/src/php-src/ext/standard/metaphone.c

Source
/*
   +----------------------------------------------------------------------+
   | Copyright © The PHP Group and Contributors.                          |
   +----------------------------------------------------------------------+
   | This source file is subject to the Modified BSD License that is      |
   | bundled with this package in the file LICENSE, and is available      |
   | through the World Wide Web at <https://www.php.net/license/>.        |
   |                                                                      |
   | SPDX-License-Identifier: BSD-3-Clause                                |
   +----------------------------------------------------------------------+
   | Author: Thies C. Arntzen <thies@thieso.net>                          |
   +----------------------------------------------------------------------+
*/

/*
  Based on CPANs "Text-Metaphone-1.96" by Michael G Schwern <schwern@pobox.com>
*/

#include "php.h"

static void metaphone(unsigned char *word, size_t word_len, zend_long max_phonemes, zend_string **phoned_word, int traditional);

/* {{{ Break english phrases down into their phonemes */
PHP_FUNCTION(metaphone)
{
  zend_string *str;
  zend_string *result = NULL;
  zend_long phones = 0;

  ZEND_PARSE_PARAMETERS_START(1, 2)
    Z_PARAM_STR(str)
    Z_PARAM_OPTIONAL
    Z_PARAM_LONG(phones)
  ZEND_PARSE_PARAMETERS_END();

  if (phones < 0) {
    zend_argument_value_error(2, "must be greater than or equal to 0");
    RETURN_THROWS();
  }

  metaphone((unsigned char *)ZSTR_VAL(str), ZSTR_LEN(str), phones, &result, 1);
  RETVAL_STR(result);
}
/* }}} */

/*
   this is now the original code by Michael G Schwern:
   i've changed it just a slightly bit (use emalloc,
   get rid of includes etc)
  - thies - 13.09.1999
*/

/*-----------------------------  */
/* this used to be "metaphone.h" */
/*-----------------------------  */

/* Special encodings */
#define  SH   'X'
#define  TH   '0'

/*-----------------------------  */
/* end of "metaphone.h"          */
/*-----------------------------  */

/*----------------------------- */
/* this used to be "metachar.h" */
/*----------------------------- */

/* Metachar.h ... little bits about characters for metaphone */
/*-- Character encoding array & accessing macros --*/
/* Stolen directly out of the book... */
static const char _codes[26] =
{
  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
/*  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 */
};


/* Note: these functions require an uppercase letter input! */
static zend_always_inline char encode(char c) {
  if (isalpha((unsigned char)c)) {
    ZEND_ASSERT(c >= 'A' && c <= 'Z');
    return _codes[(c - 'A')];
  } else {
    return 0;
  }
}

#define isvowel(c)  (encode(c) & 1)    /* AEIOU */

/* These letters are passed through unchanged */
#define NOCHANGE(c) (encode(c) & 2)   /* FJMNR */

/* These form diphthongs when preceding H */
#define AFFECTH(c)  (encode(c) & 4)    /* CGPST */

/* These make C and G soft */
#define MAKESOFT(c) (encode(c) & 8)    /* EIY */

/* These prevent GH from becoming F */
#define NOGHTOF(c)  (encode(c) & 16)  /* BDH */

/*----------------------------- */
/* end of "metachar.h"          */
/*----------------------------- */

/* I suppose I could have been using a character pointer instead of
 * accessing the array directly... */

#define Convert_Raw(c) toupper((unsigned char)c)
/* Look at the next letter in the word */
#define Read_Raw_Next_Letter (word[w_idx+1])
#define Read_Next_Letter (Convert_Raw(Read_Raw_Next_Letter))
/* Look at the current letter in the word */
#define Read_Raw_Curr_Letter (word[w_idx])
#define Read_Curr_Letter (Convert_Raw(Read_Raw_Curr_Letter))
/* Go N letters back. */
#define Look_Back_Letter(n) (w_idx >= n ? Convert_Raw(word[w_idx-n]) : '\0')
/* Previous letter.  I dunno, should this return null on failure? */
#define Read_Prev_Letter (Look_Back_Letter(1))
/* Look two letters down.  It makes sure you don't walk off the string. */
#define Read_After_Next_Letter  (Read_Raw_Next_Letter != '\0' ? Convert_Raw(word[w_idx+2]) \
                           : '\0')
#define Look_Ahead_Letter(n) (toupper((unsigned char)Lookahead((char *) word+w_idx, n)))


