LCS 问题

/**
 * Author: yiminghe
 * Date: 2008-10-24
 * Time: 15:08:32
 * Any problem ,contact yiminghe@fudan.edu.cn.
 */

import java.io.*;
import java.util.*;

/**
 * 利用 后缀树算法
 *
 * LCS 问题及其扩展,找到多个字符串的所有公共子串
 *
 */
class LCS {
    public static void main(String[] args) throws IOException {
        String[] source = {"axybcde", "cdefxy", "xyccde"};

        SuffixTreeNode st = buildSuffixTree(source);

        String result;
        result = Lcs(st.firstChild, source.length);

        if (result.equals(""))
            System.out.println("No common string!");
        else
            System.out.println("The longest common substring is : "
                    + result + " .");


        String[] commons = commonString(source);
        System.out.println(Arrays.asList(commons));
    }

    /**
     * 建立一个后缀树
     *
     * @param ss 字符串数组
     * @return 后缀树的根结点
     */
    public static SuffixTreeNode buildSuffixTree(String ss[]) {
        HashMap<String, String> belong = new HashMap<String, String>();
        belong.put("0", "");
        SuffixTreeNode SuffixTree =
                new SuffixTreeNode(-1, "", 0, belong);

        //Add suffixs...
        for (int i = 0; i < ss.length; i++) {
            System.err.print("后缀树[" + (i + 1) + "]");

            belong = new HashMap<String, String>();
            belong.put("" + (i + 1), "");
            for (int index = 0; index < ss[i].length(); index++) {
                String str = ss[i].substring(index);
                SuffixTree.insert(index, str, 0, belong);
            }
            System.err.println("  -  OK");
        }

        return SuffixTree;
    }

    /**
     * 深度遍历
     *
     * @param suffixtree 根的后缀树结点
     * @param count      字符串总数
     * @return 最长公共子串
     */
    public static String Lcs(SuffixTreeNode suffixtree, int count) {
        String result = "";
        String result2;
        while (suffixtree != null) {
            int flag = suffixtree.belongTo.size();

            if (flag == count) {
                if (suffixtree.isLeaf()) {
                    //找到最大
                    if (result.length() <
                            suffixtree.label.length())
                        result = suffixtree.label;
                } else {
                    //只是后缀的后缀
                    result2 = Lcs(suffixtree.firstChild, count);
                    //要完整的后缀
                    if (result.length() <
                            (suffixtree.label.length() + result2.length()))
                        result = suffixtree.label + result2;
                }
            }
            suffixtree = suffixtree.next;
        }
        return result;
    }


    /**
     * 找到所有的相同子串,子串间不相互包含
     *
     * @param source 字符串集合
     * @return 字符串集合
     */
    public static String[] commonString(String[] source) {
        HashSet<String> r = new HashSet<String>();
        SuffixTreeNode st = buildSuffixTree(source);
        recurCommon(r, st.firstChild, source.length);
        String[] original = r.toArray(new String[r.size()]);
        ArrayList<String> result = new ArrayList<String>();
        for (int i = 0; i < original.length; i++) {
            int j = 0;
            for (j = 0; j < original.length; j++) {

                //有和其它元素相互包含 ,舍弃
                if (i != j && original[j].endsWith(original[i])) {
                    break;
                }
            }

            if (j == original.length) {
                result.add(original[i]);
            }
        }

        return result.toArray(new String[result.size()]);
    }


    //搜集子串,并且去掉明显的嵌套子串
    private static boolean recurCommon(HashSet<String> r, SuffixTreeNode suffixtree, int count) {
        boolean result = false;
        while (suffixtree != null) {
            int flag = suffixtree.belongTo.size();

            if (flag == count) {
                result = true;
                if (suffixtree.isLeaf()) {
                    String re = suffixtree.label;
                    SuffixTreeNode temp = suffixtree;
                    while (temp.parent != null) {
                        temp = temp.parent;
                        re = temp.label + re;
                    }

                    r.add(re);

                } else {
                    //只是后缀的后缀
                    boolean has = recurCommon(r, suffixtree.firstChild, count);
                    //要完整的后缀
                    if (!has) {
                        String re = suffixtree.label;
                        SuffixTreeNode temp = suffixtree;
                        while (temp.parent != null) {
                            temp = temp.parent;
                            re = temp.label + re;
                        }

                        r.add(re);

                    }
                }
            }
            suffixtree = suffixtree.next;
        }
        return result;
    }


