Age Owner TLA Line data Source code
1 : /*
2 : * fuzzystrmatch.c
3 : *
4 : * Functions for "fuzzy" comparison of strings
5 : *
6 : * Joe Conway <mail@joeconway.com>
7 : *
8 : * contrib/fuzzystrmatch/fuzzystrmatch.c
9 : * Copyright (c) 2001-2023, PostgreSQL Global Development Group
10 : * ALL RIGHTS RESERVED;
11 : *
12 : * metaphone()
13 : * -----------
14 : * Modified for PostgreSQL by Joe Conway.
15 : * Based on CPAN's "Text-Metaphone-1.96" by Michael G Schwern <schwern@pobox.com>
16 : * Code slightly modified for use as PostgreSQL function (palloc, elog, etc).
17 : * Metaphone was originally created by Lawrence Philips and presented in article
18 : * in "Computer Language" December 1990 issue.
19 : *
20 : * Permission to use, copy, modify, and distribute this software and its
21 : * documentation for any purpose, without fee, and without a written agreement
22 : * is hereby granted, provided that the above copyright notice and this
23 : * paragraph and the following two paragraphs appear in all copies.
24 : *
25 : * IN NO EVENT SHALL THE AUTHORS OR DISTRIBUTORS BE LIABLE TO ANY PARTY FOR
26 : * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES, INCLUDING
27 : * LOST PROFITS, ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS
28 : * DOCUMENTATION, EVEN IF THE AUTHOR OR DISTRIBUTORS HAVE BEEN ADVISED OF THE
29 : * POSSIBILITY OF SUCH DAMAGE.
30 : *
31 : * THE AUTHORS AND DISTRIBUTORS SPECIFICALLY DISCLAIM ANY WARRANTIES,
32 : * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY
33 : * AND FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS
34 : * ON AN "AS IS" BASIS, AND THE AUTHOR AND DISTRIBUTORS HAS NO OBLIGATIONS TO
35 : * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
36 : *
37 : */
38 :
39 : #include "postgres.h"
40 :
41 : #include <ctype.h>
42 :
43 : #include "mb/pg_wchar.h"
44 : #include "utils/builtins.h"
45 : #include "utils/varlena.h"
46 : #include "varatt.h"
47 :
6158 tgl 48 GIC 2 : PG_MODULE_MAGIC;
6158 tgl 49 ECB :
50 : /*
51 : * Soundex
52 : */
53 : static void _soundex(const char *instr, char *outstr);
54 :
55 : #define SOUNDEX_LEN 4
56 :
57 : /* ABCDEFGHIJKLMNOPQRSTUVWXYZ */
58 : static const char *soundex_table = "01230120022455012623010202";
59 :
60 : static char
5115 tgl 61 GIC 127 : soundex_code(char letter)
5115 tgl 62 ECB : {
5115 tgl 63 GIC 127 : letter = toupper((unsigned char) letter);
5115 tgl 64 ECB : /* Defend against non-ASCII letters */
5115 tgl 65 GIC 127 : if (letter >= 'A' && letter <= 'Z')
5115 tgl 66 CBC 126 : return soundex_table[letter - 'A'];
67 1 : return letter;
5115 tgl 68 ECB : }
69 :
70 : /*
71 : * Metaphone
72 : */
73 : #define MAX_METAPHONE_STRLEN 255
74 :
75 : /*
76 : * Original code by Michael G Schwern starts here.
77 : * Code slightly modified for use as PostgreSQL function.
78 : */
79 :
80 :
81 : /**************************************************************************
82 : metaphone -- Breaks english phrases down into their phonemes.
83 :
84 : Input
85 : word -- An english word to be phonized
86 : max_phonemes -- How many phonemes to calculate. If 0, then it
87 : will phonize the entire phrase.
88 : phoned_word -- The final phonized word. (We'll allocate the
89 : memory.)
90 : Output
91 : error -- A simple error flag, returns true or false
92 :
93 : NOTES: ALL non-alpha characters are ignored, this includes whitespace,
94 : although non-alpha characters will break up phonemes.
