New (faster) representation for tags added; faster construction of parentheses
[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 wt_node_internal::wt_node_internal() { }
74
75 wt_node_internal::~wt_node_internal() {
76         delete bitmap;
77         if(right_child!=NULL) delete right_child;
78         if(left_child!=NULL) delete left_child;
79 }
80
81 uint wt_node_internal::rank(uint symbol, uint pos, uint l, wt_coder * c) {
82         bool is_set = c->is_set(symbol,l);
83         if(!is_set) {
84                 if(left_child==NULL) return 0;
85                 return left_child->rank(symbol, bitmap->rank0(pos)-1,l+1,c);
86         }
87         else {
88                 if(right_child==NULL) return 0;
89                 return right_child->rank(symbol, bitmap->rank1(pos)-1,l+1,c);
90         }
91 }
92
93 uint wt_node_internal::select(uint symbol, uint pos, uint l, wt_coder * c) {
94         bool is_set = c->is_set(symbol, l);
95         if(!is_set) {
96                 if(left_child==NULL)
97                         return (uint)(-1);
98                 uint new_pos = left_child->select(symbol, pos, l+1,c);
99                 if(new_pos+1==0) return (uint)(-1);
100                 return bitmap->select0(new_pos)+1;
101         } else {
102                 if(right_child==NULL)
103                         return (uint)(-1);
104                 uint new_pos = right_child->select(symbol, pos, l+1,c);
105                 if(new_pos+1==0) return (uint)(-1);
106                 return bitmap->select1(new_pos)+1;
107         }
108 }
109
110 uint wt_node_internal::access(uint pos) {
111         bool is_set = bitmap->access(pos);
112         if(!is_set) {
113                 assert(left_child!=NULL);
114                 return left_child->access(bitmap->rank0(pos)-1);
115         } else {
116                 assert(right_child!=NULL);
117                 return right_child->access(bitmap->rank1(pos)-1);
118         }
119 }
120
121 uint wt_node_internal::size() {
122         uint s = bitmap->size()+sizeof(wt_node_internal);
123         if(left_child!=NULL)
124                 s += left_child->size();
125         if(right_child!=NULL)
126                 s += right_child->size();
127         return s;
128 }
129
130 uint wt_node_internal::save(FILE *fp) {
131   uint wr = WT_NODE_INTERNAL_HDR;
132   wr = fwrite(&wr,sizeof(uint),1,fp);
133   if(wr!=1) return 1;
134   if(bitmap->save(fp)) return 1;
135   if(left_child!=NULL) {
136     if(left_child->save(fp)) return 1;
137   } else {
138     wr = WT_NODE_NULL_HDR;
139     wr = fwrite(&wr,sizeof(uint),1,fp);
140     if(wr!=1) return 1;
141   }
142   if(right_child!=NULL) {
143     if(right_child->save(fp)) return 1;
144   } else {
145     wr = WT_NODE_NULL_HDR;
146     wr = fwrite(&wr,sizeof(uint),1,fp);
147     if(wr!=1) return 1;
148   }
149   return 0;
150 }
151
152 wt_node_internal * wt_node_internal::load(FILE *fp) {
153   uint rd;
154   if(fread(&rd,sizeof(uint),1,fp)!=1) return NULL;
155   if(rd!=WT_NODE_INTERNAL_HDR) return NULL;
156   wt_node_internal * ret = new wt_node_internal();
157   ret->bitmap = static_bitsequence::load(fp);
158   ret->left_child = wt_node::load(fp);
159   ret->right_child = wt_node::load(fp);
160   return ret;
161 }