ruby-sfst 0.1.0
Sign up to get free protection for your applications and to get access to all the features.
- data/CHANGELOG +1 -0
- data/Manifest +31 -0
- data/README.rdoc +25 -0
- data/Rakefile +22 -0
- data/ext/sfst_machine/alphabet.C +807 -0
- data/ext/sfst_machine/alphabet.h +281 -0
- data/ext/sfst_machine/basic.C +84 -0
- data/ext/sfst_machine/basic.h +24 -0
- data/ext/sfst_machine/compact.C +616 -0
- data/ext/sfst_machine/compact.h +98 -0
- data/ext/sfst_machine/determinise.C +304 -0
- data/ext/sfst_machine/extconf.rb +4 -0
- data/ext/sfst_machine/fst-compiler.C +2375 -0
- data/ext/sfst_machine/fst-compiler.h +113 -0
- data/ext/sfst_machine/fst-compiler.yy +213 -0
- data/ext/sfst_machine/fst.C +966 -0
- data/ext/sfst_machine/fst.h +365 -0
- data/ext/sfst_machine/interface.C +1838 -0
- data/ext/sfst_machine/interface.h +94 -0
- data/ext/sfst_machine/make-compact.C +328 -0
- data/ext/sfst_machine/make-compact.h +34 -0
- data/ext/sfst_machine/mem.h +74 -0
- data/ext/sfst_machine/operators.C +1131 -0
- data/ext/sfst_machine/sfst_machine.cc +411 -0
- data/ext/sfst_machine/utf8-scanner.C +2197 -0
- data/ext/sfst_machine/utf8-scanner.ll +179 -0
- data/ext/sfst_machine/utf8.C +146 -0
- data/ext/sfst_machine/utf8.h +19 -0
- data/lib/sfst.rb +99 -0
- data/ruby-sfst.gemspec +34 -0
- data/test/test_sfst.fst +3 -0
- data/test/test_sfst.rb +119 -0
- metadata +100 -0
@@ -0,0 +1,281 @@
|
|
1
|
+
/*******************************************************************/
|
2
|
+
/* */
|
3
|
+
/* FILE alphabet.h */
|
4
|
+
/* MODULE alphabet */
|
5
|
+
/* PROGRAM SFST */
|
6
|
+
/* AUTHOR Helmut Schmid, IMS, University of Stuttgart */
|
7
|
+
/* */
|
8
|
+
/* PURPOSE finite state tools */
|
9
|
+
/* */
|
10
|
+
/*******************************************************************/
|
11
|
+
|
12
|
+
#ifndef _ALPHABET_H_
|
13
|
+
#define _ALPHABET_H_
|
14
|
+
|
15
|
+
#include <stdio.h>
|
16
|
+
|
17
|
+
#include "basic.h"
|
18
|
+
|
19
|
+
#include <set>
|
20
|
+
#include <vector>
|
21
|
+
|
22
|
+
#include <iostream>
|
23
|
+
|
24
|
+
#ifndef CODE_DATA_TYPE
|
25
|
+
typedef unsigned short Character; // data type of the symbol codes
|
26
|
+
#else
|
27
|
+
typedef unsigned CODE_DATA_TYPE Character;
|
28
|
+
#endif
|
29
|
+
|
30
|
+
// data type used to indicate whether some action is to be performed
|
31
|
+
// on the analysis level (lower) or the surface level (upper)
|
32
|
+
typedef enum {upper, lower} Level;
|
33
|
+
|
34
|
+
#ifdef SGIext
|
35
|
+
|
36
|
+
#include <ext/hash_set>
|
37
|
+
#include <ext/hash_map>
|
38
|
+
|
39
|
+
#else
|
40
|
+
|
41
|
+
#include <hash_set>
|
42
|
+
#include <hash_map>
|
43
|
+
|
44
|
+
#endif
|
45
|
+
|
46
|
+
extern char EpsilonString[]; // holds the symbol representing the empty string
|
47
|
+
// which is usually "<>"
|
48
|
+
|
49
|
+
|
50
|
+
/***************** class Label ***********************************/
|
51
|
+
|
52
|
+
class Label {
|
53
|
+
|
54
|
+
private:
|
55
|
+
// data structure where the two symbols are stored
