ruby-sfst 0.4.3 → 0.4.4
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +1 -0
- data/COPYING +280 -0
- data/Gemfile +3 -0
- data/Gemfile.lock +54 -0
- data/README.md +1 -1
- data/Rakefile +9 -18
- data/bin/console +7 -0
- data/bin/setup +6 -0
- data/ext/sfst/alphabet.cc +879 -0
- data/ext/sfst/alphabet.h +302 -0
- data/ext/sfst/basic.cc +85 -0
- data/ext/{sfst_machine → sfst}/basic.h +7 -4
- data/ext/sfst/compact.cc +629 -0
- data/ext/sfst/compact.h +100 -0
- data/ext/sfst/determinise.cc +279 -0
- data/ext/{sfst_machine → sfst}/extconf.rb +2 -1
- data/ext/sfst/fst.cc +1150 -0
- data/ext/sfst/fst.h +374 -0
- data/ext/sfst/hopcroft.cc +681 -0
- data/ext/sfst/interface.cc +1921 -0
- data/ext/sfst/interface.h +171 -0
- data/ext/sfst/make-compact.cc +323 -0
- data/ext/{sfst_machine → sfst}/make-compact.h +15 -13
- data/ext/sfst/mem.h +80 -0
- data/ext/sfst/operators.cc +1273 -0
- data/ext/{sfst_machine → sfst}/sfst_machine.cc +89 -78
- data/ext/sfst/sgi.h +72 -0
- data/ext/sfst/utf8.cc +149 -0
- data/ext/{sfst_machine → sfst}/utf8.h +7 -4
- data/lib/sfst.rb +2 -1
- data/lib/sfst/version.rb +1 -1
- data/ruby-sfst.gemspec +23 -23
- metadata +107 -35
- data/ext/sfst_machine/alphabet.cc +0 -812
- data/ext/sfst_machine/alphabet.h +0 -273
- data/ext/sfst_machine/basic.cc +0 -84
- data/ext/sfst_machine/compact.cc +0 -616
- data/ext/sfst_machine/compact.h +0 -98
- data/ext/sfst_machine/determinise.cc +0 -303
- data/ext/sfst_machine/fst.cc +0 -1000
- data/ext/sfst_machine/fst.h +0 -369
- data/ext/sfst_machine/interface.cc +0 -1842
- data/ext/sfst_machine/interface.h +0 -93
- data/ext/sfst_machine/make-compact.cc +0 -327
- data/ext/sfst_machine/mem.h +0 -74
- data/ext/sfst_machine/operators.cc +0 -1131
- data/ext/sfst_machine/sgi.h +0 -44
- data/ext/sfst_machine/utf8.cc +0 -146
- data/test/test_sfst.fst +0 -3
- data/test/test_sfst.rb +0 -114
data/ext/sfst_machine/compact.h
DELETED
@@ -1,98 +0,0 @@
|
|
1
|
-
/*******************************************************************/
|
2
|
-
/* */
|
3
|
-
/* FILE compact.h */
|
4
|
-
/* MODULE compact */
|
5
|
-
/* PROGRAM SFST */
|
6
|
-
/* AUTHOR Helmut Schmid, IMS, University of Stuttgart */
|
7
|
-
/* */
|
8
|
-
/* PURPOSE finite state tools */
|
9
|
-
/* */
|
10
|
-
/*******************************************************************/
|
11
|
-
|
12
|
-
#ifndef _COMPACT_H_
|
13
|
-
#define _COMPACT_H_
|
14
|
-
|
15
|
-
#include "alphabet.h"
|
16
|
-
|
17
|
-
#include <vector>
|
18
|
-
|
19
|
-
typedef std::vector<unsigned int> CAnalysis;
|
20
|
-
|
21
|
-
class CompactTransducer {
|
22
|
-
|
23
|
-
protected:
|
24
|
-
|
25
|
-
// the following data structures are used to store the nodes
|
26
|
-
|
27
|
-
unsigned int number_of_nodes; // number of nodes in the transducer
|
28
|
-
char *finalp; // finalp[i] is 1 if node i is final and 0 otherwise
|
29
|
-
unsigned int *first_arc; // first_arc[i] is the number of the first
|
30
|
-
// arc outgoing from node i
|
31
|
-
|
32
|
-
// the following data structures are used to store the transition arcs
|
33
|
-
|
34
|
-
unsigned int number_of_arcs; // total number of arcs in the transducer
|
35
|
-
Label *label; // the label (character pair) of arc i
|
36
|
-
unsigned int *target_node; // target node of arc i
|
37
|
-
|
38
|
-
// the following data structures are used to store the stochastic parameters
|
39
|
-
float *final_logprob;
|
40
|
-
float *arc_logprob;
|
41
|
-
|
42
|
-
// functions needed to read the transducer from a file
|
43
|
-
|
44
|
-
void read_finalp( FILE *file );
|
45
|
-
void read_first_arcs( FILE *file );
|
46
|
-
void read_target_nodes( FILE *file );
|
47
|
-
void read_labels( FILE *file );
|
48
|
-
void read_probs( FILE *file );
|
49
|
-
|
50
|
-
// functions needed to analyze data with the transducer
|
51
|
-
|
52
|
-
void analyze( unsigned int n, std::vector<Character> &ch, size_t ipos,
|
53
|
-
CAnalysis&, std::vector<CAnalysis>&);
|
54
|
-
|
55
|
-
// function selecting the simplest morphological analysis
|
56
|
-
|
57
|
-
int compute_score( CAnalysis &ana );
|
58
|
-
void disambiguate( std::vector<CAnalysis> &analyses );
|
59
|
-
|
60
|
-
// functions for longest-match analysis of input data
|
61
|
-
|
62
|
-
void longest_match2(unsigned int, char*, int, CAnalysis&, int&, CAnalysis&);
|
63
|
-
|
64
|
-
void convert( CAnalysis &cana, Analysis &ana );
|
65
|
-
|
66
|
-
public:
|
67
|
-
size_t node_count() { return number_of_nodes; };
|
68
|
-
size_t arc_count() { return number_of_arcs; };
|
69
|
-
|
70
|
-
bool both_layers; // print surface and analysis symbols
|
71
|
-
bool simplest_only; // print only the simplest analyses
|
72
|
-
|
73
|
-
Alphabet alphabet; // data structure which maps symbols to numeric codes
|
74
|
-
CompactTransducer(); // dummy constructor
|
75
|
-
CompactTransducer( FILE*, FILE *pfile=NULL ); // reads a (stochastic) transducer
|
76
|
-
~CompactTransducer(); // destroys a transducer
|
77
|
-
|
78
|
-
// the analysis function returns the set of analyses for the string "s"
|
79
|
-
// in the argument "analyses"
|
80
|
-
void analyze_string( char *s, std::vector<CAnalysis > &analyses );
|
81
|
-
|
82
|
-
void compute_probs( std::vector<CAnalysis> &analyses, std::vector<double> &prob );
|
83
|
-
char *print_analysis( CAnalysis &ana );
|
84
|
-
|
85
|
-
// longest-match analysis
|
86
|
-
const char *longest_match( char*& );
|
87
|
-
|
88
|
-
// EM training
|
89
|
-
bool train2( char *s, std::vector<double> &arcfreq, std::vector<double> &finalfreq );
|
90
|
-
bool train( char *s, std::vector<double> &arcfreq, std::vector<double> &finalfreq );
|
91
|
-
void estimate_probs( std::vector<double> &arcfreq, std::vector<double> &finalfreq );
|
92
|
-
|
93
|
-
// robust analysis
|
94
|
-
float robust_analyze_string( char *string, std::vector<CAnalysis> &analyses,
|
95
|
-
float ErrorsAllowed );
|
96
|
-
};
|
97
|
-
|
98
|
-
#endif
|
@@ -1,303 +0,0 @@
|
|
1
|
-
|
2
|
-
/*******************************************************************/
|
3
|
-
/* */
|
4
|
-
/* FILE determinise.C */
|
5
|
-
/* MODULE determinise */
|
6
|
-
/* PROGRAM SFST */
|
7
|
-
/* AUTHOR Helmut Schmid, IMS, University of Stuttgart */
|
8
|
-
/* */
|
9
|
-
/*******************************************************************/
|
10
|
-
|
11
|
-
|
12
|
-
#include "fst.h"
|
13
|
-
|
14
|
-
using std::vector;
|
15
|
-
using std::pair;
|
16
|
-
using std::set;
|
17
|
-
|
18
|
-
/***************** class NodeSet *********************************/
|
19
|
-
|
20
|
-
class NodeSet {
|
21
|
-
// This class is used to store a set of nodes.