95 : ****************************************************************************/
96 :
97 :
98 : /* I add modifications to the traditional metaphone algorithm that you
99 : might find in books. Define this if you want metaphone to behave
100 : traditionally */
101 : #undef USE_TRADITIONAL_METAPHONE
102 :
103 : /* Special encodings */
104 : #define SH 'X'
105 : #define TH '0'
106 :
107 : static char Lookahead(char *word, int how_far);
108 : static void _metaphone(char *word, int max_phonemes, char **phoned_word);
109 :
110 : /* Metachar.h ... little bits about characters for metaphone */
111 :
112 :
113 : /*-- Character encoding array & accessing macros --*/
114 : /* Stolen directly out of the book... */
115 : static const char _codes[26] = {
116 : 1, 16, 4, 16, 9, 2, 4, 16, 9, 2, 0, 2, 2, 2, 1, 4, 0, 2, 4, 4, 1, 0, 0, 0, 8, 0
117 : /* a b c d e f g h i j k l m n o p q r s t u v w x y z */
118 : };
119 :
120 : static int
5115 tgl 121 GIC 1 : getcode(char c)
5115 tgl 122 ECB : {
5115 tgl 123 GIC 1 : if (isalpha((unsigned char) c))
5115 tgl 124 ECB : {
5115 tgl 125 GIC 1 : c = toupper((unsigned char) c);
5115 tgl 126 ECB : /* Defend against non-ASCII letters */
5115 tgl 127 GIC 1 : if (c >= 'A' && c <= 'Z')
5115 tgl 128 CBC 1 : return _codes[c - 'A'];
5115 tgl 129 ECB : }
5115 tgl 130 UIC 0 : return 0;
5115 tgl 131 EUB : }
132 :
133 : #define isvowel(c) (getcode(c) & 1) /* AEIOU */
134 :
135 : /* These letters are passed through unchanged */
136 : #define NOCHANGE(c) (getcode(c) & 2) /* FJMNR */
137 :
138 : /* These form diphthongs when preceding H */
139 : #define AFFECTH(c) (getcode(c) & 4) /* CGPST */
140 :
141 : /* These make C and G soft */
142 : #define MAKESOFT(c) (getcode(c) & 8) /* EIY */
143 :
144 : /* These prevent GH from becoming F */
145 : #define NOGHTOF(c) (getcode(c) & 16) /* BDH */
146 :
5484 tgl 147 GIC 2 : PG_FUNCTION_INFO_V1(levenshtein_with_costs);
5484 tgl 148 ECB : Datum
5484 tgl 149 GIC 1 : levenshtein_with_costs(PG_FUNCTION_ARGS)
5484 tgl 150 ECB : {
4637 rhaas 151 GIC 1 : text *src = PG_GETARG_TEXT_PP(0);
4637 rhaas 152 CBC 1 : text *dst = PG_GETARG_TEXT_PP(1);
5050 bruce 153 1 : int ins_c = PG_GETARG_INT32(2);
154 1 : int del_c = PG_GETARG_INT32(3);
155 1 : int sub_c = PG_GETARG_INT32(4);
3069 rhaas 156 ECB : const char *s_data;
157 : const char *t_data;
158 : int s_bytes,
159 : t_bytes;
160 :
161 : /* Extract a pointer to the actual character data */
3069 rhaas 162 GIC 1 : s_data = VARDATA_ANY(src);
3069 rhaas 163 CBC 1 : t_data = VARDATA_ANY(dst);
2634 tgl 164 ECB : /* Determine length of each string in bytes */
3069 rhaas 165 GIC 1 : s_bytes = VARSIZE_ANY_EXHDR(src);
3069 rhaas 166 CBC 1 : t_bytes = VARSIZE_ANY_EXHDR(dst);
3069 rhaas 167 ECB :
2634 tgl 168 GIC 1 : PG_RETURN_INT32(varstr_levenshtein(s_data, s_bytes, t_data, t_bytes,
2634 tgl 169 ECB : ins_c, del_c, sub_c, false));
170 : }
171 :
172 :
5484 tgl 173 GIC 2 : PG_FUNCTION_INFO_V1(levenshtein);
5484 tgl 174 ECB : Datum
5484 tgl 175 GIC 1 : levenshtein(PG_FUNCTION_ARGS)
5484 tgl 176 ECB : {
4637 rhaas 177 GIC 1 : text *src = PG_GETARG_TEXT_PP(0);
4637 rhaas 178 CBC 1 : text *dst = PG_GETARG_TEXT_PP(1);
3069 rhaas 179 ECB : const char *s_data;
180 : const char *t_data;
181 : int s_bytes,
182 : t_bytes;
183 :
184 : /* Extract a pointer to the actual character data */
3069 rhaas 185 GIC 1 : s_data = VARDATA_ANY(src);
3069 rhaas 186 CBC 1 : t_data = VARDATA_ANY(dst);
2634 tgl 187 ECB : /* Determine length of each string in bytes */
3069 rhaas 188 GIC 1 : s_bytes = VARSIZE_ANY_EXHDR(src);
3069 rhaas 189 CBC 1 : t_bytes = VARSIZE_ANY_EXHDR(dst);
3069 rhaas 190 ECB :
2634 tgl 191 GIC 1 : PG_RETURN_INT32(varstr_levenshtein(s_data, s_bytes, t_data, t_bytes,
2634 tgl 192 ECB : 1, 1, 1, false));
193 : }
194 :
195 :
