rset/rsbetween.c

   1 /* $Id: rsbetween.c,v 1.14 2004-08-03 12:15:45 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 <rsbetween.h>
  39 #include <zebrautl.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_count_between (RSET ct); */
  52 static int r_read_between (RSFD rfd, void *buf, int *term_index);
  53 static int r_write_between (RSFD rfd, const void *buf);
  54
  55 static const struct rset_control control_between =
  56 {
  57     "between",
  58     r_create_between,
  59     r_open_between,
  60     r_close_between,
  61     r_delete_between,
  62     r_rewind_between,
  63     r_forward_between, /* rset_default_forward, */
  64     /* r_count_between, */
  65     r_read_between,
  66     r_write_between,
  67 };
  68
  69
  70 const struct rset_control *rset_kind_between = &control_between;
  71
  72 struct rset_between_info {
  73     int key_size;
  74     RSET rset_l;
  75     RSET rset_m;
  76     RSET rset_r;
  77     RSET rset_attr;
  78     int term_index_s;
  79     int (*cmp)(const void *p1, const void *p2);
  80     char *(*printer)(const void *p1, char *buf);
  81     struct rset_between_rfd *rfd_list;
  82 };
  83
  84 struct rset_between_rfd {
  85     RSFD rfd_l;
  86     RSFD rfd_m;
  87     RSFD rfd_r;
  88     RSFD rfd_attr;
  89     int  more_l;
  90     int  more_m;
  91     int  more_r;
  92     int  more_attr;
  93     int term_index_l;
  94     int term_index_m;
  95     int term_index_r;
  96     void *buf_l;
  97     void *buf_m;
  98     void *buf_r;
  99     void *buf_attr;
 100     int level;
 101     struct rset_between_rfd *next;
 102     struct rset_between_info *info;
 103 };
 104
 105 #if RSBETWEEN_DEBUG
 106 static void log2 (struct rset_between_rfd *p, char *msg, int cmp_l, int cmp_r)
 107 {
 108     char buf_l[32];
 109     char buf_m[32];
 110     char buf_r[32];
 111     logf(LOG_DEBUG,"btw: %s l=%s(%d/%d) m=%s(%d) r=%s(%d/%d), lev=%d",
 112       msg,
 113       (*p->info->printer)(p->buf_l, buf_l), p->more_l, cmp_l,
 114       (*p->info->printer)(p->buf_m, buf_m), p->more_m,
 115       (*p->info->printer)(p->buf_r, buf_r), p->more_r, cmp_r,
 116       p->level);
 117 }
 118 #endif
 119
 120 static void *r_create_between (RSET ct, const struct rset_control *sel,
 121                                void *parms)
 122 {
 123     rset_between_parms *between_parms = (rset_between_parms *) parms;
 124     struct rset_between_info *info;
 125
 126     info = (struct rset_between_info *) xmalloc (sizeof(*info));
 127     info->key_size = between_parms->key_size;
 128     info->rset_l = between_parms->rset_l;
 129     info->rset_m = between_parms->rset_m;
 130     info->rset_r = between_parms->rset_r;
 131     info->rset_attr = between_parms->rset_attr;
 132     if (rset_is_volatile(info->rset_l) ||
 133         rset_is_volatile(info->rset_m) ||
 134         rset_is_volatile(info->rset_r))
 135         ct->flags |= RSET_FLAG_VOLATILE;
 136     info->cmp = between_parms->cmp;
 137     info->printer = between_parms->printer;
 138     info->rfd_list = NULL;
 139
 140     info->term_index_s = info->rset_l->no_rset_terms;
 141     if (info->rset_m)
 142     {
 143         ct->no_rset_terms =
 144             info->rset_l->no_rset_terms +
 145             info->rset_m->no_rset_terms +
 146             info->rset_r->no_rset_terms;
 147         ct->rset_terms = (RSET_TERM *)
 148             xmalloc (sizeof (*ct->rset_terms) * ct->no_rset_terms);
 149         memcpy (ct->rset_terms, info->rset_l->rset_terms,
 150                 info->rset_l->no_rset_terms * sizeof(*ct->rset_terms));
 151         memcpy (ct->rset_terms + info->rset_l->no_rset_terms,
 152                 info->rset_m->rset_terms,
 153                 info->rset_m->no_rset_terms * sizeof(*ct->rset_terms));
 154         memcpy (ct->rset_terms + info->rset_l->no_rset_terms +
 155                 info->rset_m->no_rset_terms,
 156                 info->rset_r->rset_terms,
 157                 info->rset_r->no_rset_terms * sizeof(*ct->rset_terms));
 158     }
 159     else
 160     {
 161         ct->no_rset_terms =
 162             info->rset_l->no_rset_terms +
 163             info->rset_r->no_rset_terms;
 164         ct->rset_terms = (RSET_TERM *)
 165             xmalloc (sizeof (*ct->rset_terms) * ct->no_rset_terms);
 166         memcpy (ct->rset_terms, info->rset_l->rset_terms,
 167                 info->rset_l->no_rset_terms * sizeof(*ct->rset_terms));
