parse_text.c
author viric@llimona
Wed, 29 Aug 2007 00:19:14 +0200
changeset 14 a961bb8806b9
child 15 17a66ceb774a
permissions -rw-r--r--
first 'zparsetext'.
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
14
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
     1
#include <stdio.h>
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
     2
#include "dictre.h"
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
     3
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
     4
static void give_accent_to_word(const char *tmp)
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
     5
{
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
     6
    char def[MAXDEF];
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
     7
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
     8
    find_def(tmp, def);
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
     9
    if (def[0] != 0) /* found */
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    10
    {
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    11
        /* Print the word UNTIL a space.
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    12
         * the definition will have the form:
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    13
         *    ACCENTED_WORD NOMINATIVE1 NOMINATIVE2 ... \n */
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    14
        char *first_space;
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    15
        char *pos;
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    16
        first_space = strchr(def, ' ');
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    17
        if (first_space != 0) /* Space found */
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    18
            for(pos = def; pos < first_space; ++pos)
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    19
                putchar(*pos);
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    20
        return;
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    21
    }
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    22
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    23
    /* if first_space == 0 or word not found */
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    24
    printf("%s", tmp);
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    25
}
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    26
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    27
static void process_text(FILE *in, int pos, int length)
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    28
{
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    29
    unsigned char tmp[MAXWORD];
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    30
    int wordpos = 0;
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    31
    do
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    32
    {
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    33
        int c;
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    34
        /* Check pos only if length >= 0 */
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    35
        if (length >= 0 && pos >= length)
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    36
            break;
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    37
        c = fgetc(in);
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    38
        if (c == EOF)
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    39
            break;
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    40
        if (is_ASCII(c))
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    41
        {
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    42
            if (wordpos != 0)
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    43
            {
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    44
                tmp[wordpos] = 0;
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    45
                give_accent_to_word(tmp);
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    46
                wordpos = 0;
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    47
            }
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    48
            putchar(c);
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    49
        }
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    50
        else /* non-ASCII - we consider it russian */
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    51
        {
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    52
            tmp[wordpos++] = c;
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    53
        }
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    54
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    55
        pos += 1;
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    56
    } while(1);
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    57
}
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    58
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    59
int main()
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    60
{
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    61
    init_dictionary();
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    62
    process_text(stdin, 0, -1);
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    63
    end_dictionary();
a961bb8806b9 first 'zparsetext'.
viric@llimona
parents:
diff changeset
    64
}