rexx/lstring/soundex.c

   1 /*
   2  * $Header$
   3  * $Log$
   4  * Revision 1.1  2001/04/04 05:43:38  wang
   5  * First commit: compiles on Linux, Amiga, Windows, Windows CE, generic gcc
   6  *
   7  * Revision 1.2  1999/11/26 12:52:25  bnv
   8  * Nothing important.
   9  *
  10  * Revision 1.1  1998/07/02 17:18:00  bnv
  11  * Initial Version
  12  *
  13  */
  14
  15 #include <ctype.h>
  16 #include <string.h>
  17 #include <lstring.h>
  18
  19 /* ------------------------* Soundex *--------------------------- */
  20 #define MAX_LENGTH     20       /* max # of chars to check      */
  21 #define SOUNDEX_LENGTH  4       /* length of Soundex code       */
  22
  23 #define ALPHA_OFFSET    65      /* ALPHA_OFFSET is decimal      */
  24                                 /* value of 'A' in ASCII character set  */
  25 void
  26 Lsoundex( const PLstr to, const PLstr str )
  27 {
  28         /* ctable contains the Soundex value of each
  29            letter in alphabetical order. 0 represents
  30            letters which are to be ignored. */
  31         char            *code, *name;
  32         static char     ctable[] =  { "01230120022455012623010202" };
  33         char            prior=' ', c;
  34         short           i, y=0;
  35         word            len;
  36
  37
  38         /* ---- initialise ---- */
  39         L2STR(str);
  40         LZEROSTR(*to);
  41         Lfx(to,SOUNDEX_LENGTH);
  42
  43         len = LLEN(*str);
  44         code = LSTR(*to);
  45         name = LSTR(*str);
  46         if (len==0) return;
  47
  48         /* convert name to all upper case */
  49         if (len > MAX_LENGTH) len = MAX_LENGTH;
  50
  51         /* generate 1st character of Soundex code */
  52         code[0] = l2u[(byte)name[0]];
  53         y=1;
  54
  55
  56         if (l2u[(byte)name[0]]=='K')    /* modifications */
  57                 code[0] = 'C';
  58         else
  59         if (l2u[(byte)name[0]]=='P' && l2u[(byte)name[1]]=='H')
  60                 code[0] = 'F';
  61
  62         /* loop through the rest of name, until code complete */
  63         for (i=1; i<len; i++) {
  64                 c = l2u[(byte)name[i]];
  65                         /* skip non alpha */
  66                 if (!ISALPHA(c))
  67                         continue;
  68
  69                         /* skip succesive occurance */
  70                 if (c==prior)  continue;
  71                 prior = c;
  72
  73                         /* lookup letter in table   */
  74                 c -= ALPHA_OFFSET;
  75                 if (ctable[(byte)c]=='0')
  76                         continue;       /* ignore this letter */
  77
  78                 code[y++] = ctable[(byte)c];    /* add to code */
  79
  80                 if (y >= SOUNDEX_LENGTH) break; /* code is complete */
  81         }
  82         while (y < SOUNDEX_LENGTH) code[y++] = '0';   /* pad code with zeros */
  83         code[y] = '\0';
  84         LLEN(*to) = y;
  85 } /* Lsoundex */