4555 rhaas 196 GIC 1 : PG_FUNCTION_INFO_V1(levenshtein_less_equal_with_costs);
4555 rhaas 197 ECB : Datum
4555 rhaas 198 UIC 0 : levenshtein_less_equal_with_costs(PG_FUNCTION_ARGS)
4555 rhaas 199 EUB : {
4555 rhaas 200 UIC 0 : text *src = PG_GETARG_TEXT_PP(0);
4555 rhaas 201 UBC 0 : text *dst = PG_GETARG_TEXT_PP(1);
202 0 : int ins_c = PG_GETARG_INT32(2);
203 0 : int del_c = PG_GETARG_INT32(3);
204 0 : int sub_c = PG_GETARG_INT32(4);
205 0 : int max_d = PG_GETARG_INT32(5);
3069 rhaas 206 EUB : const char *s_data;
207 : const char *t_data;
208 : int s_bytes,
209 : t_bytes;
210 :
211 : /* Extract a pointer to the actual character data */
3069 rhaas 212 UIC 0 : s_data = VARDATA_ANY(src);
3069 rhaas 213 UBC 0 : t_data = VARDATA_ANY(dst);
2634 tgl 214 EUB : /* Determine length of each string in bytes */
3069 rhaas 215 UIC 0 : s_bytes = VARSIZE_ANY_EXHDR(src);
3069 rhaas 216 UBC 0 : t_bytes = VARSIZE_ANY_EXHDR(dst);
3069 rhaas 217 EUB :
2634 tgl 218 UIC 0 : PG_RETURN_INT32(varstr_levenshtein_less_equal(s_data, s_bytes,
2634 tgl 219 EUB : t_data, t_bytes,
220 : ins_c, del_c, sub_c,
221 : max_d, false));
222 : }
223 :
224 :
4555 rhaas 225 GIC 2 : PG_FUNCTION_INFO_V1(levenshtein_less_equal);
4555 rhaas 226 ECB : Datum
4555 rhaas 227 GIC 2 : levenshtein_less_equal(PG_FUNCTION_ARGS)
4555 rhaas 228 ECB : {
4555 rhaas 229 GIC 2 : text *src = PG_GETARG_TEXT_PP(0);
4555 rhaas 230 CBC 2 : text *dst = PG_GETARG_TEXT_PP(1);
231 2 : int max_d = PG_GETARG_INT32(2);
3069 rhaas 232 ECB : const char *s_data;
233 : const char *t_data;
234 : int s_bytes,
235 : t_bytes;
236 :
237 : /* Extract a pointer to the actual character data */
3069 rhaas 238 GIC 2 : s_data = VARDATA_ANY(src);
3069 rhaas 239 CBC 2 : t_data = VARDATA_ANY(dst);
2634 tgl 240 ECB : /* Determine length of each string in bytes */
3069 rhaas 241 GIC 2 : s_bytes = VARSIZE_ANY_EXHDR(src);
3069 rhaas 242 CBC 2 : t_bytes = VARSIZE_ANY_EXHDR(dst);
3069 rhaas 243 ECB :
2634 tgl 244 GIC 2 : PG_RETURN_INT32(varstr_levenshtein_less_equal(s_data, s_bytes,
2634 tgl 245 ECB : t_data, t_bytes,
246 : 1, 1, 1,
247 : max_d, false));
248 : }
249 :
250 :
251 : /*
252 : * Calculates the metaphone of an input string.
253 : * Returns number of characters requested
254 : * (suggested value is 4)
255 : */
7915 bruce 256 GIC 2 : PG_FUNCTION_INFO_V1(metaphone);
7915 bruce 257 ECB : Datum
7915 bruce 258 GIC 1 : metaphone(PG_FUNCTION_ARGS)
7915 bruce 259 ECB : {
5493 tgl 260 GIC 1 : char *str_i = TextDatumGetCString(PG_GETARG_DATUM(0));
5493 tgl 261 CBC 1 : size_t str_i_len = strlen(str_i);
7915 bruce 262 ECB : int reqlen;
263 : char *metaph;
264 :
265 : /* return an empty string if we receive one */
6856 mail 266 GIC 1 : if (!(str_i_len > 0))
5493 tgl 267 LBC 0 : PG_RETURN_TEXT_P(cstring_to_text(""));
6856 mail 268 EUB :
7915 bruce 269 GIC 1 : if (str_i_len > MAX_METAPHONE_STRLEN)
7199 tgl 270 LBC 0 : ereport(ERROR,
7199 tgl 271 EUB : (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
272 : errmsg("argument exceeds the maximum length of %d bytes",
273 : MAX_METAPHONE_STRLEN)));
274 :
7915 bruce 275 GIC 1 : reqlen = PG_GETARG_INT32(1);
7915 bruce 276 CBC 1 : if (reqlen > MAX_METAPHONE_STRLEN)
7199 tgl 277 LBC 0 : ereport(ERROR,
7199 tgl 278 EUB : (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
279 : errmsg("output exceeds the maximum length of %d bytes",
280 : MAX_METAPHONE_STRLEN)));
281 :
7915 bruce 282 GIC 1 : if (!(reqlen > 0))
7199 tgl 283 LBC 0 : ereport(ERROR,
7199 tgl 284 EUB : (errcode(ERRCODE_ZERO_LENGTH_CHARACTER_STRING),
285 : errmsg("output cannot be empty string")));
286 :
2061 peter_e 287 GIC 1 : _metaphone(str_i, reqlen, &metaph);
2061 peter_e 288 CBC 1 : PG_RETURN_TEXT_P(cstring_to_text(metaph));
7915 bruce 289 ECB : }
290 :
291 :
292 : /*
293 : * Original code by Michael G Schwern starts here.