|
56
|
+
struct {
|
57
|
+
Character lower;
|
58
|
+
Character upper;
|
59
|
+
} label;
|
60
|
+
|
61
|
+
public:
|
62
|
+
static const Character epsilon=0; // code of the empty symbol
|
63
|
+
|
64
|
+
// new label with two identical symbols
|
65
|
+
Label( Character c=epsilon ) { label.lower = label.upper = c; };
|
66
|
+
|
67
|
+
// new label with two different symbols
|
68
|
+
Label( Character c1, Character c2 )
|
69
|
+
{ label.lower = c1; label.upper = c2; };
|
70
|
+
|
71
|
+
// returns the indicated symbol of the label
|
72
|
+
Character get_char( Level l ) const
|
73
|
+
{ return ((l==upper)? label.upper: label.lower); };
|
74
|
+
|
75
|
+
// returns the "upper" symbol of the label (i.e. the surface symbol)
|
76
|
+
Character upper_char() const { return label.upper; };
|
77
|
+
|
78
|
+
// returns the "lower" symbol of the label (i.e. the analysis symbol)
|
79
|
+
Character lower_char() const { return label.lower; };
|
80
|
+
|
81
|
+
// replaces symbols in a label
|
82
|
+
Label replace_char( Character c, Character nc ) const {
|
83
|
+
Label l = *this;
|
84
|
+
if (l.label.lower == c)
|
85
|
+
l.label.lower = nc;
|
86
|
+
if (l.label.upper == c)
|
87
|
+
l.label.upper = nc;
|
88
|
+
return l;
|
89
|
+
};
|
90
|
+
|
91
|
+
// operators checking the equality of labels
|
92
|
+
int operator==( Label l ) const
|
93
|
+
{ return (label.lower==l.label.lower && label.upper==l.label.upper); };
|
94
|
+
int operator!=( Label l ) const
|
95
|
+
{ return !(l == *this); };
|
96
|
+
|
97
|
+
// comparison operator needed for sorting labels
|
98
|
+
int operator<( Label l ) const {
|
99
|
+
return (upper_char() < l.upper_char()); };
|
100
|
+
|
101
|
+
// check whether the label is epsilon (i.e. both symbols are epsilon)
|
102
|
+
// transitions with epsilon labels are epsilon transitions
|
103
|
+
int is_epsilon() const
|
104
|
+
{ return (label.upper == epsilon && label.lower == epsilon); };
|
105
|
+
|
106
|
+
// check whether the "upper" symbol is epsilon
|
107
|
+
int upper_is_epsilon() const
|
108
|
+
{ return (label.upper == epsilon); };
|
109
|
+
|
110
|
+
// check whether the "lower" symbol is epsilon
|
111
|
+
int lower_is_epsilon() const
|
112
|
+
{ return (label.lower == epsilon); };
|
113
|
+
|
114
|
+
// hash function needed to store labels in a hash table
|
115
|
+
struct label_hash {
|
116
|
+
size_t operator() ( const Label l ) const {
|
117
|
+
return (size_t)l.lower_char() ^
|
118
|
+
((size_t)l.upper_char() << 16) ^
|
119
|
+
((size_t)l.upper_char() >> 16);
|
120
|
+
}
|
121
|
+
};
|
122
|
+
|
123
|
+
// hash function needed to store labels in a hash table
|
124
|
+
struct label_cmp {
|
125
|
+
bool operator() ( const Label l1, const Label l2 ) const {
|
126
|
+
return (l1.lower_char() < l2.lower_char() ||
|
127
|
+
(l1.lower_char() == l2.lower_char() &&
|
128
|
+
l1.upper_char() < l2.upper_char()));
|
129
|
+
}
|
130
|
+
};
|
131
|
+
|
132
|
+
// comparison operator needed to store labels in a hash table
|
133
|
+
