2001-08-07 20:47:43 +04:00
|
|
|
/*
|
|
|
|
* fuzzystrmatch.c
|
|
|
|
*
|
|
|
|
* Functions for "fuzzy" comparison of strings
|
|
|
|
*
|
Jim C. Nasby wrote:
> Second argument to metaphone is suposed to set the limit on the
> number of characters to return, but it breaks on some phrases:
>
> usps=# select metaphone(a,3),metaphone(a,4),metaphone(a,20) from
> (select 'Hello world'::varchar AS a) a;
> HLW | HLWR | HLWRLT
>
> usps=# select metaphone(a,3),metaphone(a,4),metaphone(a,20) from
> (select 'A A COMEAUX MEMORIAL'::varchar AS a) a;
> AKM | AKMKS | AKMKSMMRL
>
> In every case I've found that does this, the 4th and 5th letters are
> always 'KS'.
Nice catch.
There was a bug in the original metaphone algorithm from CPAN. Patch
attached (while I was at it I updated my email address, changed the
copyright to PGDG, and removed an unnecessary palloc). Here's how it
looks now:
regression=# select metaphone(a,4) from (select 'A A COMEAUX
MEMORIAL'::varchar AS a) a;
metaphone
-----------
AKMK
(1 row)
regression=# select metaphone(a,5) from (select 'A A COMEAUX
MEMORIAL'::varchar AS a) a;
metaphone
-----------
AKMKS
(1 row)
Joe Conway
2003-06-25 02:59:46 +04:00
|
|
|
* Joe Conway <mail@joeconway.com>
|
|
|
|
*
|
|
|
|
* Copyright (c) 2001, 2002, 2003 by PostgreSQL Global Development Group
|
|
|
|
* ALL RIGHTS RESERVED;
|
2001-08-07 20:47:43 +04:00
|
|
|
*
|
|
|
|
* levenshtein()
|
|
|
|
* -------------
|
|
|
|
* Written based on a description of the algorithm by Michael Gilleland
|
|
|
|
* found at http://www.merriampark.com/ld.htm
|
|
|
|
* Also looked at levenshtein.c in the PHP 4.0.6 distribution for
|
|
|
|
* inspiration.
|
|
|
|
*
|
|
|
|
* metaphone()
|
|
|
|
* -----------
|
|
|
|
* Modified for PostgreSQL by Joe Conway.
|
|
|
|
* Based on CPAN's "Text-Metaphone-1.96" by Michael G Schwern <schwern@pobox.com>
|
|
|
|
* Code slightly modified for use as PostgreSQL function (palloc, elog, etc).
|
|
|
|
* Metaphone was originally created by Lawrence Philips and presented in article
|
|
|
|
* in "Computer Language" December 1990 issue.
|
|
|
|
*
|
|
|
|
* Permission to use, copy, modify, and distribute this software and its
|
|
|
|
* documentation for any purpose, without fee, and without a written agreement
|
|
|
|
* is hereby granted, provided that the above copyright notice and this
|
|
|
|
* paragraph and the following two paragraphs appear in all copies.
|
2001-10-25 09:50:21 +04:00
|
|
|
*
|
2001-08-07 20:47:43 +04:00
|
|
|
* IN NO EVENT SHALL THE AUTHORS OR DISTRIBUTORS BE LIABLE TO ANY PARTY FOR
|
|
|
|
* DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES, INCLUDING
|
|
|
|
* LOST PROFITS, ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS
|
|
|
|
* DOCUMENTATION, EVEN IF THE AUTHOR OR DISTRIBUTORS HAVE BEEN ADVISED OF THE
|
|
|
|
* POSSIBILITY OF SUCH DAMAGE.
|
2001-10-25 09:50:21 +04:00
|
|
|
*
|
2001-08-07 20:47:43 +04:00
|
|
|
* THE AUTHORS AND DISTRIBUTORS SPECIFICALLY DISCLAIM ANY WARRANTIES,
|
|
|
|
* INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY
|
|
|
|
* AND FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS
|
|
|
|
* ON AN "AS IS" BASIS, AND THE AUTHOR AND DISTRIBUTORS HAS NO OBLIGATIONS TO
|
|
|
|
* PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include "fuzzystrmatch.h"
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Calculates Levenshtein Distance between two strings.
|
|
|
|
* Uses simplest and fastest cost model only, i.e. assumes a cost of 1 for
|
|
|
|
* each deletion, substitution, or insertion.
