1 |
#define PERL_POLLUTE |
2 |
|
3 |
/* -*- Mode: C -*- |
4 |
* $Basename: WAIT.xs $ |
5 |
* $Revision: 1.6 $ |
6 |
* Author : Ulrich Pfeifer |
7 |
* Created On : Thu Aug 15 18:01:00 1996 |
8 |
* Last Modified By: Ulrich Pfeifer |
9 |
* Last Modified On: Wed Nov 5 17:01:30 1997 |
10 |
* Language : C |
11 |
* Update Count : 106 |
12 |
* Status : Unknown, Use with caution! |
13 |
* |
14 |
* (C) Copyright 1997, Ulrich Pfeifer, all rights reserved. |
15 |
* |
16 |
*/ |
17 |
|
18 |
#ifdef __cplusplus |
19 |
extern "C" { |
20 |
#endif |
21 |
#include "EXTERN.h" |
22 |
#include "perl.h" |
23 |
#include "XSUB.h" |
24 |
#include "ppport.h" |
25 |
#ifdef __cplusplus |
26 |
} |
27 |
#endif |
28 |
|
29 |
#include "soundex.h" |
30 |
#include "stemmer.h" |
31 |
#include "metaphone.h" |
32 |
#include "levenstein.h" |
33 |
|
34 |
static unsigned char *lchars = |
35 |
"abcdefghijklmnopqrstuvwxyzàáâãäåæçèéêëìíîïñòóôõöøùúûüýß"; |
36 |
static unsigned char *uchars = |
37 |
"ABCDEFGHIJKLMNOPQRSTUVWXYZÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÑÒÓÔÕÖØÙÚÛÜÝß"; |
38 |
static char *scodes = |
39 |
"01230120022455012623010202000000 00000000500000 000002"; |
40 |
static char *pcodes = |
41 |
"01230720022455012683070808000000 00000000500000 000008"; |
42 |
static unsigned char tou[256]; |
43 |
static unsigned char tol[256]; |
44 |
static unsigned char tos[256]; |
45 |
static char scd[256]; |
46 |
static char pcd[256]; |
47 |
static unsigned char *nums = "0123456789"; |
48 |
|
49 |
void init_lcuc () |
50 |
{ |
51 |
short i; |
52 |
short l = strlen(lchars); |
53 |
|
54 |
for(i=0;i<256;i++) { |
55 |
tou[i] = (unsigned char)i; |
56 |
tol[i] = (unsigned char)i; |
57 |
tos[i] = (unsigned char)' '; |
58 |
scd[i] = ' '; |
59 |
pcd[i] = ' '; |
60 |
} |
61 |
for (i=0;i<l;i++) { |
62 |
tou[lchars[i]] = uchars[i]; |
63 |
tol[uchars[i]] = lchars[i]; |
64 |
tos[uchars[i]] = uchars[i]; |
65 |
tos[lchars[i]] = lchars[i]; |
66 |
scd[uchars[i]] = scodes[i]; |
67 |
scd[lchars[i]] = scodes[i]; |
68 |
pcd[uchars[i]] = pcodes[i]; |
69 |
pcd[lchars[i]] = pcodes[i]; |
70 |
} |
71 |
for (i=0;i<10;i++) { |
72 |
tos[nums[i]] = nums[i]; |
73 |
} |
74 |
} |
75 |
|
76 |
unsigned char ToUpper(c) |
77 |
unsigned char c; |
78 |
{ |
79 |
return (tou[c]); |
80 |
} |
81 |
|
82 |
unsigned char ToLower(c) |
83 |
unsigned char c; |
84 |
{ |
85 |
return (tol[c]); |
86 |
} |
87 |
|
88 |
#define ToUpper(c) (tou[c]) |
89 |
#define ToLower(c) (tol[c]) |
90 |
#define SoundexLen 4 /* length of a soundex code */ |
91 |
#define SoundexKey "Z000" /* default key for soundex code */ |
92 |
|
93 |
bool IsAlpha(c) |
94 |
unsigned char c; |
95 |
{ |
96 |
return ((bool) scd[c] != ' '); |
97 |
} |
98 |
|
99 |
bool IsVowel(c) |
100 |
unsigned char c; |
101 |
{ |
102 |
return ((bool) scd[c] == '0'); |
103 |
} |
