Modules/soundex.c

   1 /*
   2   [Header: soundexmodule.c,v 1.2 95/05/02 15:40:45 dwwillia Exp ]
   3
   4   Perform soundex comparisons on strings.
   5
   6   Soundex is an algorithm that hashes English strings into numerical value.
   7   Strings that sound the same are hashed to the same value.  This allows
   8   for non-literal string matching.
   9
  10   From: David Wayne Williams <dwwillia@iucf.indiana.edu>
  11
  12   Apr 29 1996 - added get_soundex method that returns the soundex of a
  13                 string (chrish@qnx.com)
  14   May 2 1996  - added doc strings (chrish@qnx.com)
  15 */
  16
  17 #include <string.h>
  18 #include <ctype.h>
  19 #include "Python.h"
  20
  21 static char soundex_module__doc__[] =
  22 "Perform Soundex comparisons on strings, allowing non-literal matching.";
  23
  24 static void soundex_hash(char *str, char *result)
  25 {
  26     char *sptr = str;           /* pointer into str */
  27     char *rptr = result;        /* pointer into result */
  28
  29     if(*str == '\0')
  30     {
  31         strcpy(result,"000000");
  32         return;
  33     }
  34
  35     /*  Preserve the first character of the input string.
  36      */
  37     *(rptr++) = toupper(*(sptr++));
  38
  39     /* Translate the rest of the input string into result.  The following
  40        transformations are used:
  41
  42        1) All vowels, W, and H, are skipped.
  43
  44        2) BFPV = 1
  45           CGJKQSXZ = 2
  46           DT = 3
  47           L = 4
  48           MN = 5
  49           R = 6
  50
  51        3) Only translate the first of adjacent equal translations.  I.E.
  52           remove duplicate digits.
  53     */
  54
  55     for(;(rptr - result) < 6 &&  *sptr != '\0';sptr++)
  56     {
  57         switch (toupper(*sptr))
  58         {
  59         case 'W':
  60         case 'H':
  61         case 'A':
  62         case 'I':
  63         case 'O':
  64         case 'U':
  65         case 'Y':
  66             break;
  67         case 'B':
  68         case 'F':
  69         case 'P':
  70         case 'V':
  71             if(*(rptr - 1) != '1')
  72                 *(rptr++) = '1';
  73             break;
  74         case 'C':
  75         case 'G':
  76         case 'J':
  77         case 'K':
  78         case 'Q':
  79         case 'S':
  80         case 'X':
  81         case 'Z':
  82             if(*(rptr - 1) != '2')
  83                 *(rptr++) = '2';
  84             break;
  85         case 'D':
  86         case 'T':
  87             if(*(rptr - 1) != '3')
  88                 *(rptr++) = '3';
  89             break;
  90         case 'L':
  91             if(*(rptr - 1) != '4')
  92                 *(rptr++) = '4';
  93             break;
  94         case 'M':
  95         case 'N':
  96             if(*(rptr - 1) != '5')
  97                 *(rptr++) = '5';
  98             break;
  99         case 'R':
 100             if(*(rptr -1) != '6')
 101                 *(rptr++) = '6';
 102         default:
 103             break;
 104         }
 105     }
 106
 107     /* Pad 0's on right side of string out to 6 characters.
 108      */
 109     for(; rptr < result + 6; rptr++)
 110         *rptr = '0';
 111
 112     /* Terminate the result string.
 113      */
 114     *(result + 6) = '\0';
 115 }
 116
 117
 118 /* Return the actual soundex value.         */
 119 /* Added by Chris Herborth (chrish@qnx.com) */
 120 static char soundex_get_soundex__doc__[] =
 121         "Return the (English) Soundex hash value for a string.";
 122 static PyObject *
 123 get_soundex(PyObject *self, PyObject *args)
 124 {
 125         char *str;
 126         char sdx[7];
 127
 128         if(!PyArg_ParseTuple( args, "s", &str))
 129           return NULL;
 130
 131         soundex_hash(str, sdx);
 132
 133         return PyString_FromString(sdx);
 134 }
 135
 136 static char soundex_sound_similar__doc__[] =
 137         "Compare two strings to see if they sound similar (English).";
 138 static PyObject *
 139 sound_similar(PyObject *self, PyObject *args)
 140 {
 141     char *str1, *str2;
 142     char res1[7], res2[7];
 143
 144     if(!PyArg_ParseTuple(args, "ss", &str1, &str2))
 145         return NULL;
 146
 147     soundex_hash(str1, res1);
 148     soundex_hash(str2, res2);
 149
 150     if(!strcmp(res1,res2))
 151         return Py_BuildValue("i",1);
 152     else
 153         return Py_BuildValue("i",0);
 154 }
 155
 156 /* Python Method Table.
 157  */
 158 static PyMethodDef SoundexMethods[] =
 159 {
 160         {"sound_similar", sound_similar, 1, soundex_sound_similar__doc__},
 161         {"get_soundex", get_soundex, 1, soundex_get_soundex__doc__},
 162
 163     {NULL, NULL }               /* sentinel */
 164 };
 165
 166
 167 /* Register the method table.
 168  */
 169 DL_EXPORT(void)
 170 initsoundex()
 171 {
 172     (void) Py_InitModule4("soundex",
 173                           SoundexMethods,
 174                           soundex_module__doc__,
 175                           (PyObject *)NULL,
 176                           PYTHON_API_VERSION);
 177 }