ext/hunspell/hunzip.cxx

   1 #include <stdlib.h>
   2 #include <string.h>
   3 #include <stdio.h>
   4
   5 #include "hunzip.hxx"
   6
   7 #define CODELEN  65536
   8 #define BASEBITREC 5000
   9
  10 #define UNCOMPRESSED '\002'
  11 #define MAGIC "hz0"
  12 #define MAGIC_ENCRYPT "hz1"
  13 #define MAGICLEN (sizeof(MAGIC) - 1)
  14
  15 int Hunzip::fail(const char * err, const char * par) {
  16     fprintf(stderr, err, par);
  17     return -1;
  18 }
  19
  20 Hunzip::Hunzip(const char * file, const char * key) {
  21     bufsiz = 0;
  22     lastbit = 0;
  23     inc = 0;
  24     outc = 0;
  25     dec = NULL;
  26     fin = NULL;
  27     filename = (char *) malloc(strlen(file) + 1);
  28     if (filename) strcpy(filename, file);
  29     if (getcode(key) == -1) bufsiz = -1;
  30     else bufsiz = getbuf();
  31 }
  32
  33 int Hunzip::getcode(const char * key) {
  34     unsigned char c[2];
  35     int i, j, n, p;
  36     int allocatedbit = BASEBITREC;
  37     const char * enc = key;
  38
  39     if (!filename) return -1;
  40
  41     fin = fopen(filename, "rb");
  42     if (!fin) return -1;
  43
  44     // read magic number
  45     if ((fread(in, 1, 3, fin) < MAGICLEN)
  46         || !(strncmp(MAGIC, in, MAGICLEN) == 0 ||
  47                 strncmp(MAGIC_ENCRYPT, in, MAGICLEN) == 0)) {
  48             return fail(MSG_FORMAT, filename);
  49     }
  50
  51     // check encryption
  52     if (strncmp(MAGIC_ENCRYPT, in, MAGICLEN) == 0) {
  53         unsigned char cs;
  54         if (!key) return fail(MSG_KEY, filename);
  55         if (fread(&c, 1, 1, fin) < 1) return fail(MSG_FORMAT, filename);
  56         for (cs = 0; *enc; enc++) cs ^= *enc;
  57         if (cs != c[0]) return fail(MSG_KEY, filename);
  58         enc = key;
  59     } else key = NULL;
  60
  61     // read record count
  62     if (fread(&c, 1, 2, fin) < 2) return fail(MSG_FORMAT, filename);
  63
  64     if (key) {
  65         c[0] ^= *enc;
  66         if (*(++enc) == '\0') enc = key;
  67         c[1] ^= *enc;
  68     }
  69
  70     n = ((int) c[0] << 8) + c[1];
  71     dec = (struct bit *) malloc(BASEBITREC * sizeof(struct bit));
  72     if (!dec) return fail(MSG_MEMORY, filename);
  73     dec[0].v[0] = 0;
  74     dec[0].v[1] = 0;
  75
  76     // read codes
  77     for (i = 0; i < n; i++) {
  78         unsigned char l;
  79         if (fread(c, 1, 2, fin) < 2) return fail(MSG_FORMAT, filename);
  80         if (key) {
  81             if (*(++enc) == '\0') enc = key;
  82             c[0] ^= *enc;
  83             if (*(++enc) == '\0') enc = key;
  84             c[1] ^= *enc;
  85         }
  86         if (fread(&l, 1, 1, fin) < 1) return fail(MSG_FORMAT, filename);
  87         if (key) {
  88             if (*(++enc) == '\0') enc = key;
  89             l ^= *enc;
  90         }
  91         if (fread(in, 1, l/8+1, fin) < (size_t) l/8+1) return fail(MSG_FORMAT, filename);
  92         if (key) for (j = 0; j <= l/8; j++) {
  93             if (*(++enc) == '\0') enc = key;
  94             in[j] ^= *enc;
  95         }
  96         p = 0;
  97         for (j = 0; j < l; j++) {
  98             int b = (in[j/8] & (1 << (7 - (j % 8)))) ? 1 : 0;
  99             int oldp = p;
 100             p = dec[p].v[b];
 101             if (p == 0) {
 102                 lastbit++;
 103                 if (lastbit == allocatedbit) {
 104                     allocatedbit += BASEBITREC;
 105                     dec = (struct bit *) realloc(dec, allocatedbit * sizeof(struct bit));
 106                 }
 107                 dec[lastbit].v[0] = 0;
 108                 dec[lastbit].v[1] = 0;
 109                 dec[oldp].v[b] = lastbit;
 110                 p = lastbit;
 111             }
 112         }
 113         dec[p].c[0] = c[0];
 114         dec[p].c[1] = c[1];
 115     }
 116     return 0;
 117 }
 118
 119 Hunzip::~Hunzip()
 120 {
 121     if (dec) free(dec);
 122     if (fin) fclose(fin);
 123     if (filename) free(filename);
 124 }
 125
 126 int Hunzip::getbuf() {
 127     int p = 0;
 128     int o = 0;
 129     do {
 130         if (inc == 0) inbits = fread(in, 1, BUFSIZE, fin) * 8;
 131         for (; inc < inbits; inc++) {
 132             int b = (in[inc / 8] & (1 << (7 - (inc % 8)))) ? 1 : 0;
 133             int oldp = p;
 134             p = dec[p].v[b];
 135             if (p == 0) {
 136                 if (oldp == lastbit) {
 137                     fclose(fin);
 138                     fin = NULL;
 139                     // add last odd byte
 140                     if (dec[lastbit].c[0]) out[o++]  = dec[lastbit].c[1];
 141                     return o;
 142                 }
 143                 out[o++] = dec[oldp].c[0];
 144                 out[o++] = dec[oldp].c[1];
 145                 if (o == BUFSIZE) return o;
 146                 p = dec[p].v[b];
 147             }
 148         }
 149         inc = 0;
 150     } while (inbits == BUFSIZE * 8);
 151     return fail(MSG_FORMAT, filename);
 152 }
 153
 154 const char * Hunzip::getline() {
 155     char linebuf[BUFSIZE];
 156     int l = 0, eol = 0, left = 0, right = 0;
 157     if (bufsiz == -1) return NULL;
 158     while (l < bufsiz && !eol) {
 159         linebuf[l++] = out[outc];
 160         switch (out[outc]) {
 161             case '\t': break;
 162             case 31: { // escape
 163                 if (++outc == bufsiz) {
 164                     bufsiz = getbuf();
 165                     outc = 0;
 166                 }
 167                 linebuf[l - 1] = out[outc];
 168                 break;
 169             }
 170             case ' ': break;
 171             default: if (((unsigned char) out[outc]) < 47) {
 172                 if (out[outc] > 32) {
 173                     right = out[outc] - 31;
 174                     if (++outc == bufsiz) {
 175                         bufsiz = getbuf();
 176                         outc = 0;
 177                     }
 178                 }
 179                 if (out[outc] == 30) left = 9; else left = out[outc];
 180                 linebuf[l-1] = '\n';
 181                 eol = 1;
 182             }
 183         }
 184         if (++outc == bufsiz) {
 185             outc = 0;
 186             bufsiz = fin ? getbuf(): -1;
 187         }
 188     }
 189     if (right) strcpy(linebuf + l - 1, line + strlen(line) - right - 1);
 190     else linebuf[l] = '\0';
 191     strcpy(line + left, linebuf);
 192     return line;
 193 }