|
|
|
|
*/
|
|
|
|
PG_FUNCTION_INFO_V1(levenshtein);
|
|
|
|
Datum
|
|
|
|
levenshtein(PG_FUNCTION_ARGS)
|
|
|
|
{
|
2001-10-25 09:50:21 +04:00
|
|
|
char *str_s;
|
|
|
|
char *str_s0;
|
|
|
|
char *str_t;
|
|
|
|
int cols = 0;
|
|
|
|
int rows = 0;
|
|
|
|
int *u_cells;
|
|
|
|
int *l_cells;
|
|
|
|
int *tmp;
|
|
|
|
int i;
|
|
|
|
int j;
|
2001-08-07 20:47:43 +04:00
|
|
|
|
|
|
|
/*
|
2001-10-25 09:50:21 +04:00
|
|
|
* Fetch the arguments. str_s is referred to as the "source" cols =
|
|
|
|
* length of source + 1 to allow for the initialization column str_t
|
|
|
|
* is referred to as the "target", rows = length of target + 1 rows =
|
|
|
|
* length of target + 1 to allow for the initialization row
|
2001-08-07 20:47:43 +04:00
|
|
|
*/
|
|
|
|
str_s = DatumGetCString(DirectFunctionCall1(textout, PointerGetDatum(PG_GETARG_TEXT_P(0))));
|
|
|
|
str_t = DatumGetCString(DirectFunctionCall1(textout, PointerGetDatum(PG_GETARG_TEXT_P(1))));
|
|
|
|
|
|
|
|
cols = strlen(str_s) + 1;
|
|
|
|
rows = strlen(str_t) + 1;
|
|
|
|
|
|
|
|
/*
|
2001-10-25 09:50:21 +04:00
|
|
|
* Restrict the length of the strings being compared to something
|
|
|
|
* reasonable because we will have to perform rows * cols
|
2003-03-11 01:28:22 +03:00
|
|
|
* calculations. If longer strings need to be compared, increase
|
2001-10-25 09:50:21 +04:00
|
|
|
* MAX_LEVENSHTEIN_STRLEN to suit (but within your tolerance for speed
|
|
|
|
* and memory usage).
|
2001-08-07 20:47:43 +04:00
|
|
|
*/
|
|
|
|
if ((cols > MAX_LEVENSHTEIN_STRLEN + 1) || (rows > MAX_LEVENSHTEIN_STRLEN + 1))
|
2003-07-24 21:52:50 +04:00
|
|
|
ereport(ERROR,
|
|
|
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
|
|
|
|
errmsg("argument exceeds max length: %d",
|
|
|
|
MAX_LEVENSHTEIN_STRLEN)));
|
2001-08-07 20:47:43 +04:00
|
|
|
|
|
|
|
/*
|
2001-10-25 09:50:21 +04:00
|
|
|
* If either rows or cols is 0, the answer is the other value. This
|
|
|
|
* makes sense since it would take that many insertions the build a
|
|
|
|
* matching string
|
2001-08-07 20:47:43 +04:00
|
|
|
*/
|
|
|
|
|
|
|
|
if (cols == 0)
|
|
|
|
PG_RETURN_INT32(rows);
|
|
|
|
|
|
|
|
if (rows == 0)
|
|
|
|
PG_RETURN_INT32(cols);
|
|
|
|
|
|
|
|
/*
|
2001-10-25 09:50:21 +04:00
|
|
|
* Allocate two vectors of integers. One will be used for the "upper"
|
|
|
|
* row, the other for the "lower" row. Initialize the "upper" row to
|
|
|
|
* 0..cols
|
2001-08-07 20:47:43 +04:00
|
|
|
*/
|
|
|
|
u_cells = palloc(sizeof(int) * cols);
|
|
|
|
for (i = 0; i < cols; i++)
|
|
|
|
u_cells[i] = i;
|
|
|
|
|
|
|
|
l_cells = palloc(sizeof(int) * cols);
|
|
|
|
|
|
|
|
/*
|
2001-10-25 09:50:21 +04:00
|
|
|
* Use str_s0 to "rewind" the pointer to str_s in the nested for loop
|
|
|
|
* below
|
2001-08-07 20:47:43 +04:00
|
|
|
*/
|
|
|
|
str_s0 = str_s;
|
|
|
|
|
|
|
|
/*
|
2001-10-25 09:50:21 +04:00
|
|
|
* Loop throught the rows, starting at row 1. Row 0 is used for the
|
|
|
|
* initial "upper" row.
|
2001-08-07 20:47:43 +04:00
|
|
|
*/
|
|
|
|
for (j = 1; j < rows; j++)
|
|
|
|
{
|
|
|
|
/*
|
2001-10-25 09:50:21 +04:00
|
|
|
* We'll always start with col 1, and initialize lower row col 0
|
|
|
|
* to j
|
2001-08-07 20:47:43 +04:00
|
|
|
*/
|
|
|
|
l_cells[0] = j;
|
|
|
|
|
|
|
|
for (i = 1; i < cols; i++)
|
|
|
|
{
|
2001-10-25 09:50:21 +04:00
|
|
|
int c = 0;
|
|
|
|
int c1 = 0;
|
|
|
|
int c2 = 0;
|
|
|
|
int c3 = 0;
|
2001-08-07 20:47:43 +04:00
|
|
|
|
|
|
|
/*
|
2001-10-25 09:50:21 +04:00
|
|
|
* The "cost" value is 0 if the character at the current col
|
|
|
|
* position in the source string, matches the character at the
|
|
|
|
* current row position in the target string; cost is 1
|
|
|
|
* otherwise.
