World of Shareware - Software Farm 2/wosw_2.zip/wosw_2/CPROG/US20SRC.ZIP/CDICT.C

C/C++ Source or Header | 1992-06-26 | 5KB | 218 lines

/*    CDICT:    Compress Dictionary utility program for
        MicroSPELL 2.0

        (C)opyright May 1987,1992 by Daniel Lawrence
        All Rights Reserved
*/

#include    <stdio.h>
#include    "dopt.h"
#include    "dsfx.h"

/* globals */

char mdfile[NSTRING];        /* main dictionary text file name */
char mcfile[NSTRING];        /* compressed dictionary file name */
FILE *mdptr = NULL;        /* main dictionary file pointer */
FILE *mcptr = NULL;        /* compressed dictionary file pointer */
int sflen[NSUFFIX];        /* length of suffixes */
long letter_offset[ALPHASIZE];    /* offsets in file to letters */
unsigned char lcase[128];    /* lower case table (quick!) */

main(argc, argv)

int argc;    /* # of command line arguments */
char **argv;    /* text of command line arguments */

{
    register char *word;        /* current word */
    register int suffix;        /* suffix index */
    register char cur_first_letter;    /* current first letter scanned */
    register int index;        /* loop index */
    long total_words;        /* total number of words in this dictionary */
    char lastword[NSTRING];        /* previous word in dictionary */
    char tempword[NSTRING];        /* temporary word in dictionary */
    char *nxtmword();

    printf("CDICT Dictionary Compression Utility for MicroSPELL v%s\n",
        VERSION);

    if (argc != 3) {
        help();
        exit(EXBADOPT);
    }

    strcpy(mdfile, argv[1]);
    strcpy(mcfile, argv[2]);

    if (mopen() != TRUE) {
        printf("%%Can not open text dictionary file\n");
        exit(EXMDICT);
    }

    /* init the lower case table */
    for (index = 0; index < 128; index ++)
        if ('A' <= index && index <= 'Z')
            lcase[index] = index - 'A' + 'a';
        else
            lcase[index] = index;

    /* open the output compressed dictionary file */
    mcptr = fopen(mcfile, "wb");
    if (mcptr == NULL) {
        printf("%%Can not open compressed dictionary output file\n");
        exit(EXMDICT);
    }

    /* position past character table */
    fwrite((char *)&(letter_offset[0]), sizeof(long), ALPHASIZE, mcptr);

    /* prepare the suffix length table */
    for (suffix = 0; suffix < NSUFFIX; suffix++)
        sflen[suffix] = strlen(sfx[suffix]);

    printf("[Compressing %s => %s]\n", mdfile, mcfile);
    lastword[0] = 0;    /* null last word */

    /* scan the dictionary, compressing */
    cur_first_letter = 0;
    total_words = 0L;
    word = nxtmword();
    while (word) {
        if (lcase[word[0]] != cur_first_letter) {
            cur_first_letter = lcase[word[0]];
            letter_offset[cur_first_letter - 'a'] = ftell(mcptr);
        }
        strcpy(tempword, word);
        cmpsword(lastword, word);
        total_words++;
        strcpy(lastword, tempword);
        word = nxtmword();
    }

    /* write out letter offset table to front of file */
    fseek(mcptr, 0, 0);
    fwrite((char *)&(letter_offset[0]), sizeof(long), ALPHASIZE, mcptr);

    /* close things up */
    mclose();
    fclose(mcptr);
    printf("[%ld words in dictionary compressed]\n", total_words);
}

help()        /* tell us about cdict... */

{
    printf("\nUsage:\n\n");
    printf("    CDICT <text dictionary> <compressed dictionary>\n");
}

mopen()        /* open the main dicionary */

{
    /* if it is already open, close it down */
    if (mdptr != NULL)
        fclose(mdptr);

    /* open up the text dictionary... */
    if ((mdptr = fopen(mdfile, "r")) == NULL)
        return(FALSE);

    return(TRUE);
}

mclose()    /* close the dictionary down */

{
    /* if it is already open, close it down */
    if (mdptr != NULL)
        fclose(mdptr);
    mdptr = NULL;
}

char *nxtmword()    /* get the next word from the main dictionary */

{
    static char word[NSTRING];    /* word to return */

    /* is it already closed? */
    if (mdptr == NULL)
        return(NULL);

    /* get the next word */
    if (fgets(word, NSTRING - 1, mdptr) == NULL) {
        /* no more left!!!! close out */
        fclose(mdptr);
        mdptr = NULL;
        return(NULL);
    }

    /* all's well, dump the return, any trailing spaces and
       return the word */
    do
        word[strlen(word) - 1] = 0;
    while (word[strlen(word) - 1] == ' ');
    return(word);
}

cmpsword(lastword, word)    /* compress the given word */

char *lastword;        /* previous dictionary word */
char *word;        /* current dictionary word */

{
    register int index;    /* index into current word */
    register int same;    /* # of same characters */
    register int suffix;    /* suffix code */
    register int wlen;    /* length of current word */
    register int orig_first;    /* original first character */

    /* scan for common suffixes */
    wlen = strlen(word);
    for (suffix = 0; suffix < NSUFFIX; suffix++) {
        if (wlen < sflen[suffix])
            continue;
        if (strcmp(&word[wlen - sflen[suffix]], sfx[suffix]) == 0) {
            word[wlen - sflen[suffix]] = 0;    /* trunc it */
            break;
        }
    }

    /* If there is no suffix...suffix ends up as NSUFFIX */

    /* save the capitalization of the original lead char */
    word[0] = lcase[(orig_first = word[0])];

    /* scan for like beginning characters */
    index = 0;
    while (lastword[index] && lastword[index] == word[index])
        index++;

    same = index;
#if    ASCII
    suffix |= 128;
#endif

    if (orig_first == word[0])
        fprintf(mcptr, "%c%s%c", 'A'+same, &word[index], suffix);
    else
        fprintf(mcptr, "%c%s%c", 'A'+same+128, &word[index], suffix);
}

#if    CMS
#undef    fopen
/*    The IBM 30xx likes to tell us when file opens
    fail...it's too chatty....I like to handle these myself    */

FILE *cmsopen(file, mode)

char *file;    /* name of file to open */
char *mode;    /* mode to open it in */

{
    quiet(1);
    return(fopen(file,mode));
}
#endif