 168         memcpy (ct->rset_terms + info->rset_l->no_rset_terms,
 169                 info->rset_r->rset_terms,
 170                 info->rset_r->no_rset_terms * sizeof(*ct->rset_terms));
 171     }
 172
 173     return info;
 174 }
 175
 176 static RSFD r_open_between (RSET ct, int flag)
 177 {
 178     struct rset_between_info *info = (struct rset_between_info *) ct->buf;
 179     struct rset_between_rfd *rfd;
 180
 181     if (flag & RSETF_WRITE)
 182     {
 183         logf (LOG_FATAL, "between set type is read-only");
 184         return NULL;
 185     }
 186     rfd = (struct rset_between_rfd *) xmalloc (sizeof(*rfd));
 187     rfd->next = info->rfd_list;
 188     info->rfd_list = rfd;
 189     rfd->info = info;
 190
 191     rfd->buf_l = xmalloc (info->key_size);
 192     rfd->buf_m = xmalloc (info->key_size);
 193     rfd->buf_r = xmalloc (info->key_size);
 194     rfd->buf_attr = xmalloc (info->key_size);
 195
 196     rfd->rfd_l = rset_open (info->rset_l, RSETF_READ);
 197     rfd->rfd_m = rset_open (info->rset_m, RSETF_READ);
 198     rfd->rfd_r = rset_open (info->rset_r, RSETF_READ);
 199
 200     rfd->more_l = rset_read (info->rset_l, rfd->rfd_l, rfd->buf_l,
 201                              &rfd->term_index_l);
 202     rfd->more_m = rset_read (info->rset_m, rfd->rfd_m, rfd->buf_m,
 203                              &rfd->term_index_m);
 204     rfd->more_r = rset_read (info->rset_r, rfd->rfd_r, rfd->buf_r,
 205                              &rfd->term_index_r);
 206     if (info->rset_attr)
 207     {
 208         int dummy;
 209         rfd->rfd_attr = rset_open (info->rset_attr, RSETF_READ);
 210         rfd->more_attr = rset_read (info->rset_attr, rfd->rfd_attr,
 211                                     rfd->buf_attr, &dummy);
 212     }
 213     rfd->level=0;
 214     return rfd;
 215 }
 216
 217 static void r_close_between (RSFD rfd)
 218 {
 219     struct rset_between_info *info = ((struct rset_between_rfd*)rfd)->info;
 220     struct rset_between_rfd **rfdp;
 221
 222     for (rfdp = &info->rfd_list; *rfdp; rfdp = &(*rfdp)->next)
 223         if (*rfdp == rfd)
 224         {
 225             xfree ((*rfdp)->buf_l);
 226             xfree ((*rfdp)->buf_m);
 227             xfree ((*rfdp)->buf_r);
 228             xfree ((*rfdp)->buf_attr);
 229             rset_close (info->rset_l, (*rfdp)->rfd_l);
 230             rset_close (info->rset_m, (*rfdp)->rfd_m);
 231             rset_close (info->rset_r, (*rfdp)->rfd_r);
 232             if (info->rset_attr)
 233                 rset_close (info->rset_attr, (*rfdp)->rfd_attr);
 234
 235             *rfdp = (*rfdp)->next;
 236             xfree (rfd);
 237             return;
 238         }
 239     logf (LOG_FATAL, "r_close_between but no rfd match!");
 240     assert (0);
 241 }
 242
 243 static void r_delete_between (RSET ct)
 244 {
 245     struct rset_between_info *info = (struct rset_between_info *) ct->buf;
 246
 247     assert (info->rfd_list == NULL);
 248     xfree (ct->rset_terms);
 249     rset_delete (info->rset_l);
 250     rset_delete (info->rset_m);
 251     rset_delete (info->rset_r);
 252     if (info->rset_attr)
 253         rset_delete (info->rset_attr);
 254     xfree (info);
 255 }
 256
 257 static void r_rewind_between (RSFD rfd)
 258 {
 259     struct rset_between_info *info = ((struct rset_between_rfd*)rfd)->info;
 260     struct rset_between_rfd *p = (struct rset_between_rfd *) rfd;
 261
 262 #if RSBETWEEN_DEBUG
 263     logf (LOG_DEBUG, "rsbetween_rewind");
 264 #endif
 265     rset_rewind (info->rset_l, p->rfd_l);
 266     rset_rewind (info->rset_m, p->rfd_m);
 267     rset_rewind (info->rset_r, p->rfd_r);
 268     p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l, &p->term_index_l);
 269     p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m, &p->term_index_m);
 270     p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r, &p->term_index_r);
 271     if (info->rset_attr)
 272     {
 273         int dummy;
 274         rset_rewind (info->rset_attr, p->rfd_attr);
 275         p->more_attr = rset_read (info->rset_attr, p->rfd_attr, p->buf_attr,
 276                                   &dummy);
 277     }
 278     p->level=0;
 279 }
 280
 281
 282
 283 static int r_forward_between(RSET ct, RSFD rfd, void *buf, int *term_index,
 284                      int (*cmpfunc)(const void *p1, const void *p2),
 285                      const void *untilbuf)
 286 {
 287     struct rset_between_info *info = ((struct rset_between_rfd*)rfd)->info;