/* Allows us to safely look ahead an arbitrary # of letters */
/* I probably could have just used strlen... */
static char Lookahead(char *word, size_t how_far)
{
  size_t idx;
  for (idx = 0; word[idx] != '\0' && idx < how_far; idx++);
  /* Edge forward in the string... */

  return  word[idx];      /* idx will be either == to how_far or
                 * at the end of the string where it will be null
                 */
}


/* phonize one letter
 * We don't know the buffers size in advance. On way to solve this is to just
 * re-allocate the buffer size. We're using an extra of 2 characters (this
 * could be one though; or more too). */
#define Phonize(c)  { \
            if (p_idx >= max_buffer_len) { \
              *phoned_word = zend_string_extend(*phoned_word, 2 * sizeof(char) + max_buffer_len, 0); \
              max_buffer_len += 2; \
            } \
            ZSTR_VAL(*phoned_word)[p_idx++] = c; \
            ZSTR_LEN(*phoned_word) = p_idx; \
          }
/* Slap a null character on the end of the phoned word */
#define End_Phoned_Word() { \
              if (p_idx == max_buffer_len) { \
                *phoned_word = zend_string_extend(*phoned_word, 1 * sizeof(char) + max_buffer_len, 0); \
                max_buffer_len += 1; \
              } \
              ZSTR_VAL(*phoned_word)[p_idx] = '\0'; \
              ZSTR_LEN(*phoned_word) = p_idx; \
            }
/* How long is the phoned word? */
#define Phone_Len (p_idx)

/* Note is a letter is a 'break' in the word */
#define Isbreak(c)  (!isalpha((unsigned char)(c)))

/* {{{ metaphone */
static void metaphone(unsigned char *word, size_t word_len, zend_long max_phonemes, zend_string **phoned_word, int traditional)
{
  size_t w_idx = 0;       /* point in the phonization we're at. */
  size_t p_idx = 0;       /* end of the phoned phrase */
  size_t max_buffer_len = 0;    /* maximum length of the destination buffer */
  char curr_letter;
  ZEND_ASSERT(word != NULL);
  ZEND_ASSERT(max_phonemes >= 0);

/*-- Allocate memory for our phoned_phrase --*/
  if (max_phonemes == 0) { /* Assume largest possible */
    max_buffer_len = word_len;
    *phoned_word = zend_string_alloc(sizeof(char) * word_len + 1, 0);
  } else {
    max_buffer_len = max_phonemes;
    *phoned_word = zend_string_alloc(sizeof(char) * max_phonemes + 1, 0);
  }


/*-- The first phoneme has to be processed specially. --*/
  /* Find our first letter */
  for (; !isalpha((unsigned char)(curr_letter = Read_Raw_Curr_Letter)); w_idx++) {
    /* On the off chance we were given nothing but crap... */
    if (curr_letter == '\0') {
      End_Phoned_Word();
      return;
    }
  }

  curr_letter = Convert_Raw(curr_letter);

  switch (curr_letter) {
    /* AE becomes E */
  case 'A':
    if (Read_Next_Letter == 'E') {
      Phonize('E');
      w_idx += 2;
    }
    /* Remember, preserve vowels at the beginning */
    else {
      Phonize('A');
      w_idx++;
    }
    break;
    /* [GKP]N becomes N */
  case 'G':
  case 'K':
  case 'P':
    if (Read_Next_Letter == 'N') {
      Phonize('N');
      w_idx += 2;
    }
    break;
    /* WH becomes W,
       WR becomes R
       W if followed by a vowel */
  case 'W': {
    char next_letter = Read_Next_Letter;
    if (next_letter == 'R') {
      Phonize('R');
      w_idx += 2;
    } else if (next_letter == 'H' || isvowel(next_letter)) {
      Phonize('W');
      w_idx += 2;
    }
    /* else ignore */
    break;
  }
    /* X becomes S */
  case 'X':
    Phonize('S');
    w_idx++;
    break;
    /* Vowels are kept */
    /* We did A already
       case 'A':
       case 'a':
     */
  case 'E':
  case 'I':
  case 'O':
  case 'U':
    Phonize(curr_letter);
    w_idx++;
    break;
  default:
    /* do nothing */
    break;
  }



