rset/rsbetween.c

   1 /* $Id: rsbetween.c,v 1.16 2004-08-04 09:59:03 heikki Exp $
   2    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
   3    Index Data Aps
   4
   5 This file is part of the Zebra server.
   6
   7 Zebra is free software; you can redistribute it and/or modify it under
   8 the terms of the GNU General Public License as published by the Free
   9 Software Foundation; either version 2, or (at your option) any later
  10 version.
  11
  12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
  13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  14 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  15 for more details.
  16
  17 You should have received a copy of the GNU General Public License
  18 along with Zebra; see the file LICENSE.zebra.  If not, write to the
  19 Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
  20 02111-1307, USA.
  21 */
  22
  23
  24 /* rsbetween is (mostly) used for xml searches. It returns the hits of the
  25  * "middle" rset, that are in between the "left" and "right" rsets. For
  26  * example "Shakespeare" in between "<title>" and </title>. The thing is
  27  * complicated by the inclusion of attributes (from their own rset). If attrs
  28  * specified, they must match the "left" rset (start tag). "Hamlet" between
  29  * "<title lang=eng>" and "</title>". (This assumes that the attributes are
  30  * indexed to the same seqno as the tags).
  31 */
  32
  33 #include <stdio.h>
  34 #include <stdlib.h>
  35 #include <string.h>
  36 #include <assert.h>
  37
  38 #include <zebrautl.h>
  39 #include <rsbetween.h>
  40
  41 #define RSBETWEEN_DEBUG 0
  42
  43 static void *r_create_between(RSET ct, const struct rset_control *sel, void *parms);
  44 static RSFD r_open_between (RSET ct, int flag);
  45 static void r_close_between (RSFD rfd);
  46 static void r_delete_between (RSET ct);
  47 static void r_rewind_between (RSFD rfd);
  48 static int r_forward_between(RSET ct, RSFD rfd, void *buf, int *term_index,
  49                      int (*cmpfunc)(const void *p1, const void *p2),
  50                      const void *untilbuf);
  51 static int r_read_between (RSFD rfd, void *buf, int *term_index);
  52 static int r_write_between (RSFD rfd, const void *buf);
  53
  54 static const struct rset_control control_between =
  55 {
  56     "between",
  57     r_create_between,
  58     r_open_between,
  59     r_close_between,
  60     r_delete_between,
  61     r_rewind_between,
  62     r_forward_between, /* rset_default_forward, */
  63     rset_default_pos,
  64     r_read_between,
  65     r_write_between,
  66 };
  67
  68
  69 const struct rset_control *rset_kind_between = &control_between;
  70
  71 struct rset_between_info {
  72     int key_size;
  73     RSET rset_l;
  74     RSET rset_m;
  75     RSET rset_r;
  76     RSET rset_attr;
  77     int term_index_s;
  78     int (*cmp)(const void *p1, const void *p2);
  79     char *(*printer)(const void *p1, char *buf);
  80     struct rset_between_rfd *rfd_list;
  81 };
  82
  83 struct rset_between_rfd {
  84     RSFD rfd_l;
  85     RSFD rfd_m;
  86     RSFD rfd_r;
  87     RSFD rfd_attr;
  88     int  more_l;
  89     int  more_m;
  90     int  more_r;
  91     int  more_attr;
  92     int term_index_l;
  93     int term_index_m;
  94     int term_index_r;
  95     void *buf_l;
  96     void *buf_m;
  97     void *buf_r;
  98     void *buf_attr;
  99     int level;
 100     struct rset_between_rfd *next;
 101     struct rset_between_info *info;
 102 };
 103
 104 #if RSBETWEEN_DEBUG
 105 static void log2 (struct rset_between_rfd *p, char *msg, int cmp_l, int cmp_r)
 106 {
 107     char buf_l[32];
 108     char buf_m[32];
 109     char buf_r[32];
 110     logf(LOG_DEBUG,"btw: %s l=%s(%d/%d) m=%s(%d) r=%s(%d/%d), lev=%d",
 111       msg,
 112       (*p->info->printer)(p->buf_l, buf_l), p->more_l, cmp_l,
 113       (*p->info->printer)(p->buf_m, buf_m), p->more_m,