|
22
|
-
// Whenever a new node is added, all nodes accessible
|
23
|
-
// through epsilon transitions are added as well.
|
24
|
-
|
25
|
-
private:
|
26
|
-
set<Node*> ht;
|
27
|
-
|
28
|
-
public:
|
29
|
-
typedef set<Node*>::iterator iterator;
|
30
|
-
NodeSet() {};
|
31
|
-
void add( Node* );
|
32
|
-
bool insert(Node *node) {
|
33
|
-
pair<iterator, bool> result = ht.insert(node);
|
34
|
-
return result.second;
|
35
|
-
};
|
36
|
-
iterator begin() const { return ht.begin(); }
|
37
|
-
iterator end() const { return ht.end(); }
|
38
|
-
size_t size() const { return ht.size(); }
|
39
|
-
void clear() { ht.clear(); }
|
40
|
-
};
|
41
|
-
|
42
|
-
|
43
|
-
/***************** class NodeArray *******************************/
|
44
|
-
|
45
|
-
class NodeArray {
|
46
|
-
|
47
|
-
private:
|
48
|
-
size_t sizev;
|
49
|
-
bool final;
|
50
|
-
Node **node;
|
51
|
-
|
52
|
-
public:
|
53
|
-
NodeArray( NodeSet& );
|
54
|
-
~NodeArray() { delete[] node; };
|
55
|
-
size_t size() const { return sizev; }
|
56
|
-
bool is_final() const { return final; };
|
57
|
-
Node* &operator[]( int i ) const { return node[i]; }
|
58
|
-
};
|
59
|
-
|
60
|
-
|
61
|
-
/***************** class Transition ******************************/
|
62
|
-
|
63
|
-
class Transition {
|
64
|
-
public:
|
65
|
-
Label label;
|
66
|
-
NodeArray *nodes;
|
67
|
-
Transition(Label l, NodeArray *na) { label = l; nodes = na; };
|
68
|
-
};
|
69
|
-
|
70
|
-
|
71
|
-
/***************** class NodeMapping ****************************/
|
72
|
-
|
73
|
-
class NodeMapping {
|
74
|
-
// This class is used to map a node set from one transducer
|
75
|
-
// to a single node in another transducer
|
76
|
-
|
77
|
-
private:
|
78
|
-
struct hashf {
|
79
|
-
size_t operator()(const NodeArray *na) const {
|
80
|
-
size_t key=na->size() ^ na->is_final();
|
81
|
-
for( size_t i=0; i<na->size(); i++)
|
82
|
-
key = (key<<1) ^ (size_t)(*na)[i];
|
83
|
-
return key;
|
84
|
-
}
|
85
|
-
};
|
86
|
-
struct equalf {
|
87
|
-
int operator()(const NodeArray *na1, const NodeArray *na2) const {
|
88
|
-
if (na1->size() != na2->size() || na1->is_final() != na2->is_final())
|
89
|
-
return 0;
|
90
|
-
for( size_t i=0; i<na1->size(); i++)
|
91
|
-
if ((*na1)[i] != (*na2)[i])
|
92
|
-
return 0;
|
93
|
-
return 1;
|
94
|
-
}
|
95
|
-
};
|
96
|
-
typedef hash_map<NodeArray*, Node*, hashf, equalf> NodeMap;
|
97
|
-
NodeMap hm;
|
98
|
-
|
99
|
-
public:
|
100
|
-
typedef NodeMap::iterator iterator;
|
101
|
-
~NodeMapping();
|
102
|
-
iterator begin() { return hm.begin(); };
|
103
|
-
iterator end() { return hm.end(); };
|
104
|
-
iterator find( NodeArray *na) { return hm.find( na ); };
|
105
|
-
Node* &operator[]( NodeArray *na ) { return hm.operator[](na); };
|
106
|
-
|
107
|
-
};
|
108
|
-
|
109
|
-
|
110
|
-
/***************** class LabelMapping ****************************/
|
111
|
-
|
112
|
-
class LabelMapping {
|
113
|
-
// This class is used to map a label to a node set
|
114
|
-
|
115
|
-
private:
|
116
|
-
struct hashf {
|
117
|
-
size_t operator()(const Label l) const {
|
118
|
-
return l.lower_char() | (l.upper_char() << 16);
|
119
|
-
}
|
120
|
-
};
|
121
|
-
struct equalf {
|
122
|
-
int operator()(const Label l1, const Label l2) const {
|
123
|
-
return l1==l2;
|
124
|
-
}
|
125
|
-
};
|
126
|
-
typedef hash_map<const Label, NodeSet, hashf, equalf> LabelMap;
|
127
|
-
LabelMap lm;
|
128
|
-
|
129
|
-
public:
|
130
|
-
LabelMapping(): lm(8) {};
|
131
|
-
typedef LabelMap::iterator iterator;
|
132
|
-
iterator begin() { return lm.begin(); };
|
133
|
-
iterator end() { return lm.end(); };
|
134
|
-
size_t size() { return lm.size(); };
|
135
|
-
iterator find( Label l) { return lm.find( l ); };
|
136
|
-
NodeSet &operator[]( const Label l ) { return lm.operator[]( l ); };
|
137
|
-
|
138
|
-
};
|
139
|
-
|
140
|
-
static void determinise_node( NodeArray&, Node*, Transducer*, NodeMapping&, long );
|
141
|
-
|
142
|
-
|
143
|
-
|
144
|
-
/*******************************************************************/
|
145
|
-
/* */
|
146
|
-
/* NodeSet::add */
|
147
|
-
/* */
|
148
|
-
/*******************************************************************/
|
149
|
-
|
150
|
-
void NodeSet::add( Node *node )
|
151
|
-
|
152
|
-
{
|
153
|
-
pair<iterator, bool> result = ht.insert(node);
|
154
|
-
if (result.second) {
|
155
|
-
// new node, add nodes reachable with epsilon transitions
|
156
|
-
for( ArcsIter p(node->arcs(),ArcsIter::eps); p; p++ ) {
|
157
|
-
Arc *arc=p;
|
158
|
-
if (!arc->label().is_epsilon())
|
159
|
-
break;
|
160
|
-
add(arc->target_node());
|
161
|
-
}
|
162
|
-
}
|
163
|
-
}
|
164
|
-
|
165
|
-
|
166
|
-
/*******************************************************************/
|
167
|
-
/* */
|
168
|
-
/* NodeArray::NodeArray */
|
169
|
-
/* */
|
170
|
-
/*******************************************************************/
|
171
|
-
|
172
|
-
NodeArray::NodeArray( NodeSet &ns )
|
173
|
-
|
174
|
-
{
|
175
|
-
sizev = 0;
|
176
|
-
NodeSet::iterator it;
|
177
|
-
|
178
|
-
final = false;
|
179
|
-
node = new Node*[ns.size()];
|
180
|
-
for( it=ns.begin(); it!=ns.end(); it++ ) {
|
181
|
-
Node *nn = *it;
|
182
|
-
if (nn->arcs()->non_epsilon_transition_exists())
|
183
|
-
node[sizev++] = nn;
|
184
|
-
final |= nn->is_final();
|
185
|
-
}
|
186
|
-
std::sort(node, node+sizev);
|
187
|
-
}
|
188
|
-
|
189
|
-
|
190
|
-
/*******************************************************************/
|
191
|
-
/* */
|
192
|
-
/* NodeMapping::~NodeMapping */
|
193
|
-
/* */
|
194
|
-
/*******************************************************************/
|
195
|
-
|
196
|
-
NodeMapping::~NodeMapping()
|
197
|
-
|
198
|
-
{
|
199
|
-
// if we delete NodeArrays without removing them from NodeMapping,
|
200
|
-
// the system will crash when NodeMapping is deleted.