  /* On to the metaphoning */
  for (; (curr_letter = Read_Raw_Curr_Letter) != '\0' &&
     (max_phonemes == 0 || Phone_Len < (size_t)max_phonemes);
     w_idx++) {
    /* How many letters to skip because an earlier encoding handled
     * multiple letters */
    unsigned short int skip_letter = 0;


    /* THOUGHT:  It would be nice if, rather than having things like...
     * well, SCI.  For SCI you encode the S, then have to remember
     * to skip the C.  So the phonome SCI invades both S and C.  It would
     * be better, IMHO, to skip the C from the S part of the encoding.
     * Hell, I'm trying it.
     */

    /* Ignore non-alphas */
    if (!isalpha((unsigned char)curr_letter))
      continue;

    curr_letter = Convert_Raw(curr_letter);
    /* Note: we can't cache curr_letter from the previous loop
     * because of the skip_letter variable. */
    char prev_letter = Read_Prev_Letter;

    /* Drop duplicates, except CC */
    if (curr_letter == prev_letter &&
      curr_letter != 'C')
      continue;

    switch (curr_letter) {
      /* B -> B unless in MB */
    case 'B':
      if (prev_letter != 'M')
        Phonize('B');
      break;
      /* 'sh' if -CIA- or -CH, but not SCH, except SCHW.
       * (SCHW is handled in S)
       *  S if -CI-, -CE- or -CY-
       *  dropped if -SCI-, SCE-, -SCY- (handed in S)
       *  else K
       */
    case 'C': {
      char next_letter = Read_Next_Letter;
      if (MAKESOFT(next_letter)) { /* C[IEY] */
        if (next_letter == 'I' && Read_After_Next_Letter == 'A') { /* CIA */
          Phonize(SH);
        }
        /* SC[IEY] */
        else if (prev_letter == 'S') {
          /* Dropped */
        } else {
          Phonize('S');
        }
      } else if (next_letter == 'H') {
        if ((!traditional) && (prev_letter == 'S' || Read_After_Next_Letter == 'R')) { /* Christ, School */
          Phonize('K');
        } else {
          Phonize(SH);
        }
        skip_letter++;
      } else {
        Phonize('K');
      }
      break;
    }
      /* J if in -DGE-, -DGI- or -DGY-
       * else T
       */
    case 'D':
      if (Read_Next_Letter == 'G' &&
        MAKESOFT(Read_After_Next_Letter)) {
        Phonize('J');
        skip_letter++;
      } else
        Phonize('T');
      break;
      /* F if in -GH and not B--GH, D--GH, -H--GH, -H---GH
       * else dropped if -GNED, -GN,
       * else dropped if -DGE-, -DGI- or -DGY- (handled in D)
       * else J if in -GE-, -GI, -GY and not GG
       * else K
       */
    case 'G': {
      char next_letter = Read_Next_Letter;
      if (next_letter == 'H') {
        if (!(NOGHTOF(Look_Back_Letter(3)) ||
            Look_Back_Letter(4) == 'H')) {
          Phonize('F');
          skip_letter++;
        } else {
          /* silent */
        }
      } else if (next_letter == 'N') {