|
2001-08-07 20:47:43 +04:00
|
|
|
*/
|
|
|
|
c = ((CHAREQ(str_s, str_t)) ? 0 : 1);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* c1 is upper right cell plus 1
|
|
|
|
*/
|
|
|
|
c1 = u_cells[i] + 1;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* c2 is lower left cell plus 1
|
|
|
|
*/
|
|
|
|
c2 = l_cells[i - 1] + 1;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* c3 is cell diagonally above to the left plus "cost"
|
|
|
|
*/
|
|
|
|
c3 = u_cells[i - 1] + c;
|
|
|
|
|
|
|
|
/*
|
2001-10-25 09:50:21 +04:00
|
|
|
* The lower right cell is set to the minimum of c1, c2, c3
|
2001-08-07 20:47:43 +04:00
|
|
|
*/
|
|
|
|
l_cells[i] = (c1 < c2 ? c1 : c2) < c3 ? (c1 < c2 ? c1 : c2) : c3;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Increment the pointer to str_s
|
|
|
|
*/
|
|
|
|
NextChar(str_s);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2001-10-25 09:50:21 +04:00
|
|
|
* Lower row now becomes the upper row, and the upper row gets
|
|
|
|
* reused as the new lower row.
|
2001-08-07 20:47:43 +04:00
|
|
|
*/
|
|
|
|
tmp = u_cells;
|
|
|
|
u_cells = l_cells;
|
|
|
|
l_cells = tmp;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Increment the pointer to str_t
|
|
|
|
*/
|
|
|
|
NextChar(str_t);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Rewind the pointer to str_s
|
|
|
|
*/
|
|
|
|
str_s = str_s0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2001-10-25 09:50:21 +04:00
|
|
|
* Because the final value (at position row, col) was swapped from the
|
|
|
|
* lower row to the upper row, that's where we'll find it.
|
2001-08-07 20:47:43 +04:00
|
|
|
*/
|
|
|
|
PG_RETURN_INT32(u_cells[cols - 1]);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Calculates the metaphone of an input string.
|
|
|
|
* Returns number of characters requested
|
|
|
|
* (suggested value is 4)
|
|
|
|
*/
|
|
|
|
PG_FUNCTION_INFO_V1(metaphone);
|
|
|
|
Datum
|
|
|
|
metaphone(PG_FUNCTION_ARGS)
|
|
|
|
{
|
|
|
|
int reqlen;
|
2001-10-25 09:50:21 +04:00
|
|
|
char *str_i;
|
2001-08-07 20:47:43 +04:00
|
|
|
size_t str_i_len;
|
2001-10-25 09:50:21 +04:00
|
|
|
char *metaph;
|
|
|
|
text *result_text;
|
2001-08-07 20:47:43 +04:00
|
|
|
int retval;
|
|
|
|
|
|
|
|
str_i = DatumGetCString(DirectFunctionCall1(textout, PointerGetDatum(PG_GETARG_TEXT_P(0))));
|
|
|
|
str_i_len = strlen(str_i);
|
|
|
|
|
|
|
|
if (str_i_len > MAX_METAPHONE_STRLEN)
|
2003-07-24 21:52:50 +04:00
|
|
|
ereport(ERROR,
|
|
|
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
|
|
|
|
errmsg("argument exceeds max length: %d",
|
|
|
|
MAX_METAPHONE_STRLEN)));
|
|
|
|
|
2001-08-07 20:47:43 +04:00
|
|
|
if (!(str_i_len > 0))
|
2003-07-24 21:52:50 +04:00
|
|
|
ereport(ERROR,
|
|
|
|
(errcode(ERRCODE_ZERO_LENGTH_CHARACTER_STRING),
|
|
|
|
errmsg("argument is empty string")));
|
2001-08-07 20:47:43 +04:00
|
|
|
|
|
|
|
reqlen = PG_GETARG_INT32(1);
|
|
|
|
if (reqlen > MAX_METAPHONE_STRLEN)
|
2003-07-24 21:52:50 +04:00
|
|
|
ereport(ERROR,
|
|
|
|
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
|
|
|
|
errmsg("output length exceeds max length: %d",
|
|
|
|
MAX_METAPHONE_STRLEN)));
|
|
|
|
|
2001-08-07 20:47:43 +04:00
|
|
|
if (!(reqlen > 0))
|
2003-07-24 21:52:50 +04:00
|
|
|
ereport(ERROR,
|
|
|
|
(errcode(ERRCODE_ZERO_LENGTH_CHARACTER_STRING),
|
|
|
|
errmsg("output cannot be empty string")));
|
|
|
|
|
2001-08-07 20:47:43 +04:00
|
|
|
|
|
|
|
retval = _metaphone(str_i, reqlen, &metaph);
|
|
|
|
if (retval == META_SUCCESS)
|
|
|
|
{
|
|
|
|
result_text = DatumGetTextP(DirectFunctionCall1(textin, CStringGetDatum(metaph)));
|
|
|
|
PG_RETURN_TEXT_P(result_text);
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
2003-07-24 21:52:50 +04:00
|
|
|
/* internal error */
|
2001-08-07 20:47:43 +04:00
|
|
|
elog(ERROR, "metaphone: failure");
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Keep the compiler quiet
|
|
|
|
*/
|
|
|
|
PG_RETURN_NULL();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2001-10-25 09:50:21 +04:00
|
|
|
/*
|
2001-08-07 20:47:43 +04:00
|
|
|
* Original code by Michael G Schwern starts here.