 288     struct rset_between_rfd *p = (struct rset_between_rfd *) rfd;
 289     int rc;
 290 #if RSBETWEEN_DEBUG
 291     log2( p, "fwd: before forward", 0,0);
 292 #endif
 293     /* It is enough to forward the m pointer here, the read will */
 294     /* naturally forward the l, m, and attr pointers */
 295     if (p->more_m)
 296         p->more_m=rset_forward(info->rset_m,p->rfd_m, p->buf_m,
 297                         &p->term_index_m, info->cmp,untilbuf);
 298 #if RSBETWEEN_DEBUG
 299     log2( p, "fwd: after forward M", 0,0);
 300 #endif
 301     rc = r_read_between(rfd, buf, term_index);
 302 #if RSBETWEEN_DEBUG
 303     log2( p, "fwd: after forward", 0,0);
 304 #endif
 305     return rc;
 306 }
 307
 308 /*
 309 static int r_count_between (RSET ct)
 310 {
 311     return 0;
 312 }
 313 */
 314
 315
 316
 317 static int r_read_between (RSFD rfd, void *buf, int *term_index)
 318 {
 319     struct rset_between_rfd *p = (struct rset_between_rfd *) rfd;
 320     struct rset_between_info *info = p->info;
 321     int cmp_l=0;
 322     int cmp_r=0;
 323     int attr_match = 0;
 324
 325     while (p->more_m)
 326     {
 327 #if RSBETWEEN_DEBUG
 328         log2( p, "start of loop", cmp_l, cmp_r);
 329 #endif
 330
 331         /* forward L until past m, count levels, note rec boundaries */
 332         if (p->more_l)
 333             cmp_l= (*info->cmp)(p->buf_l, p->buf_m);
 334         else
 335         {
 336             p->level = 0;
 337             cmp_l=2; /* past this record */
 338         }
 339 #if RSBETWEEN_DEBUG
 340         log2( p, "after first L", cmp_l, cmp_r);
 341 #endif
 342
 343         while (cmp_l < 0)   /* l before m */
 344         {
 345             if (cmp_l == -2)
 346                 p->level=0; /* earlier record */
 347             if (cmp_l == -1)
 348             {
 349                 p->level++; /* relevant start tag */
 350
 351                 if (!info->rset_attr)
 352                     attr_match = 1;
 353                 else
 354                 {
 355                     int cmp_attr;
 356                     int dummy_term;
 357                     attr_match = 0;
 358                     while (p->more_attr)
 359                     {
 360                         cmp_attr = (*info->cmp)(p->buf_attr, p->buf_l);
 361                         if (cmp_attr == 0)
 362                         {
 363                             attr_match = 1;
 364                             break;
 365                         }
 366                         else if (cmp_attr > 0)
 367                             break;
 368                         else if (cmp_attr==-1)
 369                             p->more_attr = rset_read (info->rset_attr, p->rfd_attr,
 370                                                   p->buf_attr, &dummy_term);
 371                             /* if we had a forward that went all the way to
 372                              * the seqno, we could use that. But fwd only goes
 373                              * to the sysno */
 374                         else if (cmp_attr==-2)
 375                         {
 376                             p->more_attr = rset_forward(
 377                                       info->rset_attr, p->rfd_attr,
 378                                       p->buf_attr, &dummy_term,
 379                                       info->cmp, p->buf_l);
 380 #if RSBETWEEN_DEBUG
 381                             logf(LOG_DEBUG, "btw: after frowarding attr m=%d",p->more_attr);
 382 #endif
 383                         }
 384                     } /* while more_attr */
 385                 }
 386             }
 387 #define NEWCODE 1
 388 #if NEWCODE
 389             if (cmp_l==-2)
 390             {
 391                 if (p->more_l)
 392                 {
 393                     p->more_l=rset_forward(
 394                                       info->rset_l, p->rfd_l,
 395                                       p->buf_l, &p->term_index_l,
 396                                       info->cmp, p->buf_m);
 397                     if (p->more_l)
 398                         cmp_l= (*info->cmp)(p->buf_l, p->buf_m);
 399                     else
 400                         cmp_l=2;
 401 #if RSBETWEEN_DEBUG
 402                     log2( p, "after forwarding L", cmp_l, cmp_r);
 403 #endif
 404                 }
 405             } else
 406             {