        char after_next_letter = Read_After_Next_Letter;
        if (Isbreak(after_next_letter) ||
          (after_next_letter == 'E' &&
           Look_Ahead_Letter(3) == 'D')) {
          /* dropped */
        } else
          Phonize('K');
      } else if (MAKESOFT(next_letter) &&
             prev_letter != 'G') {
        Phonize('J');
      } else {
        Phonize('K');
      }
      break;
    }
      /* H if before a vowel and not after C,G,P,S,T */
    case 'H':
      if (isvowel(Read_Next_Letter) &&
        !AFFECTH(prev_letter))
        Phonize('H');
      break;
      /* dropped if after C
       * else K
       */
    case 'K':
      if (prev_letter != 'C')
        Phonize('K');
      break;
      /* F if before H
       * else P
       */
    case 'P':
      if (Read_Next_Letter == 'H') {
        Phonize('F');
      } else {
        Phonize('P');
      }
      break;
      /* K
       */
    case 'Q':
      Phonize('K');
      break;
      /* 'sh' in -SH-, -SIO- or -SIA- or -SCHW-
       * else S
       */
    case 'S': {
      char next_letter = Read_Next_Letter;
      char after_next_letter;
      if (next_letter == 'I' &&
        ((after_next_letter = Read_After_Next_Letter) == 'O' ||
         after_next_letter == 'A')) {
        Phonize(SH);
      } else if (next_letter == 'H') {
        Phonize(SH);
        skip_letter++;
      } else if ((!traditional) && (next_letter == 'C' && Look_Ahead_Letter(2) == 'H' && Look_Ahead_Letter(3) == 'W')) {
        Phonize(SH);
        skip_letter += 2;
      } else {
        Phonize('S');
      }
      break;
    }
      /* 'sh' in -TIA- or -TIO-
       * else 'th' before H
       * else T
       */
    case 'T': {
      char next_letter = Read_Next_Letter;
      char after_next_letter;
      if (next_letter == 'I' &&
        ((after_next_letter = Read_After_Next_Letter) == 'O' ||
         after_next_letter == 'A')) {
        Phonize(SH);
      } else if (next_letter == 'H') {
        Phonize(TH);
        skip_letter++;
      } else if (!(next_letter == 'C' && Read_After_Next_Letter == 'H')) {
        Phonize('T');
      }
      break;
    }
      /* F */
    case 'V':
      Phonize('F');
      break;
      /* W before a vowel, else dropped */
    case 'W':
      if (isvowel(Read_Next_Letter))
        Phonize('W');
      break;
      /* KS */
    case 'X':
      Phonize('K');
      Phonize('S');
      break;
      /* Y if followed by a vowel */
    case 'Y':
      if (isvowel(Read_Next_Letter))
        Phonize('Y');
      break;
      /* S */
    case 'Z':
      Phonize('S');
      break;
      /* No transformation */
    case 'F':
    case 'J':
    case 'L':
    case 'M':
    case 'N':
    case 'R':
      Phonize(curr_letter);
      break;
    default:
      /* nothing */
      break;
    }           /* END SWITCH */

