rset/rsbetween.c

   1 /* $Id: rsbetween.c,v 1.17 2004-08-06 14:09:02 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 static void r_pos_between (RSFD rfd, double *current, double *total);
  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_pos_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     zint hits;
 104 };
 105
 106 #if RSBETWEEN_DEBUG
 107 static void log2 (struct rset_between_rfd *p, char *msg, int cmp_l, int cmp_r)
 108 {
 109     char buf_l[32];
 110     char buf_m[32];
 111     char buf_r[32];
 112     logf(LOG_DEBUG,"btw: %s l=%s(%d/%d) m=%s(%d) r=%s(%d/%d), lev=%d",
 113       msg,
 114       (*p->info->printer)(p->buf_l, buf_l), p->more_l, cmp_l,
 115       (*p->info->printer)(p->buf_m, buf_m), p->more_m,
 116       (*p->info->printer)(p->buf_r, buf_r), p->more_r, cmp_r,
 117       p->level);
 118 }
 119 #endif
 120
 121 static void *r_create_between (RSET ct, const struct rset_control *sel,
 122                                void *parms)
 123 {
 124     rset_between_parms *between_parms = (rset_between_parms *) parms;
 125     struct rset_between_info *info;
 126
 127     info = (struct rset_between_info *) xmalloc (sizeof(*info));
 128     info->key_size = between_parms->key_size;
 129     info->rset_l = between_parms->rset_l;
 130     info->rset_m = between_parms->rset_m;
 131     info->rset_r = between_parms->rset_r;
 132     info->rset_attr = between_parms->rset_attr;
 133     if (rset_is_volatile(info->rset_l) ||
 134         rset_is_volatile(info->rset_m) ||
 135         rset_is_volatile(info->rset_r))
 136         ct->flags |= RSET_FLAG_VOLATILE;
 137     info->cmp = between_parms->cmp;
 138     info->printer = between_parms->printer;
 139     info->rfd_list = NULL;
 140
 141     info->term_index_s = info->rset_l->no_rset_terms;
 142     if (info->rset_m)
 143     {
 144         ct->no_rset_terms =
 145             info->rset_l->no_rset_terms +
 146             info->rset_m->no_rset_terms +
 147             info->rset_r->no_rset_terms;
 148         ct->rset_terms = (RSET_TERM *)
 149             xmalloc (sizeof (*ct->rset_terms) * ct->no_rset_terms);
 150         memcpy (ct->rset_terms, info->rset_l->rset_terms,
 151                 info->rset_l->no_rset_terms * sizeof(*ct->rset_terms));
 152         memcpy (ct->rset_terms + info->rset_l->no_rset_terms,
 153                 info->rset_m->rset_terms,
 154                 info->rset_m->no_rset_terms * sizeof(*ct->rset_terms));
 155         memcpy (ct->rset_terms + info->rset_l->no_rset_terms +
 156                 info->rset_m->no_rset_terms,
 157                 info->rset_r->rset_terms,
 158                 info->rset_r->no_rset_terms * sizeof(*ct->rset_terms));
 159     }
 160     else
 161     {
 162         ct->no_rset_terms =
 163             info->rset_l->no_rset_terms +
 164             info->rset_r->no_rset_terms;
 165         ct->rset_terms = (RSET_TERM *)
 166             xmalloc (sizeof (*ct->rset_terms) * ct->no_rset_terms);
 167         memcpy (ct->rset_terms, info->rset_l->rset_terms,
 168                 info->rset_l->no_rset_terms * sizeof(*ct->rset_terms));
 169         memcpy (ct->rset_terms + info->rset_l->no_rset_terms,
 170                 info->rset_r->rset_terms,
 171                 info->rset_r->no_rset_terms * sizeof(*ct->rset_terms));
 172     }
 173
 174     return info;
 175 }
 176
 177 static RSFD r_open_between (RSET ct, int flag)
 178 {
 179     struct rset_between_info *info = (struct rset_between_info *) ct->buf;
 180     struct rset_between_rfd *rfd;
 181
 182     if (flag & RSETF_WRITE)
 183     {
 184         logf (LOG_FATAL, "between set type is read-only");
 185         return NULL;
 186     }
 187     rfd = (struct rset_between_rfd *) xmalloc (sizeof(*rfd));
 188     rfd->next = info->rfd_list;
 189     info->rfd_list = rfd;
 190     rfd->info = info;
 191
 192     rfd->buf_l = xmalloc (info->key_size);
 193     rfd->buf_m = xmalloc (info->key_size);
 194     rfd->buf_r = xmalloc (info->key_size);
 195     rfd->buf_attr = xmalloc (info->key_size);
 196
 197     rfd->rfd_l = rset_open (info->rset_l, RSETF_READ);