294 : * Code slightly modified for use as PostgreSQL
295 : * function (palloc, etc).
296 : */
297 :
298 : /* I suppose I could have been using a character pointer instead of
299 : * accessing the array directly... */
300 :
301 : /* Look at the next letter in the word */
302 : #define Next_Letter (toupper((unsigned char) word[w_idx+1]))
303 : /* Look at the current letter in the word */
304 : #define Curr_Letter (toupper((unsigned char) word[w_idx]))
305 : /* Go N letters back. */
306 : #define Look_Back_Letter(n) \
307 : (w_idx >= (n) ? toupper((unsigned char) word[w_idx-(n)]) : '\0')
308 : /* Previous letter. I dunno, should this return null on failure? */
309 : #define Prev_Letter (Look_Back_Letter(1))
310 : /* Look two letters down. It makes sure you don't walk off the string. */
311 : #define After_Next_Letter \
312 : (Next_Letter != '\0' ? toupper((unsigned char) word[w_idx+2]) : '\0')
313 : #define Look_Ahead_Letter(n) toupper((unsigned char) Lookahead(word+w_idx, n))
314 :
315 :
316 : /* Allows us to safely look ahead an arbitrary # of letters */
317 : /* I probably could have just used strlen... */
318 : static char
7836 bruce 319 UIC 0 : Lookahead(char *word, int how_far)
7836 bruce 320 EUB : {
7836 bruce 321 UIC 0 : char letter_ahead = '\0'; /* null by default */
7836 bruce 322 EUB : int idx;
323 :
7836 bruce 324 UIC 0 : for (idx = 0; word[idx] != '\0' && idx < how_far; idx++);
7836 bruce 325 EUB : /* Edge forward in the string... */
326 :
6385 bruce 327 UIC 0 : letter_ahead = word[idx]; /* idx will be either == to how_far or at the
6385 bruce 328 EUB : * end of the string */
7915 bruce 329 UIC 0 : return letter_ahead;
7915 bruce 330 EUB : }
331 :
332 :
333 : /* phonize one letter */
334 : #define Phonize(c) do {(*phoned_word)[p_idx++] = c;} while (0)
335 : /* Slap a null character on the end of the phoned word */
336 : #define End_Phoned_Word do {(*phoned_word)[p_idx] = '\0';} while (0)
337 : /* How long is the phoned word? */
338 : #define Phone_Len (p_idx)
339 :
340 : /* Note is a letter is a 'break' in the word */
341 : #define Isbreak(c) (!isalpha((unsigned char) (c)))
342 :
343 :
344 : static void
5050 bruce 345 GIC 1 : _metaphone(char *word, /* IN */
7836 bruce 346 ECB : int max_phonemes,
347 : char **phoned_word) /* OUT */
348 : {
7836 bruce 349 GIC 1 : int w_idx = 0; /* point in the phonization we're at. */
7836 bruce 350 CBC 1 : int p_idx = 0; /* end of the phoned phrase */
7836 bruce 351 ECB :
352 : /*-- Parameter checks --*/
353 :
354 : /*
355 : * Shouldn't be necessary, but left these here anyway jec Aug 3, 2001
356 : */
357 :
358 : /* Negative phoneme length is meaningless */
7915 bruce 359 GIC 1 : if (!(max_phonemes > 0))
7199 tgl 360 ECB : /* internal error */
7915 bruce 361 UIC 0 : elog(ERROR, "metaphone: Requested output length must be > 0");
7915 bruce 362 EUB :
363 : /* Empty/null string is meaningless */
7915 bruce 364 GIC 1 : if ((word == NULL) || !(strlen(word) > 0))
7199 tgl 365 ECB : /* internal error */
7915 bruce 366 UIC 0 : elog(ERROR, "metaphone: Input string length must be > 0");
7836 bruce 367 EUB :
368 : /*-- Allocate memory for our phoned_phrase --*/
7836 bruce 369 GIC 1 : if (max_phonemes == 0)
7836 bruce 370 ECB : { /* Assume largest possible */