    w_idx += skip_letter;
  }              /* END FOR */

  End_Phoned_Word();
}                /* END metaphone */
/* }}} */

Coverage Report

Created: 2026-06-02 06:36

Line	Count	Source
1		/*
2		+----------------------------------------------------------------------+
3		\| Copyright © The PHP Group and Contributors. \|
4		+----------------------------------------------------------------------+
5		\| This source file is subject to the Modified BSD License that is \|
6		\| bundled with this package in the file LICENSE, and is available \|
7		\| through the World Wide Web at <https://www.php.net/license/>. \|
8		\| \|
9		\| SPDX-License-Identifier: BSD-3-Clause \|
10		+----------------------------------------------------------------------+
11		\| Author: Thies C. Arntzen <thies@thieso.net> \|
12		+----------------------------------------------------------------------+
13		*/
14
15		/*
16		Based on CPANs "Text-Metaphone-1.96" by Michael G Schwern <schwern@pobox.com>
17		*/
18
19		#include "php.h"
20
21		static void metaphone(unsigned char word, size_t word_len, zend_long max_phonemes, zend_string *phoned_word, int traditional);
22
23		/* {{{ Break english phrases down into their phonemes */
24		PHP_FUNCTION(metaphone)
25	0	{
26	0	zend_string *str;
27	0	zend_string *result = NULL;
28	0	zend_long phones = 0;
29
30	0	ZEND_PARSE_PARAMETERS_START(1, 2)
31	0	Z_PARAM_STR(str)
32	0	Z_PARAM_OPTIONAL
33	0	Z_PARAM_LONG(phones)
34	0	ZEND_PARSE_PARAMETERS_END();
35
36	0	if (phones < 0) {
37	0	zend_argument_value_error(2, "must be greater than or equal to 0");
38	0	RETURN_THROWS();
39	0	}
40
41	0	metaphone((unsigned char *)ZSTR_VAL(str), ZSTR_LEN(str), phones, &result, 1);
42	0	RETVAL_STR(result);
43	0	}
44		/* }}} */
45
46		/*
47		this is now the original code by Michael G Schwern:
48		i've changed it just a slightly bit (use emalloc,
49		get rid of includes etc)
50		- thies - 13.09.1999
51		*/
52
53		/----------------------------- /
54		/* this used to be "metaphone.h" */
55		/----------------------------- /
56
57		/* Special encodings */
58		#define SH 'X'
59		#define TH '0'
60
61		/----------------------------- /
62		/* end of "metaphone.h" */
63		/----------------------------- /
64
65		/----------------------------- /
66		/* this used to be "metachar.h" */
67		/----------------------------- /
68
69		/* Metachar.h ... little bits about characters for metaphone */
70		/-- Character encoding array & accessing macros --/
71		/* Stolen directly out of the book... */
72		static const char _codes[26] =
73		{
74		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
75		/* 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 */
76		};
77
78
79		/* Note: these functions require an uppercase letter input! */
80	0	static zend_always_inline char encode(char c) {
81	0	if (isalpha((unsigned char)c)) {
82	0	ZEND_ASSERT(c >= 'A' && c <= 'Z');
83	0	return _codes[(c - 'A')];
84	0	} else {
85	0	return 0;
86	0	}
87	0	}
88
89	0	#define isvowel(c) (encode(c) & 1) /* AEIOU */
90
91		/* These letters are passed through unchanged */
92		#define NOCHANGE(c) (encode(c) & 2) /* FJMNR */
93
94		/* These form diphthongs when preceding H */
95	0	#define AFFECTH(c) (encode(c) & 4) /* CGPST */
96
97		/* These make C and G soft */
98	0	#define MAKESOFT(c) (encode(c) & 8) /* EIY */
99
100		/* These prevent GH from becoming F */
101	0	#define NOGHTOF(c) (encode(c) & 16) /* BDH */
102
103		/----------------------------- /
104		/* end of "metachar.h" */
105		/----------------------------- /
106