|
|
|
|
* Code slightly modified for use as PostgreSQL
|
|
|
|
* function (palloc, etc). Original includes
|
|
|
|
* are rolled into fuzzystrmatch.h
|
|
|
|
*------------------------------------------------------------------*/
|
|
|
|
|
|
|
|
/* I suppose I could have been using a character pointer instead of
|
2003-03-11 01:28:22 +03:00
|
|
|
* accessing the array directly... */
|
2001-08-07 20:47:43 +04:00
|
|
|
|
|
|
|
/* Look at the next letter in the word */
|
2001-12-31 02:09:42 +03:00
|
|
|
#define Next_Letter (toupper((unsigned char) word[w_idx+1]))
|
2001-08-07 20:47:43 +04:00
|
|
|
/* Look at the current letter in the word */
|
2001-12-31 02:09:42 +03:00
|
|
|
#define Curr_Letter (toupper((unsigned char) word[w_idx]))
|
2001-08-07 20:47:43 +04:00
|
|
|
/* Go N letters back. */
|
2001-12-31 02:09:42 +03:00
|
|
|
#define Look_Back_Letter(n) \
|
|
|
|
(w_idx >= (n) ? toupper((unsigned char) word[w_idx-(n)]) : '\0')
|
2001-08-07 20:47:43 +04:00
|
|
|
/* Previous letter. I dunno, should this return null on failure? */
|
|
|
|
#define Prev_Letter (Look_Back_Letter(1))
|
|
|
|
/* Look two letters down. It makes sure you don't walk off the string. */
|
2001-12-31 02:09:42 +03:00
|
|
|
#define After_Next_Letter \
|
|
|
|
(Next_Letter != '\0' ? toupper((unsigned char) word[w_idx+2]) : '\0')
|
|
|
|
#define Look_Ahead_Letter(n) toupper((unsigned char) Lookahead(word+w_idx, n))
|
2001-08-07 20:47:43 +04:00
|
|
|
|
2001-10-25 09:50:21 +04:00
|
|
|
|
2001-08-07 20:47:43 +04:00
|
|
|
/* Allows us to safely look ahead an arbitrary # of letters */
|
|
|
|
/* I probably could have just used strlen... */
|
2001-10-25 09:50:21 +04:00
|
|
|
char
|
|
|
|
Lookahead(char *word, int how_far)
|
|
|
|
{
|
|
|
|
char letter_ahead = '\0'; /* null by default */
|
|
|
|
int idx;
|
|
|
|
|
|
|
|
for (idx = 0; word[idx] != '\0' && idx < how_far; idx++);
|
|
|
|
/* Edge forward in the string... */
|
|
|
|
|
|
|
|
letter_ahead = word[idx]; /* idx will be either == to how_far or at
|
|
|
|
* the end of the string */
|
2001-08-07 20:47:43 +04:00
|
|
|
return letter_ahead;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/* phonize one letter */
|
2001-10-25 05:29:37 +04:00
|
|
|
#define Phonize(c) do {(*phoned_word)[p_idx++] = c;} while (0)
|
2001-08-07 20:47:43 +04:00
|
|
|
/* Slap a null character on the end of the phoned word */
|
2001-10-25 09:50:21 +04:00
|
|
|
#define End_Phoned_Word do {(*phoned_word)[p_idx] = '\0';} while (0)
|
2001-08-07 20:47:43 +04:00
|
|
|
/* How long is the phoned word? */
|
|
|
|
#define Phone_Len (p_idx)
|
|
|
|
|
|
|
|
/* Note is a letter is a 'break' in the word */
|
2001-12-31 02:09:42 +03:00
|
|
|
#define Isbreak(c) (!isalpha((unsigned char) (c)))
|
2001-10-25 09:50:21 +04:00
|
|
|
|
|
|
|
|
|
|
|
int
|
|
|
|
_metaphone(
|
|
|
|
/* IN */
|
|
|
|
char *word,
|
|
|
|
int max_phonemes,
|
|
|
|
/* OUT */
|
|
|
|
char **phoned_word
|
|
|
|
)
|
|
|
|
{
|
|
|
|
int w_idx = 0; /* point in the phonization we're at. */
|
|
|
|
int p_idx = 0; /* end of the phoned phrase */
|
|
|
|
|
2001-08-07 20:47:43 +04:00
|
|
|
/*-- Parameter checks --*/
|
|
|
|
|
|
|
|
/*
|
2001-10-25 09:50:21 +04:00
|
|
|
* Shouldn't be necessary, but left these here anyway jec Aug 3, 2001
|
2001-08-07 20:47:43 +04:00
|
|
|
*/
|
|
|
|
|
|
|
|
/* Negative phoneme length is meaningless */
|
|
|
|
if (!(max_phonemes > 0))
|
2003-07-24 21:52:50 +04:00
|
|
|
/* internal error */
|
2001-08-07 20:47:43 +04:00
|
|
|
elog(ERROR, "metaphone: Requested output length must be > 0");
|
|
|
|
|
|
|
|
/* Empty/null string is meaningless */
|
|
|
|
if ((word == NULL) || !(strlen(word) > 0))
|
2003-07-24 21:52:50 +04:00
|
|
|
/* internal error */
|
2001-08-07 20:47:43 +04:00
|
|
|
elog(ERROR, "metaphone: Input string length must be > 0");
