cpp/galago/contrib/libstemmer_c/runtime/utilities.c

   1
   2 #include <stdio.h>
   3 #include <stdlib.h>
   4 #include <string.h>
   5
   6 #include "header.h"
   7
   8 #define unless(C) if(!(C))
   9
  10 #define CREATE_SIZE 1
  11
  12 extern symbol * create_s(void) {
  13     symbol * p;
  14     void * mem = malloc(HEAD + (CREATE_SIZE + 1) * sizeof(symbol));
  15     if (mem == NULL) return NULL;
  16     p = (symbol *) (HEAD + (char *) mem);
  17     CAPACITY(p) = CREATE_SIZE;
  18     SET_SIZE(p, CREATE_SIZE);
  19     return p;
  20 }
  21
  22 extern void lose_s(symbol * p) {
  23     if (p == NULL) return;
  24     free((char *) p - HEAD);
  25 }
  26
  27 /*
  28    new_p = X_skip_utf8(p, c, lb, l, n); skips n characters forwards from p + c
  29    if n +ve, or n characters backwards from p +c - 1 if n -ve. new_p is the new
  30    position, or 0 on failure.
  31
  32    -- used to implement hop and next in the utf8 case.
  33 */
  34
  35 extern int skip_utf8(const symbol * p, int c, int lb, int l, int n) {
  36     int b;
  37     if (n >= 0) {
  38         for (; n > 0; n--) {
  39             if (c >= l) return -1;
  40             b = p[c++];
  41             if (b >= 0xC0) {   /* 1100 0000 */
  42                 while (c < l) {
  43                     b = p[c];
  44                     if (b >= 0xC0 || b < 0x80) break;
  45                     /* break unless b is 10------ */
  46                     c++;
  47                 }
  48             }
  49         }
  50     } else {
  51         for (; n < 0; n++) {
  52             if (c <= lb) return -1;
  53             b = p[--c];
  54             if (b >= 0x80) {   /* 1000 0000 */
  55                 while (c > lb) {
  56                     b = p[c];
  57                     if (b >= 0xC0) break; /* 1100 0000 */
  58                     c--;
  59                 }
  60             }
  61         }
  62     }
  63     return c;
  64 }
  65
  66 /* Code for character groupings: utf8 cases */
  67
  68 static int get_utf8(const symbol * p, int c, int l, int * slot) {
  69     int b0, b1;
  70     if (c >= l) return 0;
  71     b0 = p[c++];
  72     if (b0 < 0xC0 || c == l) {   /* 1100 0000 */
  73         * slot = b0; return 1;
  74     }
  75     b1 = p[c++];
  76     if (b0 < 0xE0 || c == l) {   /* 1110 0000 */
  77         * slot = (b0 & 0x1F) << 6 | (b1 & 0x3F); return 2;
  78     }
  79     * slot = (b0 & 0xF) << 12 | (b1 & 0x3F) << 6 | (*p & 0x3F); return 3;
  80 }
  81
  82 static int get_b_utf8(const symbol * p, int c, int lb, int * slot) {
  83     int b0, b1;
  84     if (c <= lb) return 0;
  85     b0 = p[--c];
  86     if (b0 < 0x80 || c == lb) {   /* 1000 0000 */
  87         * slot = b0; return 1;
  88     }
  89     b1 = p[--c];
  90     if (b1 >= 0xC0 || c == lb) {   /* 1100 0000 */
  91         * slot = (b1 & 0x1F) << 6 | (b0 & 0x3F); return 2;
  92     }
  93     * slot = (*p & 0xF) << 12 | (b1 & 0x3F) << 6 | (b0 & 0x3F); return 3;
  94 }
  95
  96 extern int in_grouping_U(struct SN_env * z, const unsigned char * s, int min, int max) {
  97     int ch;
  98     int w = get_utf8(z->p, z->c, z->l, & ch);
  99     unless (w) return 0;
 100     if (ch > max || (ch -= min) < 0 || (s[ch >> 3] & (0X1 << (ch & 0X7))) == 0) return 0;
 101     z->c += w; return 1;
 102 }
 103
 104 extern int in_grouping_b_U(struct SN_env * z, const unsigned char * s, int min, int max) {
 105     int ch;
 106     int w = get_b_utf8(z->p, z->c, z->lb, & ch);
 107     unless (w) return 0;