107		/* I suppose I could have been using a character pointer instead of
108		* accessing the array directly... */
109
110	0	#define Convert_Raw(c) toupper((unsigned char)c)
111		/* Look at the next letter in the word */
112	0	#define Read_Raw_Next_Letter (word[w_idx+1])
113	0	#define Read_Next_Letter (Convert_Raw(Read_Raw_Next_Letter))
114		/* Look at the current letter in the word */
115	0	#define Read_Raw_Curr_Letter (word[w_idx])
116		#define Read_Curr_Letter (Convert_Raw(Read_Raw_Curr_Letter))
117		/* Go N letters back. */
118	0	#define Look_Back_Letter(n) (w_idx >= n ? Convert_Raw(word[w_idx-n]) : '\0')
119		/* Previous letter. I dunno, should this return null on failure? */
120	0	#define Read_Prev_Letter (Look_Back_Letter(1))
121		/* Look two letters down. It makes sure you don't walk off the string. */
122	0	#define Read_After_Next_Letter (Read_Raw_Next_Letter != '\0' ? Convert_Raw(word[w_idx+2]) \
123	0	: '\0')
124	0	#define Look_Ahead_Letter(n) (toupper((unsigned char)Lookahead((char *) word+w_idx, n)))
125
126
127		/* Allows us to safely look ahead an arbitrary # of letters */
128		/* I probably could have just used strlen... */
129		static char Lookahead(char *word, size_t how_far)
130	0	{
131	0	size_t idx;
132	0	for (idx = 0; word[idx] != '\0' && idx < how_far; idx++);
133		/* Edge forward in the string... */
134
135	0	return word[idx]; /* idx will be either == to how_far or
136		* at the end of the string where it will be null
137		*/
138	0	}
139
140
141		/* phonize one letter
142		* We don't know the buffers size in advance. On way to solve this is to just
143		* re-allocate the buffer size. We're using an extra of 2 characters (this
144		* could be one though; or more too). */
145	0	#define Phonize(c) { \
146	0	if (p_idx >= max_buffer_len) { \
147	0	phoned_word = zend_string_extend(phoned_word, 2 * sizeof(char) + max_buffer_len, 0); \
148	0	max_buffer_len += 2; \
149	0	} \
150	0	ZSTR_VAL(*phoned_word)[p_idx++] = c; \
151	0	ZSTR_LEN(*phoned_word) = p_idx; \
152	0	}
153		/* Slap a null character on the end of the phoned word */
154	0	#define End_Phoned_Word() { \
155	0	if (p_idx == max_buffer_len) { \
156	0	phoned_word = zend_string_extend(phoned_word, 1 * sizeof(char) + max_buffer_len, 0); \
157	0	max_buffer_len += 1; \
158	0	} \
159	0	ZSTR_VAL(*phoned_word)[p_idx] = '\0'; \
160	0	ZSTR_LEN(*phoned_word) = p_idx; \
161	0	}
162		/* How long is the phoned word? */
163	0	#define Phone_Len (p_idx)
164
165		/* Note is a letter is a 'break' in the word */
166	0	#define Isbreak(c) (!isalpha((unsigned char)(c)))
167
168		/* {{{ metaphone */
169		static void metaphone(unsigned char word, size_t word_len, zend_long max_phonemes, zend_string *phoned_word, int traditional)
170	0	{
171	0	size_t w_idx = 0; /* point in the phonization we're at. */
172	0	size_t p_idx = 0; /* end of the phoned phrase */
173	0	size_t max_buffer_len = 0; /* maximum length of the destination buffer */
174	0	char curr_letter;
175	0	ZEND_ASSERT(word != NULL);
176	0	ZEND_ASSERT(max_phonemes >= 0);
177
178		/-- Allocate memory for our phoned_phrase --/
179	0	if (max_phonemes == 0) { /* Assume largest possible */
180	0	max_buffer_len = word_len;
181	0	phoned_word = zend_string_alloc(sizeof(char) word_len + 1, 0);
182	0	} else {
183	0	max_buffer_len = max_phonemes;
184	0	phoned_word = zend_string_alloc(sizeof(char) max_phonemes + 1, 0);
185	0	}
186
187
188		/-- The first phoneme has to be processed specially. --/
189		/* Find our first letter */
190	0	for (; !isalpha((unsigned char)(curr_letter = Read_Raw_Curr_Letter)); w_idx++) {