|
201
|
-
for( iterator it=hm.begin(); it!=hm.end(); ) {
|
202
|
-
NodeArray *na=it->first;
|
203
|
-
iterator old = it++;
|
204
|
-
hm.erase(old);
|
205
|
-
delete na;
|
206
|
-
}
|
207
|
-
}
|
208
|
-
|
209
|
-
|
210
|
-
/*******************************************************************/
|
211
|
-
/* */
|
212
|
-
/* compute_transitions */
|
213
|
-
/* */
|
214
|
-
/*******************************************************************/
|
215
|
-
|
216
|
-
static void compute_transitions( NodeArray &na, vector<Transition> &t )
|
217
|
-
|
218
|
-
{
|
219
|
-
LabelMapping lmap;
|
220
|
-
|
221
|
-
// for all nodes in the current set
|
222
|
-
for( size_t i=0; i<na.size(); i++) {
|
223
|
-
Node *n = na[i]; // old node
|
224
|
-
|
225
|
-
// For each non-epsilon transition, add the target node
|
226
|
-
// to the respective node set.
|
227
|
-
for( ArcsIter p(n->arcs(),ArcsIter::non_eps); p; p++ ) {
|
228
|
-
Arc *arc=p;
|
229
|
-
lmap[arc->label()].add(arc->target_node());
|
230
|
-
}
|
231
|
-
}
|
232
|
-
|
233
|
-
t.reserve(lmap.size());
|
234
|
-
for( LabelMapping::iterator it=lmap.begin(); it!=lmap.end(); it++ )
|
235
|
-
t.push_back(Transition(it->first, new NodeArray( it->second )));
|
236
|
-
}
|
237
|
-
|
238
|
-
|
239
|
-
/*******************************************************************/
|
240
|
-
/* */
|
241
|
-
/* determinise_node */
|
242
|
-
/* */
|
243
|
-
/*******************************************************************/
|
244
|
-
|
245
|
-
static void determinise_node( NodeArray &na, Node *node, Transducer *a,
|
246
|
-
NodeMapping &map, long depth )
|
247
|
-
{
|
248
|
-
if (depth > 10000)
|
249
|
-
fprintf(stderr,"\r%ld",depth);
|
250
|
-
node->set_final(na.is_final());
|
251
|
-
|
252
|
-
vector<Transition> t;
|
253
|
-
compute_transitions( na, t );
|
254
|
-
|
255
|
-
for( size_t i=0; i<t.size(); i++ ) {
|
256
|
-
NodeMapping::iterator it=map.find(t[i].nodes);
|
257
|
-
if (it == map.end()) {
|
258
|
-
// new node set
|
259
|
-
Node *target_node = a->new_node();
|
260
|
-
map[t[i].nodes] = target_node;
|
261
|
-
node->add_arc( t[i].label, target_node, a );
|
262
|
-
determinise_node( *t[i].nodes, target_node, a, map, depth+1 );
|
263
|
-
}
|
264
|
-
else {
|
265
|
-
delete t[i].nodes;
|
266
|
-
node->add_arc( t[i].label, it->second, a );
|
267
|
-
}
|
268
|
-
}
|
269
|
-
}
|
270
|
-
|
271
|
-
|
272
|
-
/*******************************************************************/
|
273
|
-
/* */
|
274
|
-
/* Transducer::determinise */
|
275
|
-
/* */
|
276
|
-
/*******************************************************************/
|
277
|
-
|
278
|
-
Transducer &Transducer::determinise()
|
279
|
-
|
280
|
-
{
|
281
|
-
// initialisations
|
282
|
-
NodeMapping map;
|
283
|
-
|
284
|
-
Transducer *a = new Transducer();
|
285
|
-
a->alphabet.copy(alphabet);
|
286
|
-
|
287
|
-
// creation of the initial node set consisting of all nodes
|
288
|
-
// reachable from the start node via epsilon transitions.
|
289
|
-
NodeArray *na;
|
290
|
-
{
|
291
|
-
NodeSet ns;
|
292
|
-
ns.add(root_node());
|
293
|
-
na = new NodeArray(ns);
|
294
|
-
}
|
295
|
-
|
296
|
-
// map the node set to the new root node
|
297
|
-
map[na] = a->root_node();
|
298
|
-
|
299
|
-
// determinise the transducer recursively
|
300
|
-
determinise_node( *na, a->root_node(), a, map, 0);
|
301
|
-
a->deterministic = 1;
|
302
|
-
return *a;
|
303
|
-
}
|
data/ext/sfst_machine/fst.cc
DELETED
@@ -1,1000 +0,0 @@
|
|
1
|
-
|
2
|
-
/*******************************************************************/
|
3
|
-
/* */
|
4
|
-
/* FILE fst.C */
|
5
|
-
/* MODULE fst */
|
6
|
-
/* PROGRAM SFST */
|
7
|
-
/* AUTHOR Helmut Schmid, IMS, University of Stuttgart */
|
8
|
-
/* */
|
9
|
-
/* PURPOSE basic FST functions */
|
10
|
-
/* */
|
11
|
-
/*******************************************************************/
|
12
|
-
|
13
|
-
#include "fst.h"
|
14
|
-
|
15
|
-
using std::vector;
|
16
|
-
using std::istream;
|
17
|
-
using std::ostream;
|
18
|
-
using std::cerr;
|
19
|
-
|
20
|
-
const int BUFFER_SIZE=100000;
|
21
|
-
|
22
|
-
|
23
|
-
/*******************************************************************/
|
24
|
-
/* */
|
25
|
-
/* Arcs::size */
|
26
|
-
/* */
|
27
|
-
/*******************************************************************/
|
28
|
-
|
29
|
-
int Arcs::size() const
|
30
|
-
|
31
|
-
{
|
32
|
-
int n=0;
|
33
|
-
for( Arc *p=first_arcp; p; p=p->next ) n++;
|
34
|
-
for( Arc *p=first_epsilon_arcp; p; p=p->next ) n++;
|
35
|
-
return n;
|
36
|
-
}
|
37
|
-
|
38
|
-
|
39
|
-
/*******************************************************************/
|
40
|
-
/* */
|
41
|
-
/* Arcs::target_node */
|
42
|
-
/* */
|
43
|
-
/*******************************************************************/
|
44
|
-
|
45
|
-
Node *Arcs::target_node( Label l )
|
46
|
-
|
47
|
-
{
|
48
|
-
Arc *arc;
|
49
|
-
|
50
|
-
for( arc=first_arcp; arc; arc=arc->next)
|
51
|
-
if (arc->label() == l)
|
52
|
-
return arc->target_node();
|
53
|
-
|
54
|
-
return NULL;
|
55
|
-
}
|
56
|
-
|
57
|
-
const Node *Arcs::target_node( Label l ) const
|
58
|
-
|
59
|
-
{
|
60
|
-
const Arc *arc;
|
61
|
-
|
62
|
-
for( arc=first_arcp; arc; arc=arc->next)
|
63
|
-
if (arc->label() == l)
|
64
|
-
return arc->target_node();
|
65
|
-
|
66
|
-
return NULL;
|
67
|
-
}
|
68
|
-
|
69
|
-
|
70
|
-
/*******************************************************************/
|
71
|
-
/* */
|
72
|
-
/* Arcs::add_arc */
|
73
|
-
/* */
|
74
|
-
/*******************************************************************/
|
75
|
-
|
76
|
-
void Arcs::add_arc( Label l, Node *node, Transducer *a )
|
77
|
-
|
78
|
-
{
|
79
|
-
Arc *arc=a->new_arc( l, node );
|
80
|
-
|
81
|
-
if (l.is_epsilon()) {
|
82
|
-
arc->next = first_epsilon_arcp;
|
83
|
-
first_epsilon_arcp = arc;
