f5312db058ad0e55140b1cfde383e2b2c118e34f
[SXSI/XMLTree.git] / libcds / src / static_sequence / wt_node_internal.cpp
1 /* wt_node_internal.cpp
2  * Copyright (C) 2008, Francisco Claude, all rights reserved.
3  *
4  * wt_node_internal
5  *
6  * This library is free software; you can redistribute it and/or
7  * modify it under the terms of the GNU Lesser General Public
8  * License as published by the Free Software Foundation; either
9  * version 2.1 of the License, or (at your option) any later version.
10  *
11  * This library is distributed in the hope that it will be useful,
12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
14  * Lesser General Public License for more details.
15  *
16  * You should have received a copy of the GNU Lesser General Public
17  * License along with this library; if not, write to the Free Software
18  * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
19  *
20  */
21  
22 #include <wt_node_internal.h>
23
24 wt_node_internal::wt_node_internal(uint * symbols, uint n, uint l, wt_coder * c, static_bitsequence_builder * bmb) {
25         uint * ibitmap = new uint[n/W+1];
26         for(uint i=0;i<n/W+1;i++)
27                 ibitmap[i]=0;
28         for(uint i=0;i<n;i++) 
29                 if(c->is_set(symbols[i],l))
30                         bitset(ibitmap,i);
31         bitmap = bmb->build(ibitmap, n);
32   delete [] ibitmap;
33         uint count_right = bitmap->rank1(n-1);
34         uint count_left = n-count_right+1;
35         uint * left = new uint[count_left+1];
36         uint * right = new uint[count_right+1];
37         count_right = count_left = 0;
38         bool match_left = true, match_right = true;
39         for(uint i=0;i<n;i++) {
40                 if(bitmap->access(i)) {
41                         right[count_right++]=symbols[i];
42                         if(count_right>1)
43                                 if(right[count_right-1]!=right[count_right-2])
44                                         match_right = false;
45                 }
46                 else {
47                         left[count_left++]=symbols[i];
48                         if(count_left>1)
49                                 if(left[count_left-1]!=left[count_left-2])
50                                         match_left = false;
51                 }
52         }
53         if(count_left>0) {
54                 if(match_left/* && c->done(left[0],l+1)*/)
55                         left_child = new wt_node_leaf(left[0], count_left);
56                 else
57                         left_child = new wt_node_internal(left, count_left, l+1, c, bmb);
58         } else {
59                 left_child = NULL;
60         }
61         if(count_right>0) {
62                 if(match_right/* && c->done(right[0],l+1)*/)
63                         right_child = new wt_node_leaf(right[0], count_right);
64                 else
65                         right_child = new wt_node_internal(right, count_right, l+1, c, bmb);
66         } else {
67                 right_child = NULL;
68         }
69         delete [] left;
70         delete [] right;
71 }
72
73 // Deletes symbols array!
74 wt_node_internal::wt_node_internal(uchar * symbols, uint n, uint l, wt_coder * c, static_bitsequence_builder * bmb) {
75         uint * ibitmap = new uint[n/W+1];
76         for(uint i=0;i<n/W+1;i++)
77                 ibitmap[i]=0;
78         for(uint i=0;i<n;i++) 
79                 if(c->is_set((uint)symbols[i],l))
80                         bitset(ibitmap,i);
81         bitmap = bmb->build(ibitmap, n);
82   delete [] ibitmap;
83         uint count_right = bitmap->rank1(n-1);
84         uint count_left = n-count_right+1;
85         uchar * left = new uchar[count_left+1];
86         uchar * right = new uchar[count_right+1];
87         count_right = count_left = 0;
88         bool match_left = true, match_right = true;
89         for(uint i=0;i<n;i++) {
90                 if(bitmap->access(i)) {
91                         right[count_right++]=symbols[i];
92                         if(count_right>1)
93                                 if(right[count_right-1]!=right[count_right-2])
94                                         match_right = false;
95                 }
96                 else {
97                         left[count_left++]=symbols[i];
98                         if(count_left>1)
99                                 if(left[count_left-1]!=left[count_left-2])
100                                         match_left = false;
101                 }
102         }
103