 198     rfd->rfd_m = rset_open (info->rset_m, RSETF_READ);
 199     rfd->rfd_r = rset_open (info->rset_r, RSETF_READ);
 200
 201     rfd->more_l = rset_read (info->rset_l, rfd->rfd_l, rfd->buf_l,
 202                              &rfd->term_index_l);
 203     rfd->more_m = rset_read (info->rset_m, rfd->rfd_m, rfd->buf_m,
 204                              &rfd->term_index_m);
 205     rfd->more_r = rset_read (info->rset_r, rfd->rfd_r, rfd->buf_r,
 206                              &rfd->term_index_r);
 207     if (info->rset_attr)
 208     {
 209         int dummy;
 210         rfd->rfd_attr = rset_open (info->rset_attr, RSETF_READ);
 211         rfd->more_attr = rset_read (info->rset_attr, rfd->rfd_attr,
 212                                     rfd->buf_attr, &dummy);
 213     }
 214     rfd->level=0;
 215     rfd->hits=0;
 216     return rfd;
 217 }
 218
 219 static void r_close_between (RSFD rfd)
 220 {
 221     struct rset_between_info *info = ((struct rset_between_rfd*)rfd)->info;
 222     struct rset_between_rfd **rfdp;
 223
 224     for (rfdp = &info->rfd_list; *rfdp; rfdp = &(*rfdp)->next)
 225         if (*rfdp == rfd)
 226         {
 227             xfree ((*rfdp)->buf_l);
 228             xfree ((*rfdp)->buf_m);
 229             xfree ((*rfdp)->buf_r);
 230             xfree ((*rfdp)->buf_attr);
 231             rset_close (info->rset_l, (*rfdp)->rfd_l);
 232             rset_close (info->rset_m, (*rfdp)->rfd_m);
 233             rset_close (info->rset_r, (*rfdp)->rfd_r);
 234             if (info->rset_attr)
 235                 rset_close (info->rset_attr, (*rfdp)->rfd_attr);
 236
 237             *rfdp = (*rfdp)->next;
 238             xfree (rfd);
 239             return;
 240         }
 241     logf (LOG_FATAL, "r_close_between but no rfd match!");
 242     assert (0);
 243 }
 244
 245 static void r_delete_between (RSET ct)
 246 {
 247     struct rset_between_info *info = (struct rset_between_info *) ct->buf;
 248
 249     assert (info->rfd_list == NULL);
 250     xfree (ct->rset_terms);
 251     rset_delete (info->rset_l);
 252     rset_delete (info->rset_m);
 253     rset_delete (info->rset_r);
 254     if (info->rset_attr)
 255         rset_delete (info->rset_attr);
 256     xfree (info);
 257 }
 258
 259 static void r_rewind_between (RSFD rfd)
 260 {
 261     struct rset_between_info *info = ((struct rset_between_rfd*)rfd)->info;
 262     struct rset_between_rfd *p = (struct rset_between_rfd *) rfd;
 263
 264 #if RSBETWEEN_DEBUG
 265     logf (LOG_DEBUG, "rsbetween_rewind");
 266 #endif
 267     rset_rewind (info->rset_l, p->rfd_l);
 268     rset_rewind (info->rset_m, p->rfd_m);
 269     rset_rewind (info->rset_r, p->rfd_r);
 270     p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l, &p->term_index_l);
 271     p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m, &p->term_index_m);
 272     p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r, &p->term_index_r);
 273     if (info->rset_attr)
 274     {
 275         int dummy;
 276         rset_rewind (info->rset_attr, p->rfd_attr);
 277         p->more_attr = rset_read (info->rset_attr, p->rfd_attr, p->buf_attr,
 278                                   &dummy);
 279     }
 280     p->level=0;
 281     p->hits=0;
 282 }
 283
 284
 285
 286 static int r_forward_between(RSET ct, RSFD rfd, void *buf, int *term_index,
 287                      int (*cmpfunc)(const void *p1, const void *p2),
 288                      const void *untilbuf)
 289 {
 290     struct rset_between_info *info = ((struct rset_between_rfd*)rfd)->info;
 291     struct rset_between_rfd *p = (struct rset_between_rfd *) rfd;
 292     int rc;
 293 #if RSBETWEEN_DEBUG
 294     log2( p, "fwd: before forward", 0,0);
 295 #endif
 296     /* It is enough to forward the m pointer here, the read will */
 297     /* naturally forward the l, m, and attr pointers */
 298     if (p->more_m)
 299         p->more_m=rset_forward(info->rset_m,p->rfd_m, p->buf_m,
 300                         &p->term_index_m, info->cmp,untilbuf);
 301 #if RSBETWEEN_DEBUG
 302     log2( p, "fwd: after forward M", 0,0);
