文本选择器，筛选出可能为新词的词汇

CnTextSelector.java

package grid.text.selector;

import grid.common.TextUtils;


public class CnTextSelector extends CommonTextSelector {

    public CnTextSelector(String document,int minSelectLen,int maxSelectLen) {
        super(document,minSelectLen,maxSelectLen);
    }

    protected void adjustCurLen() {
        while (pos < docLen && !TextUtils.isCnLetter(document.charAt(pos))) {
            pos++;
        }
        for (int i = 0; i < maxSelectLen && pos + i < docLen; i++) {
            if (!TextUtils.isCnLetter(document.charAt(pos + i))) {
                curLen = i;
                if (curLen < minSelectLen) {
                    pos++;
                    adjustCurLen();
                }
                return;
            }
        }

        curLen = pos + maxSelectLen > docLen ? docLen - pos : maxSelectLen;
    }
}

CommonTextSelector.java

package grid.text.selector;


public class CommonTextSelector implements TextSelector {

    protected String document;

    protected int pos = 0;

    protected int maxSelectLen = 5;

    protected int minSelectLen = 2;

    protected int curLen;

    protected final int docLen;

    public CommonTextSelector(String document,int maxSelectLen) {
        this.document = document;
        this.minSelectLen = minSelectLen;
        this.maxSelectLen = maxSelectLen;
        docLen = document.length();
        adjustCurLen();
    }

    public void select() {
        pos += ++curLen;
        adjustCurLen();
    }

    protected void adjustCurLen() {
        curLen = pos + maxSelectLen > docLen ? docLen - pos : maxSelectLen;
    }

    public String next() {
        if (curLen < minSelectLen) {
            pos++;
            adjustCurLen();
        }

        if (pos + curLen <= docLen && curLen >= minSelectLen) {
            return document.substring(pos,pos + curLen--);
        } else {
            curLen--;
            // return document.substring(pos,docLen);
            return "";
        }
    }

    public boolean end() {
        return curLen < minSelectLen && curLen + pos >= docLen - 1;
    }

    @Override
    public int getCurPos() {
        return pos;
    }
}

TextSelector.java

package grid.text.selector;


public interface TextSelector {
    public boolean end();

    public void select();

    public String next();

    public int getCurPos();

}

测试代码

（编辑：晋中站长网）

【声明】本站内容均来自网络，其相关言论仅代表作者个人观点，不代表本站立场。若无意侵犯到您的权利，请及时与联系站长删除相关内容!