2118 tgl 371 UIC 0 : *phoned_word = palloc(sizeof(char) * strlen(word) + 1);
7836 bruce 372 EUB : }
373 : else
374 : {
7915 bruce 375 GIC 1 : *phoned_word = palloc(sizeof(char) * max_phonemes + 1);
7915 bruce 376 ECB : }
377 :
378 : /*-- The first phoneme has to be processed specially. --*/
379 : /* Find our first letter */
7770 tgl 380 GIC 1 : for (; !isalpha((unsigned char) (Curr_Letter)); w_idx++)
7836 bruce 381 ECB : {
382 : /* On the off chance we were given nothing but crap... */
7836 bruce 383 UIC 0 : if (Curr_Letter == '\0')
7836 bruce 384 EUB : {
7832 bruce 385 UIC 0 : End_Phoned_Word;
2061 peter_e 386 UBC 0 : return;
7915 bruce 387 EUB : }
388 : }
389 :
7836 bruce 390 GIC 1 : switch (Curr_Letter)
7836 bruce 391 ECB : {
392 : /* AE becomes E */
7915 bruce 393 UIC 0 : case 'A':
7836 bruce 394 UBC 0 : if (Next_Letter == 'E')
7836 bruce 395 EUB : {
7915 bruce 396 UIC 0 : Phonize('E');
7836 bruce 397 UBC 0 : w_idx += 2;
7915 bruce 398 EUB : }
399 : /* Remember, preserve vowels at the beginning */
400 : else
401 : {
7915 bruce 402 UIC 0 : Phonize('A');
7915 bruce 403 UBC 0 : w_idx++;
7915 bruce 404 EUB : }
7915 bruce 405 UIC 0 : break;
7836 bruce 406 EUB : /* [GKP]N becomes N */
7915 bruce 407 GIC 1 : case 'G':
7915 bruce 408 ECB : case 'K':
409 : case 'P':
7836 bruce 410 GIC 1 : if (Next_Letter == 'N')
7836 bruce 411 ECB : {
7915 bruce 412 UIC 0 : Phonize('N');
7836 bruce 413 UBC 0 : w_idx += 2;
7915 bruce 414 EUB : }
7915 bruce 415 GIC 1 : break;
7836 bruce 416 ECB :
417 : /*
418 : * WH becomes H, WR becomes R W if followed by a vowel
419 : */
7915 bruce 420 UIC 0 : case 'W':
7836 bruce 421 UBC 0 : if (Next_Letter == 'H' ||
422 0 : Next_Letter == 'R')
7915 bruce 423 EUB : {
7836 bruce 424 UIC 0 : Phonize(Next_Letter);
7836 bruce 425 UBC 0 : w_idx += 2;
7915 bruce 426 EUB : }
7836 bruce 427 UIC 0 : else if (isvowel(Next_Letter))
7836 bruce 428 EUB : {
7836 bruce 429 UIC 0 : Phonize('W');
7836 bruce 430 UBC 0 : w_idx += 2;
7915 bruce 431 EUB : }
432 : /* else ignore */
7915 bruce 433 UIC 0 : break;
7836 bruce 434 EUB : /* X becomes S */
7915 bruce 435 UIC 0 : case 'X':
7915 bruce 436 UBC 0 : Phonize('S');
437 0 : w_idx++;
438 0 : break;
7836 bruce 439 EUB : /* Vowels are kept */
440 :
441 : /*
442 : * We did A already case 'A': case 'a':
443 : */
7915 bruce 444 UIC 0 : case 'E':
7915 bruce 445 EUB : case 'I':
446 : case 'O':
447 : case 'U':
7915 bruce 448 UIC 0 : Phonize(Curr_Letter);
7915 bruce 449 UBC 0 : w_idx++;
450 0 : break;
451 0 : default:
7915 bruce 452 EUB : /* do nothing */
7915 bruce 453 UIC 0 : break;
7915 bruce 454 EUB : }
455 :
456 :
457 :
458 : /* On to the metaphoning */
7836 bruce 459 GIC 6 : for (; Curr_Letter != '\0' &&
7836 bruce 460 CBC 5 : (max_phonemes == 0 || Phone_Len < max_phonemes);
461 5 : w_idx++)
7836 bruce 462 ECB : {
463 : /*
464 : * How many letters to skip because an earlier encoding handled
465 : * multiple letters
466 : */
7836 bruce 467 GIC 5 : unsigned short int skip_letter = 0;
7836 bruce 468 ECB :
469 :
470 : /*
471 : * THOUGHT: It would be nice if, rather than having things like...
472 : * well, SCI. For SCI you encode the S, then have to remember to skip
473 : * the C. So the phonome SCI invades both S and C. It would be
474 : * better, IMHO, to skip the C from the S part of the encoding. Hell,
475 : * I'm trying it.
