appendixE.tr

.ce 2
Приложение E
.sp 1
.
\fBФайл main.c\fP
.sp 1
.
.ps 10p
.vs 11p
.
.fam C
   1    #include <stdio.h>
   2    #include <locale.h>
   3    #include <string.h>
   4    #include <wctype.h>
   5    #include <wchar.h>
   6    #include <getopt.h>
   7    
   8    #include "markov.h"
   9    #include "hashtb.h"
  10    #include "xstdlib.h"
  11    
  12    #define DEF_EOS ".;!?"
  13    #define DEF_IGN "'\"[]{}(),:"
  14    
  15    enum
  16    {
  17        EM = 0x0100,
  18        EA,
  19        IM,
  20        IA,
  21    };
  22    
  23    struct arguments
  24    {
  25        wchar_t *eos;
  26        wchar_t *ign;
  27        char *word;
  28    } args;
  29    
  30    typedef struct dictionary {
  31        char **word;
  32        int pos;
  33        int len;
  34    } DICT;
  35    
  36    void argsinit(int argc, char **argv);
  37    char *getword(void);
  38    int isweos(wchar_t wc);
  39    int iswign(wchar_t wc);
  40    
  41    int main(int argc, char **argv)
  42    {
  43        int i, size;
  44        int prenw = -1, numword;
  45        MARK mword;
  46        char *word;
  47        DICT dict = {NULL, 0, 0};
  48    
  49        setlocale(LC_CTYPE, "");
  50    
  51        mword = minit();
  52        argsinit(argc, argv);
  53    
  54        for (i = 1; (word = getword()) != NULL; ++i) {
  55            if ((numword = getnum(word)) == -1) {
  56                if (dict.len == dict.pos) {
  57                    dict.len += 1000;
  58                    size = (dict.len + 1) * sizeof(char**);
  59                    dict.word = xrealloc(dict.word, size);
  60                }
  61    
  62                numword = dict.pos;
  63                dict.word[dict.pos] = word;
  64                ++dict.pos;
  65            }
  66    
  67            addentry(numword, word);
  68            mcount(mword, prenw, numword);
  69    
  70            prenw = numword;
  71        }
  72    
  73        if (dict.pos <= 1) {
  74            fprintf(stderr, "File is not parsed\n");
  75            exit(EXIT_FAILURE);
  76        }
  77    
  78        xrealloc(dict.word, dict.len * sizeof(char**));
  79        normalize(mword);
  80    
  81        numword = 0;
  82    
  83        if (args.word != NULL && (numword = getnum(args.word)) == -1) {
  84            fprintf(stderr, "The word \"%s\" not found\n", args.word);
  85            exit(EXIT_FAILURE);
  86        }
  87    
  88        for (i = 0; i < 10; ++i) {
  89            printf("%s ", dict.word[numword]);
  90    
  91            if ((numword = getel(mword, numword)) == -1)
  92                break;
  93        }
  94        putchar('\n');
  95    
  96        return 0;
  97    }
  98    
  99    char *getword(void)
 100    {
 101        int i;
 102        wchar_t wc;
 103        char *word;
 104        int lenword = 128;
 105    
 106        word = xmalloc(lenword);
 107    
 108        for (; (wc = fgetwc(stdin)) != WEOF && iswspace(wc);)
 109            ;
 110    
 111        if (isweos(wc)) {
 112            i = wctomb(word, wc);
 113            word[i] = '\0';
 114            xrealloc(word, lenword - i - 1);
 115    
 116            return word;
 117        }
 118    
 119        ungetwc(wc, stdin);
 120    
 121        for (i = 0; ((wc = fgetwc(stdin)) != WEOF);) {
 122            if (lenword < i + MB_CUR_MAX + 1) {
 123                lenword *= 2;
 124                xrealloc(word, lenword);
 125            }
 126    
 127            if (iswign(wc))
 128                continue;
 129            else if (iswalpha(wc) || wc == '-') {
 130                wc = towlower(wc);
 131                i += wctomb(&(word[i]), wc);
 132            }
