rset/rsbetween.c

   1 /* $Id: rsbetween.c,v 1.13 2004-06-08 15:05:16 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 static int r_count_between (RSET ct)
 309 {
 310     return 0;
 311 }
 312
 313
 314
 315 static int r_read_between (RSFD rfd, void *buf, int *term_index)
 316 {
 317     struct rset_between_rfd *p = (struct rset_between_rfd *) rfd;
 318     struct rset_between_info *info = p->info;
 319     int cmp_l=0;
 320     int cmp_r=0;
 321     int attr_match = 0;
 322
 323     while (p->more_m)
 324     {
 325 #if RSBETWEEN_DEBUG
 326         log2( p, "start of loop", cmp_l, cmp_r);
 327 #endif
 328
 329         /* forward L until past m, count levels, note rec boundaries */
 330         if (p->more_l)
 331             cmp_l= (*info->cmp)(p->buf_l, p->buf_m);
 332         else
 333         {
 334             p->level = 0;
 335             cmp_l=2; /* past this record */
 336         }
 337 #if RSBETWEEN_DEBUG
 338         log2( p, "after first L", cmp_l, cmp_r);
 339 #endif
 340
 341         while (cmp_l < 0)   /* l before m */
 342         {
 343             if (cmp_l == -2)
 344                 p->level=0; /* earlier record */
 345             if (cmp_l == -1)
 346             {
 347                 p->level++; /* relevant start tag */
 348
 349                 if (!info->rset_attr)
 350                     attr_match = 1;
 351                 else
 352                 {
 353                     int cmp_attr;
 354                     int dummy_term;
 355                     attr_match = 0;
 356                     while (p->more_attr)
 357                     {
 358                         cmp_attr = (*info->cmp)(p->buf_attr, p->buf_l);
 359                         if (cmp_attr == 0)
 360                         {
 361                             attr_match = 1;
 362                             break;
 363                         }
 364                         else if (cmp_attr > 0)
 365                             break;
 366                         else if (cmp_attr==-1)
 367                             p->more_attr = rset_read (info->rset_attr, p->rfd_attr,
 368                                                   p->buf_attr, &dummy_term);
 369                             /* if we had a forward that went all the way to
 370                              * the seqno, we could use that. But fwd only goes
 371                              * to the sysno */
 372                         else if (cmp_attr==-2)
 373                         {
 374                             p->more_attr = rset_forward(
 375                                       info->rset_attr, p->rfd_attr,
 376                                       p->buf_attr, &dummy_term,
 377                                       info->cmp, p->buf_l);
 378 #if RSBETWEEN_DEBUG
 379                             logf(LOG_DEBUG, "btw: after frowarding attr m=%d",p->more_attr);
 380 #endif
 381                         }
 382                     } /* while more_attr */
 383                 }
 384             }
 385 #define NEWCODE 1
 386 #if NEWCODE
 387             if (cmp_l==-2)
 388             {
 389                 if (p->more_l)
 390                 {
 391                     p->more_l=rset_forward(
 392                                       info->rset_l, p->rfd_l,
 393                                       p->buf_l, &p->term_index_l,
 394                                       info->cmp, p->buf_m);
 395                     if (p->more_l)
 396                         cmp_l= (*info->cmp)(p->buf_l, p->buf_m);
 397                     else
 398                         cmp_l=2;
 399 #if RSBETWEEN_DEBUG
 400                     log2( p, "after forwarding L", cmp_l, cmp_r);
 401 #endif
 402                 }
 403             } else
 404             {
 405                 p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l,
 406                               &p->term_index_l);
 407             }
 408 #else
 409             p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l,
 410                               &p->term_index_l);
 411 #endif
 412             if (p->more_l)
 413             {
