X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=rset%2Frsbetween.c;h=56cc894725fe0bfc9654888d11af8c750ca1c35b;hp=5c5f8eb1b20b60595ede30a6effc2cf4c6d05344;hb=deff57cfa9d9b39c4a4f1c9b82a64c6e61d821a4;hpb=3ee1a8c3e7c63fe8fe350249e4763a029ae81737
diff --git a/rset/rsbetween.c b/rset/rsbetween.c
index 5c5f8eb..56cc894 100644
--- a/rset/rsbetween.c
+++ b/rset/rsbetween.c
@@ -1,6 +1,6 @@
-/* $Id: rsbetween.c,v 1.8 2002-11-11 15:05:29 heikki Exp $
- Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
- Index Data Aps
+/* $Id: rsbetween.c,v 1.36 2005-03-30 09:25:24 adam Exp $
+ Copyright (C) 1995-2005
+ Index Data ApS
This file is part of the Zebra server.
@@ -21,370 +21,316 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
*/
+/* rsbetween is (mostly) used for xml searches. It returns the hits of the
+ * "middle" rset, that are in between the "left" and "right" rsets. For
+ * example "Shakespeare" in between "" and . The thing is
+ * complicated by the inclusion of attributes (from their own rset). If attrs
+ * specified, they must match the "left" rset (start tag). "Hamlet" between
+ * "" and "". (This assumes that the attributes are
+ * indexed to the same seqno as the tags).
+ *
+*/
#include
#include
#include
#include
-#include
-#include
+#include
+#include
-static void *r_create_between(RSET ct, const struct rset_control *sel, void *parms);
-static RSFD r_open_between (RSET ct, int flag);
-static void r_close_between (RSFD rfd);
-static void r_delete_between (RSET ct);
-static void r_rewind_between (RSFD rfd);
-static int r_count_between (RSET ct);
-static int r_read_between (RSFD rfd, void *buf, int *term_index);
-static int r_write_between (RSFD rfd, const void *buf);
-static const struct rset_control control_between =
+static RSFD r_open (RSET ct, int flag);
+static void r_close (RSFD rfd);
+static void r_delete (RSET ct);
+static int r_forward(RSFD rfd, void *buf,
+ TERMID *term, const void *untilbuf);
+static int r_read(RSFD rfd, void *buf, TERMID *term );
+static int r_write(RSFD rfd, const void *buf);
+static void r_pos(RSFD rfd, double *current, double *total);
+static void r_get_terms(RSET ct, TERMID *terms, int maxterms, int *curterm);
+
+static const struct rset_control control =
{
"between",
- r_create_between,
- r_open_between,
- r_close_between,
- r_delete_between,
- r_rewind_between,
- r_count_between,
- r_read_between,
- r_write_between,
+ r_delete,
+ r_get_terms,
+ r_open,
+ r_close,
+ r_forward,
+ r_pos,
+ r_read,
+ r_write,
};
-const struct rset_control *rset_kind_between = &control_between;
+const struct rset_control *rset_kind_between = &control;
+
+#define STARTTAG 0
+#define HIT 1
+#define STOPTAG 2
+#define ATTRTAG 3
struct rset_between_info {
- int key_size;
- RSET rset_l;
- RSET rset_m;
- RSET rset_r;
- RSET rset_attr;
- int term_index_s;
- int (*cmp)(const void *p1, const void *p2);
- char *(*printer)(const void *p1, char *buf);
- struct rset_between_rfd *rfd_list;
+ RSET andset; /* the multi-and of the above */
+ TERMID startterm; /* pseudo terms for detecting which one we read from */
+ TERMID stopterm;
+ TERMID attrterm;
};
struct rset_between_rfd {
- RSFD rfd_l;
- RSFD rfd_m;
- RSFD rfd_r;
- RSFD rfd_attr;
- int more_l;
- int more_m;
- int more_r;
- int more_attr;
- int term_index_l;
- int term_index_m;
- int term_index_r;
- void *buf_l;
- void *buf_m;
- void *buf_r;
- void *buf_attr;
- int level;
- struct rset_between_rfd *next;
- struct rset_between_info *info;
+ RSFD andrfd;
+ void *recbuf; /* a key that tells which record we are in */
+ void *startbuf; /* the start tag */
+ int startbufok; /* we have seen the first start tag */
+ void *attrbuf; /* the attr tag. If these two match, we have attr match */
+ int attrbufok; /* we have seen the first attr tag, can compare */
+ int depth; /* number of start-tags without end-tags */
+ int attrdepth; /* on what depth the attr matched */
+ zint hits;
};
-static void *r_create_between (RSET ct, const struct rset_control *sel,
- void *parms)
+static int log_level = 0;
+static int log_level_initialized = 0;
+
+
+/* make sure that the rset has a term attached. If not, create one */
+/* we need these terms for the tags, to distinguish what we read */
+static void checkterm( RSET rs, char *tag, NMEM nmem)
{
- rset_between_parms *between_parms = (rset_between_parms *) parms;
