sys/src/cmd/dict/roget.c

   1 #include <u.h>
   2 #include <libc.h>
   3 #include <bio.h>
   4 #include <ctype.h>
   5 #include "dict.h"
   6
   7 /* Roget's Thesaurus from project Gutenberg */
   8
   9 static long Last = 0;
  10
  11 void
  12 rogetprintentry(Entry e, int cmd)
  13 {
  14         int spc;
  15         char c, *p;
  16
  17         spc = 0;
  18         p = e.start;
  19
  20         if(cmd == 'h'){
  21                 while(!isspace(*p) && p < e.end)
  22                         p++;
  23                 while(strncmp(p, " -- ", 4) != 0 && p < e.end){
  24                         while(isspace(*p) && p < e.end)
  25                                 p++;
  26                         if (*p == '[' || *p == '{'){
  27                                 c = (*p == '[')? ']': '}';
  28                                 while(*p != c && p < e.end)
  29                                         p++;
  30                                 p++;
  31                                 continue;
  32                         }
  33                         if (isdigit(*p) || ispunct(*p)){
  34                                 while(!isspace(*p) && p < e.end)
  35                                         p++;
  36                                 continue;
  37                         }
  38
  39
  40                         if (isspace(*p))
  41                                 spc = 1;
  42                         else
  43                         if (spc){
  44                                 outchar(' ');
  45                                 spc = 0;
  46                         }
  47
  48                         while(!isspace(*p) && p < e.end)
  49                                 outchar(*p++);
  50                 }
  51                 return;
  52         }
  53
  54         while(p < e.end && !isspace(*p))
  55                 p++;
  56         while(p < e.end && isspace(*p))
  57                 p++;
  58
  59         while (p < e.end){
  60                 if (p < e.end -4 && strncmp(p, " -- ", 4) == 0){        /* first line */
  61                         outnl(2);
  62                         p += 4;
  63                         spc = 0;
  64                 }
  65
  66                 if (p < e.end -2 && strncmp(p, "[ ", 2) == 0){          /* twiddle layout */
  67                         outchars(" [");
  68                         continue;
  69                 }
  70
  71                 if (p < e.end -4 && strncmp(p, "&c (", 4) == 0){        /* usefull xref */
  72                         if (spc)
  73                                 outchar(' ');
  74                         outchar('/');
  75                         while(p < e.end && *p != '(')
  76                                 p++;
  77                         p++;
  78                         while(p < e.end && *p != ')')
  79                                 outchar(*p++);
  80                         p++;
  81                         while(p < e.end && isspace(*p))
  82                                 p++;
  83                         while(p < e.end && isdigit(*p))
  84                                 p++;
  85                         outchar('/');
  86                         continue;
  87                 }
  88
  89                 if (p < e.end -3 && strncmp(p, "&c ", 3) == 0){         /* less usefull xref */
  90                         while(p < e.end && !isdigit(*p))
  91                                 p++;
  92                         while(p < e.end && isdigit(*p))
  93                                 p++;
  94                         continue;
  95                 }
  96
  97                 if (*p == '\n' && p < (e.end -1)){                      /* their newlines */
  98                         spc = 0;
  99                         p++;
 100                         if (isspace(*p)){                               /* their continuation line */
 101                                 while (isspace(*p))
 102                                         p++;
 103                                 p--;
 104                         }
 105                         else{
 106                                 outnl(2);
 107                         }
 108                 }
 109                 if (spc && *p != ';' && *p != '.' &&
 110                     *p != ',' && !isspace(*p)){                         /* drop spaces before punct */
 111                         spc = 0;
 112                         outchar(' ');
 113                 }
 114                 if (isspace(*p))
 115                         spc = 1;
 116                 else
 117                         outchar(*p);
 118                 p++;
 119         }
 120         outnl(0);
 121 }
 122
 123 long
 124 rogetnextoff(long fromoff)
 125 {
 126         int i;
 127         vlong l;
 128         char *p;
 129
 130         Bseek(bdict, fromoff, 0);
 131         Brdline(bdict, '\n');
 132         while ((p = Brdline(bdict, '\n')) != nil){
 133                 l = Blinelen(bdict);
 134                 if (!isdigit(*p))
 135                         continue;
 136                 for (i = 0; i < l-4; i++)
 137                         if (strncmp(p+i, " -- ", 4) == 0)
 138                                 return Boffset(bdict)-l;
 139         }
 140         return Boffset(bdict);
 141 }
 142
 143 void
 144 rogetprintkey(void)
 145 {
 146         Bprint(bout, "No pronunciation key.\n");
 147 }