|
| 1 | +#include <string> |
| 2 | +//#include <tr1/unordered_map> |
| 3 | +#include <limits> |
| 4 | +#include <map> |
| 5 | +#include <vector> |
| 6 | +#include <iostream> |
| 7 | +#include <stdexcept> |
| 8 | + |
| 9 | +using std::vector; |
| 10 | +using std::string; |
| 11 | +using std::map; |
| 12 | +using std::make_pair; |
| 13 | +using std::cout; |
| 14 | +using std::endl; |
| 15 | +using std::out_of_range; |
| 16 | +using std::ostream; |
| 17 | +//typedef tr1::unordered_map map; |
| 18 | + |
| 19 | +// TODO: upgrade it to process trace. Rule: char-->elem string-->elem_list |
| 20 | +class SuffixTree |
| 21 | +{ |
| 22 | +public: |
| 23 | + // active point is initialized as (root, None, 0), remainder initialized as 1 |
| 24 | + SuffixTree(string str):test_str(str), pos(0), root(test_str), active_point(&root, 0, 0), remainder(0), ls() {} |
| 25 | + int construct(void); |
| 26 | + |
| 27 | + // return -1 if no such sub exist, return the beginning postion of this substring in thr original string if it exist |
| 28 | + int search(string sub); |
| 29 | + |
| 30 | + // return the length of the longest prefix of sub which can be matched in suffix tree |
| 31 | + template <class Iterator> |
| 32 | + Iterator inc_search(Iterator sub) |
| 33 | + { |
| 34 | + typedef typename Iterator::value_type T; // extract real type |
| 35 | + |
| 36 | + Iterator result = sub; |
| 37 | + Node* node = &root; |
| 38 | + Edge* edge = NULL; |
| 39 | + int pos = 0; // the iter's pos at edge |
| 40 | + int edge_len = -1; |
| 41 | + bool flag = true; |
| 42 | + |
| 43 | + |
| 44 | + while (flag) { |
| 45 | + if (edge == NULL) { |
| 46 | + edge = node->find_edge(*result); |
| 47 | + if (edge == NULL) { |
| 48 | + flag = false; |
| 49 | + } |
| 50 | + else { |
| 51 | + result++; |
| 52 | + pos = 1; // the second element of the edge |
| 53 | + edge_len = edge->length(); |
| 54 | + } |
| 55 | + } |
| 56 | + else { |
| 57 | + if (pos >= edge_len) { |
| 58 | + node = edge->endpoint; |
| 59 | + edge = NULL; |
| 60 | + edge_len = 0; |
| 61 | + } |
| 62 | + else { |
| 63 | + if (*result == (*edge)[pos]) { |
| 64 | + result++; |
| 65 | + pos++; |
| 66 | + } |
| 67 | + else |
| 68 | + flag = false; |
| 69 | + } |
| 70 | + } |
| 71 | + } |
| 72 | + |
| 73 | + return result; |
| 74 | + } |
| 75 | + |
| 76 | + int print_tree(void); |
| 77 | +private: |
| 78 | + string test_str; |
| 79 | + |
| 80 | + struct Node; |
| 81 | + typedef struct Node Node; |
| 82 | + |
| 83 | + struct Edge{ |
| 84 | + // the begin and end pos of this edge, note that INT_MAX stands for #(the changing end pos of this entire string) |
| 85 | + int begin, end; |
| 86 | + // Is there a better way to find test_str? |
| 87 | + string& test_node_str; |
| 88 | + |
| 89 | + Node * endpoint; |
| 90 | + |
| 91 | + Edge(int b, int e, string& str): |
| 92 | + test_node_str(str) |
| 93 | + { |
| 94 | + begin = b; |
| 95 | + end = e; |
| 96 | + endpoint = NULL; |
| 97 | + //std::cout << "Edge initialized" << std::endl; |
| 98 | + } |
| 99 | + |
| 100 | + void change_edge(int b, int e) |
| 101 | + { |
| 102 | + begin = b; |
| 103 | + end = e; |
| 104 | + } |
| 105 | + |
| 106 | + int length(void) |
| 107 | + { |
| 108 | + |
| 109 | + if (end > test_node_str.size()) |
| 110 | + return test_node_str.size() - begin; |
| 111 | + else |
| 112 | + return end - begin + 1; |