|
84
|
-
}
|
85
|
-
else {
|
86
|
-
arc->next = first_arcp;
|
87
|
-
first_arcp = arc;
|
88
|
-
}
|
89
|
-
}
|
90
|
-
|
91
|
-
|
92
|
-
/*******************************************************************/
|
93
|
-
/* */
|
94
|
-
/* Arcs::remove_arc */
|
95
|
-
/* */
|
96
|
-
/*******************************************************************/
|
97
|
-
|
98
|
-
int Arcs::remove_arc( Arc *arc )
|
99
|
-
|
100
|
-
{
|
101
|
-
Arc **p = (arc->label().is_epsilon()) ? &first_epsilon_arcp : &first_arcp;
|
102
|
-
for( ; *p; p=&(*p)->next )
|
103
|
-
if (*p == arc) {
|
104
|
-
*p = arc->next;
|
105
|
-
return 1;
|
106
|
-
}
|
107
|
-
return 0;
|
108
|
-
}
|
109
|
-
|
110
|
-
|
111
|
-
/*******************************************************************/
|
112
|
-
/* */
|
113
|
-
/* Node::init */
|
114
|
-
/* */
|
115
|
-
/*******************************************************************/
|
116
|
-
|
117
|
-
void Node::init()
|
118
|
-
|
119
|
-
{
|
120
|
-
final = false;
|
121
|
-
visited = 0;
|
122
|
-
arcsp.init();
|
123
|
-
forwardp = NULL;
|
124
|
-
}
|
125
|
-
|
126
|
-
|
127
|
-
/*******************************************************************/
|
128
|
-
/* */
|
129
|
-
/* Node::clear_visited */
|
130
|
-
/* */
|
131
|
-
/*******************************************************************/
|
132
|
-
|
133
|
-
void Node::clear_visited( NodeHashSet &nodeset )
|
134
|
-
|
135
|
-
{
|
136
|
-
if (nodeset.find( this ) == nodeset.end()) {
|
137
|
-
visited = 0;
|
138
|
-
nodeset.insert( this );
|
139
|
-
fprintf(stderr," %lu", nodeset.size());
|
140
|
-
for( ArcsIter p(arcs()); p; p++ ) {
|
141
|
-
Arc *arc=p;
|
142
|
-
arc->target_node()->clear_visited( nodeset );
|
143
|
-
}
|
144
|
-
}
|
145
|
-
}
|
146
|
-
|
147
|
-
|
148
|
-
/*******************************************************************/
|
149
|
-
/* */
|
150
|
-
/* NodeNumbering::number_node */
|
151
|
-
/* */
|
152
|
-
/*******************************************************************/
|
153
|
-
|
154
|
-
void NodeNumbering::number_node( Node *node, Transducer &a )
|
155
|
-
|
156
|
-
{
|
157
|
-
if (!node->was_visited( a.vmark )) {
|
158
|
-
nummap[node] = nodes.size();
|
159
|
-
nodes.push_back(node);
|
160
|
-
for( ArcsIter p(node->arcs()); p; p++ ) {
|
161
|
-
Arc *arc=p;
|
162
|
-
number_node( arc->target_node(), a );
|
163
|
-
}
|
164
|
-
}
|
165
|
-
}
|
166
|
-
|
167
|
-
|
168
|
-
/*******************************************************************/
|
169
|
-
/* */
|
170
|
-
/* NodeNumbering::NodeNumbering */
|
171
|
-
/* */
|
172
|
-
/*******************************************************************/
|
173
|
-
|
174
|
-
NodeNumbering::NodeNumbering( Transducer &a )
|
175
|
-
|
176
|
-
{
|
177
|
-
a.incr_vmark();
|
178
|
-
number_node( a.root_node(), a );
|
179
|
-
}
|
180
|
-
|
181
|
-
|
182
|
-
/*******************************************************************/
|
183
|
-
/* */
|
184
|
-
/* Transducer::new_node */
|
185
|
-
/* */
|
186
|
-
/*******************************************************************/
|
187
|
-
|
188
|
-
Node *Transducer::new_node()
|
189
|
-
|
190
|
-
{
|
191
|
-
Node *node=(Node*)mem.alloc( sizeof(Node) );
|
192
|
-
|
193
|
-
node->init();
|
194
|
-
return node;
|
195
|
-
}
|
196
|
-
|
197
|
-
|
198
|
-
/*******************************************************************/
|
199
|
-
/* */
|
200
|
-
/* Transducer::new_arc */
|
201
|
-
/* */
|
202
|
-
/*******************************************************************/
|
203
|
-
|
204
|
-
Arc *Transducer::new_arc( Label l, Node *target )
|
205
|
-
|
206
|
-
{
|
207
|
-
Arc *arc=(Arc*)mem.alloc( sizeof(Arc) );
|
208
|
-
|
209
|
-
arc->init( l, target);
|
210
|
-
return arc;
|
211
|
-
}
|
212
|
-
|
213
|
-
|
214
|
-
/*******************************************************************/
|
215
|
-
/* */
|
216
|
-
/* Transducer::add_string */
|
217
|
-
/* */
|
218
|
-
/*******************************************************************/
|
219
|
-
|
220
|
-
void Transducer::add_string( char *s, bool extended, Alphabet *a )
|
221
|
-
|
222
|
-
{
|
223
|
-
if (a == NULL)
|
224
|
-
a = &alphabet;
|
225
|
-
|
226
|
-
Node *node=root_node();
|
227
|
-
Label l;
|
228
|
-
while (!(l = a->next_label(s, extended)).is_epsilon()) {
|
229
|
-
a->insert(l);
|
230
|
-
Arcs *arcs=node->arcs();
|
231
|
-
node = arcs->target_node( l );
|
232
|
-
if (node == NULL) {
|
233
|
-
node = new_node();
|
234
|
-
arcs->add_arc( l, node, this );
|
235
|
-
}
|
236
|
-
}
|
237
|
-
node->set_final(1);
|
238
|
-
}
|
239
|
-
|
240
|
-
|
241
|
-
/*******************************************************************/
|
242
|
-
/* */
|
243
|
-
/* Transducer::Transducer */
|
244
|
-
/* */
|
245
|
-
/*******************************************************************/
|
246
|
-
|
247
|
-
Transducer::Transducer( vector<Label> &path )
|
248
|
-
: root(), mem()
|
249
|
-
{
|
250
|
-
Node *node=root_node();
|
251
|
-
|
252
|
-
vmark = 0;
|
253
|
-
deterministic = minimised = true;
|
254
|
-
for( size_t i=0; i<path.size(); i++ ) {
|
255
|
-
Arcs *arcs=node->arcs();
|
256
|
-
node = new_node();
|
257
|
-
arcs->add_arc( path[i], node, this );
|
258
|
-
}
|
259
|
-
node->set_final(1);
|
260
|
-
}
|
261
|
-
|
262
|
-
|
263
|
-
/*******************************************************************/
|
264
|
-
/* */
|
265
|
-
/* Transducer::Transducer */
|
266
|
-
/* */
|
267
|
-
/*******************************************************************/
|
268
|
-
|
269
|
-
Transducer::Transducer( istream &is, const Alphabet *a, bool verbose )
|
270
|
-
: root(), mem()
|
271
|
-
{
|
272
|
-
bool extended=false;
|
273
|
-
int n=0;
|
274
|
-
char buffer[10000];
|
275
|
-
|
276
|
-
vmark = 0;
|
277
|
-
deterministic = true;
|
278
|
-
minimised = false;
|
279
|
-
if (a) {