 114       (*p->info->printer)(p->buf_r, buf_r), p->more_r, cmp_r,
 115       p->level);
 116 }
 117 #endif
 118
 119 static void *r_create_between (RSET ct, const struct rset_control *sel,
 120                                void *parms)
 121 {
 122     rset_between_parms *between_parms = (rset_between_parms *) parms;
 123     struct rset_between_info *info;
 124
 125     info = (struct rset_between_info *) xmalloc (sizeof(*info));
 126     info->key_size = between_parms->key_size;
 127     info->rset_l = between_parms->rset_l;
 128     info->rset_m = between_parms->rset_m;
 129     info->rset_r = between_parms->rset_r;
 130     info->rset_attr = between_parms->rset_attr;
 131     if (rset_is_volatile(info->rset_l) ||
 132         rset_is_volatile(info->rset_m) ||
 133         rset_is_volatile(info->rset_r))
 134         ct->flags |= RSET_FLAG_VOLATILE;
 135     info->cmp = between_parms->cmp;
 136     info->printer = between_parms->printer;
 137     info->rfd_list = NULL;
 138
 139     info->term_index_s = info->rset_l->no_rset_terms;
 140     if (info->rset_m)
 141     {
 142         ct->no_rset_terms =
 143             info->rset_l->no_rset_terms +
 144             info->rset_m->no_rset_terms +
 145             info->rset_r->no_rset_terms;
 146         ct->rset_terms = (RSET_TERM *)
 147             xmalloc (sizeof (*ct->rset_terms) * ct->no_rset_terms);
 148         memcpy (ct->rset_terms, info->rset_l->rset_terms,
 149                 info->rset_l->no_rset_terms * sizeof(*ct->rset_terms));
 150         memcpy (ct->rset_terms + info->rset_l->no_rset_terms,
 151                 info->rset_m->rset_terms,
 152                 info->rset_m->no_rset_terms * sizeof(*ct->rset_terms));
 153         memcpy (ct->rset_terms + info->rset_l->no_rset_terms +
 154                 info->rset_m->no_rset_terms,
 155                 info->rset_r->rset_terms,
 156                 info->rset_r->no_rset_terms * sizeof(*ct->rset_terms));
 157     }
 158     else
 159     {
 160         ct->no_rset_terms =
 161             info->rset_l->no_rset_terms +
 162             info->rset_r->no_rset_terms;
 163         ct->rset_terms = (RSET_TERM *)
 164             xmalloc (sizeof (*ct->rset_terms) * ct->no_rset_terms);
 165         memcpy (ct->rset_terms, info->rset_l->rset_terms,
 166                 info->rset_l->no_rset_terms * sizeof(*ct->rset_terms));
 167         memcpy (ct->rset_terms + info->rset_l->no_rset_terms,
 168                 info->rset_r->rset_terms,
 169                 info->rset_r->no_rset_terms * sizeof(*ct->rset_terms));
 170     }
 171
 172     return info;
 173 }
 174
 175 static RSFD r_open_between (RSET ct, int flag)
 176 {
 177     struct rset_between_info *info = (struct rset_between_info *) ct->buf;
 178     struct rset_between_rfd *rfd;
 179
 180     if (flag & RSETF_WRITE)
 181     {
 182         logf (LOG_FATAL, "between set type is read-only");
 183         return NULL;
 184     }
 185     rfd = (struct rset_between_rfd *) xmalloc (sizeof(*rfd));
 186     rfd->next = info->rfd_list;
 187     info->rfd_list = rfd;
 188     rfd->info = info;
 189
 190     rfd->buf_l = xmalloc (info->key_size);
 191     rfd->buf_m = xmalloc (info->key_size);
 192     rfd->buf_r = xmalloc (info->key_size);
 193     rfd->buf_attr = xmalloc (info->key_size);
 194
 195     rfd->rfd_l = rset_open (info->rset_l, RSETF_READ);
 196     rfd->rfd_m = rset_open (info->rset_m, RSETF_READ);
 197     rfd->rfd_r = rset_open (info->rset_r, RSETF_READ);
 198
 199     rfd->more_l = rset_read (info->rset_l, rfd->rfd_l, rfd->buf_l,
 200                              &rfd->term_index_l);
 201     rfd->more_m = rset_read (info->rset_m, rfd->rfd_m, rfd->buf_m,
 202                              &rfd->term_index_m);
 203     rfd->more_r = rset_read (info->rset_r, rfd->rfd_r, rfd->buf_r,
 204                              &rfd->term_index_r);
 205     if (info->rset_attr)