| 113 | + } |
| 114 | + |
| 115 | + // needed by map |
| 116 | + friend bool operator<(const Edge& me, const Edge& other) |
| 117 | + { |
| 118 | + return me.begin < other.begin; |
| 119 | + } |
| 120 | + |
| 121 | + char operator[](int i) |
| 122 | + { |
| 123 | + i += begin; |
| 124 | + if (i > end) |
| 125 | + throw out_of_range("Edge [] out of range."); |
| 126 | + |
| 127 | + return test_node_str[i]; |
| 128 | + } |
| 129 | + |
| 130 | + friend ostream& operator<<(ostream& os, Edge& edge) |
| 131 | + { |
| 132 | + int end = edge.test_node_str.size()-1; |
| 133 | + if (end >= edge.end) |
| 134 | + end = edge.end; |
| 135 | + |
| 136 | + char c; |
| 137 | + for (int i=edge.begin; i<=end; i++) { |
| 138 | + c = edge.test_node_str[i]; |
| 139 | + os << c; |
| 140 | + } |
| 141 | + if (end != edge.end) |
| 142 | + os << '#'; |
| 143 | + |
| 144 | + return os; |
| 145 | + } |
| 146 | + |
| 147 | + bool is_none(void) { return begin == 0 && end == 0; } |
| 148 | + }; |
| 149 | + typedef struct Edge Edge; |
| 150 | + |
| 151 | + struct Node{ |
| 152 | + string& test_node_str; |
| 153 | + map<int, int> testmap; |
| 154 | + map<Edge*, bool> edges; |
| 155 | + // find the edge quicky by storing the leading char of this edge |
| 156 | + map<char, Edge*> findedges; |
| 157 | + Node* suffix_link; |
| 158 | + |
| 159 | + friend class LinkState; |
| 160 | + |
| 161 | + Node(string& str) : |
| 162 | + test_node_str(str), suffix_link(NULL) { edges.clear(); findedges.clear(); } |
| 163 | + |
| 164 | + void add_edge(Edge* edge) { |
| 165 | + if (edge->endpoint == NULL) |
| 166 | + edge->endpoint = new Node(test_node_str); |
| 167 | + make_pair(edge, true); |
| 168 | + edges.insert(make_pair(edge, true)); |
| 169 | + findedges.insert(make_pair(test_node_str[edge->begin], edge)); |
| 170 | + //cout << "edge added. Now we have " << edges.size() << "edges." << endl; |
| 171 | + } |
| 172 | + |
| 173 | + void del_edge(Edge* edge) { |
| 174 | + map<Edge*, bool>::iterator iter = edges.find(edge); |
| 175 | + |
| 176 | + if (iter == edges.end()) |
| 177 | + throw out_of_range("edge don't exit"); |
| 178 | + else { |
| 179 | + // note we should erase the findedges too |
| 180 | + edges.erase(edge); |
| 181 | + //cout << "delete" << (*edge)[0] << endl; |
| 182 | + findedges.erase((*edge)[0]); |
| 183 | + //cout << "edge deleted. Now we have " << edges.size() << "edges." << endl; |
| 184 | + } |
| 185 | + |
| 186 | + } |
| 187 | + |
| 188 | + // find edge by the first char |
| 189 | + Edge* find_edge(char c) |
| 190 | + { |
| 191 | + //cout << "finding edge"; |
| 192 | + map<char, Edge*>::iterator iter = findedges.find(c); |
| 193 | + //cout << "founded?" << endl; |
| 194 | + if (iter != findedges.end()) |
| 195 | + return iter->second; |
| 196 | + else |
| 197 | + return NULL; |
| 198 | + } |
| 199 | + |
| 200 | + bool isleaf() { return edges.empty(); } |
| 201 | + |
| 202 | + bool operator==(Node& other) |
| 203 | + { |
| 204 | + return (this) == (&other); |
| 205 | + } |
| 206 | + |
| 207 | + friend ostream& operator<<(ostream& os, Node& node) |
| 208 | + { |
| 209 | + map<Edge*, bool>::iterator iter; |
| 210 | + map<char, Edge*>::iterator iter_f; |
| 211 | + |
| 212 | + for (iter=node.edges.begin(); iter!=node.edges.end(); ++iter) |
| 213 | + os << iter->first << '\t'; |
| 214 | + os << endl; |
