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 usage(void)
  27 {
  28         fprint(2, "usage: %s [-D] [-d dictname]\n", argv0);
  29         exits("usage");
  30 }
  31
  32 void
  33 main(int argc, char **argv)
  34 {
  35         int i;
  36         long a, ae;
  37         char *p;
  38         Entry e;
  39
  40         Binit(&boutbuf, 1, OWRITE);
  41         dict = &dicts[0];
  42         ARGBEGIN {
  43                 case 'd':
  44                         p = EARGF(usage());
  45                         dict = 0;
  46
  47                         for(i=0; dicts[i].name; i++) {
  48                                 if(strcmp(p, dicts[i].name)==0) {
  49                                         dict = &dicts[i];
  50                                         break;
  51                                 }
  52                         }
  53                         if(dict == nil) {
  54                                 err("unknown dictionary: %s", p);
  55                                 exits("nodict");
  56                         }
  57                         break;
  58                 case 'D':
  59                         debug++;
  60                         break;
  61                 default:
  62                         usage();
  63         }ARGEND
  64         USED(argc,argv);
  65         bdict = Bopen(dict->path, OREAD);
  66         if(!bdict) {
  67                 err("can't open dictionary %s", dict->path);
  68                 exits("nodict");
  69         }
  70         ae = Bseek(bdict, 0, 2);
  71         for(a = 0; a < ae; a = (*dict->nextoff)(a+1)) {
  72                 linelen = 0;
  73                 e = getentry(a);
  74                 Bprint(bout, "%ld\t", a);
  75                 linelen = 4;    /* only has to be approx right */
  76                 (*dict->printentry)(e, 'h');
  77         }
  78         exits(0);
  79 }
  80
  81 Entry
  82 getentry(long b)
  83 {
  84         long e, n, dtop;
  85         static Entry ans;
  86         static int anslen = 0;
  87
  88         e = (*dict->nextoff)(b+1);
  89         ans.doff = b;
  90         if(e < 0) {
  91                 dtop = Bseek(bdict, 0L, 2);
  92                 if(b < dtop) {
  93                         e = dtop;
  94                 } else {
  95                         err("couldn't seek to entry");
  96                         ans.start = 0;
  97                         ans.end = 0;
  98                 }
  99         }
 100         n = e-b;
 101         if(n) {
 102                 if(n > anslen) {
 103                         if((ans.start = realloc(ans.start, n)) == nil)
 104                                 sysfatal("realloc: %r");
 105                         anslen = n;
 106                 }
 107                 Bseek(bdict, b, 0);
 108                 n = Bread(bdict, ans.start, n);
 109                 ans.end = ans.start + n;
 110         }
 111         return ans;
 112 }