191		/* On the off chance we were given nothing but crap... */
192	0	if (curr_letter == '\0') {
193	0	End_Phoned_Word();
194	0	return;
195	0	}
196	0	}
197
198	0	curr_letter = Convert_Raw(curr_letter);
199
200	0	switch (curr_letter) {
201		/* AE becomes E */
202	0	case 'A':
203	0	if (Read_Next_Letter == 'E') {
204	0	Phonize('E');
205	0	w_idx += 2;
206	0	}
207		/* Remember, preserve vowels at the beginning */
208	0	else {
209	0	Phonize('A');
210	0	w_idx++;
211	0	}
212	0	break;
213		/* [GKP]N becomes N */
214	0	case 'G':
215	0	case 'K':
216	0	case 'P':
217	0	if (Read_Next_Letter == 'N') {
218	0	Phonize('N');
219	0	w_idx += 2;
220	0	}
221	0	break;
222		/* WH becomes W,
223		WR becomes R
224		W if followed by a vowel */
225	0	case 'W': {
226	0	char next_letter = Read_Next_Letter;
227	0	if (next_letter == 'R') {
228	0	Phonize('R');
229	0	w_idx += 2;
230	0	} else if (next_letter == 'H' \|\| isvowel(next_letter)) {
231	0	Phonize('W');
232	0	w_idx += 2;
233	0	}
234		/* else ignore */
235	0	break;
236	0	}
237		/* X becomes S */
238	0	case 'X':
239	0	Phonize('S');
240	0	w_idx++;
241	0	break;
242		/* Vowels are kept */
243		/* We did A already
244		case 'A':
245		case 'a':
246		*/
247	0	case 'E':
248	0	case 'I':
249	0	case 'O':
250	0	case 'U':
251	0	Phonize(curr_letter);
252	0	w_idx++;
253	0	break;
254	0	default:
255		/* do nothing */
256	0	break;
257	0	}
258
259
260
261		/* On to the metaphoning */
262	0	for (; (curr_letter = Read_Raw_Curr_Letter) != '\0' &&
263	0	(max_phonemes == 0 \|\| Phone_Len < (size_t)max_phonemes);
264	0	w_idx++) {
265		/* How many letters to skip because an earlier encoding handled
266		* multiple letters */
267	0	unsigned short int skip_letter = 0;
268
269
270		/* THOUGHT: It would be nice if, rather than having things like...
271		* well, SCI. For SCI you encode the S, then have to remember
272		* to skip the C. So the phonome SCI invades both S and C. It would
273		* be better, IMHO, to skip the C from the S part of the encoding.
274		* Hell, I'm trying it.
275		*/
276
277		/* Ignore non-alphas */
278	0	if (!isalpha((unsigned char)curr_letter))
279	0	continue;
280
281	0	curr_letter = Convert_Raw(curr_letter);
282		/* Note: we can't cache curr_letter from the previous loop
283		* because of the skip_letter variable. */
284	0	char prev_letter = Read_Prev_Letter;
285
286		/* Drop duplicates, except CC */
287	0	if (curr_letter == prev_letter &&
288	0	curr_letter != 'C')
289	0	continue;
290
291	0	switch (curr_letter) {
292		/* B -> B unless in MB */
293	0	case 'B':
294	0	if (prev_letter != 'M')
295	0	Phonize('B');
296	0	break;
297		/* 'sh' if -CIA- or -CH, but not SCH, except SCHW.
298		* (SCHW is handled in S)
299		* S if -CI-, -CE- or -CY-
300		* dropped if -SCI-, SCE-, -SCY- (handed in S)
301		* else K
302		*/
303	0	case 'C': {
304	0	char next_letter = Read_Next_Letter;
305	0	if (MAKESOFT(next_letter)) { /* C[IEY] */
306	0	if (next_letter == 'I' && Read_After_Next_Letter == 'A') { /* CIA */
307	0	Phonize(SH);
308	0	}
309		/* SC[IEY] */
310	0	else if (prev_letter == 'S') {
311		/* Dropped */
312	0	} else {
313	0	Phonize('S');
314	0	}
315	0	} else if (next_letter == 'H') {
316	0	if ((!traditional) && (prev_letter == 'S' \|\| Read_After_Next_Letter == 'R')) { /* Christ, School */
317	0	Phonize('K');
318	0	} else {
319	0	Phonize(SH);
320	0	}
321	0	skip_letter++;
322	0	} else {
323	0	Phonize('K');
324	0	}
325	0	break;
326	0	}
327		/* J if in -DGE-, -DGI- or -DGY-
328		* else T
329		*/
330	0	case 'D':
331	0	if (Read_Next_Letter == 'G' &&