|
2001-10-25 09:50:21 +04:00
|
|
|
|
2001-08-07 20:47:43 +04:00
|
|
|
/*-- Allocate memory for our phoned_phrase --*/
|
2001-10-25 09:50:21 +04:00
|
|
|
if (max_phonemes == 0)
|
|
|
|
{ /* Assume largest possible */
|
|
|
|
*phoned_word = palloc(sizeof(char) * strlen(word) +1);
|
2001-08-07 20:47:43 +04:00
|
|
|
if (!*phoned_word)
|
|
|
|
return META_ERROR;
|
2001-10-25 09:50:21 +04:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
2001-08-07 20:47:43 +04:00
|
|
|
*phoned_word = palloc(sizeof(char) * max_phonemes + 1);
|
|
|
|
if (!*phoned_word)
|
|
|
|
return META_ERROR;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*-- The first phoneme has to be processed specially. --*/
|
|
|
|
/* Find our first letter */
|
2001-12-31 02:09:42 +03:00
|
|
|
for (; !isalpha((unsigned char) (Curr_Letter)); w_idx++)
|
2001-10-25 09:50:21 +04:00
|
|
|
{
|
2001-08-07 20:47:43 +04:00
|
|
|
/* On the off chance we were given nothing but crap... */
|
2001-10-25 09:50:21 +04:00
|
|
|
if (Curr_Letter == '\0')
|
|
|
|
{
|
2001-10-29 22:41:54 +03:00
|
|
|
End_Phoned_Word;
|
|
|
|
return META_SUCCESS; /* For testing */
|
2001-08-07 20:47:43 +04:00
|
|
|
}
|
2001-10-25 09:50:21 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
switch (Curr_Letter)
|
|
|
|
{
|
|
|
|
/* AE becomes E */
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'A':
|
2001-10-25 09:50:21 +04:00
|
|
|
if (Next_Letter == 'E')
|
|
|
|
{
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize('E');
|
2001-10-25 09:50:21 +04:00
|
|
|
w_idx += 2;
|
2001-08-07 20:47:43 +04:00
|
|
|
}
|
|
|
|
/* Remember, preserve vowels at the beginning */
|
2001-10-25 09:50:21 +04:00
|
|
|
else
|
|
|
|
{
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize('A');
|
|
|
|
w_idx++;
|
|
|
|
}
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
/* [GKP]N becomes N */
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'G':
|
|
|
|
case 'K':
|
|
|
|
case 'P':
|
2001-10-25 09:50:21 +04:00
|
|
|
if (Next_Letter == 'N')
|
|
|
|
{
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize('N');
|
2001-10-25 09:50:21 +04:00
|
|
|
w_idx += 2;
|
2001-08-07 20:47:43 +04:00
|
|
|
}
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* WH becomes H, WR becomes R W if followed by a vowel
|
|
|
|
*/
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'W':
|
2001-10-25 09:50:21 +04:00
|
|
|
if (Next_Letter == 'H' ||
|
|
|
|
Next_Letter == 'R')
|
2001-08-07 20:47:43 +04:00
|
|
|
{
|
2001-10-25 09:50:21 +04:00
|
|
|
Phonize(Next_Letter);
|
|
|
|
w_idx += 2;
|
2001-08-07 20:47:43 +04:00
|
|
|
}
|
2001-10-25 09:50:21 +04:00
|
|
|
else if (isvowel(Next_Letter))
|
|
|
|
{
|
|
|
|
Phonize('W');
|
|
|
|
w_idx += 2;
|
2001-08-07 20:47:43 +04:00
|
|
|
}
|
|
|
|
/* else ignore */
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
/* X becomes S */
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'X':
|
|
|
|
Phonize('S');
|
|
|
|
w_idx++;
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
/* Vowels are kept */
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We did A already case 'A': case 'a':
|
|
|
|
*/
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'E':
|
|
|
|
case 'I':
|
|
|
|
case 'O':
|
|
|
|
case 'U':
|
|
|
|
Phonize(Curr_Letter);
|
|
|
|
w_idx++;
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
/* do nothing */
|
|
|
|
break;
|
|
|
|
}
|
2001-10-25 09:50:21 +04:00
|
|
|
|
|
|
|
|
|
|
|
|
2001-08-07 20:47:43 +04:00
|
|
|
/* On to the metaphoning */
|
2001-10-25 09:50:21 +04:00
|
|
|
for (; Curr_Letter != '\0' &&
|
|
|
|
(max_phonemes == 0 || Phone_Len < max_phonemes);
|
|
|
|
w_idx++)
|
|
|
|
{
|
|
|
|
/*
|
2003-03-11 01:28:22 +03:00
|
|
|
* How many letters to skip because an earlier encoding handled
|
2001-10-25 09:50:21 +04:00