|
280
|
-
alphabet.copy(*a);
|
281
|
-
extended = true;
|
282
|
-
}
|
283
|
-
while (is.getline(buffer, 10000)) {
|
284
|
-
if (verbose && ++n % 10000 == 0) {
|
285
|
-
if (n == 10000)
|
286
|
-
cerr << "\n";
|
287
|
-
cerr << "\r" << n << " words";
|
288
|
-
}
|
289
|
-
// delete final whitespace characters
|
290
|
-
int l;
|
291
|
-
for( l=strlen(buffer)-1; l>=0; l-- )
|
292
|
-
if ((buffer[l] != ' ' && buffer[l] != '\t' && buffer[l] != '\r') ||
|
293
|
-
(l > 0 && buffer[l-1] == '\\'))
|
294
|
-
break;
|
295
|
-
buffer[l+1] = 0;
|
296
|
-
|
297
|
-
add_string(buffer, extended);
|
298
|
-
}
|
299
|
-
if (verbose && n >= 10000)
|
300
|
-
cerr << "\n";
|
301
|
-
}
|
302
|
-
|
303
|
-
|
304
|
-
/*******************************************************************/
|
305
|
-
/* */
|
306
|
-
/* Transducer::Transducer */
|
307
|
-
/* */
|
308
|
-
/*******************************************************************/
|
309
|
-
|
310
|
-
Transducer::Transducer( char *s, const Alphabet *a, bool extended )
|
311
|
-
: root(), mem()
|
312
|
-
{
|
313
|
-
vmark = 0;
|
314
|
-
deterministic = minimised = true;
|
315
|
-
if (a)
|
316
|
-
alphabet.copy(*a);
|
317
|
-
add_string(s, extended);
|
318
|
-
}
|
319
|
-
|
320
|
-
|
321
|
-
/*******************************************************************/
|
322
|
-
/* */
|
323
|
-
/* Transducer::clear */
|
324
|
-
/* */
|
325
|
-
/*******************************************************************/
|
326
|
-
|
327
|
-
void Transducer::clear()
|
328
|
-
|
329
|
-
{
|
330
|
-
vmark = 0;
|
331
|
-
deterministic = minimised = false;
|
332
|
-
root.init();
|
333
|
-
mem.clear();
|
334
|
-
alphabet.clear();
|
335
|
-
}
|
336
|
-
|
337
|
-
|
338
|
-
/*******************************************************************/
|
339
|
-
/* */
|
340
|
-
/* Transducer::store_symbols */
|
341
|
-
/* */
|
342
|
-
/*******************************************************************/
|
343
|
-
|
344
|
-
void Transducer::store_symbols(Node *node, SymbolMap &symbol, LabelSet &labels)
|
345
|
-
|
346
|
-
{
|
347
|
-
if (!node->was_visited( vmark )) {
|
348
|
-
Arcs *arcs=node->arcs();
|
349
|
-
for( ArcsIter p(arcs); p; p++ ) {
|
350
|
-
Arc *arc=p;
|
351
|
-
Label l=arc->label();
|
352
|
-
|
353
|
-
labels.insert(l);
|
354
|
-
|
355
|
-
Character c = l.upper_char();
|
356
|
-
if (symbol.find(c) == symbol.end()) {
|
357
|
-
const char *s = alphabet.code2symbol(c);
|
358
|
-
if (s)
|
359
|
-
symbol[c] = fst_strdup(s);
|
360
|
-
}
|
361
|
-
|
362
|
-
c = l.lower_char();
|
363
|
-
if (symbol.find(c) == symbol.end()) {
|
364
|
-
const char *s = alphabet.code2symbol(c);
|
365
|
-
if (s)
|
366
|
-
symbol[c] = fst_strdup(s);
|
367
|
-
}
|
368
|
-
|
369
|
-
store_symbols( arc->target_node(), symbol, labels );
|
370
|
-
}
|
371
|
-
}
|
372
|
-
}
|
373
|
-
|
374
|
-
|
375
|
-
/*******************************************************************/
|
376
|
-
/* */
|
377
|
-
/* Transducer::minimise_alphabet */
|
378
|
-
/* */
|
379
|
-
/*******************************************************************/
|
380
|
-
|
381
|
-
void Transducer::minimise_alphabet()
|
382
|
-
|
383
|
-
{
|
384
|
-
SymbolMap symbols;
|
385
|
-
LabelSet labels;
|
386
|
-
incr_vmark();
|
387
|
-
store_symbols(root_node(), symbols, labels);
|
388
|
-
alphabet.clear();
|
389
|
-
for( SymbolMap::iterator it=symbols.begin(); it!=symbols.end(); it++ ) {
|
390
|
-
alphabet.add_symbol( it->second, it->first );
|
391
|
-
free(it->second);
|
392
|
-
}
|
393
|
-
for( LabelSet::iterator it=labels.begin(); it!=labels.end(); it++ )
|
394
|
-
alphabet.insert(*it);
|
395
|
-
}
|
396
|
-
|
397
|
-
|
398
|
-
/*******************************************************************/
|
399
|
-
/* */
|
400
|
-
/* Transducer::minimise */
|
401
|
-
/* */
|
402
|
-
/*******************************************************************/
|
403
|
-
|
404
|
-
Transducer &Transducer::minimise( bool verbose )
|
405
|
-
|
406
|
-
{
|
407
|
-
if (minimised)
|
408
|
-
return copy();
|
409
|
-
|
410
|
-
Transducer *a1, *a2;
|
411
|
-
|
412
|
-
a1 = &reverse();
|
413
|
-
a2 = &a1->determinise();
|
414
|
-
delete a1;
|
415
|
-
|
416
|
-
a1 = &a2->reverse();
|
417
|
-
delete a2;
|
418
|
-
|
419
|
-
a2 = &a1->determinise();
|
420
|
-
delete a1;
|
421
|
-
|
422
|
-
a2->minimised = true;
|
423
|
-
a2->minimise_alphabet();
|
424
|
-
|
425
|
-
return *a2;
|
426
|
-
}
|
427
|
-
|
428
|
-
|
429
|
-
/*******************************************************************/
|
430
|
-
/* */
|
431
|
-
/* Transducer::enumerate_paths_node */
|
432
|
-
/* */
|
433
|
-
/*******************************************************************/
|
434
|
-
|
435
|
-
void Transducer::enumerate_paths_node( Node *node, vector<Label> &path,
|
436
|
-
NodeHashSet &previous,
|
437
|
-
vector<Transducer*> &result )
|
438
|
-
{
|
439
|
-
if (node->is_final())
|
440
|
-
result.push_back(new Transducer(path));
|
441
|
-
|
442
|
-
for( ArcsIter it_arc(node->arcs()); it_arc; it_arc++ ) {
|
443
|
-
Arc *arc=it_arc;
|
444
|
-
|
445
|
-
NodeHashSet::iterator it_node=previous.insert(node).first;
|
446
|
-
path.push_back(arc->label());
|
447
|
-
enumerate_paths_node( arc->target_node(), path, previous, result );
|
448
|
-
path.pop_back();
|
449
|
-
previous.erase(it_node);
|
450
|
-
}
|
451
|
-
}
|
452
|
-
|
453
|
-
|
454
|
-
/*******************************************************************/
|
455
|
-
/* */
|
456
|
-
/* Transducer::enumerate_paths */
|
457
|
-
/* */
|
458
|
-
/*******************************************************************/
|
459
|
-
|
460
|
-
bool Transducer::enumerate_paths( vector<Transducer*> &result )
|
461
|
-
|
462
|
-
{
|
463
|
-
if (is_infinitely_ambiguous())
|
464
|
-
return true;
|
465
|
-
for( size_t i=0; i<result.size(); i++ )
|
466
|
-