 206     {
 207         int dummy;
 208         rfd->rfd_attr = rset_open (info->rset_attr, RSETF_READ);
 209         rfd->more_attr = rset_read (info->rset_attr, rfd->rfd_attr,
 210                                     rfd->buf_attr, &dummy);
 211     }
 212     rfd->level=0;
 213     return rfd;
 214 }
 215
 216 static void r_close_between (RSFD rfd)
 217 {
 218     struct rset_between_info *info = ((struct rset_between_rfd*)rfd)->info;
 219     struct rset_between_rfd **rfdp;
 220
 221     for (rfdp = &info->rfd_list; *rfdp; rfdp = &(*rfdp)->next)
 222         if (*rfdp == rfd)
 223         {
 224             xfree ((*rfdp)->buf_l);
 225             xfree ((*rfdp)->buf_m);
 226             xfree ((*rfdp)->buf_r);
 227             xfree ((*rfdp)->buf_attr);
 228             rset_close (info->rset_l, (*rfdp)->rfd_l);
 229             rset_close (info->rset_m, (*rfdp)->rfd_m);
 230             rset_close (info->rset_r, (*rfdp)->rfd_r);
 231             if (info->rset_attr)
 232                 rset_close (info->rset_attr, (*rfdp)->rfd_attr);
 233
 234             *rfdp = (*rfdp)->next;
 235             xfree (rfd);
 236             return;
 237         }
 238     logf (LOG_FATAL, "r_close_between but no rfd match!");
 239     assert (0);
 240 }
 241
 242 static void r_delete_between (RSET ct)
 243 {
 244     struct rset_between_info *info = (struct rset_between_info *) ct->buf;
 245
 246     assert (info->rfd_list == NULL);
 247     xfree (ct->rset_terms);
 248     rset_delete (info->rset_l);
 249     rset_delete (info->rset_m);
 250     rset_delete (info->rset_r);
 251     if (info->rset_attr)
 252         rset_delete (info->rset_attr);
 253     xfree (info);
 254 }
 255
 256 static void r_rewind_between (RSFD rfd)
 257 {
 258     struct rset_between_info *info = ((struct rset_between_rfd*)rfd)->info;
 259     struct rset_between_rfd *p = (struct rset_between_rfd *) rfd;
 260
 261 #if RSBETWEEN_DEBUG
 262     logf (LOG_DEBUG, "rsbetween_rewind");
 263 #endif
 264     rset_rewind (info->rset_l, p->rfd_l);
 265     rset_rewind (info->rset_m, p->rfd_m);
 266     rset_rewind (info->rset_r, p->rfd_r);
 267     p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l, &p->term_index_l);
 268     p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m, &p->term_index_m);
 269     p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r, &p->term_index_r);
 270     if (info->rset_attr)
 271     {
 272         int dummy;
 273         rset_rewind (info->rset_attr, p->rfd_attr);
 274         p->more_attr = rset_read (info->rset_attr, p->rfd_attr, p->buf_attr,
 275                                   &dummy);
 276     }
 277     p->level=0;
 278 }
 279
 280
 281
 282 static int r_forward_between(RSET ct, RSFD rfd, void *buf, int *term_index,
 283                      int (*cmpfunc)(const void *p1, const void *p2),
 284                      const void *untilbuf)
 285 {
 286     struct rset_between_info *info = ((struct rset_between_rfd*)rfd)->info;
 287     struct rset_between_rfd *p = (struct rset_between_rfd *) rfd;
 288     int rc;
 289 #if RSBETWEEN_DEBUG
 290     log2( p, "fwd: before forward", 0,0);
 291 #endif
 292     /* It is enough to forward the m pointer here, the read will */
 293     /* naturally forward the l, m, and attr pointers */
 294     if (p->more_m)
 295         p->more_m=rset_forward(info->rset_m,p->rfd_m, p->buf_m,
 296                         &p->term_index_m, info->cmp,untilbuf);
 297 #if RSBETWEEN_DEBUG
 298     log2( p, "fwd: after forward M", 0,0);
 299 #endif
 300     rc = r_read_between(rfd, buf, term_index);
 301 #if RSBETWEEN_DEBUG
 302     log2( p, "fwd: after forward", 0,0);
 303 #endif
 304     return rc;
 305 }
 306
 307
 308
 309
 310 static int r_read_between (RSFD rfd, void *buf, int *term_index)
 311 {
 312     struct rset_between_rfd *p = (struct rset_between_rfd *) rfd;
 313     struct rset_between_info *info = p->info;