332	0	MAKESOFT(Read_After_Next_Letter)) {
333	0	Phonize('J');
334	0	skip_letter++;
335	0	} else
336	0	Phonize('T');
337	0	break;
338		/* F if in -GH and not B--GH, D--GH, -H--GH, -H---GH
339		* else dropped if -GNED, -GN,
340		* else dropped if -DGE-, -DGI- or -DGY- (handled in D)
341		* else J if in -GE-, -GI, -GY and not GG
342		* else K
343		*/
344	0	case 'G': {
345	0	char next_letter = Read_Next_Letter;
346	0	if (next_letter == 'H') {
347	0	if (!(NOGHTOF(Look_Back_Letter(3)) \|\|
348	0	Look_Back_Letter(4) == 'H')) {
349	0	Phonize('F');
350	0	skip_letter++;
351	0	} else {
352		/* silent */
353	0	}
354	0	} else if (next_letter == 'N') {
355	0	char after_next_letter = Read_After_Next_Letter;
356	0	if (Isbreak(after_next_letter) \|\|
357	0	(after_next_letter == 'E' &&
358	0	Look_Ahead_Letter(3) == 'D')) {
359		/* dropped */
360	0	} else
361	0	Phonize('K');
362	0	} else if (MAKESOFT(next_letter) &&
363	0	prev_letter != 'G') {
364	0	Phonize('J');
365	0	} else {
366	0	Phonize('K');
367	0	}
368	0	break;
369	0	}
370		/* H if before a vowel and not after C,G,P,S,T */
371	0	case 'H':
372	0	if (isvowel(Read_Next_Letter) &&
373	0	!AFFECTH(prev_letter))
374	0	Phonize('H');
375	0	break;
376		/* dropped if after C
377		* else K
378		*/
379	0	case 'K':
380	0	if (prev_letter != 'C')
381	0	Phonize('K');
382	0	break;
383		/* F if before H
384		* else P
385		*/
386	0	case 'P':
387	0	if (Read_Next_Letter == 'H') {
388	0	Phonize('F');
389	0	} else {
390	0	Phonize('P');
391	0	}
392	0	break;
393		/* K
394		*/
395	0	case 'Q':
396	0	Phonize('K');
397	0	break;
398		/* 'sh' in -SH-, -SIO- or -SIA- or -SCHW-
399		* else S
400		*/
401	0	case 'S': {
402	0	char next_letter = Read_Next_Letter;
403	0	char after_next_letter;
404	0	if (next_letter == 'I' &&
405	0	((after_next_letter = Read_After_Next_Letter) == 'O' \|\|
406	0	after_next_letter == 'A')) {
407	0	Phonize(SH);
408	0	} else if (next_letter == 'H') {
409	0	Phonize(SH);
410	0	skip_letter++;
411	0	} else if ((!traditional) && (next_letter == 'C' && Look_Ahead_Letter(2) == 'H' && Look_Ahead_Letter(3) == 'W')) {
412	0	Phonize(SH);
413	0	skip_letter += 2;
414	0	} else {
415	0	Phonize('S');
416	0	}
417	0	break;
418	0	}
419		/* 'sh' in -TIA- or -TIO-
420		* else 'th' before H
421		* else T
422		*/
423	0	case 'T': {
424	0	char next_letter = Read_Next_Letter;
425	0	char after_next_letter;
426	0	if (next_letter == 'I' &&
427	0	((after_next_letter = Read_After_Next_Letter) == 'O' \|\|
428	0	after_next_letter == 'A')) {
429	0	Phonize(SH);
430	0	} else if (next_letter == 'H') {
431	0	Phonize(TH);
432	0	skip_letter++;
433	0	} else if (!(next_letter == 'C' && Read_After_Next_Letter == 'H')) {
434	0	Phonize('T');
435	0	}
436	0	break;
437	0	}
438		/* F */
439	0	case 'V':
440	0	Phonize('F');
441	0	break;
442		/* W before a vowel, else dropped */
443	0	case 'W':
444	0	if (isvowel(Read_Next_Letter))
445	0	Phonize('W');
446	0	break;
447		/* KS */
448	0	case 'X':
449	0	Phonize('K');
450	0	Phonize('S');
451	0	break;
452		/* Y if followed by a vowel */
453	0	case 'Y':
454	0	if (isvowel(Read_Next_Letter))
455	0	Phonize('Y');
456	0	break;
457		/* S */
458	0	case 'Z':
459	0	Phonize('S');
460	0	break;
461		/* No transformation */
462	0	case 'F':
463	0	case 'J':
464	0	case 'L':
465	0	case 'M':
466	0	case 'N':
467	0	case 'R':
468	0	Phonize(curr_letter);
469	0	break;
470	0	default:
471		/* nothing */
472	0	break;
473	0	} /* END SWITCH */
474
475	0	w_idx += skip_letter;
476	0	} /* END FOR */
477
478	0	End_Phoned_Word();
479	0	} /* END metaphone */
480		/* }}} */