struct label_eq {
|
134
|
+
bool operator() ( const Label l1, const Label l2 ) const {
|
135
|
+
return (l1.lower_char() == l2.lower_char() &&
|
136
|
+
l1.upper_char() == l2.upper_char());
|
137
|
+
}
|
138
|
+
};
|
139
|
+
};
|
140
|
+
|
141
|
+
typedef std::vector<Label> Analysis;
|
142
|
+
|
143
|
+
|
144
|
+
/***************** class Alphabet *******************************/
|
145
|
+
|
146
|
+
class Alphabet {
|
147
|
+
|
148
|
+
// string comparison operators needed to stored strings in a hash table
|
149
|
+
struct eqstr {
|
150
|
+
bool operator()(const char* s1, const char* s2) const {
|
151
|
+
return strcmp(s1, s2) == 0;
|
152
|
+
}
|
153
|
+
};
|
154
|
+
|
155
|
+
// data structure storing labels without repetitions (i.e. as a set)
|
156
|
+
typedef std::set<Label, Label::label_cmp> LabelSet;
|
157
|
+
|
158
|
+
// hash table used to map the symbols to their codes
|
159
|
+
typedef __gnu_cxx::hash_map<const char*, Character, __gnu_cxx::hash<const char*>,eqstr> SymbolMap;
|
160
|
+
|
161
|
+
// hash table used to map the codes back to the symbols
|
162
|
+
typedef __gnu_cxx::hash_map<Character, char*> CharMap;
|
163
|
+
|
164
|
+
private:
|
165
|
+
SymbolMap sm; // maps symbols to codes
|
166
|
+
|
167
|
+
CharMap cm; // maps codes to symbols
|
168
|
+
LabelSet ls; // set of labels known to the alphabet
|
169
|
+
|
170
|
+
// add a new symbol with symbol code c
|
171
|
+
void add( const char *symbol, Character c );
|
172
|
+
|
173
|
+
public:
|
174
|
+
bool utf8;
|
175
|
+
|
176
|
+
// iterators over the set of known labels
|
177
|
+
typedef LabelSet::iterator iterator;
|
178
|
+
typedef LabelSet::const_iterator const_iterator;
|
179
|
+
Alphabet();
|
180
|
+
~Alphabet() { clear(); };
|
181
|
+
const_iterator begin() const { return ls.begin(); };
|
182
|
+
const_iterator end() const { return ls.end(); };
|
183
|
+
size_t size() const { return ls.size(); };
|
184
|
+
|
185
|
+
void clear();
|
186
|
+
void clear_char_pairs() { ls.clear(); };
|
187
|
+
|
188
|
+
// lookup a label in the alphabet
|
189
|
+
iterator find( Label l ) { return ls.find(l); };
|
190
|
+
|
191
|
+
// insert a label in the alphabet
|
192
|
+
void insert( Label l ) { if (!l.is_epsilon()) ls.insert(l); };
|
193
|
+
|
194
|
+
// insert the known symbols from another alphabet
|
195
|
+
void insert_symbols( const Alphabet& );
|
196
|
+
|
197
|
+
// insert the labels and known symbols from another alphabet
|
198
|
+
void copy( const Alphabet& );
|
199
|
+
|
200
|
+
// create the alphabet of a transducer obtained by a composition operation
|
201
|
+
void compose( const Alphabet &la, const Alphabet &ua );
|
202
|
+
|
203
|
+
// add a symbol to the alphabet and return its code
|
204
|
+
Character add_symbol(const char *symbol);
|
205
|
+
|
206
|
+
// add a symbol to the alphabet with a given code
|
207
|
+
void add_symbol(const char *symbol, Character c );
|
208
|
+
|
209
|
+
// create a new marker symbol and return its code
|
210
|
+
Character new_marker( void );
|
211
|
+