| 215 | + |
| 216 | + for (iter_f=node.findedges.begin(); iter_f!=node.findedges.end(); ++iter_f) |
| 217 | + os << iter_f->first << "-->" << iter_f->second << endl; |
| 218 | + |
| 219 | + return os; |
| 220 | + } |
| 221 | + }; |
| 222 | + //typedef struct Node Node; |
| 223 | + |
| 224 | + class ActivePoint{ |
| 225 | + public: |
| 226 | + Node* active_node; |
| 227 | + char active_edge; |
| 228 | + int active_length; |
| 229 | + |
| 230 | + ActivePoint(Node* node, char edge, int length): |
| 231 | + active_node(node), active_edge(edge), active_length(length) { std::cout << "ActivePoint initialized" << std::endl; } |
| 232 | + }; |
| 233 | + |
| 234 | + Node root; |
| 235 | + ActivePoint active_point; |
| 236 | + |
| 237 | + Node* get_active_node(void) { return active_point.active_node; } |
| 238 | + void set_active_node(Node* node) { active_point.active_node = node; cout << "Active node set as " << node << endl; } |
| 239 | + char get_active_edge(void) { return active_point.active_edge; } |
| 240 | + void set_active_edge(char edge) { active_point.active_edge = edge; } |
| 241 | + int get_active_length(void) { return active_point.active_length; } |
| 242 | + void set_active_length(int len) { active_point.active_length = len; } |
| 243 | + void inc_active_len() { active_point.active_length++; } |
| 244 | + void dec_active_len() { active_point.active_length--; } |
| 245 | + |
| 246 | + // how many suffixes is to be inserted? |
| 247 | + int remainder; |
| 248 | + // how many characters inserted? |
| 249 | + int pos; |
| 250 | + char get_ele(int i) { return test_str[i]; } |
| 251 | + // insert a char from pos to suffix tree |
| 252 | + int insert(); |
| 253 | + int insert_rule1(); |
| 254 | + int insert_rule3(); |
| 255 | + int print_node(Node* node, int level); |
| 256 | + |
| 257 | + |
| 258 | + Node* seperate_edge(Node * node, Edge* edge, int rule); |
| 259 | + |
| 260 | + // check if we can change active node |
| 261 | + void check_an(void) |
| 262 | + { |
| 263 | + Node* node = get_active_node(); |
| 264 | + Edge* edge = node->find_edge(get_active_edge()); |
| 265 | + |
| 266 | + if (edge == NULL) |
| 267 | + return; |
| 268 | + |
| 269 | + int edge_size = edge->end - edge->begin + 1; |
| 270 | + |
| 271 | + // update |
| 272 | + if (edge_size == get_active_length()) { |
| 273 | + set_active_node(edge->endpoint); |
| 274 | + set_active_edge(0); |
| 275 | + set_active_length(0); |
| 276 | + } |
| 277 | + } |
| 278 | + |
| 279 | + // this class indicate when shall we insert a suffix link |
| 280 | + // ls should be a singleton |
| 281 | + class LinkState |
| 282 | + { |
| 283 | + bool first; |
| 284 | + |
| 285 | + Node* prev, *curr; |
| 286 | + |
| 287 | + public: |
| 288 | + LinkState() : first(true), prev(NULL), curr(NULL) {} |
| 289 | + |
| 290 | + void ins_link(Node* node) |
| 291 | + { |
| 292 | + prev = curr; |
| 293 | + curr = node; |
| 294 | + |
| 295 | + if (!first) { |
| 296 | + prev->suffix_link = curr; |
| 297 | + cout << "Suffix link added from prev " << prev << " to curr " << curr << endl; |
| 298 | + } |
| 299 | + |
| 300 | + first = false; |
| 301 | + } |
| 302 | + |
| 303 | + void clear(void) |
| 304 | + { |
| 305 | + first = true; |
| 306 | + prev = curr = NULL; |
| 307 | + } |
| 308 | + }; |
| 309 | + LinkState ls; |
| 310 | +}; |
0 commit comments