 303 #endif
 304     rc = r_read_between(rfd, buf, term_index);
 305 #if RSBETWEEN_DEBUG
 306     log2( p, "fwd: after forward", 0,0);
 307 #endif
 308     return rc;
 309 }
 310
 311
 312
 313
 314 static int r_read_between (RSFD rfd, void *buf, int *term_index)
 315 {
 316     struct rset_between_rfd *p = (struct rset_between_rfd *) rfd;
 317     struct rset_between_info *info = p->info;
 318     int cmp_l=0;
 319     int cmp_r=0;
 320     int attr_match = 0;
 321
 322     while (p->more_m)
 323     {
 324 #if RSBETWEEN_DEBUG
 325         log2( p, "start of loop", cmp_l, cmp_r);
 326 #endif
 327
 328         /* forward L until past m, count levels, note rec boundaries */
 329         if (p->more_l)
 330             cmp_l= (*info->cmp)(p->buf_l, p->buf_m);
 331         else
 332         {
 333             p->level = 0;
 334             cmp_l=2; /* past this record */
 335         }
 336 #if RSBETWEEN_DEBUG
 337         log2( p, "after first L", cmp_l, cmp_r);
 338 #endif
 339
 340         while (cmp_l < 0)   /* l before m */
 341         {
 342             if (cmp_l == -2)
 343                 p->level=0; /* earlier record */
 344             if (cmp_l == -1)
 345             {
 346                 p->level++; /* relevant start tag */
 347
 348                 if (!info->rset_attr)
 349                     attr_match = 1;
 350                 else
 351                 {
 352                     int cmp_attr;
 353                     int dummy_term;
 354                     attr_match = 0;
 355                     while (p->more_attr)
 356                     {
 357                         cmp_attr = (*info->cmp)(p->buf_attr, p->buf_l);
 358                         if (cmp_attr == 0)
 359                         {
 360                             attr_match = 1;
 361                             break;
 362                         }
 363                         else if (cmp_attr > 0)
 364                             break;
 365                         else if (cmp_attr==-1)
 366                             p->more_attr = rset_read (info->rset_attr, p->rfd_attr,
 367                                                   p->buf_attr, &dummy_term);
 368                             /* if we had a forward that went all the way to
 369                              * the seqno, we could use that. But fwd only goes
 370                              * to the sysno */
 371                         else if (cmp_attr==-2)
 372                         {
 373                             p->more_attr = rset_forward(
 374                                       info->rset_attr, p->rfd_attr,
 375                                       p->buf_attr, &dummy_term,
 376                                       info->cmp, p->buf_l);
 377 #if RSBETWEEN_DEBUG
 378                             logf(LOG_DEBUG, "btw: after frowarding attr m=%d",p->more_attr);
 379 #endif
 380                         }
 381                     } /* while more_attr */
 382                 }
 383             }
 384 #define NEWCODE 1
 385 #if NEWCODE
 386             if (cmp_l==-2)
 387             {
 388                 if (p->more_l)
 389                 {
 390                     p->more_l=rset_forward(
 391                                       info->rset_l, p->rfd_l,
 392                                       p->buf_l, &p->term_index_l,
 393                                       info->cmp, p->buf_m);
 394                     if (p->more_l)
 395                         cmp_l= (*info->cmp)(p->buf_l, p->buf_m);
 396                     else
 397                         cmp_l=2;
 398 #if RSBETWEEN_DEBUG
 399                     log2( p, "after forwarding L", cmp_l, cmp_r);
 400 #endif
 401                 }
 402             } else
 403             {
 404                 p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l,
 405                               &p->term_index_l);
 406             }
 407 #else
 408             p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l,
 409                               &p->term_index_l);
 410 #endif
 411             if (p->more_l)
 412             {