 133            else if (wc == '\n' && word[i - 1] == '-')
 134                --i;
 135            else if ((iswspace(wc) || isweos(wc)) && i > 0) {
 136                ungetwc(wc, stdin);
 137    
 138                break;
 139            }
 140        }
 141    
 142        xrealloc(word, i + 1);
 143        word[i] = '\0';
 144    
 145        if (i == 0)
 146            return NULL;
 147        else
 148            return word;
 149    }
 150    
 151    int isweos(wchar_t wc)
 152    {
 153        int i;
 154    
 155        for (i = 0; args.eos[i] != '\0'; ++i) {
 156            if (wc == args.eos[i])
 157                return ++i;
 158        }
 159    
 160        return 0;
 161    }
 162    
 163    int iswign(wchar_t wc)
 164    {
 165        int i;
 166    
 167        for (i = 0; args.ign[i] != '\0'; ++i) {
 168            if (wc == args.ign[i])
 169                return ++i;
 170        }
 171    
 172        return 0;
 173    }
 174    
 175    void printhelp(void)
 176    {
 177        printf("Usage: markov [OPTION]...\n\n");
 178    
 179        printf("OPTIONS:\n");
 180        printf("--eosA EOS\t\tadd End-Of-Sentence characters\n");
 181        printf("--eosM EOS\t\tmodify End-Of-Sentence characters\n");
 182        printf("-h, --help\t\tgive this help list and exit\n");
 183        printf("--ignA IGN\t\tadd IGNored characters\n");
 184        printf("--ignM IGN\t\tmodify IGNored characters\n");
 185        printf("\n");
 186    
 187        printf("DEFAULT:\n");
 188        printf("EOS: %s\n", DEF_EOS);
 189        printf("IGN: %s\n", DEF_IGN);
 190    }
 191    
 192    void argsinit(int argc, char **argv)
 193    {
 194        int size;
 195        int neos;
 196        int nign;
 197    
 198        int opt;
 199        char sopt[] = "hw:";
 200        const struct option lopt[] = {
 201            {"help", no_argument,       NULL, 'h'},
 202            {"eosM", required_argument, NULL, EM },
 203            {"eosA", required_argument, NULL, EA },
 204            {"ignM", required_argument, NULL, IM },
 205            {"ignA", required_argument, NULL, IA },
 206            {NULL,   0,                 NULL, 0  },
 207        };
 208    
 209        size = sizeof (wchar_t);
 210        args.eos = xmalloc((strlen(DEF_EOS) + 1) * size);
 211        args.ign = xmalloc((strlen(DEF_IGN) + 1) * size);
 212    
 213        neos = mbstowcs(args.eos, DEF_EOS, sizeof (DEF_EOS));
 214        nign = mbstowcs(args.ign, DEF_IGN, sizeof (DEF_IGN));
 215    
 216        for (opterr = 0; (opt = getopt_long(argc, argv, sopt, lopt, NULL)) != -1;) {
 217            switch (opt) {
 218            case 'h':
 219                printhelp();
 220                exit(EXIT_SUCCESS);
 221    
 222            case 'w':
 223                args.word = optarg;
 224    
 225                break;
 226    
 227            case EM:
 228                neos = 0;
 229            case EA:
 230                args.eos = realloc(args.eos, (neos + 1025) * size);
 231                neos += mbstowcs(args.eos + neos, optarg, 1024);
 232                args.eos = realloc(args.eos, (neos + 1) * size);
 233    
 234                break;
 235    
 236            case IM:
 237                nign = 0;
 238            case IA:
 239                args.ign = realloc(args.ign, (nign + 1025) * size);
 240                nign += mbstowcs(args.ign + nign, optarg, 1024);
 241                args.ign = realloc(args.ign, (nign + 1) * size);
 242    
 243                break;
 244    
 245            default:
 246                fprintf(stderr, "markov: Unknown argument:\n");
 247                exit(EXIT_FAILURE);
 248            }
 249        }
 250    }
.sp 1
.fam T
.ps 13
.vs 16