delete result[i];
|
467
|
-
result.clear();
|
468
|
-
|
469
|
-
vector<Label> path;
|
470
|
-
NodeHashSet previous;
|
471
|
-
enumerate_paths_node( root_node(), path, previous, result );
|
472
|
-
return false;
|
473
|
-
}
|
474
|
-
|
475
|
-
|
476
|
-
|
477
|
-
|
478
|
-
/*******************************************************************/
|
479
|
-
/* */
|
480
|
-
/* Transducer::print_strings_node */
|
481
|
-
/* */
|
482
|
-
/*******************************************************************/
|
483
|
-
|
484
|
-
int Transducer::print_strings_node(Node *node, char *buffer, int pos,
|
485
|
-
FILE *file, bool with_brackets )
|
486
|
-
{
|
487
|
-
int result = 0;
|
488
|
-
|
489
|
-
if (node->was_visited( vmark )) {
|
490
|
-
if (node->forward() != NULL) { // cycle detected
|
491
|
-
cerr << "Warning: cyclic analyses (cycle aborted)\n";
|
492
|
-
return 0;
|
493
|
-
}
|
494
|
-
node->set_forward(node); // used like a flag for loop detection
|
495
|
-
}
|
496
|
-
if (pos == BUFFER_SIZE)
|
497
|
-
throw "Output string in function print_strings_node is too long";
|
498
|
-
if (node->is_final()) {
|
499
|
-
buffer[pos] = '\0';
|
500
|
-
fprintf(file,"%s\n", buffer);
|
501
|
-
result = 1;
|
502
|
-
}
|
503
|
-
for( ArcsIter i(node->arcs()); i; i++ ) {
|
504
|
-
int p=pos;
|
505
|
-
Arc *arc=i;
|
506
|
-
Label l=arc->label();
|
507
|
-
alphabet.write_label(l, buffer, &p, with_brackets);
|
508
|
-
result |= print_strings_node(arc->target_node(), buffer, p,
|
509
|
-
file, with_brackets );
|
510
|
-
}
|
511
|
-
node->set_forward(NULL);
|
512
|
-
|
513
|
-
return result;
|
514
|
-
}
|
515
|
-
|
516
|
-
|
517
|
-
/*******************************************************************/
|
518
|
-
/* */
|
519
|
-
/* Transducer::print_strings */
|
520
|
-
/* */
|
521
|
-
/*******************************************************************/
|
522
|
-
|
523
|
-
int Transducer::print_strings( FILE *file, bool with_brackets )
|
524
|
-
|
525
|
-
{
|
526
|
-
char buffer[BUFFER_SIZE];
|
527
|
-
incr_vmark();
|
528
|
-
return print_strings_node( root_node(), buffer, 0, file, with_brackets );
|
529
|
-
}
|
530
|
-
|
531
|
-
|
532
|
-
/*******************************************************************/
|
533
|
-
/* */
|
534
|
-
/* Transducer::analyze_string */
|
535
|
-
/* */
|
536
|
-
/*******************************************************************/
|
537
|
-
|
538
|
-
bool Transducer::analyze_string( char *string, FILE *file, bool with_brackets )
|
539
|
-
|
540
|
-
{
|
541
|
-
vector<Character> input;
|
542
|
-
alphabet.string2symseq( string, input );
|
543
|
-
vector<Label> labels;
|
544
|
-
for( size_t i=0; i<input.size(); i++ )
|
545
|
-
labels.push_back(Label(input[i]));
|
546
|
-
|
547
|
-
Transducer a1(labels);
|
548
|
-
Transducer *a2=&(*this || a1);
|
549
|
-
Transducer *a3=&(a2->lower_level());
|
550
|
-
delete a2;
|
551
|
-
a2 = &a3->minimise();
|
552
|
-
delete a3;
|
553
|
-
|
554
|
-
a2->alphabet.copy(alphabet);
|
555
|
-
bool result = a2->print_strings( file, with_brackets );
|
556
|
-
delete a2;
|
557
|
-
return result;
|
558
|
-
}
|
559
|
-
|
560
|
-
|
561
|
-
/*******************************************************************/
|
562
|
-
/* */
|
563
|
-
/* Transducer::generate_string */
|
564
|
-
/* */
|
565
|
-
/*******************************************************************/
|
566
|
-
|
567
|
-
bool Transducer::generate_string( char *string, FILE *file, bool with_brackets)
|
568
|
-
|
569
|
-
{
|
570
|
-
Transducer a1(string, &alphabet, false);
|
571
|
-
Transducer *a2=&(a1 || *this);
|
572
|
-
Transducer *a3=&(a2->upper_level());
|
573
|
-
delete a2;
|
574
|
-
a2 = &a3->minimise();
|
575
|
-
delete a3;
|
576
|
-
|
577
|
-
a2->alphabet.copy(alphabet);
|
578
|
-
bool result = a2->print_strings( file, with_brackets );
|
579
|
-
delete a2;
|
580
|
-
return result;
|
581
|
-
}
|
582
|
-
|
583
|
-
|
584
|
-
/*******************************************************************/
|
585
|
-
/* */
|
586
|
-
/* complete */
|
587
|
-
/* */
|
588
|
-
/*******************************************************************/
|
589
|
-
|
590
|
-
static void complete( Node *node, Alphabet &alphabet, int vmark)
|
591
|
-
|
592
|
-
{
|
593
|
-
if (node->was_visited( vmark ))
|
594
|
-
return;
|
595
|
-
for( ArcsIter p(node->arcs()); p; p++ ) {
|
596
|
-
Arc *arc=p;
|
597
|
-
if (!arc->label().is_epsilon())
|
598
|
-
alphabet.insert(arc->label());
|
599
|
-
complete(arc->target_node(), alphabet, vmark);
|
600
|
-
}
|
601
|
-
}
|
602
|
-
|
603
|
-
|
604
|
-
/*******************************************************************/
|
605
|
-
/* */
|
606
|
-
/* Transducer::complete_alphabet */
|
607
|
-
/* */
|
608
|
-
/*******************************************************************/
|
609
|
-
|
610
|
-
void Transducer::complete_alphabet()
|
611
|
-
|
612
|
-
{
|
613
|
-
incr_vmark();
|
614
|
-
complete(root_node(), alphabet, vmark);
|
615
|
-
}
|
616
|
-
|
617
|
-
|
618
|
-
/*******************************************************************/
|
619
|
-
/* */
|
620
|
-
/* print_node */
|
621
|
-
/* */
|
622
|
-
/*******************************************************************/
|
623
|
-
|
624
|
-
static void print_node( ostream &s, Node *node, NodeNumbering &index,
|
625
|
-
long vmark, Alphabet &abc )
|
626
|
-
|
627
|
-
{
|
628
|
-
if (!node->was_visited( vmark )) {
|
629
|
-
Arcs *arcs=node->arcs();
|
630
|
-
for( ArcsIter p(arcs); p; p++ ) {
|
631
|
-
Arc *arc=p;
|
632
|
-
s << index[node] << "\t" << index[arc->target_node()];
|
633
|
-
s << "\t" << abc.write_char(arc->label().lower_char());
|
634
|
-
s << "\t" << abc.write_char(arc->label().upper_char());
|
635
|
-
s << "\n";
|
636
|
-
}
|
637
|
-
if (node->is_final())
|
638
|
-
s << index[node] << "\n";
|
639
|
-