104 |
|
105 |
static char SCode(c) |
106 |
unsigned char c; |
107 |
{ |
108 |
return (scd[c]); |
109 |
} |
110 |
|
111 |
#define IsAlpha(c) (scd[c] != ' ') |
112 |
#define IsVowel(c) (scd[c] == '0') |
113 |
#define SCode(c) (scd[c]) |
114 |
|
115 |
char PCode(c) |
116 |
unsigned char c; |
117 |
{ |
118 |
return (pcd[c]); |
119 |
} |
120 |
|
121 |
void SoundexCode (Name, Key) |
122 |
unsigned char *Name; |
123 |
unsigned char *Key; |
124 |
{ |
125 |
unsigned char LastLetter; |
126 |
int Index; |
127 |
|
128 |
/* set default key */ |
129 |
strcpy(Key, SoundexKey); |
130 |
|
131 |
/* keep first letter */ |
132 |
Key[0] = *Name; |
133 |
LastLetter = *Name; |
134 |
Name++; |
135 |
|
136 |
/* scan rest of string */ |
137 |
for (Index = 1; (Index < SoundexLen) && *Name; Name++) |
138 |
{ |
139 |
/* use only letters */ |
140 |
if (IsAlpha(*Name)) |
141 |
{ |
142 |
/* ignore duplicate successive chars */ |
143 |
if (LastLetter != *Name) |
144 |
{ |
145 |
/* new LastLetter */ |
146 |
LastLetter = *Name; |
147 |
|
148 |
/* ignore letters with code 0 */ |
149 |
if (!IsVowel(*Name) && (SCode(*Name) != 0)) |
150 |
{ |
151 |
Key[Index] = SCode(*Name); |
152 |
Index++; |
153 |
} |
154 |
} |
155 |
} |
156 |
} |
157 |
} |
158 |
|
159 |
static unsigned char * isolc (s, l) |
160 |
unsigned char * s; |
161 |
int l; |
162 |
{ |
163 |
int i; |
164 |
for (i=0;i<l;i++) { |
165 |
s[i] = tol[s[i]]; |
166 |
} |
167 |
return(s); |
168 |
} |
169 |
|
170 |
static unsigned char * isouc (s, l) |
171 |
unsigned char * s; |
172 |
int l; |
173 |
{ |
174 |
int i; |
175 |
|
176 |
for (i=0;i<l;i++) { |
177 |
s[i] = tou[s[i]]; |
178 |
} |
179 |
return(s); |
180 |
} |
181 |
|
182 |
static unsigned char * isotr (s, l) |
183 |
unsigned char * s; |
184 |
int l; |
185 |
{ |
186 |
int i; |
187 |
|
188 |
for (i=0;i<l;i++) { |
189 |
s[i] = tos[s[i]]; |
190 |
} |
191 |
return(s); |
192 |
} |
193 |
|
194 |
MODULE = WAIT PACKAGE = WAIT::Filter |
195 |
|
196 |
PROTOTYPES: ENABLE |
197 |
|
198 |
BOOT: |
199 |
init_lcuc (); |
200 |
|
201 |
|
202 |
char * |
203 |
isolc(word) |
204 |
char * word |
205 |
CODE: |
206 |
{ |
207 |
char *copy; |
208 |
ST(0) = sv_mortalcopy(ST(0)); |
209 |
copy = (char *)SvPV(ST(0),PL_na); |
210 |
(void) isolc(copy, (int)PL_na); |
211 |
} |
212 |
|
213 |
char * |
214 |
isouc(word) |
215 |
char * word |
216 |
CODE: |
217 |
{ |
218 |
char *copy; |
219 |
ST(0) = sv_mortalcopy(ST(0)); |
220 |
copy = (char *)SvPV(ST(0),PL_na); |
221 |
(void) isouc(copy, (int)PL_na); |
222 |
} |
223 |
|
224 |
char * |
225 |
isotr(word) |
226 |
char * word |
227 |
CODE: |
228 |
{ |
229 |
char *copy; |
230 |
ST(0) = sv_mortalcopy(ST(0)); |
231 |
copy = (char *)SvPV(ST(0),PL_na); |
232 |
(void) isotr(copy, (int)PL_na); |
233 |
} |
234 |
|
235 |
char * |
236 |
disolc(word) |
237 |
char * word |
