Merge branch 'master' into develop

Author: xtaci

Makefile

@@ -3,7 +3,7 @@ CC=gcc
 CPP=g++
 AR=ar
 RANLIB=ranlib
-CFLAGS= -g -Wall -Wno-unused-function
+CFLAGS= -g -Wall -Wno-unused-function -std=gnu++0x
 SRCDIR = ./src
 INCLUDEDIR = -I./include -I.
 DEPS = 
@@ -68,7 +68,8 @@ PROGRAMS =  m_based_demo \
 			bubble_sort_demo \
 			selection_sort_demo	\
 			8queue_demo \
-			palindrome_demo
+			palindrome_demo	\
+			suffix_array_demo
 
 all: $(PROGRAMS)
 

README.md

@@ -55,7 +55,7 @@
     Red-black tree
     Interval tree
     Prefix Tree(Trie)
-    *Suffix Tree(未实现)*
+    Suffix Tree
     B-Tree
 
     Hash by multiplication
@@ -96,4 +96,4 @@
     wycg1984: for K-Means
     xmuliang: for HeapSort, Kruskal MST
     wyh267: for base64, LRU, bubble sort, selection sort
-    ZhangYou0122 : Push-Relabel algorithm
+    ZhangYou0122 : Push-Relabel algorithm, SuffixTree

include/suffix_array.h

@@ -0,0 +1,92 @@
+/*
+ * suffix array algorithm
+ *
+ * Features: suffix array can sort all the suffixs in time complexity O(n*log^2(n)), and use memory in O(n).
+ * And suffix array can get two suffixs' longest common prefix(lcp) in O(log(n)) complexity. 
+ * You can test it by running suffix_array_demo.cpp
+ * Want to get more detailed information about suffix array? Please google SUFF_AR_ENG.pdf
+ *
+ * author: [email protected]
+ */
+
+#ifndef _SUFFIX_ARRAY_H
+#define _SUFFIX_ARRAY_H
+
+#include <algorithm>
+#include <vector>
+#include <string>
+#include <math.h>
+
+using namespace std;
+
+namespace alg {
+	class SuffixArray {
+	private:
+		vector<vector<int> > bucket;
+		vector<int> suffix;
+		int N, L, K;
+		const string& str;
+		void suffix_sort();
+		void update_bucket();
+		
+		bool less_than(int a, int b) { 
+			if(K==0) return str[a]<str[b];
+			else {
+				if(bucket[K-1][a]==bucket[K-1][b]) return bucket[K-1][a+L/2]<bucket[K-1][b+L/2];
+				else return bucket[K-1][a]<bucket[K-1][b];
+			}
+		}
+
+		bool equal(int a, int b) {
+			return !less_than(a,b) && !less_than(b,a);
+		}
+
+	public:
+		explicit SuffixArray(const string& s) : N(s.size()), L(0), K(0), str(s) { suffix_sort();}
+		// return the sorted suffix
+		int operator [] (int i) { return suffix[i];}
+		// Given two suffixs of string, return the longest common prefix length
+		int lcp_length(int x, int y);
+	};
+
+	void SuffixArray::suffix_sort() {
+		// init suffix
+		suffix.resize(N);
+		for(int i=0;i<N;i++) suffix[i]=i;
+		// init bucket
+		bucket.resize(ceil(log2(N))+1);
+		for(size_t k=0;k<bucket.size();k++) bucket[k].resize(N+N);
+	
+		for(L=1,K=0;(L>>1)<N;L<<=1,K++) {
+			sort(suffix.begin(), suffix.end(), bind(&SuffixArray::less_than, *this, placeholders::_1, placeholders::_2));
+			update_bucket();
+		}
+	}
+
+
+	void SuffixArray::update_bucket() {
+		int seq=0;
+		bucket[K][suffix[0]]=0;
+		for(int i=1;i<N;i++) {
+			if(!equal(suffix[i],suffix[i-1])) seq++;
+			bucket[K][suffix[i]]=seq;
+		}
+		fill(bucket[K].begin()+N, bucket[K].end(), -1);
+	}
+
+	int SuffixArray::lcp_length(int x, int y) {
+		if(x==y) return N-x;
+		int ret=0;
+		for(int k=K-1;k>=0 && x<N && y<N;k--) {
+			if(bucket[k][x]==bucket[k][y]) {
+				x += (1<<k);
+				y += (1<<k);
+				ret += (1<<k);
+			}
+		}
+		return ret;
+	}
+}
+
+#endif
+

src/suffix_array_demo.cpp

@@ -0,0 +1,37 @@
+#include <iostream>
+#include <string>
+#include <math.h>
+
+#include "suffix_array.h"
+
+using namespace std;
+using namespace alg;
+
+void print(string::iterator b, string::iterator e) {
+	for(auto it=b;it!=e;++it) cout<<*it;
+}
+
+int main()
+{
+	string str;
+	while(cin>>str) {
+		SuffixArray sa(str);
+		cout<<endl;
+		cout<<"sorted suffixs are:"<<endl;
+		for(size_t i=0;i<str.size();i++) {
+			print(str.begin()+sa[i], str.end());
+			cout<<endl;
+		}
+		cout<<endl;
+		cout<<"The length of the longest common prefix of two suffixs ";
+		int i=rand()%str.size();
+		int j=rand()%str.size();
+		print(str.begin()+i,str.end());
+		cout<<" and ";
+		print(str.begin()+j,str.end());
+		cout<<" is ";
+		cout<<sa.lcp_length(i,j)<<endl;
+		cout<<endl;
+	}
+	return 0;
+}