 108     if (ch > max || (ch -= min) < 0 || (s[ch >> 3] & (0X1 << (ch & 0X7))) == 0) return 0;
 109     z->c -= w; return 1;
 110 }
 111
 112 extern int out_grouping_U(struct SN_env * z, const unsigned char * s, int min, int max) {
 113     int ch;
 114     int w = get_utf8(z->p, z->c, z->l, & ch);
 115     unless (w) return 0;
 116     unless (ch > max || (ch -= min) < 0 || (s[ch >> 3] & (0X1 << (ch & 0X7))) == 0) return 0;
 117     z->c += w; return 1;
 118 }
 119
 120 extern int out_grouping_b_U(struct SN_env * z, const unsigned char * s, int min, int max) {
 121     int ch;
 122     int w = get_b_utf8(z->p, z->c, z->lb, & ch);
 123     unless (w) return 0;
 124     unless (ch > max || (ch -= min) < 0 || (s[ch >> 3] & (0X1 << (ch & 0X7))) == 0) return 0;
 125     z->c -= w; return 1;
 126 }
 127
 128 /* Code for character groupings: non-utf8 cases */
 129
 130 extern int in_grouping(struct SN_env * z, const unsigned char * s, int min, int max) {
 131     int ch;
 132     if (z->c >= z->l) return 0;
 133     ch = z->p[z->c];
 134     if (ch > max || (ch -= min) < 0 || (s[ch >> 3] & (0X1 << (ch & 0X7))) == 0) return 0;
 135     z->c++; return 1;
 136 }
 137
 138 extern int in_grouping_b(struct SN_env * z, const unsigned char * s, int min, int max) {
 139     int ch;
 140     if (z->c <= z->lb) return 0;
 141     ch = z->p[z->c - 1];
 142     if (ch > max || (ch -= min) < 0 || (s[ch >> 3] & (0X1 << (ch & 0X7))) == 0) return 0;
 143     z->c--; return 1;
 144 }
 145
 146 extern int out_grouping(struct SN_env * z, const unsigned char * s, int min, int max) {
 147     int ch;
 148     if (z->c >= z->l) return 0;
 149     ch = z->p[z->c];
 150     unless (ch > max || (ch -= min) < 0 || (s[ch >> 3] & (0X1 << (ch & 0X7))) == 0) return 0;
 151     z->c++; return 1;
 152 }
 153
 154 extern int out_grouping_b(struct SN_env * z, const unsigned char * s, int min, int max) {
 155     int ch;
 156     if (z->c <= z->lb) return 0;
 157     ch = z->p[z->c - 1];
 158     unless (ch > max || (ch -= min) < 0 || (s[ch >> 3] & (0X1 << (ch & 0X7))) == 0) return 0;
 159     z->c--; return 1;
 160 }
 161
 162 extern int eq_s(struct SN_env * z, int s_size, const symbol * s) {
 163     if (z->l - z->c < s_size || memcmp(z->p + z->c, s, s_size * sizeof(symbol)) != 0) return 0;
 164     z->c += s_size; return 1;
 165 }
 166
 167 extern int eq_s_b(struct SN_env * z, int s_size, const symbol * s) {
 168     if (z->c - z->lb < s_size || memcmp(z->p + z->c - s_size, s, s_size * sizeof(symbol)) != 0) return 0;
 169     z->c -= s_size; return 1;
 170 }
 171
 172 extern int eq_v(struct SN_env * z, const symbol * p) {
 173     return eq_s(z, SIZE(p), p);
 174 }
 175
 176 extern int eq_v_b(struct SN_env * z, const symbol * p) {
 177     return eq_s_b(z, SIZE(p), p);
 178 }
 179
 180 extern int find_among(struct SN_env * z, const struct among * v, int v_size) {
 181
 182     int i = 0;
 183     int j = v_size;
 184
 185     int c = z->c; int l = z->l;
 186     symbol * q = z->p + c;
 187
 188     const struct among * w;
 189
 190     int common_i = 0;
 191     int common_j = 0;
 192
 193     int first_key_inspected = 0;
 194
 195     while(1) {
 196         int k = i + ((j - i) >> 1);
 197         int diff = 0;
 198         int common = common_i < common_j ? common_i : common_j; /* smaller */
 199         w = v + k;
 200         {
 201             int i; for (i = common; i < w->s_size; i++) {
 202                 if (c + common == l) { diff = -1; break; }
 203                 diff = q[common] - w->s[i];
 204                 if (diff != 0) break;
