1 /*
2 +----------------------------------------------------------------------+
3 | Copyright (c) The PHP Group |
4 +----------------------------------------------------------------------+
5 | This source file is subject to version 3.01 of the PHP license, |
6 | that is bundled with this package in the file LICENSE, and is |
7 | available through the world-wide-web at the following url: |
8 | https://www.php.net/license/3_01.txt |
9 | If you did not receive a copy of the PHP license and are unable to |
10 | obtain it through the world-wide-web, please send a note to |
11 | license@php.net so we can mail you a copy immediately. |
12 +----------------------------------------------------------------------+
13 | Author: Bjørn Borud - Guardian Networks AS <borud@guardian.no> |
14 +----------------------------------------------------------------------+
15 */
16
17 #include "php.h"
18 #include <stdlib.h>
19 #include <ctype.h>
20
21 /* Simple soundex algorithm as described by Knuth in TAOCP, vol 3 */
22 /* {{{ Calculate the soundex key of a string */
PHP_FUNCTION(soundex)23 PHP_FUNCTION(soundex)
24 {
25 char *str;
26 size_t i, _small, str_len, code, last;
27 char soundex[4 + 1];
28
29 static const char soundex_table[26] =
30 {0, /* A */
31 '1', /* B */
32 '2', /* C */
33 '3', /* D */
34 0, /* E */
35 '1', /* F */
36 '2', /* G */
37 0, /* H */
38 0, /* I */
39 '2', /* J */
40 '2', /* K */
41 '4', /* L */
42 '5', /* M */
43 '5', /* N */
44 0, /* O */
45 '1', /* P */
46 '2', /* Q */
47 '6', /* R */
48 '2', /* S */
49 '3', /* T */
50 0, /* U */
51 '1', /* V */
52 0, /* W */
53 '2', /* X */
54 0, /* Y */
55 '2'}; /* Z */
56
57 ZEND_PARSE_PARAMETERS_START(1, 1)
58 Z_PARAM_STRING(str, str_len)
59 ZEND_PARSE_PARAMETERS_END();
60
61 /* build soundex string */
62 last = -1;
63 for (i = 0, _small = 0; i < str_len && _small < 4; i++) {
64 /* convert chars to upper case and strip non-letter chars */
65 /* BUG: should also map here accented letters used in non */
66 /* English words or names (also found in English text!): */
67 /* esstsett, thorn, n-tilde, c-cedilla, s-caron, ... */
68 code = toupper((int)(unsigned char)str[i]);
69 if (code >= 'A' && code <= 'Z') {
70 if (_small == 0) {
71 /* remember first valid char */
72 soundex[_small++] = (char)code;
73 last = soundex_table[code - 'A'];
74 }
75 else {
76 /* ignore sequences of consonants with same soundex */
77 /* code in trail, and vowels unless they separate */
78 /* consonant letters */
79 code = soundex_table[code - 'A'];
80 if (code != last) {
81 if (code != 0) {
82 soundex[_small++] = (char)code;
83 }
84 last = code;
85 }
86 }
87 }
88 }
89 /* pad with '0' and terminate with 0 ;-) */
90 while (_small < 4) {
91 soundex[_small++] = '0';
92 }
93 soundex[_small] = '\0';
94
95 RETURN_STRINGL(soundex, _small);
96 }
97 /* }}} */
98
