viric@14: #include <stdio.h>
viric@14: #include <sys/stat.h>
viric@14: #include <sys/types.h>
viric@14: #include <sys/mman.h>
viric@14: #include <fcntl.h>
viric@14: #include "dictre.h"
viric@14: 
viric@17: const static char indexext[] = ".index";
viric@17: const static char dictext[] = ".dict";
viric@14: 
viric@14: int get_filesize(const char *fname)
viric@14: {
viric@14:     struct stat st;
viric@14:     int res;
viric@14:     res = stat(fname, &st);
viric@14:     if (res == -1)
viric@14:     {
viric@14:         fprintf(stderr, "Problem stating the file %s\n", fname);
viric@14:         perror("Error:");
viric@14:         exit(-1);
viric@14:     }
viric@14: 
viric@14:     return st.st_size;
viric@14: }
viric@14: 
viric@17: void init_dictionary(struct Dict *d, const char *base)
viric@14: {
viric@17:     char *filename;
viric@17: 
viric@17:     filename = (char *) malloc(strlen(base) + 10);
viric@17: 
viric@17:     /* Prepare .index filename and open it*/
viric@17:     strcpy(filename, base);
viric@17:     strcat(filename, indexext);
viric@17: 
viric@17:     d->indexsize = get_filesize(filename);
viric@17:     d->indexfd = open(filename, O_RDONLY);
viric@17:     if (d->indexfd == -1)
viric@14:     {
viric@17:         fprintf(stderr, "Problem opening the file %s\n", filename);
viric@14:         perror("Error:");
viric@14:         exit(-1);
viric@14:     }
viric@17:     d->index = (unsigned char *) mmap(0, d->indexsize, PROT_READ, MAP_SHARED,
viric@17:             d->indexfd, 0);
viric@14: 
viric@17:     /* Prepare .dict filename and open it*/
viric@17:     strcpy(filename, base);
viric@17:     strcat(filename, dictext);
viric@17:     d->defs = fopen(filename, "r");
viric@17:     if (d->defs == 0)
viric@14:     {
viric@17:         fprintf(stderr, "Problem opening the file %s\n", filename);
viric@14:         perror("Error:");
viric@14:         exit(-1);
viric@14:     }
viric@14: 
viric@21:     d->trim_first_line = 0;
viric@21:     d->trim_last_newlines = 0;
viric@21: 
viric@17:     free(filename);
viric@14: }
viric@14: 
viric@17: void end_dictionary(struct Dict *d)
viric@14: {
viric@17:     munmap(d->index, d->indexsize);
viric@17:     close(d->indexfd);
viric@17:     fclose(d->defs);
viric@14: }
viric@14: 
viric@21: static int trim_first_line(char *def, int len)
viric@21: {
viric@21:     int new_line_pos;
viric@21:     int i,j;
viric@21:     for(i=0; i < len; ++i)
viric@21:     {
viric@24:         /* prepare ltrimming when finding the first newline character */
viric@21:         if (def[i] == '\n')
viric@21:         {
viric@24:             /* Remove spaces after the first newline */
viric@24:             for(i+=1; def[i] == ' '; ++i);
viric@24: 
viric@21:             /* Break */
viric@24:             memmove(def, def + i,
viric@21:                     len - i - 1);
viric@21:             def[len-i-1] = 0;
viric@24:             return len-i-1+1/*\0*/;
viric@21:         }
viric@21:     }
viric@21:     return len;
viric@21: }
viric@21: 
viric@21: static int trim_last_newlines(char *def, int len)
viric@21: {
viric@21:     int new_line_pos;
viric@21:     int i,j;
viric@22:     if (len < 2)
viric@22:         return len;
viric@22: 
viric@22:     for(i=len-2; i >= 0; --i)
viric@21:     {
viric@21:         if (def[i] != '\n' && def[i] != '\r')
viric@21:         {
viric@21:             def[i+1] = '\0';
viric@21:             return i + 1;
viric@21:         }
viric@21:     }
viric@21:     return len;
viric@21: }
viric@21: 
viric@17: static void fill_def(struct Dict *d, int offset, int length, char * def)
viric@14: {
viric@17:     fseek(d->defs, offset, SEEK_SET);
viric@17:     fread(def, 1, length, d->defs);
viric@21:     def[length] = 0;
viric@21:     if (d->trim_first_line)
viric@21:     {
viric@21:         length = trim_first_line(def, length + 1/*\0*/) - 1 /*\0*/;
viric@21:     }
viric@21:     if (d->trim_last_newlines)
viric@21:     {
viric@21:         length = trim_last_newlines(def, length+1) - 1; /* math as above*/
viric@21:     }
viric@17: }
viric@17: 
viric@21: static int pointer_at_end(struct Dict *d, const unsigned char *ptr)
viric@17: {
viric@17:     if (ptr >= (d->index + d->indexsize))
viric@14:         return 1;
viric@14:     return 0;
viric@14: }
viric@14: 
viric@21: static const char * skip_until_newline(struct Dict *d, const char *from)
viric@14: {
viric@17:     if (pointer_at_end(d, from))
viric@14:         return 0;
viric@14:     while(*from != '\n' && *from != 0)