for( ArcsIter p(arcs); p; p++ ) {
|
640
|
-
Arc *arc=p;
|
641
|
-
print_node( s, arc->target_node(), index, vmark, abc );
|
642
|
-
}
|
643
|
-
}
|
644
|
-
}
|
645
|
-
|
646
|
-
|
647
|
-
/*******************************************************************/
|
648
|
-
/* */
|
649
|
-
/* operator<< */
|
650
|
-
/* */
|
651
|
-
/*******************************************************************/
|
652
|
-
|
653
|
-
ostream &operator<<( ostream &s, Transducer &a )
|
654
|
-
|
655
|
-
{
|
656
|
-
NodeNumbering index(a);
|
657
|
-
a.incr_vmark();
|
658
|
-
print_node( s, a.root_node(), index, a.vmark, a.alphabet );
|
659
|
-
return s;
|
660
|
-
}
|
661
|
-
|
662
|
-
|
663
|
-
/*******************************************************************/
|
664
|
-
/* */
|
665
|
-
/* store_node_info */
|
666
|
-
/* */
|
667
|
-
/*******************************************************************/
|
668
|
-
|
669
|
-
static void store_node_info( FILE *file, Node *node )
|
670
|
-
|
671
|
-
{
|
672
|
-
// write final flag
|
673
|
-
char c=node->is_final();
|
674
|
-
fwrite(&c,sizeof(c),1,file);
|
675
|
-
|
676
|
-
// write the number of arcs
|
677
|
-
int nn = node->arcs()->size();
|
678
|
-
if (nn > 65535)
|
679
|
-
throw "Error: in function store_node\n";
|
680
|
-
unsigned short n=(unsigned short)nn;
|
681
|
-
fwrite(&n,sizeof(n),1,file);
|
682
|
-
}
|
683
|
-
|
684
|
-
|
685
|
-
/*******************************************************************/
|
686
|
-
/* */
|
687
|
-
/* store_arc_label */
|
688
|
-
/* */
|
689
|
-
/*******************************************************************/
|
690
|
-
|
691
|
-
static void store_arc_label( FILE *file, Arc *arc )
|
692
|
-
|
693
|
-
{
|
694
|
-
Label l=arc->label();
|
695
|
-
Character lc=l.lower_char();
|
696
|
-
Character uc=l.upper_char();
|
697
|
-
fwrite(&lc,sizeof(lc),1,file);
|
698
|
-
fwrite(&uc,sizeof(uc),1,file);
|
699
|
-
}
|
700
|
-
|
701
|
-
|
702
|
-
/*******************************************************************/
|
703
|
-
/* */
|
704
|
-
/* store_node */
|
705
|
-
/* */
|
706
|
-
/*******************************************************************/
|
707
|
-
|
708
|
-
static void store_node( FILE *file, Node *node, NodeNumbering &index,
|
709
|
-
long vmark )
|
710
|
-
{
|
711
|
-
if (!node->was_visited( vmark )) {
|
712
|
-
|
713
|
-
store_node_info( file, node );
|
714
|
-
|
715
|
-
// write the arcs
|
716
|
-
for( ArcsIter p(node->arcs()); p; p++ ) {
|
717
|
-
Arc *arc=p;
|
718
|
-
store_arc_label( file, arc );
|
719
|
-
unsigned int t=index[arc->target_node()];
|
720
|
-
fwrite(&t,sizeof(t),1,file);
|
721
|
-
store_node(file, arc->target_node(), index, vmark );
|
722
|
-
}
|
723
|
-
}
|
724
|
-
}
|
725
|
-
|
726
|
-
|
727
|
-
/*******************************************************************/
|
728
|
-
/* */
|
729
|
-
/* store_lowmem_node */
|
730
|
-
/* */
|
731
|
-
/*******************************************************************/
|
732
|
-
|
733
|
-
static void store_lowmem_node( FILE *file, Node *node, NodeNumbering &index,
|
734
|
-
vector<unsigned int> &startpos)
|
735
|
-
{
|
736
|
-
store_node_info( file, node );
|
737
|
-
|
738
|
-
// write the arcs
|
739
|
-
for( ArcsIter p(node->arcs()); p; p++ ) {
|
740
|
-
Arc *arc=p;
|
741
|
-
store_arc_label( file, arc );
|
742
|
-
unsigned int t=startpos[index[arc->target_node()]];
|
743
|
-
fwrite(&t,sizeof(t),1,file);
|
744
|
-
}
|
745
|
-
}
|
746
|
-
|
747
|
-
|
748
|
-
/*******************************************************************/
|
749
|
-
/* */
|
750
|
-
/* Transducer::store_lowmem */
|
751
|
-
/* */
|
752
|
-
/*******************************************************************/
|
753
|
-
|
754
|
-
void Transducer::store_lowmem( FILE *file )
|
755
|
-
|
756
|
-
{
|
757
|
-
fputc('l',file);
|
758
|
-
alphabet.store(file);
|
759
|
-
|
760
|
-
// storing size of index table
|
761
|
-
NodeNumbering index(*this);
|
762
|
-
|
763
|
-
// compute the start position of the first node
|
764
|
-
unsigned int pos=(unsigned int)ftell(file);
|
765
|
-
vector<unsigned int> startpos;
|
766
|
-
for( size_t i=0; i<index.number_of_nodes(); i++ ) {
|
767
|
-
startpos.push_back(pos);
|
768
|
-
Node *node=index.get_node(i);
|
769
|
-
Arcs *arcs=node->arcs();
|
770
|
-
pos += sizeof(char) // size of final flag
|
771
|
-
+ sizeof(unsigned short) // size of number of arcs
|
772
|
-
+ arcs->size() * (sizeof(Character) * 2 + sizeof(unsigned int)); // size of n arcs
|
773
|
-
}
|
774
|
-
|
775
|
-
// storing nodes
|
776
|
-
for( size_t i=0; i<index.number_of_nodes(); i++ )
|
777
|
-
store_lowmem_node( file, index.get_node(i), index, startpos );
|
778
|
-
}
|
779
|
-
|
780
|
-
|
781
|
-
/*******************************************************************/
|
782
|
-
/* */
|
783
|
-
/* Transducer::store */
|
784
|
-
/* */
|
785
|
-
/*******************************************************************/
|
786
|
-
|
787
|
-
void Transducer::store( FILE *file )
|
788
|
-
|
789
|
-
{
|
790
|
-
fputc('a',file);
|
791
|
-
|
792
|
-
NodeNumbering index(*this);
|
793
|
-
incr_vmark();
|
794
|
-
unsigned int n=index.number_of_nodes();
|
795
|
-
fwrite(&n,sizeof(n),1,file);
|
796
|
-
store_node( file, root_node(), index, vmark );
|
797
|
-
|
798
|
-
alphabet.store(file);
|
799
|
-
}
|
800
|
-
|
801
|
-
|
802
|
-
/*******************************************************************/
|
803
|
-
/* */
|
804
|
-
/* read_node */
|
805
|
-
/* */
|
806
|
-
/*******************************************************************/
|
807
|
-
|
808
|
-
static void read_node( FILE *file, Node *node, Node **p, Transducer *a )
|
809
|
-
{
|
810
|
-
char c;
|
811
|
-
fread(&c,sizeof(c),1,file);
|
812
|
-
node->set_final(c);
|
813
|
-
|
814
|
-
unsigned short n;
|
815
|
-
fread( &n, sizeof(n), 1, file);