238 |
CODE: |
239 |
{ |
240 |
(void) isolc(word, (int)PL_na); |
241 |
} |
242 |
|
243 |
char * |
244 |
disouc(word) |
245 |
char * word |
246 |
CODE: |
247 |
{ |
248 |
(void) isouc(word, (int)PL_na); |
249 |
} |
250 |
|
251 |
char * |
252 |
disotr(word) |
253 |
char * word |
254 |
CODE: |
255 |
{ |
256 |
(void) isotr(word, (int)PL_na); |
257 |
} |
258 |
|
259 |
char * |
260 |
Soundex(word) |
261 |
char * word |
262 |
CODE: |
263 |
{ |
264 |
char key[5]; |
265 |
Soundex (word, key); |
266 |
ST(0) = sv_newmortal(); |
267 |
sv_setpv((SV *) ST(0), key); |
268 |
} |
269 |
|
270 |
char * |
271 |
Phonix (word) |
272 |
char * word |
273 |
CODE: |
274 |
{ |
275 |
char key[9]; |
276 |
Phonix (word, key); |
277 |
ST(0) = sv_newmortal(); |
278 |
sv_setpv((SV *) ST(0), key); |
279 |
} |
280 |
|
281 |
char * |
282 |
Stem (word) |
283 |
char * word |
284 |
CODE: |
285 |
{ |
286 |
char copy[80]; |
287 |
strncpy(copy, word, 79); |
288 |
if (Stem(copy)) { |
289 |
ST(0) = sv_newmortal(); |
290 |
sv_setpv((SV *) ST(0), copy); |
291 |
} |
292 |
} |
293 |
|
294 |
char * |
295 |
Metaphone (word) |
296 |
char * word |
297 |
CODE: |
298 |
{ |
299 |
char metaph[80]; |
300 |
metaph[0] = '\0'; |
301 |
phonetic(word,metaph,79); |
302 |
ST(0) = sv_newmortal(); |
303 |
sv_setpv((SV *) ST(0), metaph); |
304 |
} |
305 |
|
306 |
void |
307 |
split_pos(ipair) |
308 |
SV * ipair; |
309 |
PPCODE: |
310 |
{ |
311 |
AV * aipair = (AV *) SvRV(ipair); |
312 |
char * word = (char *)SvPV(*av_fetch(aipair, 0, 0),PL_na); |
313 |
int offset = (av_len(aipair)?SvIV(*av_fetch(aipair, 1, 0)):0); |
314 |
char * begin = word; |
315 |
SV * pair[2]; |
316 |
|
317 |
pair[0] = newSV((STRLEN)20); |
318 |
pair[1] = newSV((STRLEN)0); |
319 |
|
320 |
while (*word) { |
321 |
char * start; |
322 |
AV * apair; |
323 |
SV * ref; |
324 |
while (*word && isspace(*word)) word++; |
325 |
if (!*word) break; |
326 |
start = word; |
327 |
while (*word && !isspace(*word)) word++; |
328 |
EXTEND(sp, 1); |
329 |
sv_setpvn(pair[0], start, (STRLEN)(word-start)); |
330 |
sv_setiv(pair[1],offset + start - begin); |
331 |
apair = av_make(2, pair); |
332 |
ref = newRV_inc((SV*) apair); |
333 |
SvREFCNT_dec(apair); |
334 |
PUSHs(sv_2mortal(ref)); |
335 |
} |
336 |
/* free pair */ |
337 |
SvREFCNT_dec(pair[0]); |
338 |
SvREFCNT_dec(pair[1]); |
339 |
} |
340 |
|
341 |
MODULE = WAIT PACKAGE = WAIT::Table |
342 |
int |
343 |
max(left,right=0) |
344 |
int left |
345 |
int right |
346 |
CODE: |
347 |
{ |
348 |
RETVAL = (left>right)?left:right; |
349 |
ST(0) = sv_newmortal(); |
350 |
sv_setiv(ST(0), (IV)RETVAL); |
351 |
} |
352 |
|
353 |
|
354 |
MODULE = WAIT PACKAGE = WAIT::Metric |
355 |
|
356 |
int |
357 |
WLD(word,towards,mode=' ',limit=0) |
358 |
char * word |
359 |
char * towards |
360 |
char mode |
361 |
int limit |