 205                 common++;
 206             }
 207         }
 208         if (diff < 0) { j = k; common_j = common; }
 209                  else { i = k; common_i = common; }
 210         if (j - i <= 1) {
 211             if (i > 0) break; /* v->s has been inspected */
 212             if (j == i) break; /* only one item in v */
 213
 214             /* - but now we need to go round once more to get
 215                v->s inspected. This looks messy, but is actually
 216                the optimal approach.  */
 217
 218             if (first_key_inspected) break;
 219             first_key_inspected = 1;
 220         }
 221     }
 222     while(1) {
 223         w = v + i;
 224         if (common_i >= w->s_size) {
 225             z->c = c + w->s_size;
 226             if (w->function == 0) return w->result;
 227             {
 228                 int res = w->function(z);
 229                 z->c = c + w->s_size;
 230                 if (res) return w->result;
 231             }
 232         }
 233         i = w->substring_i;
 234         if (i < 0) return 0;
 235     }
 236 }
 237
 238 /* find_among_b is for backwards processing. Same comments apply */
 239
 240 extern int find_among_b(struct SN_env * z, const struct among * v, int v_size) {
 241
 242     int i = 0;
 243     int j = v_size;
 244
 245     int c = z->c; int lb = z->lb;
 246     symbol * q = z->p + c - 1;
 247
 248     const struct among * w;
 249
 250     int common_i = 0;
 251     int common_j = 0;
 252
 253     int first_key_inspected = 0;
 254
 255     while(1) {
 256         int k = i + ((j - i) >> 1);
 257         int diff = 0;
 258         int common = common_i < common_j ? common_i : common_j;
 259         w = v + k;
 260         {
 261             int i; for (i = w->s_size - 1 - common; i >= 0; i--) {
 262                 if (c - common == lb) { diff = -1; break; }
 263                 diff = q[- common] - w->s[i];
 264                 if (diff != 0) break;
 265                 common++;
 266             }
 267         }
 268         if (diff < 0) { j = k; common_j = common; }
 269                  else { i = k; common_i = common; }
 270         if (j - i <= 1) {
 271             if (i > 0) break;
 272             if (j == i) break;
 273             if (first_key_inspected) break;
 274             first_key_inspected = 1;
 275         }
 276     }
 277     while(1) {
 278         w = v + i;
 279         if (common_i >= w->s_size) {
 280             z->c = c - w->s_size;
 281             if (w->function == 0) return w->result;
 282             {
 283                 int res = w->function(z);
 284                 z->c = c - w->s_size;
 285                 if (res) return w->result;
 286             }
 287         }
 288         i = w->substring_i;
 289         if (i < 0) return 0;
 290     }
 291 }
 292
 293
 294 /* Increase the size of the buffer pointed to by p to at least n symbols.
 295  * If insufficient memory, returns NULL and frees the old buffer.
 296  */
 297 static symbol * increase_size(symbol * p, int n) {
 298     symbol * q;
 299     int new_size = n + 20;
 300     void * mem = realloc((char *) p - HEAD,
 301                          HEAD + (new_size + 1) * sizeof(symbol));
 302     if (mem == NULL) {
 303         lose_s(p);
 304         return NULL;
 305     }
 306     q = (symbol *) (HEAD + (char *)mem);
 307     CAPACITY(q) = new_size;
 308     return q;
 309 }
 310
 311 /* to replace symbols between c_bra and c_ket in z->p by the
 312    s_size symbols at s.