|
|
|
* multiple letters
|
2001-08-07 20:47:43 +04:00
|
|
|
*/
|
2001-10-25 09:50:21 +04:00
|
|
|
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.
|
|
|
|
*/
|
|
|
|
|
2001-08-07 20:47:43 +04:00
|
|
|
/* Ignore non-alphas */
|
2001-12-31 02:09:42 +03:00
|
|
|
if (!isalpha((unsigned char) (Curr_Letter)))
|
2001-08-07 20:47:43 +04:00
|
|
|
continue;
|
2001-10-25 09:50:21 +04:00
|
|
|
|
2001-08-07 20:47:43 +04:00
|
|
|
/* Drop duplicates, except CC */
|
2001-10-25 09:50:21 +04:00
|
|
|
if (Curr_Letter == Prev_Letter &&
|
|
|
|
Curr_Letter != 'C')
|
2001-08-07 20:47:43 +04:00
|
|
|
continue;
|
2001-10-25 09:50:21 +04:00
|
|
|
|
|
|
|
switch (Curr_Letter)
|
|
|
|
{
|
|
|
|
/* B -> B unless in MB */
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'B':
|
2001-10-25 09:50:21 +04:00
|
|
|
if (Prev_Letter != 'M')
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize('B');
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* '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
|
|
|
|
*/
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'C':
|
2001-10-25 09:50:21 +04:00
|
|
|
if (MAKESOFT(Next_Letter))
|
|
|
|
{ /* C[IEY] */
|
|
|
|
if (After_Next_Letter == 'A' &&
|
|
|
|
Next_Letter == 'I')
|
|
|
|
{ /* CIA */
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize(SH);
|
|
|
|
}
|
|
|
|
/* SC[IEY] */
|
2001-10-25 09:50:21 +04:00
|
|
|
else if (Prev_Letter == 'S')
|
|
|
|
{
|
|
|
|
/* Dropped */
|
2001-08-07 20:47:43 +04:00
|
|
|
}
|
2001-10-25 09:50:21 +04:00
|
|
|
else
|
|
|
|
Phonize('S');
|
2001-08-07 20:47:43 +04:00
|
|
|
}
|
2001-10-25 09:50:21 +04:00
|
|
|
else if (Next_Letter == 'H')
|
|
|
|
{
|
2001-08-07 20:47:43 +04:00
|
|
|
#ifndef USE_TRADITIONAL_METAPHONE
|
2001-10-25 09:50:21 +04:00
|
|
|
if (After_Next_Letter == 'R' ||
|
|
|
|
Prev_Letter == 'S')
|
|
|
|
{ /* Christ, School */
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize('K');
|
|
|
|
}
|
2001-10-25 09:50:21 +04:00
|
|
|
else
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize(SH);
|
|
|
|
#else
|
|
|
|
Phonize(SH);
|
|
|
|
#endif
|
|
|
|
skip_letter++;
|
|
|
|
}
|
2001-10-25 09:50:21 +04:00
|
|
|
else
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize('K');
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* J if in -DGE-, -DGI- or -DGY- else T
|
|
|
|
*/
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'D':
|
2001-10-25 09:50:21 +04:00
|
|
|
if (Next_Letter == 'G' &&
|
|
|
|
MAKESOFT(After_Next_Letter))
|
|
|
|
{
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize('J');
|
|
|
|
skip_letter++;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
Phonize('T');
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* 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
|
|
|
|
*/
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'G':
|
2001-10-25 09:50:21 +04:00
|
|
|
if (Next_Letter == 'H')
|
|
|
|
{
|
|
|
|
if (!(NOGHTOF(Look_Back_Letter(3)) ||
|
|
|
|
Look_Back_Letter(4) == 'H'))
|
|
|
|
{
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize('F');
|
|
|
|
skip_letter++;
|
|
|
|
}
|
2001-10-25 09:50:21 +04:00
|
|
|
else
|
|
|
|
{
|
2001-08-07 20:47:43 +04:00
|
|
|
/* silent */
|
|
|
|
}
|
|
|
|
}
|
2001-10-25 09:50:21 +04:00
|
|
|
else if (Next_Letter == 'N')
|
|
|
|
{
|
|
|
|
if (Isbreak(After_Next_Letter) ||
|
|
|
|
(After_Next_Letter == 'E' &&
|
|
|
|
Look_Ahead_Letter(3) == 'D'))
|
|
|
|
{
|
2001-08-07 20:47:43 +04:00
|
|
|
/* dropped */
|
|
|
|
}
|
|
|
|
else
|
|
|
|
Phonize('K');
|
|
|
|
}
|
2001-10-25 09:50:21 +04:00
|
|
|
else if (MAKESOFT(Next_Letter) &&
|
|
|
|
Prev_Letter != 'G')
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize('J');
|
2001-10-25 09:50:21 +04:00
|
|
|
else
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize('K');