 413                 cmp_l= (*info->cmp)(p->buf_l, p->buf_m);
 414             }
 415             else
 416                 cmp_l=2;
 417 #if RSBETWEEN_DEBUG
 418             log2( p, "end of L loop", cmp_l, cmp_r);
 419 #endif
 420         } /* forward L */
 421
 422
 423         /* forward R until past m, count levels */
 424 #if RSBETWEEN_DEBUG
 425         log2( p, "Before moving R", cmp_l, cmp_r);
 426 #endif
 427         if (p->more_r)
 428             cmp_r= (*info->cmp)(p->buf_r, p->buf_m);
 429         else
 430             cmp_r=2;
 431 #if RSBETWEEN_DEBUG
 432         log2( p, "after first R", cmp_l, cmp_r);
 433 #endif
 434         while (cmp_r < 0)   /* r before m */
 435         {
 436              /* -2, earlier record, don't count level */
 437             if (cmp_r == -1)
 438                 p->level--; /* relevant end tag */
 439             if (p->more_r)
 440             {
 441 #if NEWCODE
 442                 if (cmp_r==-2)
 443                 {
 444                     p->more_r=rset_forward(
 445                                       info->rset_r, p->rfd_r,
 446                                       p->buf_r, &p->term_index_r,
 447                                       info->cmp, p->buf_m);
 448                 } else
 449                 {
 450                     p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r,
 451                                        &p->term_index_r);
 452                 }
 453                 if (p->more_r)
 454                     cmp_r= (*info->cmp)(p->buf_r, p->buf_m);
 455
 456 #else
 457                 p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r,
 458                                        &p->term_index_r);
 459                 cmp_r= (*info->cmp)(p->buf_r, p->buf_m);
 460 #endif
 461             }
 462             else
 463                 cmp_r=2;
 464 #if RSBETWEEN_DEBUG
 465         log2( p, "End of R loop", cmp_l, cmp_r);
 466 #endif
 467         } /* forward R */
 468
 469         if ( ( p->level <= 0 ) && ! p->more_l)
 470             return 0; /* no more start tags, nothing more to find */
 471
 472         if ( attr_match && p->level > 0)  /* within a tag pair (or deeper) */
 473         {
 474             memcpy (buf, p->buf_m, info->key_size);
 475             *term_index = p->term_index_m;
 476 #if RSBETWEEN_DEBUG
 477             log2( p, "Returning a hit (and forwarding m)", cmp_l, cmp_r);
 478 #endif
 479             p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m,
 480                                    &p->term_index_m);
 481             if (cmp_l == 2)
 482                 p->level = 0;
 483             p->hits++;
 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
 532
 533 static void r_pos_between (RSFD rfd, double *current, double *total)
 534 {
 535     struct rset_between_rfd *p = (struct rset_between_rfd *) rfd;
 536     struct rset_between_info *info = p->info;
 537     double lcur,ltot;
 538     double mcur,mtot;
 539     double rcur,rtot;
 540     double r;
 541     ltot=-1; rtot=-1;
 542     rset_pos(info->rset_l, p->rfd_l,  &lcur, &ltot);
 543     rset_pos(info->rset_m, p->rfd_m,  &mcur, &mtot);
 544     rset_pos(info->rset_r, p->rfd_r,  &rcur, &rtot);
 545     if ( (ltot<0) && (mtot<0) && (rtot<0) ) { /*no position */
 546         *current=mcur;  /* return same as you got */
 547         *total=mtot;    /* probably -1 for not available */
 548     }
 549     if ( ltot<0) { ltot=0; lcur=0;} /* if only one useful, use it */
 550     if ( mtot<0) { mtot=0; mcur=0;}
 551     if ( rtot<0) { rtot=0; rcur=0;}
 552     if ( ltot+mtot+rtot < 1 ) { /* empty rset */
 553         *current=0;
 554         *total=0;
 555         return;
 556     }
 557     r=1.0*(lcur+mcur+rcur)/(ltot+mtot+rtot); /* weighed average of l and r */
 558     *current=p->hits;
 559     *total=*current/r ;
 560 #if RSBETWEEN_DEBUG
 561     yaz_log(LOG_DEBUG,"betw_pos: (%s/%s) %0.1f/%0.1f= %0.4f ",
 562                     info->rset_l->control->desc, info->rset_r->control->desc,
 563                     *current, *total, r);
 564 #endif
 565 }