sys/src/cmd/dict/mkindex.c

   1 #include <u.h>
   2 #include <libc.h>
   3 #include <bio.h>
   4 #include "dict.h"
   5
   6 /*
   7  * Use this to start making an index for a new dictionary.
   8  * Get the dictionary-specific nextoff and printentry(_,'h')
   9  * commands working, add a record to the dicts[] array below,
  10  * and run this program to get a list of offset,headword
  11  * pairs
  12  */
  13 Biobuf  boutbuf;
  14 Biobuf  *bdict;
  15 Biobuf  *bout = &boutbuf;
  16 int     linelen;
  17 int     breaklen = 2000;
  18 int     outinhibit;
  19 int     debug;
  20
  21 Dict    *dict;  /* current dictionary */
  22
  23 Entry   getentry(long);
  24
  25 void
  26 main(int argc, char **argv)
  27 {
  28         int i;
  29         long a, ae;
  30         char *p;
  31         Entry e;
  32
  33         Binit(&boutbuf, 1, OWRITE);
  34         dict = &dicts[0];
  35         ARGBEGIN {
  36                 case 'd':
  37                         p = ARGF();
  38                         dict = 0;
  39                         if(p) {
  40                                 for(i=0; dicts[i].name; i++)
  41                                         if(strcmp(p, dicts[i].name)==0) {
  42                                                 dict = &dicts[i];
  43                                                 break;
  44                                         }
  45                         }
  46                         if(!dict) {
  47                                 err("unknown dictionary: %s", p);
  48                                 exits("nodict");
  49                         }
  50                         break;
  51                 case 'D':
  52                         debug++;
  53                         break;
  54         ARGEND }
  55         USED(argc,argv);
  56         bdict = Bopen(dict->path, OREAD);
  57         ae = Bseek(bdict, 0, 2);
  58         if(!bdict) {
  59                 err("can't open dictionary %s", dict->path);
  60                 exits("nodict");
  61         }
  62         for(a = 0; a < ae; a = (*dict->nextoff)(a+1)) {
  63                 linelen = 0;
  64                 e = getentry(a);
  65                 Bprint(bout, "%ld\t", a);
  66                 linelen = 4;    /* only has to be approx right */
  67                 (*dict->printentry)(e, 'h');
  68         }
  69         exits(0);
  70 }
  71
  72 Entry
  73 getentry(long b)
  74 {
  75         long e, n, dtop;
  76         static Entry ans;
  77         static int anslen = 0;
  78
  79         e = (*dict->nextoff)(b+1);
  80         ans.doff = b;
  81         if(e < 0) {
  82                 dtop = Bseek(bdict, 0L, 2);
  83                 if(b < dtop) {
  84                         e = dtop;
  85                 } else {
  86                         err("couldn't seek to entry");
  87                         ans.start = 0;
  88                         ans.end = 0;
  89                 }
  90         }
  91         n = e-b;
  92         if(n) {
  93                 if(n > anslen) {
  94                         ans.start = realloc(ans.start, n);
  95                         if(!ans.start) {
  96                                 err("out of memory");
  97                                 exits("nomem");
  98                         }
  99                         anslen = n;
 100                 }
 101                 Bseek(bdict, b, 0);
 102                 n = Bread(bdict, ans.start, n);
 103                 ans.end = ans.start + n;
 104         }
 105         return ans;
 106 }