sets a real value. The attribute's type must support real.
     *
     * @param attrNum The number of the attribute containing the real value.
     * @param attrVal The value to be set as a real value.
     */
    public void set_real(int attrNum, double attrVal) {
        //System.out.println("Gets to InstanceReader::set_real--0");//DEBUG BY JL
        prepare_to_set(attrNum, false, true);
        //if this column is mapped to the weight, set the weight.
        if(attrNum==fileSchema.get_weight_column())
            weight = attrVal;
        //System.out.println("Gets to InstanceReader::set_real--1");//DEBUG BY JL
        
        //map attribute number if needed,  if it doesn't map, then just return
        int mapNum = assimMap[attrNum];
        //System.out.println("Gets to InstanceReader::set_real--2");//DEBUG BY JL
        if (vals == null)System.out.println("InstanceReader::set_real--vals == null");//DEBUG BY JL
        if (vals[attrNum] == null)System.out.println("InstanceReader::set_real--vals["+attrNum+"] == null");//DEBUG BY JL
        
        if(mapNum == unmapped)
            return;
        else if(mapNum==mapToLabel)
            //label is a nominal column--should have failed in check above
            Error.err("ABORT_IF_REACHED");
        else
            get_schema().attr_info(mapNum).cast_to_real().set_real_val(vals[attrNum], attrVal);
        //System.out.println("Gets to InstanceReader::set_real--3");//DEBUG BY JL
        
    }
    
    /** Sets an attribute's value to unknown. Works on any type.
     *
     * @param attrNum The number of the attribute for which the unknown value will be inserted.
     */
    public void set_unknown(int attrNum) {
        prepare_to_set(attrNum,false,false);
        if(attrNum == fileSchema.get_weight_column()){
            Error.err("Mcerr:WARNING:setting an unknown weight value ("
            + fileSchema.attrInfos[attrNum].name()+").");
            weight = 0.0;
        }
        
        //map attribute number if needed.  If it doesn't map, then just return
        int mapNum = assimMap[attrNum];
        
        if(mapNum == unmapped)
            return;
        else if(mapNum == mapToLabel)
            //a later process will remove this instance
            get_schema().label_info().set_unknown(vals[attrNum]);
        else
            get_schema().attr_info(mapNum).set_unknown(vals[attrNum]);
    }
    
}