476 : */
477 :
478 : /* Ignore non-alphas */
7770 tgl 479 GIC 5 : if (!isalpha((unsigned char) (Curr_Letter)))
7915 bruce 480 LBC 0 : continue;
7836 bruce 481 EUB :
482 : /* Drop duplicates, except CC */
7836 bruce 483 GIC 5 : if (Curr_Letter == Prev_Letter &&
7836 bruce 484 LBC 0 : Curr_Letter != 'C')
7915 bruce 485 UBC 0 : continue;
7836 bruce 486 EUB :
7836 bruce 487 GIC 5 : switch (Curr_Letter)
7836 bruce 488 ECB : {
489 : /* B -> B unless in MB */
7915 bruce 490 GIC 1 : case 'B':
7836 bruce 491 CBC 1 : if (Prev_Letter != 'M')
7915 bruce 492 LBC 0 : Phonize('B');
7915 bruce 493 GBC 1 : break;
7836 bruce 494 ECB :
495 : /*
496 : * 'sh' if -CIA- or -CH, but not SCH, except SCHW. (SCHW is
497 : * handled in S) S if -CI-, -CE- or -CY- dropped if -SCI-,
498 : * SCE-, -SCY- (handed in S) else K
499 : */
7915 bruce 500 UIC 0 : case 'C':
7836 bruce 501 UBC 0 : if (MAKESOFT(Next_Letter))
7836 bruce 502 EUB : { /* C[IEY] */
7836 bruce 503 UIC 0 : if (After_Next_Letter == 'A' &&
7836 bruce 504 UBC 0 : Next_Letter == 'I')
7836 bruce 505 EUB : { /* CIA */
7915 bruce 506 UIC 0 : Phonize(SH);
7915 bruce 507 EUB : }
508 : /* SC[IEY] */
7836 bruce 509 UIC 0 : else if (Prev_Letter == 'S')
7836 bruce 510 EUB : {
511 : /* Dropped */
512 : }
513 : else
7836 bruce 514 UIC 0 : Phonize('S');
7915 bruce 515 EUB : }
7836 bruce 516 UIC 0 : else if (Next_Letter == 'H')
7836 bruce 517 EUB : {
518 : #ifndef USE_TRADITIONAL_METAPHONE
7836 bruce 519 UIC 0 : if (After_Next_Letter == 'R' ||
7836 bruce 520 UBC 0 : Prev_Letter == 'S')
7836 bruce 521 EUB : { /* Christ, School */
7915 bruce 522 UIC 0 : Phonize('K');
7915 bruce 523 EUB : }
524 : else
7915 bruce 525 UIC 0 : Phonize(SH);
7915 bruce 526 EUB : #else
527 : Phonize(SH);
528 : #endif
7915 bruce 529 UIC 0 : skip_letter++;
7915 bruce 530 EUB : }
531 : else
7915 bruce 532 UIC 0 : Phonize('K');
7915 bruce 533 UBC 0 : break;
7836 bruce 534 EUB :
535 : /*
536 : * J if in -DGE-, -DGI- or -DGY- else T
537 : */
7915 bruce 538 UIC 0 : case 'D':
7836 bruce 539 UBC 0 : if (Next_Letter == 'G' &&
540 0 : MAKESOFT(After_Next_Letter))
7836 bruce 541 EUB : {
7915 bruce 542 UIC 0 : Phonize('J');
7915 bruce 543 UBC 0 : skip_letter++;
7915 bruce 544 EUB : }
545 : else
7915 bruce 546 UIC 0 : Phonize('T');
7915 bruce 547 UBC 0 : break;
7836 bruce 548 EUB :
549 : /*
550 : * F if in -GH and not B--GH, D--GH, -H--GH, -H---GH else
551 : * dropped if -GNED, -GN, else dropped if -DGE-, -DGI- or
552 : * -DGY- (handled in D) else J if in -GE-, -GI, -GY and not GG
553 : * else K
554 : */
7915 bruce 555 GIC 1 : case 'G':
7836 bruce 556 CBC 1 : if (Next_Letter == 'H')
7836 bruce 557 ECB : {
7836 bruce 558 UIC 0 : if (!(NOGHTOF(Look_Back_Letter(3)) ||
7836 bruce 559 UBC 0 : Look_Back_Letter(4) == 'H'))
7836 bruce 560 EUB : {
7915 bruce 561 UIC 0 : Phonize('F');
7915 bruce 562 UBC 0 : skip_letter++;
7915 bruce 563 EUB : }
564 : else
565 : {
566 : /* silent */
567 : }
568 : }
7836 bruce 569 GIC 1 : else if (Next_Letter == 'N')
7836 bruce 570 ECB : {
7836 bruce 571 UIC 0 : if (Isbreak(After_Next_Letter) ||
7836 bruce 572 UBC 0 : (After_Next_Letter == 'E' &&
573 0 : Look_Ahead_Letter(3) == 'D'))
7836 bruce 574 EUB : {
575 : /* dropped */
576 : }
577 : else
7915 bruce 578 UIC 0 : Phonize('K');
7915 bruce 579 EUB : }
7836 bruce 580 GIC 1 : else if (MAKESOFT(Next_Letter) &&