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
/* H if before a vowel and not after C,G,P,S,T */
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'H':
|
2001-10-25 09:50:21 +04:00
|
|
|
if (isvowel(Next_Letter) &&
|
|
|
|
!AFFECTH(Prev_Letter))
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize('H');
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* dropped if after C else K
|
|
|
|
*/
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'K':
|
2001-10-25 09:50:21 +04:00
|
|
|
if (Prev_Letter != 'C')
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize('K');
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* F if before H else P
|
|
|
|
*/
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'P':
|
2001-10-25 09:50:21 +04:00
|
|
|
if (Next_Letter == 'H')
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize('F');
|
2001-10-25 09:50:21 +04:00
|
|
|
else
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize('P');
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* K
|
|
|
|
*/
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'Q':
|
|
|
|
Phonize('K');
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* 'sh' in -SH-, -SIO- or -SIA- or -SCHW- else S
|
|
|
|
*/
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'S':
|
2001-10-25 09:50:21 +04:00
|
|
|
if (Next_Letter == 'I' &&
|
|
|
|
(After_Next_Letter == 'O' ||
|
|
|
|
After_Next_Letter == 'A'))
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize(SH);
|
2001-10-25 09:50:21 +04:00
|
|
|
else if (Next_Letter == 'H')
|
|
|
|
{
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize(SH);
|
|
|
|
skip_letter++;
|
|
|
|
}
|
|
|
|
#ifndef USE_TRADITIONAL_METAPHONE
|
2001-10-25 09:50:21 +04:00
|
|
|
else if (Next_Letter == 'C' &&
|
|
|
|
Look_Ahead_Letter(2) == 'H' &&
|
|
|
|
Look_Ahead_Letter(3) == 'W')
|
|
|
|
{
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize(SH);
|
|
|
|
skip_letter += 2;
|
|
|
|
}
|
|
|
|
#endif
|
2001-10-25 09:50:21 +04:00
|
|
|
else
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize('S');
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* 'sh' in -TIA- or -TIO- else 'th' before H else T
|
|
|
|
*/
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'T':
|
2001-10-25 09:50:21 +04:00
|
|
|
if (Next_Letter == 'I' &&
|
|
|
|
(After_Next_Letter == 'O' ||
|
|
|
|
After_Next_Letter == 'A'))
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize(SH);
|
2001-10-25 09:50:21 +04:00
|
|
|
else if (Next_Letter == 'H')
|
|
|
|
{
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize(TH);
|
|
|
|
skip_letter++;
|
|
|
|
}
|
2001-10-25 09:50:21 +04:00
|
|
|
else
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize('T');
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
/* F */
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'V':
|
|
|
|
Phonize('F');
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
/* W before a vowel, else dropped */
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'W':
|
2001-10-25 09:50:21 +04:00
|
|
|
if (isvowel(Next_Letter))
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize('W');
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
/* KS */
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'X':
|
|
|
|
Phonize('K');
|
Jim C. Nasby wrote:
> Second argument to metaphone is suposed to set the limit on the
> number of characters to return, but it breaks on some phrases:
>
> usps=# select metaphone(a,3),metaphone(a,4),metaphone(a,20) from
> (select 'Hello world'::varchar AS a) a;
> HLW | HLWR | HLWRLT
>
> usps=# select metaphone(a,3),metaphone(a,4),metaphone(a,20) from
> (select 'A A COMEAUX MEMORIAL'::varchar AS a) a;
> AKM | AKMKS | AKMKSMMRL
>
> In every case I've found that does this, the 4th and 5th letters are
> always 'KS'.