104         delete [] symbols; 
105         symbols = 0;
106
107         if(count_left>0) {
108                 if(match_left/* && c->done(left[0],l+1)*/)
109                 {
110                     left_child = new wt_node_leaf((uint)left[0], count_left);
111                     delete [] left;
112                     left = 0;
113                 }
114                 else
115                 {
116                     left_child = new wt_node_internal(left, count_left, l+1, c, bmb);
117                     left = 0; // Already deleted
118                 }
119         } else {
120                 left_child = NULL;
121         }
122         if(count_right>0) {
123                 if(match_right/* && c->done(right[0],l+1)*/)
124                 {
125                     right_child = new wt_node_leaf((uint)right[0], count_right);
126                     delete [] right;
127                     right = 0;
128                 }
129                 else 
130                 {
131                     right_child = new wt_node_internal(right, count_right, l+1, c, bmb);
132                     right = 0; // Already deleted
133                 }
134         } else {
135                 right_child = NULL;
136         }
137 //      delete [] left; // already deleted
138 //      delete [] right;
139 }
140
141
142 wt_node_internal::wt_node_internal() { }
143
144 wt_node_internal::~wt_node_internal() {
145         delete bitmap;
146         if(right_child!=NULL) delete right_child;
147         if(left_child!=NULL) delete left_child;
148 }
149
150 uint wt_node_internal::rank(uint symbol, uint pos, uint l, wt_coder * c) {
151         bool is_set = c->is_set(symbol,l);
152         if(!is_set) {
153                 if(left_child==NULL) return 0;
154                 return left_child->rank(symbol, bitmap->rank0(pos)-1,l+1,c);
155         }
156         else {
157                 if(right_child==NULL) return 0;
158                 return right_child->rank(symbol, bitmap->rank1(pos)-1,l+1,c);
159         }
160 }
161
162 uint wt_node_internal::select(uint symbol, uint pos, uint l, wt_coder * c) {
163         bool is_set = c->is_set(symbol, l);
164         uint ret = 0;
165         if(!is_set) {
166                 if(left_child==NULL)
167                         return (uint)(-1);
168                 uint new_pos = left_child->select(symbol, pos, l+1,c);
169                 if(new_pos+1==0) return (uint)(-1);
170                 ret = bitmap->select0(new_pos)+1;
171         } else {
172                 if(right_child==NULL)
173                         return (uint)(-1);
174                 uint new_pos = right_child->select(symbol, pos, l+1,c);
175                 if(new_pos+1==0) return (uint)(-1);
176                 ret = bitmap->select1(new_pos)+1;
177         }
178         if(ret==0) return (uint)-1;
179         return ret;
180 }
181
182 uint wt_node_internal::access(uint pos) {
183         bool is_set = bitmap->access(pos);
184         if(!is_set) {
185                 assert(left_child!=NULL);
186                 return left_child->access(bitmap->rank0(pos)-1);
187         } else {
188                 assert(right_child!=NULL);
189                 return right_child->access(bitmap->rank1(pos)-1);
190         }
191 }
192
193 uint wt_node_internal::size() {
194         uint s = bitmap->size()+sizeof(wt_node_internal);
195         if(left_child!=NULL)
196                 s += left_child->size();
197         if(right_child!=NULL)
198                 s += right_child->size();
199         return s;
200 }
201
202 uint wt_node_internal::save(FILE *fp) {
203   uint wr = WT_NODE_INTERNAL_HDR;
204   wr = fwrite(&wr,sizeof(uint),1,fp);
205   if(wr!=1) return 1;
206   if(bitmap->save(fp)) return 1;
207   if(left_child!=NULL) {
208     if(left_child->save(fp)) return 1;
209   } else {
210     wr = WT_NODE_NULL_HDR;
211     wr = fwrite(&wr,sizeof(uint),1,fp);
212     if(wr!=1) return 1;
213   }
214   if(right_child!=NULL) {
215     if(right_child->save(fp)) return 1;
216   } else {
217     wr = WT_NODE_NULL_HDR;
218     wr = fwrite(&wr,sizeof(uint),1,fp);
219     if(wr!=1) return 1;
220   }
221   return 0;
222 }
223
224 wt_node_internal * wt_node_internal::load(FILE *fp) {
225   uint rd;
226   if(fread(&rd,sizeof(uint),1,fp)!=1) return NULL;
227   if(rd!=WT_NODE_INTERNAL_HDR) return NULL;
228   wt_node_internal * ret = new wt_node_internal();
229   ret->bitmap = static_bitsequence::load(fp);
230   ret->left_child = wt_node::load(fp);
231   ret->right_child = wt_node::load(fp);
232   return ret;
233 }