7836 bruce 581 LBC 0 : Prev_Letter != 'G')
7915 bruce 582 UBC 0 : Phonize('J');
7836 bruce 583 EUB : else
7915 bruce 584 GIC 1 : Phonize('K');
7915 bruce 585 CBC 1 : break;
7836 bruce 586 ECB : /* H if before a vowel and not after C,G,P,S,T */
7915 bruce 587 UIC 0 : case 'H':
7836 bruce 588 UBC 0 : if (isvowel(Next_Letter) &&
589 0 : !AFFECTH(Prev_Letter))
7915 590 0 : Phonize('H');
591 0 : break;
7836 bruce 592 EUB :
593 : /*
594 : * dropped if after C else K
595 : */
7915 bruce 596 UIC 0 : case 'K':
7836 bruce 597 UBC 0 : if (Prev_Letter != 'C')
7915 598 0 : Phonize('K');
599 0 : break;
7836 bruce 600 EUB :
601 : /*
602 : * F if before H else P
603 : */
7915 bruce 604 UIC 0 : case 'P':
7836 bruce 605 UBC 0 : if (Next_Letter == 'H')
7915 606 0 : Phonize('F');
7836 bruce 607 EUB : else
7915 bruce 608 UIC 0 : Phonize('P');
7915 bruce 609 UBC 0 : break;
7836 bruce 610 EUB :
611 : /*
612 : * K
613 : */
7915 bruce 614 UIC 0 : case 'Q':
7915 bruce 615 UBC 0 : Phonize('K');
616 0 : break;
7836 bruce 617 EUB :
618 : /*
619 : * 'sh' in -SH-, -SIO- or -SIA- or -SCHW- else S
620 : */
7915 bruce 621 UIC 0 : case 'S':
7836 bruce 622 UBC 0 : if (Next_Letter == 'I' &&
623 0 : (After_Next_Letter == 'O' ||
624 0 : After_Next_Letter == 'A'))
7915 625 0 : Phonize(SH);
7836 626 0 : else if (Next_Letter == 'H')
7836 bruce 627 EUB : {
7915 bruce 628 UIC 0 : Phonize(SH);
7915 bruce 629 UBC 0 : skip_letter++;
7915 bruce 630 EUB : }
631 : #ifndef USE_TRADITIONAL_METAPHONE
7836 bruce 632 UIC 0 : else if (Next_Letter == 'C' &&
7836 bruce 633 UBC 0 : Look_Ahead_Letter(2) == 'H' &&
634 0 : Look_Ahead_Letter(3) == 'W')
7836 bruce 635 EUB : {
7915 bruce 636 UIC 0 : Phonize(SH);
7915 bruce 637 UBC 0 : skip_letter += 2;
7915 bruce 638 EUB : }
639 : #endif
640 : else
7915 bruce 641 UIC 0 : Phonize('S');
7915 bruce 642 UBC 0 : break;
7836 bruce 643 EUB :
644 : /*
645 : * 'sh' in -TIA- or -TIO- else 'th' before H else T
646 : */
7915 bruce 647 UIC 0 : case 'T':
7836 bruce 648 UBC 0 : if (Next_Letter == 'I' &&
649 0 : (After_Next_Letter == 'O' ||
650 0 : After_Next_Letter == 'A'))
7915 651 0 : Phonize(SH);
7836 652 0 : else if (Next_Letter == 'H')
7836 bruce 653 EUB : {
7915 bruce 654 UIC 0 : Phonize(TH);
7915 bruce 655 UBC 0 : skip_letter++;
7915 bruce 656 EUB : }
657 : else
7915 bruce 658 UIC 0 : Phonize('T');
7915 bruce 659 UBC 0 : break;
7836 bruce 660 EUB : /* F */
7915 bruce 661 UIC 0 : case 'V':
7915 bruce 662 UBC 0 : Phonize('F');
663 0 : break;
7836 bruce 664 EUB : /* W before a vowel, else dropped */
7915 bruce 665 UIC 0 : case 'W':
7836 bruce 666 UBC 0 : if (isvowel(Next_Letter))
7915 667 0 : Phonize('W');
668 0 : break;
7836 bruce 669 EUB : /* KS */
7915 bruce 670 UIC 0 : case 'X':
7915 bruce 671 UBC 0 : Phonize('K');
7229 672 0 : if (max_phonemes == 0 || Phone_Len < max_phonemes)
673 0 : Phonize('S');
7915 674 0 : break;
7836 bruce 675 EUB : /* Y if followed by a vowel */
7915 bruce 676 UIC 0 : case 'Y':
7836 bruce 677 UBC 0 : if (isvowel(Next_Letter))
7915 678 0 : Phonize('Y');
679 0 : break;
7836 bruce 680 EUB : /* S */
7915 bruce 681 UIC 0 : case 'Z':
7915 bruce 682 UBC 0 : Phonize('S');
683 0 : break;
7836 bruce 684 EUB : /* No transformation */
7915 bruce 685 GIC 1 : case 'F':
7915 bruce 686 ECB : case 'J':
687 : case 'L':