 314     int cmp_l=0;
 315     int cmp_r=0;
 316     int attr_match = 0;
 317
 318     while (p->more_m)
 319     {
 320 #if RSBETWEEN_DEBUG
 321         log2( p, "start of loop", cmp_l, cmp_r);
 322 #endif
 323
 324         /* forward L until past m, count levels, note rec boundaries */
 325         if (p->more_l)
 326             cmp_l= (*info->cmp)(p->buf_l, p->buf_m);
 327         else
 328         {
 329             p->level = 0;
 330             cmp_l=2; /* past this record */
 331         }
 332 #if RSBETWEEN_DEBUG
 333         log2( p, "after first L", cmp_l, cmp_r);
 334 #endif
 335
 336         while (cmp_l < 0)   /* l before m */
 337         {
 338             if (cmp_l == -2)
 339                 p->level=0; /* earlier record */
 340             if (cmp_l == -1)
 341             {
 342                 p->level++; /* relevant start tag */
 343
 344                 if (!info->rset_attr)
 345                     attr_match = 1;
 346                 else
 347                 {
 348                     int cmp_attr;
 349                     int dummy_term;
 350                     attr_match = 0;
 351                     while (p->more_attr)
 352                     {
 353                         cmp_attr = (*info->cmp)(p->buf_attr, p->buf_l);
 354                         if (cmp_attr == 0)
 355                         {
 356                             attr_match = 1;
 357                             break;
 358                         }
 359                         else if (cmp_attr > 0)
 360                             break;
 361                         else if (cmp_attr==-1)
 362                             p->more_attr = rset_read (info->rset_attr, p->rfd_attr,
 363                                                   p->buf_attr, &dummy_term);
 364                             /* if we had a forward that went all the way to
 365                              * the seqno, we could use that. But fwd only goes
 366                              * to the sysno */
 367                         else if (cmp_attr==-2)
 368                         {
 369                             p->more_attr = rset_forward(
 370                                       info->rset_attr, p->rfd_attr,
 371                                       p->buf_attr, &dummy_term,
 372                                       info->cmp, p->buf_l);
 373 #if RSBETWEEN_DEBUG
 374                             logf(LOG_DEBUG, "btw: after frowarding attr m=%d",p->more_attr);
 375 #endif
 376                         }
 377                     } /* while more_attr */
 378                 }
 379             }
 380 #define NEWCODE 1
 381 #if NEWCODE
 382             if (cmp_l==-2)
 383             {
 384                 if (p->more_l)
 385                 {
 386                     p->more_l=rset_forward(
 387                                       info->rset_l, p->rfd_l,
 388                                       p->buf_l, &p->term_index_l,
 389                                       info->cmp, p->buf_m);
 390                     if (p->more_l)
 391                         cmp_l= (*info->cmp)(p->buf_l, p->buf_m);
 392                     else
 393                         cmp_l=2;
 394 #if RSBETWEEN_DEBUG
 395                     log2( p, "after forwarding L", cmp_l, cmp_r);
 396 #endif
 397                 }
 398             } else
 399             {
 400                 p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l,
 401                               &p->term_index_l);
 402             }
 403 #else
 404             p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l,
 405                               &p->term_index_l);
 406 #endif
 407             if (p->more_l)
 408             {
 409                 cmp_l= (*info->cmp)(p->buf_l, p->buf_m);
 410             }
 411             else
 412                 cmp_l=2;
 413 #if RSBETWEEN_DEBUG
 414             log2( p, "end of L loop", cmp_l, cmp_r);