void delete_markers();
|
212
|
+
|
213
|
+
// compute the complement of a symbol set
|
214
|
+
void complement( std::vector<Character> &sym );
|
215
|
+
|
216
|
+
// return the code of the argument symbol
|
217
|
+
int symbol2code( const char *s ) const {
|
218
|
+
SymbolMap::const_iterator p = sm.find(s);
|
219
|
+
if (p != sm.end()) return p->second;
|
220
|
+
return EOF;
|
221
|
+
};
|
222
|
+
|
223
|
+
// return the symbol for the given symbol code
|
224
|
+
const char *code2symbol( Character c ) const {
|
225
|
+
CharMap::const_iterator p=cm.find(c);
|
226
|
+
if (p == cm.end())
|
227
|
+
return NULL;
|
228
|
+
else
|
229
|
+
return p->second;
|
230
|
+
};
|
231
|
+
|
232
|
+
// write the symbol for the given symbol code into a string
|
233
|
+
void write_char( Character c, char *buffer, int *pos,
|
234
|
+
bool with_brackets=true ) const;
|
235
|
+
|
236
|
+
// write the symbol pair of a given label into a string
|
237
|
+
void write_label( Label l, char *buffer, int *pos,
|
238
|
+
bool with_brackets=true ) const;
|
239
|
+
|
240
|
+
// write the symbol for the given symbol code into a buffer and return
|
241
|
+
// a pointer to it
|
242
|
+
// the flag "with_brackets" indicates whether the angle brackets
|
243
|
+
// surrounding multi-character symbols are to be printed or not
|
244
|
+
const char *write_char( Character c, bool with_brackets=true ) const;
|
245
|
+
|
246
|
+
// write the symbol pair of a given label into a string
|
247
|
+
// and return a pointer to it
|
248
|
+
const char *write_label( Label l, bool with_brackets=true ) const;
|
249
|
+
|
250
|
+
// scan the next multi-character symbol in the argument string
|
251
|
+
int next_mcsym( char*&, int extended=1 );
|
252
|
+
|
253
|
+
// scan the next symbol in the argument string
|
254
|
+
int next_code( char*&, int extended=1 );
|
255
|
+
|
256
|
+
// convert a character string into a symbol or label sequence
|
257
|
+
void string2symseq( char*, std::vector<Character>& );
|
258
|
+
void string2labelseq( char*, std::vector<Label>& );
|
259
|
+
|
260
|
+
// scan the next label in the argument string
|
261
|
+
Label next_label( char*&, int extended=1 );
|
262
|
+
|
263
|
+
// store the alphabet in the argument file (in binary form)
|
264
|
+
void store( FILE* ) const;
|
265
|
+
|
266
|
+
// read the alphabet from the argument file
|
267
|
+
void read( FILE* );
|
268
|
+
|
269
|
+
// disambiguation and printing of analyses
|
270
|
+
int compute_score( Analysis &ana );
|
271
|
+
void disambiguate( std::vector<Analysis> &analyses );
|
272
|
+
char *print_analysis( Analysis &ana, bool both_layers );
|
273
|
+
|
274
|
+
friend std::ostream &operator<<(std::ostream&, const Alphabet&);
|
275
|
+
};
|
276
|
+
|
277
|
+
// write the alphabet to the output stream (in readable form)
|
278
|
+
std::ostream &operator<<(std::ostream&, Alphabet&);
|
279
|
+
|
280
|
+
|
281
|
+
#endif
|
@@ -0,0 +1,84 @@
|
|
1
|
+
|
2
|
+
/*******************************************************************/
|
3