 313    Returns 0 on success, -1 on error.
 314    Also, frees z->p (and sets it to NULL) on error.
 315 */
 316 extern int replace_s(struct SN_env * z, int c_bra, int c_ket, int s_size, const symbol * s, int * adjptr)
 317 {
 318     int adjustment;
 319     int len;
 320     if (z->p == NULL) {
 321         z->p = create_s();
 322         if (z->p == NULL) return -1;
 323     }
 324     adjustment = s_size - (c_ket - c_bra);
 325     len = SIZE(z->p);
 326     if (adjustment != 0) {
 327         if (adjustment + len > CAPACITY(z->p)) {
 328             z->p = increase_size(z->p, adjustment + len);
 329             if (z->p == NULL) return -1;
 330         }
 331         memmove(z->p + c_ket + adjustment,
 332                 z->p + c_ket,
 333                 (len - c_ket) * sizeof(symbol));
 334         SET_SIZE(z->p, adjustment + len);
 335         z->l += adjustment;
 336         if (z->c >= c_ket)
 337             z->c += adjustment;
 338         else
 339             if (z->c > c_bra)
 340                 z->c = c_bra;
 341     }
 342     unless (s_size == 0) memmove(z->p + c_bra, s, s_size * sizeof(symbol));
 343     if (adjptr != NULL)
 344         *adjptr = adjustment;
 345     return 0;
 346 }
 347
 348 static int slice_check(struct SN_env * z) {
 349
 350     if (z->bra < 0 ||
 351         z->bra > z->ket ||
 352         z->ket > z->l ||
 353         z->p == NULL ||
 354         z->l > SIZE(z->p)) /* this line could be removed */
 355     {
 356 #if 0
 357         fprintf(stderr, "faulty slice operation:\n");
 358         debug(z, -1, 0);
 359 #endif
 360         return -1;
 361     }
 362     return 0;
 363 }
 364
 365 extern int slice_from_s(struct SN_env * z, int s_size, const symbol * s) {
 366     if (slice_check(z)) return -1;
 367     return replace_s(z, z->bra, z->ket, s_size, s, NULL);
 368 }
 369
 370 extern int slice_from_v(struct SN_env * z, const symbol * p) {
 371     return slice_from_s(z, SIZE(p), p);
 372 }
 373
 374 extern int slice_del(struct SN_env * z) {
 375     return slice_from_s(z, 0, 0);
 376 }
 377
 378 extern int insert_s(struct SN_env * z, int bra, int ket, int s_size, const symbol * s) {
 379     int adjustment;
 380     if (replace_s(z, bra, ket, s_size, s, &adjustment))
 381         return -1;
 382     if (bra <= z->bra) z->bra += adjustment;
 383     if (bra <= z->ket) z->ket += adjustment;
 384     return 0;
 385 }
 386
 387 extern int insert_v(struct SN_env * z, int bra, int ket, const symbol * p) {
 388     int adjustment;
 389     if (replace_s(z, bra, ket, SIZE(p), p, &adjustment))
 390         return -1;
 391     if (bra <= z->bra) z->bra += adjustment;
 392     if (bra <= z->ket) z->ket += adjustment;
 393     return 0;
 394 }
 395
 396 extern symbol * slice_to(struct SN_env * z, symbol * p) {
 397     if (slice_check(z)) {
 398         lose_s(p);
 399         return NULL;
 400     }
 401     {
 402         int len = z->ket - z->bra;
 403         if (CAPACITY(p) < len) {
 404             p = increase_size(p, len);
 405             if (p == NULL)
 406                 return NULL;
 407         }
 408         memmove(p, z->p + z->bra, len * sizeof(symbol));
 409         SET_SIZE(p, len);
 410     }
 411     return p;
 412 }
 413
 414 extern symbol * assign_to(struct SN_env * z, symbol * p) {
 415     int len = z->l;
 416     if (CAPACITY(p) < len) {
 417         p = increase_size(p, len);
 418         if (p == NULL)
 419             return NULL;
 420     }
 421     memmove(p, z->p, len * sizeof(symbol));
 422     SET_SIZE(p, len);
 423     return p;
 424 }
 425
 426 #if 0
 427 extern void debug(struct SN_env * z, int number, int line_count) {
 428     int i;
 429     int limit = SIZE(z->p);
 430     /*if (number >= 0) printf("%3d (line %4d): '", number, line_count);*/
 431     if (number >= 0) printf("%3d (line %4d): [%d]'", number, line_count,limit);
 432     for (i = 0; i <= limit; i++) {
 433         if (z->lb == i) printf("{");
 434         if (z->bra == i) printf("[");
 435         if (z->c == i) printf("|");
 436         if (z->ket == i) printf("]");
 437         if (z->l == i) printf("}");
 438         if (i < limit)
 439         {   int ch = z->p[i];
 440             if (ch == 0) ch = '#';
 441             printf("%c", ch);
 442         }
 443     }
 444     printf("'\n");
 445 }
 446 #endif