 414                 cmp_l= (*info->cmp)(p->buf_l, p->buf_m);
 415             }
 416             else
 417                 cmp_l=2;
 418 #if RSBETWEEN_DEBUG
 419             log2( p, "end of L loop", cmp_l, cmp_r);
 420 #endif
 421         } /* forward L */
 422
 423
 424         /* forward R until past m, count levels */
 425 #if RSBETWEEN_DEBUG
 426         log2( p, "Before moving R", cmp_l, cmp_r);
 427 #endif
 428         if (p->more_r)
 429             cmp_r= (*info->cmp)(p->buf_r, p->buf_m);
 430         else
 431             cmp_r=2;
 432 #if RSBETWEEN_DEBUG
 433         log2( p, "after first R", cmp_l, cmp_r);
 434 #endif
 435         while (cmp_r < 0)   /* r before m */
 436         {
 437              /* -2, earlier record, don't count level */
 438             if (cmp_r == -1)
 439                 p->level--; /* relevant end tag */
 440             if (p->more_r)
 441             {
 442 #if NEWCODE
 443                 if (cmp_r==-2)
 444                 {
 445                     p->more_r=rset_forward(
 446                                       info->rset_r, p->rfd_r,
 447                                       p->buf_r, &p->term_index_r,
 448                                       info->cmp, p->buf_m);
 449                 } else
 450                 {
 451                     p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r,
 452                                        &p->term_index_r);
 453                 }
 454                 if (p->more_r)
 455                     cmp_r= (*info->cmp)(p->buf_r, p->buf_m);
 456
 457 #else
 458                 p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r,
 459                                        &p->term_index_r);
 460                 cmp_r= (*info->cmp)(p->buf_r, p->buf_m);
 461 #endif
 462             }
 463             else
 464                 cmp_r=2;
 465 #if RSBETWEEN_DEBUG
 466         log2( p, "End of R loop", cmp_l, cmp_r);
 467 #endif
 468         } /* forward R */
 469
 470         if ( ( p->level <= 0 ) && ! p->more_l)
 471             return 0; /* no more start tags, nothing more to find */
 472
 473         if ( attr_match && p->level > 0)  /* within a tag pair (or deeper) */
 474         {
 475             memcpy (buf, p->buf_m, info->key_size);
 476             *term_index = p->term_index_m;
 477 #if RSBETWEEN_DEBUG
 478             log2( p, "Returning a hit (and forwarding m)", cmp_l, cmp_r);
 479 #endif
 480             p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m,
 481                                    &p->term_index_m);
 482             if (cmp_l == 2)
 483                 p->level = 0;
 484             return 1;
 485         }
 486         else if ( ! p->more_l )  /* not in data, no more starts */
 487         {
 488 #if RSBETWEEN_DEBUG
 489             log2( p, "no more starts, exiting without a hit", cmp_l, cmp_r);
 490 #endif
 491             return 0;  /* ergo, nothing can be found. stop scanning */
 492         }
 493 #if NEWCODE
 494         if (cmp_l == 2)
 495         {
 496             p->level = 0;
 497             p->more_m=rset_forward(
 498                               info->rset_m, p->rfd_m,
 499                               p->buf_m, &p->term_index_m,
 500                               info->cmp, p->buf_l);
 501         } else
 502         {
 503             p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m,
 504                                &p->term_index_m);
 505         }
 506 #else
 507         if (cmp_l == 2)
 508             p->level = 0;
 509         p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m,
 510                                &p->term_index_m);
 511 #endif
 512 #if RSBETWEEN_DEBUG
 513         log2( p, "End of M loop", cmp_l, cmp_r);
 514 #endif
 515     } /* while more_m */
 516
 517 #if RSBETWEEN_DEBUG
 518     log2( p, "Exiting, nothing more in m", cmp_l, cmp_r);
 519 #endif
 520     return 0;  /* no more data possible */
 521
 522
 523 }  /* r_read */
 524
 525
 526 static int r_write_between (RSFD rfd, const void *buf)
 527 {
 528     logf (LOG_FATAL, "between set type is read-only");
 529     return -1;
 530 }
 531