Nice catch.
There was a bug in the original metaphone algorithm from CPAN. Patch
attached (while I was at it I updated my email address, changed the
copyright to PGDG, and removed an unnecessary palloc). Here's how it
looks now:
regression=# select metaphone(a,4) from (select 'A A COMEAUX
MEMORIAL'::varchar AS a) a;
metaphone
-----------
AKMK
(1 row)
regression=# select metaphone(a,5) from (select 'A A COMEAUX
MEMORIAL'::varchar AS a) a;
metaphone
-----------
AKMKS
(1 row)
Joe Conway
2003-06-25 02:59:46 +04:00
|
|
|
if (max_phonemes == 0 || Phone_Len < max_phonemes)
|
|
|
|
Phonize('S');
|
2001-08-07 20:47:43 +04:00
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
/* Y if followed by a vowel */
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'Y':
|
2001-10-25 09:50:21 +04:00
|
|
|
if (isvowel(Next_Letter))
|
2001-08-07 20:47:43 +04:00
|
|
|
Phonize('Y');
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
/* S */
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'Z':
|
|
|
|
Phonize('S');
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
/* No transformation */
|
2001-08-07 20:47:43 +04:00
|
|
|
case 'F':
|
|
|
|
case 'J':
|
|
|
|
case 'L':
|
|
|
|
case 'M':
|
|
|
|
case 'N':
|
|
|
|
case 'R':
|
|
|
|
Phonize(Curr_Letter);
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
/* nothing */
|
|
|
|
break;
|
2001-10-25 09:50:21 +04:00
|
|
|
} /* END SWITCH */
|
|
|
|
|
2001-08-07 20:47:43 +04:00
|
|
|
w_idx += skip_letter;
|
2001-10-25 09:50:21 +04:00
|
|
|
} /* END FOR */
|
|
|
|
|
2001-08-07 20:47:43 +04:00
|
|
|
End_Phoned_Word;
|
2001-10-25 09:50:21 +04:00
|
|
|
|
|
|
|
return (META_SUCCESS);
|
|
|
|
} /* END metaphone */
|
2001-08-07 22:16:01 +04:00
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
* SQL function: soundex(text) returns text
|
|
|
|
*/
|
|
|
|
PG_FUNCTION_INFO_V1(soundex);
|
|
|
|
|
|
|
|
Datum
|
|
|
|
soundex(PG_FUNCTION_ARGS)
|
|
|
|
{
|
|
|
|
char outstr[SOUNDEX_LEN + 1];
|
|
|
|
char *arg;
|
|
|
|
|
|
|
|
arg = _textout(PG_GETARG_TEXT_P(0));
|
|
|
|
|
|
|
|
_soundex(arg, outstr);
|
|
|
|
|
|
|
|
PG_RETURN_TEXT_P(_textin(outstr));
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
_soundex(const char *instr, char *outstr)
|
|
|
|
{
|
|
|
|
int count;
|
|
|
|
|
|
|
|
AssertArg(instr);
|
|
|
|
AssertArg(outstr);
|
|
|
|
|
|
|
|
outstr[SOUNDEX_LEN] = '\0';
|
|
|
|
|
|
|
|
/* Skip leading non-alphabetic characters */
|
|
|
|
while (!isalpha((unsigned char) instr[0]) && instr[0])
|
|
|
|
++instr;
|
|
|
|
|
|
|
|
/* No string left */
|
|
|
|
if (!instr[0])
|
|
|
|
{
|
|
|
|
outstr[0] = (char) 0;
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Take the first letter as is */
|
|
|
|
*outstr++ = (char) toupper((unsigned char) *instr++);
|
|
|
|
|
|
|
|
count = 1;
|
|
|
|
while (*instr && count < SOUNDEX_LEN)
|
|
|
|
{
|
|
|
|
if (isalpha((unsigned char) *instr) &&
|
|
|
|
soundex_code(*instr) != soundex_code(*(instr - 1)))
|
|
|
|
{
|
|
|
|
*outstr = soundex_code(instr[0]);
|
|
|
|
if (*outstr != '0')
|
|
|
|
{
|
|
|
|
++outstr;
|
|
|
|
++count;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
++instr;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Fill with 0's */
|
|
|
|
while (count < SOUNDEX_LEN)
|
|
|
|
{
|
|
|
|
*outstr = '0';
|
|
|
|
++outstr;
|
|
|
|
++count;
|
|
|
|
}
|
|
|
|
}
|