|
+
/* */
|
4
|
+
/* FILE basic.C */
|
5
|
+
/* MODULE basic */
|
6
|
+
/* PROGRAM SFST */
|
7
|
+
/* AUTHOR Helmut Schmid, IMS, University of Stuttgart */
|
8
|
+
/* */
|
9
|
+
/* PURPOSE */
|
10
|
+
/* */
|
11
|
+
/*******************************************************************/
|
12
|
+
|
13
|
+
#include <stdlib.h>
|
14
|
+
#include <string.h>
|
15
|
+
|
16
|
+
#include "basic.h"
|
17
|
+
|
18
|
+
bool Switch_Bytes=false;
|
19
|
+
|
20
|
+
|
21
|
+
|
22
|
+
/*******************************************************************/
|
23
|
+
/* */
|
24
|
+
/* fst_strdup */
|
25
|
+
/* */
|
26
|
+
/*******************************************************************/
|
27
|
+
|
28
|
+
char* fst_strdup(const char* pString)
|
29
|
+
|
30
|
+
{
|
31
|
+
char* pStringCopy = (char*)malloc(strlen(pString) + 1);
|
32
|
+
if (pStringCopy == NULL) {
|
33
|
+
fprintf(stderr, "\nError: out of memory (malloc failed)\naborted.\n");
|
34
|
+
exit(1);
|
35
|
+
}
|
36
|
+
strcpy(pStringCopy, pString);
|
37
|
+
return pStringCopy;
|
38
|
+
}
|
39
|
+
|
40
|
+
|
41
|
+
/*******************************************************************/
|
42
|
+
/* */
|
43
|
+
/* read_string */
|
44
|
+
/* */
|
45
|
+
/*******************************************************************/
|
46
|
+
|
47
|
+
int read_string( char *buffer, int size, FILE *file )
|
48
|
+
|
49
|
+
{
|
50
|
+
for( int i=0; i<size; i++ ) {
|
51
|
+
int c=fgetc(file);
|
52
|
+
if (c == EOF || c == 0) {
|
53
|
+
buffer[i] = 0;
|
54
|
+
return (c==0);
|
55
|
+
}
|
56
|
+
buffer[i] = (char)c;
|
57
|
+
}
|
58
|
+
buffer[size-1] = 0;
|
59
|
+
return 0;
|
60
|
+
}
|
61
|
+
|
62
|
+
|
63
|
+
/*******************************************************************/
|
64
|
+
/* */
|
65
|
+
/* read_num */
|
66
|
+
/* */
|
67
|
+
/*******************************************************************/
|
68
|
+
|
69
|
+
size_t read_num( void *p, size_t n, FILE *file )
|
70
|
+
|
71
|
+
{
|
72
|
+
char *pp=(char*)p;
|
73
|
+
size_t result=fread( pp, 1, n, file );
|
74
|
+
if (Switch_Bytes) {
|
75
|
+
size_t e=n/2;
|
76
|
+
for( size_t i=0; i<e; i++ ) {
|
77
|
+
char tmp=pp[i];
|
78
|
+
pp[i] = pp[--n];
|
79
|
+
pp[n] = tmp;
|
80
|
+
}
|
81
|
+
}
|
82
|
+
return result;
|
83
|
+
}
|
84
|
+
|
@@ -0,0 +1,24 @@
|
|
1
|
+
|
2
|
+
/*******************************************************************/
|
3
|
+
/* */
|
4
|
+
/* FILE basic.h */
|
5
|
+
/* MODULE basic */
|
6
|
+
/* PROGRAM SFST */
|
7
|
+
/* AUTHOR Helmut Schmid, IMS, University of Stuttgart */
|
8
|
+
/* */
|
9
|
+
/* PURPOSE */
|
10
|
+
/* */
|
11
|
+
/*******************************************************************/
|
12
|
+
|
13
|
+
#ifndef _BASIC_H_
|
14
|
+
#define _BASIC_H_
|
15
|
+
|
16
|
+
#include <stdio.h>
|
17
|
+
|
18
|
+
extern bool Switch_Bytes;
|
19
|
+
|
20
|
+
char* fst_strdup(const char* pString);
|
21
|
+
int read_string( char *buffer, int size, FILE *file );
|
22
|
+
size_t read_num( void *p, size_t size, FILE *file );
|
23
|
+
|
24
|
+
#endif
|