 407                 p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l,
 408                               &p->term_index_l);
 409             }
 410 #else
 411             p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l,
 412                               &p->term_index_l);
 413 #endif
 414             if (p->more_l)
 415             {
 416                 cmp_l= (*info->cmp)(p->buf_l, p->buf_m);
 417             }
 418             else
 419                 cmp_l=2;
 420 #if RSBETWEEN_DEBUG
 421             log2( p, "end of L loop", cmp_l, cmp_r);
 422 #endif
 423         } /* forward L */
 424
 425
 426         /* forward R until past m, count levels */
 427 #if RSBETWEEN_DEBUG
 428         log2( p, "Before moving R", cmp_l, cmp_r);
 429 #endif
 430         if (p->more_r)
 431             cmp_r= (*info->cmp)(p->buf_r, p->buf_m);
 432         else
 433             cmp_r=2;
 434 #if RSBETWEEN_DEBUG
 435         log2( p, "after first R", cmp_l, cmp_r);
 436 #endif
 437         while (cmp_r < 0)   /* r before m */
 438         {
 439              /* -2, earlier record, don't count level */
 440             if (cmp_r == -1)
 441                 p->level--; /* relevant end tag */
 442             if (p->more_r)
 443             {
 444 #if NEWCODE
 445                 if (cmp_r==-2)
 446                 {
 447                     p->more_r=rset_forward(
 448                                       info->rset_r, p->rfd_r,
 449                                       p->buf_r, &p->term_index_r,
 450                                       info->cmp, p->buf_m);
 451                 } else
 452                 {
 453                     p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r,
 454                                        &p->term_index_r);
 455                 }
 456                 if (p->more_r)
 457                     cmp_r= (*info->cmp)(p->buf_r, p->buf_m);
 458
 459 #else
 460                 p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r,
 461                                        &p->term_index_r);
 462                 cmp_r= (*info->cmp)(p->buf_r, p->buf_m);
 463 #endif
 464             }
 465             else
 466                 cmp_r=2;
 467 #if RSBETWEEN_DEBUG
 468         log2( p, "End of R loop", cmp_l, cmp_r);
 469 #endif
 470         } /* forward R */
 471
 472         if ( ( p->level <= 0 ) && ! p->more_l)
 473             return 0; /* no more start tags, nothing more to find */
 474
 475         if ( attr_match && p->level > 0)  /* within a tag pair (or deeper) */
 476         {
 477             memcpy (buf, p->buf_m, info->key_size);
 478             *term_index = p->term_index_m;
 479 #if RSBETWEEN_DEBUG
 480             log2( p, "Returning a hit (and forwarding m)", cmp_l, cmp_r);
 481 #endif
 482             p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m,
 483                                    &p->term_index_m);
 484             if (cmp_l == 2)
 485                 p->level = 0;
 486             return 1;
 487         }
 488         else if ( ! p->more_l )  /* not in data, no more starts */
 489         {
 490 #if RSBETWEEN_DEBUG
 491             log2( p, "no more starts, exiting without a hit", cmp_l, cmp_r);
 492 #endif
 493             return 0;  /* ergo, nothing can be found. stop scanning */
 494         }
 495 #if NEWCODE
 496         if (cmp_l == 2)
 497         {
 498             p->level = 0;
 499             p->more_m=rset_forward(
 500                               info->rset_m, p->rfd_m,
 501                               p->buf_m, &p->term_index_m,
 502                               info->cmp, p->buf_l);
 503         } else
 504         {
 505             p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m,
 506                                &p->term_index_m);
 507         }
 508 #else
 509         if (cmp_l == 2)
 510             p->level = 0;
 511         p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m,
 512                                &p->term_index_m);
 513 #endif
 514 #if RSBETWEEN_DEBUG
 515         log2( p, "End of M loop", cmp_l, cmp_r);
 516 #endif
 517     } /* while more_m */
 518
 519 #if RSBETWEEN_DEBUG
 520     log2( p, "Exiting, nothing more in m", cmp_l, cmp_r);
 521 #endif
 522     return 0;  /* no more data possible */
 523
 524
 525 }  /* r_read */
 526
 527
 528 static int r_write_between (RSFD rfd, const void *buf)
 529 {
 530     logf (LOG_FATAL, "between set type is read-only");
 531     return -1;
 532 }
 533