JAVA使用前缀树(Tire树)实现敏感词过滤、词典搜索

/**
 * trie tree
 *
 * @author lovely dog
 * @date 2020/10/20
 */
public class Node {
    /**
     * 子节点
     */
    private Map<Character, Node> nextNodes = new HashMap<>();

    public void addNext(Character key, Node node){
        nextNodes.put(key, node);
    }

    public Node getNext(Character key){
        return nextNodes.get(key);
    }

    public boolean isLastCharacter(){
        return nextNodes.isEmpty();
    }
}

/**
 * trie tree searcher
 *
 * @author lovely dog
 * @date 2020/10/20
 */
public class TrieSearcher {

    private Node root = new Node();

    /**
     * 添加词
     *
     * @param word 词
     */
    public void addWord(String word) {
        Node tmpNode = root;
        for (char c : word.toCharArray()) {
            Node node = tmpNode.getNext(c);
            if (null == node) {
                node = new Node();
                tmpNode.addNext(c, node);
            }
            tmpNode = node;
        }
    }

    /**
     * 替换词
     *
     * @param text         待处理文本
     * @param afterReplace 替换后的词
     * @return 处理后的文本
     */
    public String replace(String text, String afterReplace) {
        StringBuilder result = new StringBuilder(text.length());
        Node tmpNode = root;
        int begin = 0, pos = 0;
        while (pos < text.length()) {
            char c = text.charAt(pos);
            tmpNode = tmpNode.getNext(c);
            if (null == tmpNode) {
                result.append(text.charAt(begin));
                begin++;
                pos = begin;
                tmpNode = root;
            } else if (tmpNode.isLastCharacter()) {
                // 匹配完成, 进行替换
                result.append(afterReplace);
                pos++;
                begin = pos;
                tmpNode = root;
            } else {
                // 匹配上向后移
                pos++;
            }
        }
        result.append(text.substring(begin));
        return result.toString();
    }

    /**
     * 查找
     *
     * @param text 待处理文本
     * @return 统计数据 key: word value: count
     */
    public Map<String, Integer> find(String text) {
        Map<String, Integer> resultMap = new HashMap<>(16);
        Node tmpNode = root;
        StringBuilder word = new StringBuilder();
        int begin = 0, pos = 0;
        while (pos < text.length()) {
            char c = text.charAt(pos);
            tmpNode = tmpNode.getNext(c);
            if (null == tmpNode) {
                begin++;
                pos = begin;
                tmpNode = root;
            } else if (tmpNode.isLastCharacter()) {
                // 匹配完成
                String w = word.append(c).toString();
                resultMap.put(w, resultMap.getOrDefault(w, 0) + 1);
                pos++;
                begin = pos;
                tmpNode = root;
                word = new StringBuilder();
            } else {
                // 匹配上向后移
                word.append(c);
                pos++;
            }
        }
        return resultMap;
    }
}

public class Main {
    public static void main(String[] args) {
        TrieSearcher trieSearcher = new TrieSearcher();
        Stream.of("猪狗", "小狗", "小猫", "小猪", "垃圾", "狗东西").forEach(trieSearcher::addWord);
        String sentence = "你好，小狗，小猪，今天天气真好。";
        System.out.println(trieSearcher.replace(sentence, "***"));
        System.out.println(trieSearcher.find(sentence));
    }
}