根据字符数/Byte数切分字符串

/**
     * 按照长度拆分字符串(若分段小于segment个，则填充""至数组),取实际段数与默认段数较大的作为返回的数量
     * @param srcString
     * @param seg_and_len [默认段数;每段长度;是否按byte划分], 默认为[20;3800;1]；第三位默认为1，表示按照byte划分.
     * @return 
     */
    public static List<String> splitXmlString(String srcString, int... seg_and_len){
        List<String> retList = new ArrayList<String>();
        
        int _SEGMENT = 20;
        int _LEN = 3800;
        
        int default_segment = _SEGMENT;
        int field_len = _LEN;
        int isSplitByByte = 1;
        
        //检验每段长度的数值
        if(seg_and_len.length==0){
            default_segment = _SEGMENT;
            field_len = _LEN;
            isSplitByByte = 1;
            
        }else if(seg_and_len.length==1) {//仅有默认段数
            default_segment = (seg_and_len[0]>_SEGMENT) ? seg_and_len[0] : _SEGMENT;
            field_len = _LEN;
            isSplitByByte = 1;
            
        }else if(seg_and_len.length==2) {//有默认段数与每段长度
            default_segment = (seg_and_len[0]>_SEGMENT) ? seg_and_len[0] : _SEGMENT;
            field_len = (seg_and_len[1]>0) ? seg_and_len[1] : _LEN;
            isSplitByByte = 1;
            
        }else if(seg_and_len.length==3) {//有默认段数、每段长度
            default_segment = (seg_and_len[0]>_SEGMENT) ? seg_and_len[0] : _SEGMENT;
            field_len = (seg_and_len[1]>0) ? seg_and_len[1] : _LEN;
            isSplitByByte = (seg_and_len[2]==0) ? seg_and_len[2] : 1;
            
        }
        
        //检验输入字符串
        if(null==srcString||srcString.equals("")){
            for(int i=0;i<default_segment;i++){ //当输入没有的时候，返回default_segment个空字符串对象
                retList.add(i,"");
            }
            return retList;
        }
        
        switch(isSplitByByte){
            case 0://按照字符数划分
                //实际应分段数
                int relSegment = (int) Math.ceil( (double)srcString.length()/field_len );

                //检验分段数
                int seg_limit = (default_segment > relSegment) ? default_segment:relSegment;
                
                for(int i=0;i<relSegment;i++){
                    if(i!=relSegment-1) {
                        retList.add( srcString.substring(i*field_len, i*field_len+field_len) );
                    }else{ //对实际分段的最后一段，填充剩下的所有字符串.
                        retList.add( srcString.substring(i*field_len, srcString.length()) );
                    }
                }
                //若分段小于seg_limit个，则填充""至数组
                for(int i=relSegment;i<seg_limit;i++){
                    retList.add(i,"");
                }
                break;
                
            case 1://按照byte数划分
                retList = subStringSplitByByte(srcString, field_len);
                int rel_size = (null==retList||retList.isEmpty())? 0 : retList.size();
                
                if(rel_size < default_segment) {
                    for(int i=rel_size;i<default_segment;i++){
                        retList.add(i,"");
                    }
                }
                
                break;
                
            default:
                break;
        }       
        
        
        return retList;
    }
    /**
     * 根据字节长度长度len, 分割字符串src,
     * @param src
     * @param byte_len
     * @return  由于中文的原因，返回的字符串数组，每组至多byte_len为字节.
     */
    public static List<String> subStringSplitByByte(String src ,int byte_len){
        List<String> retList = new ArrayList<String>();
        
        if(null==src||src.isEmpty()){
            return retList;
        }
        if(byte_len < 2) byte_len=src.length();
        
        try {
            byte[] bytes = src.getBytes("gbk");
            
            int hanzi_cnt = 0;
            int pos = 0;
            int l_s = 0;
            int r_s = 0;
              
            for(int i=0;i<bytes.length;i++){
                if(bytes[i]<0) {
                    hanzi_cnt++;
                    r_s += 1;
                }else {
                    r_s += 2;
                }
                
                if( (i+1-pos) % byte_len==0) {        
                    if(hanzi_cnt==0 || hanzi_cnt%2==0 ){
                        retList.add(src.substring(l_s/2, r_s/2 ));
                        pos = i+1;
                        hanzi_cnt = 0;
                        l_s = r_s;
                        continue;
                    
                    }else{ // hanzi_cnt%2==1
                        retList.add(src.substring(l_s/2, r_s/2));
                        pos = i;
                        hanzi_cnt = 1;
                        l_s = r_s;
                        continue;
                    }             
                }
                
            }
            if(l_s/2 < src.length()){
                retList.add(src.substring(l_s/2, src.length() ));
            }
            
            
        } catch (UnsupportedEncodingException e) {
            logger.error("", e);
        }
        
        
        return retList;
    }

 

转载于:https://my.oschina.net/u/3202695/blog/847392