viric@14:     {
viric@14:         ++from;
viric@17:         if(pointer_at_end(d, from))
viric@14:             return 0;
viric@14:     }
viric@14:     return from;
viric@14: }
viric@14: 
viric@14: static int compare(const unsigned char *word, const unsigned char *test)
viric@14: {
viric@14:     int i;
viric@14: 
viric@14:     /*printf("Comparing %s to %.20s\n", word, test);*/
viric@14:     for(i=0; word[i] != 0 && test[i] != 0; ++i)
viric@14:     {
viric@14:         if (word[i] != test[i])
viric@14:         {
viric@14:             break;
viric@14:         }
viric@14:     }
viric@14:     if (word[i] == 0 && test[i] == '\t')
viric@14:         return 0;
viric@14:     else if (word[i] == 0)
viric@14:         return -1;
viric@14:     else if (test[i] == '\t')
viric@14:         return 1;
viric@14:     else if (word[i] > test[i])
viric@14:         return 1;
viric@14:     else if (word[i] < test[i])
viric@14:         return -1;
viric@14: 
viric@14:     /* It should never reach this. */
viric@14:     return -1;
viric@14: }
viric@14: 
viric@21: static const char * search_next(struct Dict *d, const char *word, const char *from)
viric@21: {
viric@21:     const char *ret;
viric@24:     ret = skip_until_newline(d, from) + 1;
viric@24:     if (ret == (char *) 1) /* pointer at end */
viric@24:         return 0;
viric@24:     if (compare(word, ret) == 0)
viric@21:         return ret;
viric@21:     return 0;
viric@21: }
viric@21: 
viric@21: static const char * bin_search(struct Dict *d, const char *word)
viric@14: {
viric@14:     int step, pivot;
viric@21:     const char *ret;
viric@21:     const char *test;
viric@21:     int comparision;
viric@21:     int found_once = 0;
viric@14: 
viric@17:     pivot = d->indexsize / 2;
viric@17:     step = d->indexsize / 2;
viric@14: 
viric@14:     do
viric@14:     {
viric@17:         test = d->index + pivot;
viric@17:         test = skip_until_newline(d, test);
viric@14:         if (test == 0)
viric@14:             return 0;
viric@14:         test += 1; /* skip exactly the new line */
viric@14: 
viric@14:         comparision = compare(word, test);
viric@21:         if (comparision <= 0)
viric@14:         {
viric@21:             if (comparision == 0)
viric@21:                 found_once = 1;
viric@21:             /* If == 0, we don't know that it's the FIRST
viric@21:              * match possible in the dictionary.
viric@21:              * We want all possible matches. */
viric@14:             step = step / 2;
viric@14:             pivot = pivot - step;
viric@14:         } else if (comparision > 0)
viric@14:         {
viric@14:             step = step / 2;
viric@14:             pivot = pivot + step;
viric@14:         }
viric@14:     } while(step > 0);
viric@21: 
viric@21:     if (!found_once)
viric@21:         return 0;
viric@21: 
viric@21:     if (comparision == 0) /* last comparision */
viric@21:     {
viric@21:         ret = skip_until_newline(d, d->index + pivot) + 1;
viric@21:     } else
viric@21:     {
viric@21:         ret = skip_until_newline(d, test) + 1;
viric@21:     }
viric@21:     return ret;
viric@14: }
viric@14: 
viric@21: static int my_get_int(const char **pos)
viric@14: {
viric@14:     int i;
viric@21:     const char *start;
viric@14:     int val;
viric@14: 
viric@14:     start = *pos;
viric@14:     for(i=0; start[i] != '\t' && start[i] != '\n'; ++i)
viric@14:         ;
viric@14:     val = str2int_len(start, i);
viric@14:     *pos += i + 1;
viric@14:     return val;
viric@14: }
viric@14: 
viric@17: void find_def(struct Dict *d, const char *word, char * def)
viric@14: {
viric@14:     int offset, len;
viric@21:     const char *found, *pos;
viric@21:     int wordlen;
viric@14: 
viric@21:     def[0] = 0;
viric@21:     /* we will get a pointer to the offset for the ints*/
viric@21:     found = bin_search(d, word);
viric@21:     if (found == 0)
viric@21:         return;
viric@21:     wordlen = strlen(word);
viric@21:     do
viric@14:     {
viric@21:         found += wordlen+1;
viric@21:         pos = found;
viric@21:         offset = my_get_int(&pos); /* increments pos */
viric@21:         len = my_get_int(&pos); /* increments pos */
viric@21:         fill_def(d, offset, len, def);
viric@21:         found = search_next(d, word, found);
viric@21:         if (!found)
viric@21:             break;
viric@21:         strcat(def, ", ");
viric@21:         def += strlen(def);
viric@21:     } while(1);
viric@14: }