viric@0: #include <stdio.h>
viric@0: 
viric@0: #include "dictre.h"
viric@0: 
viric@0: enum
viric@0: {
viric@0:     MAX=500000
viric@0: };
viric@0: 
viric@0: struct Word words[MAX];
viric@0: int nwords;
viric@0: struct Def defs[MAX];
viric@0: int ndefs;
viric@0: int dont_touch[20];
viric@0: int ndont_touch;
viric@0: 
viric@5: void init_load()
viric@0: {
viric@0:     ndefs = 0;
viric@0:     nwords = 0;
viric@0:     ndont_touch = 0;
viric@0: }
viric@0: 
viric@0: static void new_word(struct Word *from)
viric@0: {
viric@0:     memcpy(&words[nwords], from, sizeof(*from));
viric@0:     nwords++;
viric@0: }
viric@0: 
viric@0: static void new_dont_touch(int n)
viric@0: {
viric@0:     dont_touch[ndont_touch++] = n;
viric@0: }
viric@0: 
viric@0: static int new_def(char *def, int offset, int length)
viric@0: {
viric@0:     defs[ndefs].d = def;
viric@0:     defs[ndefs].offset = offset;
viric@0:     defs[ndefs].length = length;
viric@0:     return ndefs++;
viric@0: }
viric@0: 
viric@0: static int search_def(int offset, int length)
viric@0: {
viric@0:     int i;
viric@0: 
viric@0:     for(i=0; i < ndefs; ++i)
viric@0:     {
viric@0:         if (defs[i].offset == offset &&
viric@0:                 defs[i].length == length)
viric@0:             return i;
viric@0:     }
viric@0:     return -1;
viric@0: }
viric@0: 
viric@0: static void print_word(struct Word *w)
viric@0: {
viric@0:     printf("%s\t%i\n", w->w, w->def);
viric@0: }
viric@0: 
viric@0: void load_dictionary(FILE *index, FILE *fdefs)
viric@0: {
viric@0:     struct Word w;
viric@0:     int last_offset = 0;
viric@5:     int def_avoided = 0;
viric@6:     int numword = 0;;
viric@6:     static int dispnwords = 0;
viric@0: 
viric@0:     do {
viric@0:         int offset, length;
viric@0:         char *defstr;
viric@0:         w.w = get_word(index);
viric@6:         /*numword++;
viric@6:         printf("words: %i\n", numword);*/
viric@0:         if (w.w == 0)
viric@0:             break;
viric@5:         /*printf("Word: %s\n", w.w);*/
viric@0:         offset = get_int(index);
viric@0:         length = get_int(index);
viric@0:         if (offset > last_offset)
viric@0:         {
viric@0:             w.def = -1;
viric@0:             last_offset = offset;
viric@0:         }
viric@0:         else
viric@0:             w.def = search_def(offset, length);
viric@5:         if (w.def == -1) 
viric@0:         {
viric@5:             /* New definition */
viric@5:             int newindex, repindex;
viric@0:             defstr = get_def(fdefs, offset, length);
viric@5:             newindex = new_def(defstr, offset, length);
viric@6:             /*
viric@6:             printf("Length %i (%s): %i\n", newindex, w.w, length);
viric@6:             */
viric@5:             
viric@5:             /* Store it in the hash for repeated defs */
viric@5:             repindex = def_repeated(&defs[newindex]);
viric@5:             if (repindex != -1) 
viric@5:             {
viric@5:                 def_avoided += 1;
viric@6:                 /*
viric@6:                 printf("Repeated def avoided %i (for def %i)"
viric@6:                         " (%s)\n%i %s\n%i %s\n",
viric@6:                         def_avoided, repindex, w.w,
viric@6:                         length, defstr,
viric@6:                         defs[repindex].length, defs[repindex].d);
viric@6:                         */
viric@5:                 remove_def(newindex);
viric@5:                 newindex = repindex;
viric@5:             } else
viric@5:                 new_hashdef(&defs[newindex], newindex);
viric@5: 
viric@5:             /* Store the final index */
viric@5:             w.def = newindex;
viric@0:         }
viric@0:         /* sizeof -1  instead of strlen() */
viric@0:         if (strncmp(w.w, "00database", sizeof("00database") - 1) == 0)
viric@0:                 new_dont_touch(w.def);
viric@0:         new_word(&w);
viric@6: 
viric@6:         /* stdout Display */
viric@6:         dispnwords++;
viric@6:         if (dispnwords >= 1000)
viric@6:         {
viric@6:             dispnwords = 0;
viric@6:             printf("Loaded: %i Repeated definitions avoided: %i\n", nwords,
viric@6:                     def_avoided);
viric@6:         }
viric@6: 
viric@0:     } while(1);
viric@0: }
viric@0: 
viric@0: void print_words()
viric@0: {
viric@0:     int i;
viric@0: 
viric@0:     for(i=0; i < nwords; ++i)
viric@0:         print_word(&words[i]);
viric@0: }