688 : case 'M':
689 : case 'N':
690 : case 'R':
7915 bruce 691 GIC 1 : Phonize(Curr_Letter);
7915 bruce 692 CBC 1 : break;
693 2 : default:
7915 bruce 694 ECB : /* nothing */
7915 bruce 695 GIC 2 : break;
7836 bruce 696 ECB : } /* END SWITCH */
697 :
7915 bruce 698 GIC 5 : w_idx += skip_letter;
7836 bruce 699 ECB : } /* END FOR */
700 :
7915 bruce 701 GIC 1 : End_Phoned_Word;
2118 tgl 702 ECB : } /* END metaphone */
703 :
704 :
705 : /*
706 : * SQL function: soundex(text) returns text
707 : */
7915 bruce 708 GIC 3 : PG_FUNCTION_INFO_V1(soundex);
7915 bruce 709 ECB :
710 : Datum
7915 bruce 711 GIC 7 : soundex(PG_FUNCTION_ARGS)
7915 bruce 712 ECB : {
713 : char outstr[SOUNDEX_LEN + 1];
714 : char *arg;
715 :
2219 noah 716 GIC 7 : arg = text_to_cstring(PG_GETARG_TEXT_PP(0));
7915 bruce 717 ECB :
7915 bruce 718 GIC 7 : _soundex(arg, outstr);
7915 bruce 719 ECB :
5493 tgl 720 GIC 7 : PG_RETURN_TEXT_P(cstring_to_text(outstr));
7915 bruce 721 ECB : }
722 :
723 : static void
7915 bruce 724 GIC 13 : _soundex(const char *instr, char *outstr)
7915 bruce 725 ECB : {
726 : int count;
727 :
163 peter 728 GNC 13 : Assert(instr);
729 13 : Assert(outstr);
7915 bruce 730 ECB :
7915 bruce 731 GIC 13 : outstr[SOUNDEX_LEN] = '\0';
7915 bruce 732 ECB :
733 : /* Skip leading non-alphabetic characters */
7915 bruce 734 GIC 13 : while (!isalpha((unsigned char) instr[0]) && instr[0])
7915 bruce 735 LBC 0 : ++instr;
7915 bruce 736 EUB :
737 : /* No string left */
7915 bruce 738 GIC 13 : if (!instr[0])
7915 bruce 739 ECB : {
7915 bruce 740 UIC 0 : outstr[0] = (char) 0;
7915 bruce 741 UBC 0 : return;
7915 bruce 742 EUB : }
743 :
744 : /* Take the first letter as is */
7915 bruce 745 GIC 13 : *outstr++ = (char) toupper((unsigned char) *instr++);
7915 bruce 746 ECB :
7915 bruce 747 GIC 13 : count = 1;
7915 bruce 748 CBC 60 : while (*instr && count < SOUNDEX_LEN)
7915 bruce 749 ECB : {
7915 bruce 750 GIC 93 : if (isalpha((unsigned char) *instr) &&
7915 bruce 751 CBC 46 : soundex_code(*instr) != soundex_code(*(instr - 1)))
7915 bruce 752 ECB : {
7915 bruce 753 GIC 35 : *outstr = soundex_code(instr[0]);
7915 bruce 754 CBC 35 : if (*outstr != '0')
7915 bruce 755 ECB : {
7915 bruce 756 GIC 23 : ++outstr;
7915 bruce 757 CBC 23 : ++count;
7915 bruce 758 ECB : }
759 : }
7915 bruce 760 GIC 47 : ++instr;
7915 bruce 761 ECB : }
762 :
763 : /* Fill with 0's */
7915 bruce 764 GIC 29 : while (count < SOUNDEX_LEN)
7915 bruce 765 ECB : {
7915 bruce 766 GIC 16 : *outstr = '0';
7915 bruce 767 CBC 16 : ++outstr;
768 16 : ++count;
7915 bruce 769 ECB : }
770 : }
771 :
6647 neilc 772 GIC 2 : PG_FUNCTION_INFO_V1(difference);
6647 neilc 773 ECB :
774 : Datum
6647 neilc 775 GIC 3 : difference(PG_FUNCTION_ARGS)
6647 neilc 776 ECB : {
777 : char sndx1[SOUNDEX_LEN + 1],
778 : sndx2[SOUNDEX_LEN + 1];
779 : int i,
780 : result;
781 :
2219 noah 782 GIC 3 : _soundex(text_to_cstring(PG_GETARG_TEXT_PP(0)), sndx1);
2219 noah 783 CBC 3 : _soundex(text_to_cstring(PG_GETARG_TEXT_PP(1)), sndx2);
6647 neilc 784 ECB :
6647 neilc 785 GIC 3 : result = 0;
6385 bruce 786 CBC 15 : for (i = 0; i < SOUNDEX_LEN; i++)
6385 bruce 787 ECB : {
6647 neilc 788 GIC 12 : if (sndx1[i] == sndx2[i])
6647 neilc 789 CBC 6 : result++;
6647 neilc 790 ECB : }
791 :
6647 neilc 792 GIC 3 : PG_RETURN_INT32(result);
6647 neilc 793 ECB : }
|