}

class SuffixTreeNode {
    //原字符串的位置
    //公共就没意义了
    int index;
    //后缀值
    String label;
    //兄弟关系
    SuffixTreeNode next;
    //第一个孩子关系
    SuffixTreeNode firstChild = null;
    //父亲
    SuffixTreeNode parent = null;
    //树的层数
    int level;

    //属于哪个字符串
    HashMap<String, String> belongTo = null;

    SuffixTreeNode(int i, String s,
                   int level, HashMap<String, String> flag) {
        this.index = i;
        this.label = s;
        this.level = level;

        if (belongTo == null)
            belongTo = new HashMap<String, String>();
//Put subject-to information to belongTo...
        belongTo.putAll(flag);
    }


    void setChilden(SuffixTreeNode n) {
        this.firstChild = n;
        if (n != null)
            n.parent = this;
    }

    boolean isLeaf() {
        return (this.firstChild == null);
    }

    /**
     * 在当前结点下插入 新的后缀树结点
     *
     * @param ind    index
     * @param str    insert_str
     * @param level  level
     * @param belong belong
     */
    public void insert(int ind, String str,
                       int level, HashMap<String, String> belong) {
        SuffixTreeNode newnode, firstChild, prev;
        String strtemp, prefix;
        int index_i;

        //第一次   只有根结点
        if (this.isLeaf()) {
            newnode = new SuffixTreeNode(ind, str,
                    level + 1, belong);
            this.setChilden(newnode);
            return;
        }

        firstChild = this.firstChild;

        if (firstChild.label.charAt(0) > str.charAt(0)) {
            newnode = new SuffixTreeNode(ind, str,
                    level + 1, belong);
            this.setChilden(newnode);
            newnode.next = firstChild;
            return;
        }
        prev = firstChild;

        //合适的子结点插入位置
        while ((firstChild != null) &&
                (firstChild.label.charAt(0) <
                        str.charAt(0))) {
            prev = firstChild;
            firstChild = firstChild.next;
        }
        if (firstChild == null) {
            newnode = new SuffixTreeNode(ind, str,
                    level + 1, belong);
            newnode.parent = this;
            prev.next = newnode;
            return;
        }
        if (firstChild.label.charAt(0) > str.charAt(0)) {
            newnode = new SuffixTreeNode(ind, str,
                    level + 1, belong);
            prev.next = newnode;
            newnode.parent = this;
            newnode.next = firstChild;
            return;
        }
        //与 str 完全相同
        if (str.equals(firstChild.label)) {
            //公共前缀属性共有
            firstChild.belongTo.putAll(belong);
            return;
        }
        //首字母相同
        int minLength = Math.min(firstChild.label.length(), str.length());
        for (index_i = 1; index_i < minLength; index_i++) {
            if (firstChild.label.charAt(index_i) !=
                    str.charAt(index_i)) {
                break;
            }
        }

        //temp 较短 ，或与 str 完全相同
        if (index_i == firstChild.label.length()) {


            //str 比 temp 长的部分
            strtemp = str.substring(index_i);
            firstChild.insert(ind, strtemp, level + 1, belong);
            //公共前缀属性共有
            firstChild.belongTo.putAll(belong);
            return;
        }

        //str 较短,或者 与 temp 中间 有不同元素
        //原来的 temp 前缀 共有
        prefix = firstChild.label.substring(0, index_i);
        strtemp = firstChild.label.substring(index_i);

        //原来 temp 的 与 str 不同的后缀 分离
        prev = new SuffixTreeNode(firstChild.index, strtemp,
                level + 1, firstChild.belongTo);
        prev.setChilden(firstChild.firstChild);

        firstChild.setChilden(prev);
        firstChild.index = -1;
        firstChild.label = prefix;
        firstChild.belongTo.putAll(belong);
        prev.lowDown();


        //加入 原来 str 与 temp 不同的后缀
        if (index_i < str.length()) {
            strtemp = str.substring(index_i);

            firstChild.insert(ind, strtemp, level + 1, belong);
        }

    }


    void print() {

    }

    /**
     * 加入中间树结点，对原树结点中保存的层次信息进行刷新
     */
    void lowDown() {
        SuffixTreeNode temp;
        this.level++;
        if (this.isLeaf())
            return;
        temp = this.firstChild;
        while (temp != null) {
            temp.lowDown();
            temp = temp.next;
        }
    }
}