|
816
|
-
|
817
|
-
for( int i=0; i<n; i++ ) {
|
818
|
-
Character lc,uc;
|
819
|
-
unsigned int t;
|
820
|
-
fread(&lc,sizeof(lc),1,file);
|
821
|
-
fread(&uc,sizeof(uc),1,file);
|
822
|
-
fread(&t,sizeof(t),1,file);
|
823
|
-
if (ferror(file))
|
824
|
-
throw "Error encountered while reading transducer from file";
|
825
|
-
if (p[t])
|
826
|
-
node->add_arc( Label(lc,uc), p[t], a );
|
827
|
-
else {
|
828
|
-
p[t] = a->new_node();
|
829
|
-
node->add_arc( Label(lc,uc), p[t], a );
|
830
|
-
read_node(file, p[t], p, a );
|
831
|
-
}
|
832
|
-
}
|
833
|
-
}
|
834
|
-
|
835
|
-
|
836
|
-
/*******************************************************************/
|
837
|
-
/* */
|
838
|
-
/* Transducer::read_transducer_binary */
|
839
|
-
/* */
|
840
|
-
/*******************************************************************/
|
841
|
-
|
842
|
-
void Transducer::read_transducer_binary( FILE *file )
|
843
|
-
|
844
|
-
{
|
845
|
-
if (fgetc(file) != 'a')
|
846
|
-
throw "Error: wrong file format (not a standard transducer)\n";
|
847
|
-
|
848
|
-
vmark = deterministic = 0;
|
849
|
-
unsigned int n;
|
850
|
-
fread(&n,sizeof(n),1,file); // number of nodes
|
851
|
-
if (ferror(file))
|
852
|
-
throw "Error encountered while reading transducer from file";
|
853
|
-
|
854
|
-
Node **p=new Node*[n]; // maps indices to nodes
|
855
|
-
p[0] = root_node();
|
856
|
-
for( unsigned int i=1; i<n; i++)
|
857
|
-
p[i] = NULL;
|
858
|
-
read_node( file, root_node(), p, this );
|
859
|
-
delete[] p;
|
860
|
-
|
861
|
-
alphabet.read(file);
|
862
|
-
|
863
|
-
vmark = 1;
|
864
|
-
deterministic = minimised = 1;
|
865
|
-
}
|
866
|
-
|
867
|
-
|
868
|
-
/*******************************************************************/
|
869
|
-
/* */
|
870
|
-
/* error_message */
|
871
|
-
/* */
|
872
|
-
/*******************************************************************/
|
873
|
-
|
874
|
-
static void error_message( size_t line )
|
875
|
-
|
876
|
-
{
|
877
|
-
static char message[1000];
|
878
|
-
sprintf(message, "Error: in line %u of text transducer file",
|
879
|
-
(unsigned int)line);
|
880
|
-
throw message;
|
881
|
-
}
|
882
|
-
|
883
|
-
|
884
|
-
/*******************************************************************/
|
885
|
-
/* */
|
886
|
-
/* Transducer::create_node */
|
887
|
-
/* */
|
888
|
-
/*******************************************************************/
|
889
|
-
|
890
|
-
Node *Transducer::create_node( vector<Node*> &node, char *s, size_t line )
|
891
|
-
|
892
|
-
{
|
893
|
-
char *p;
|
894
|
-
long n = strtol(s, &p, 10);
|
895
|
-
|
896
|
-
if (s == p || n < 0)
|
897
|
-
error_message( line );
|
898
|
-
if ((long)node.size() <= n)
|
899
|
-
node.resize(n+1, NULL);
|
900
|
-
if (node[n] == NULL)
|
901
|
-
node[n] = new Node;
|
902
|
-
|
903
|
-
return node[n];
|
904
|
-
}
|
905
|
-
|
906
|
-
|
907
|
-
/*******************************************************************/
|
908
|
-
/* */
|
909
|
-
/* next_string */
|
910
|
-
/* */
|
911
|
-
/*******************************************************************/
|
912
|
-
|
913
|
-
static char *next_string( char* &s, size_t line )
|
914
|
-
|
915
|
-
{
|
916
|
-
// scan the input up to the next tab or newline character
|
917
|
-
// and unquote symbols preceded by a backslash
|
918
|
-
char *p = s;
|
919
|
-
char *q = s;
|
920
|
-
while (*q!=0 && *q!='\t' && *q!='\n' && *q!='\r') {
|
921
|
-
if (*q == '\\')
|
922
|
-
q++;
|
923
|
-
*(p++) = *(q++);
|
924
|
-
}
|
925
|
-
if (p == s)
|
926
|
-
error_message(line); // no string found
|
927
|
-
|
928
|
-
char *result=s;
|
929
|
-
// skip over following whitespace
|
930
|
-
while (*q == ' ' || *q == '\t' || *q == '\n' || *q == '\r')
|
931
|
-
q++;
|
932
|
-
|
933
|
-
if (*q == 0)
|
934
|
-
s = NULL; // end of string was reached
|
935
|
-
else
|
936
|
-
s = q; // move the string pointer s
|
937
|
-
|
938
|
-
*p = 0; // mark the end of the result string
|
939
|
-
|
940
|
-
return result;
|
941
|
-
}
|
942
|
-
|
943
|
-
|
944
|
-
/*******************************************************************/
|
945
|
-
/* */
|
946
|
-
/* Transducer::read_transducer_text */
|
947
|
-
/* */
|
948
|
-
/*******************************************************************/
|
949
|
-
|
950
|
-
void Transducer::read_transducer_text( FILE *file )
|
951
|
-
|
952
|
-
{
|
953
|
-
vector<Node*> nodes;
|
954
|
-
nodes.push_back(root_node());
|
955
|
-
|
956
|
-
vmark = deterministic = 0;
|
957
|
-
char buffer[10000];
|
958
|
-
for( size_t line=0; fgets(buffer, 10000, file ); line++ ) {
|
959
|
-
char *p = buffer;
|
960
|
-
char *s = next_string(p, line);
|
961
|
-
Node *node = create_node( nodes, s, line );
|
962
|
-
if (p == NULL)
|
963
|
-
node->set_final(true);
|
964
|
-
else {
|
965
|
-
s = next_string(p, line);
|
966
|
-
Node *target = create_node( nodes, s, line );
|
967
|
-
|
968
|
-
s = next_string(p, line);
|
969
|
-
Character lc = alphabet.add_symbol(s);
|
970
|
-
s = next_string(p, line);
|
971
|
-
Character uc = alphabet.add_symbol(s);
|
972
|
-
Label l(lc,uc);
|
973
|
-
if (l == Label::epsilon)
|
974
|
-
error_message( line );
|
975
|
-
|
976
|
-
alphabet.insert(l);
|
977
|
-
node->add_arc( l, target, this );
|
978
|
-
}
|
979
|
-
}
|
980
|
-
|
981
|
-
vmark = 1;
|
982
|
-
deterministic = minimised = 1;
|
983
|
-
}
|
984
|
-
|
985
|
-
|
986
|
-
/*******************************************************************/
|
987
|
-
/* */
|
988
|
-
/* Transducer::Transducer */
|
989
|
-
/* */
|
990
|
-
/*******************************************************************/
|
991
|
-
|
992
|
-
Transducer::Transducer( FILE *file, bool binary )
|
993
|
-
|
994
|
-
{
|
995
|
-
if (binary)
|
996
|
-
read_transducer_binary( file );
|
997
|
-
else
|
998
|
-
read_transducer_text( file );
|
999
|
-
}
|
1000
|
-
|