 415 #endif
 416         } /* forward L */
 417
 418
 419         /* forward R until past m, count levels */
 420 #if RSBETWEEN_DEBUG
 421         log2( p, "Before moving R", cmp_l, cmp_r);
 422 #endif
 423         if (p->more_r)
 424             cmp_r= (*info->cmp)(p->buf_r, p->buf_m);
 425         else
 426             cmp_r=2;
 427 #if RSBETWEEN_DEBUG
 428         log2( p, "after first R", cmp_l, cmp_r);
 429 #endif
 430         while (cmp_r < 0)   /* r before m */
 431         {
 432              /* -2, earlier record, don't count level */
 433             if (cmp_r == -1)
 434                 p->level--; /* relevant end tag */
 435             if (p->more_r)
 436             {
 437 #if NEWCODE
 438                 if (cmp_r==-2)
 439                 {
 440                     p->more_r=rset_forward(
 441                                       info->rset_r, p->rfd_r,
 442                                       p->buf_r, &p->term_index_r,
 443                                       info->cmp, p->buf_m);
 444                 } else
 445                 {
 446                     p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r,
 447                                        &p->term_index_r);
 448                 }
 449                 if (p->more_r)
 450                     cmp_r= (*info->cmp)(p->buf_r, p->buf_m);
 451
 452 #else
 453                 p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r,
 454                                        &p->term_index_r);
 455                 cmp_r= (*info->cmp)(p->buf_r, p->buf_m);
 456 #endif
 457             }
 458             else
 459                 cmp_r=2;
 460 #if RSBETWEEN_DEBUG
 461         log2( p, "End of R loop", cmp_l, cmp_r);
 462 #endif
 463         } /* forward R */
 464
 465         if ( ( p->level <= 0 ) && ! p->more_l)
 466             return 0; /* no more start tags, nothing more to find */
 467
 468         if ( attr_match && p->level > 0)  /* within a tag pair (or deeper) */
 469         {
 470             memcpy (buf, p->buf_m, info->key_size);
 471             *term_index = p->term_index_m;
 472 #if RSBETWEEN_DEBUG
 473             log2( p, "Returning a hit (and forwarding m)", cmp_l, cmp_r);
 474 #endif
 475             p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m,
 476                                    &p->term_index_m);
 477             if (cmp_l == 2)
 478                 p->level = 0;
 479             return 1;
 480         }
 481         else if ( ! p->more_l )  /* not in data, no more starts */
 482         {
 483 #if RSBETWEEN_DEBUG
 484             log2( p, "no more starts, exiting without a hit", cmp_l, cmp_r);
 485 #endif
 486             return 0;  /* ergo, nothing can be found. stop scanning */
 487         }
 488 #if NEWCODE
 489         if (cmp_l == 2)
 490         {
 491             p->level = 0;
 492             p->more_m=rset_forward(
 493                               info->rset_m, p->rfd_m,
 494                               p->buf_m, &p->term_index_m,
 495                               info->cmp, p->buf_l);
 496         } else
 497         {
 498             p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m,
 499                                &p->term_index_m);
 500         }
 501 #else
 502         if (cmp_l == 2)
 503             p->level = 0;
 504         p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m,
 505                                &p->term_index_m);
 506 #endif
 507 #if RSBETWEEN_DEBUG
 508         log2( p, "End of M loop", cmp_l, cmp_r);
 509 #endif
 510     } /* while more_m */
 511
 512 #if RSBETWEEN_DEBUG
 513     log2( p, "Exiting, nothing more in m", cmp_l, cmp_r);
 514 #endif
 515     return 0;  /* no more data possible */
 516
 517
 518 }  /* r_read */
 519
 520
 521 static int r_write_between (RSFD rfd, const void *buf)
 522 {
 523     logf (LOG_FATAL, "between set type is read-only");
 524     return -1;
 525 }
 526