- struct rset_between_info *info;
-
- info = (struct rset_between_info *) xmalloc (sizeof(*info));
- info->key_size = between_parms->key_size;
- info->rset_l = between_parms->rset_l;
- info->rset_m = between_parms->rset_m;
- info->rset_r = between_parms->rset_r;
- info->rset_attr = between_parms->rset_attr;
- if (rset_is_volatile(info->rset_l) ||
- rset_is_volatile(info->rset_m) ||
- rset_is_volatile(info->rset_r))
- ct->flags |= RSET_FLAG_VOLATILE;
- info->cmp = between_parms->cmp;
- info->printer = between_parms->printer;
- info->rfd_list = NULL;
+ if (!rs->term)
+ {
+ rs->term=
+ rset_term_create(tag,strlen(tag),"",0,nmem);
+ rs->term->rset = rs;
+ }
+}
+
+
+RSET rsbetween_create( NMEM nmem, const struct key_control *kcontrol,
+ int scope,
+ RSET rset_l, RSET rset_m, RSET rset_r, RSET rset_attr)
+{
+ RSET rnew = rset_create_base(&control, nmem, kcontrol, scope,0);
+ struct rset_between_info *info=
+ (struct rset_between_info *) nmem_malloc(rnew->nmem,sizeof(*info));
+ RSET rsetarray[4];
+ int n = 4;
- info->term_index_s = info->rset_l->no_rset_terms;
- if (info->rset_m)
+ if (!log_level_initialized)
+ {
+ log_level = yaz_log_module_level("rsbetween");
+ log_level_initialized = 1;
+ }
+ rsetarray[STARTTAG] = rset_l;
+ rsetarray[HIT] = rset_m;
+ rsetarray[STOPTAG] = rset_r;
+ rsetarray[ATTRTAG] = rset_attr;
+
+ /* make sure we have decent terms for all rsets. Create dummies if needed*/
+ checkterm( rsetarray[STARTTAG], "(start)",nmem);
+ checkterm( rsetarray[STOPTAG], "(start)",nmem);
+ info->startterm = rsetarray[STARTTAG]->term;
+ info->stopterm = rsetarray[STOPTAG]->term;
+
+ if (rset_attr)
{
- ct->no_rset_terms =
- info->rset_l->no_rset_terms +
- info->rset_m->no_rset_terms +
- info->rset_r->no_rset_terms;
- ct->rset_terms = (RSET_TERM *)
- xmalloc (sizeof (*ct->rset_terms) * ct->no_rset_terms);
- memcpy (ct->rset_terms, info->rset_l->rset_terms,
- info->rset_l->no_rset_terms * sizeof(*ct->rset_terms));
- memcpy (ct->rset_terms + info->rset_l->no_rset_terms,
- info->rset_m->rset_terms,
- info->rset_m->no_rset_terms * sizeof(*ct->rset_terms));
- memcpy (ct->rset_terms + info->rset_l->no_rset_terms +
- info->rset_m->no_rset_terms,
- info->rset_r->rset_terms,
- info->rset_r->no_rset_terms * sizeof(*ct->rset_terms));
+ checkterm( rsetarray[ATTRTAG], "(start)",nmem);
+ info->attrterm = rsetarray[ATTRTAG]->term;
+ n = 4;
}
else
{
- ct->no_rset_terms =
- info->rset_l->no_rset_terms +
- info->rset_r->no_rset_terms;
- ct->rset_terms = (RSET_TERM *)
- xmalloc (sizeof (*ct->rset_terms) * ct->no_rset_terms);
- memcpy (ct->rset_terms, info->rset_l->rset_terms,
- info->rset_l->no_rset_terms * sizeof(*ct->rset_terms));
- memcpy (ct->rset_terms + info->rset_l->no_rset_terms,
- info->rset_r->rset_terms,
- info->rset_r->no_rset_terms * sizeof(*ct->rset_terms));
+ info->attrterm = NULL;
+ n = 3;
}
+ info->andset = rsmulti_and_create( nmem, kcontrol, scope, n, rsetarray);
+ rnew->priv = info;
+ yaz_log(log_level,"create rset at %p",rnew);
+ return rnew;
+}
+
- return info;
+static void r_delete (RSET ct)
+{
+ struct rset_between_info *info = (struct rset_between_info *) ct->priv;
+ yaz_log(log_level,"delete rset at %p",ct);
+ rset_delete(info->andset);
}
-static RSFD r_open_between (RSET ct, int flag)
+
+static RSFD r_open (RSET ct, int flag)
{
- struct rset_between_info *info = (struct rset_between_info *) ct->buf;
- struct rset_between_rfd *rfd;
+ struct rset_between_info *info = (struct rset_between_info *) ct->priv;
+ RSFD rfd;
+ struct rset_between_rfd *p;
if (flag & RSETF_WRITE)
{
- logf (LOG_FATAL, "between set type is read-only");
- return NULL;
+ yaz_log (YLOG_FATAL, "between set type is read-only");
+ return NULL;
}
- rfd = (struct rset_between_rfd *) xmalloc (sizeof(*rfd));
- rfd->next = info->rfd_list;
- info->rfd_list = rfd;
- rfd->info = info;
-
- rfd->buf_l = xmalloc (info->key_size);
- rfd->buf_m = xmalloc (info->key_size);
- rfd->buf_r = xmalloc (info->key_size);
- rfd->buf_attr = xmalloc (info->key_size);
-
- rfd->rfd_l = rset_open (info->rset_l, RSETF_READ);
- rfd->rfd_m = rset_open (info->rset_m, RSETF_READ);
- rfd->rfd_r = rset_open (info->rset_r, RSETF_READ);
-
- rfd->more_l = rset_read (info->rset_l, rfd->rfd_l, rfd->buf_l,
- &rfd->term_index_l);
- rfd->more_m = rset_read (info->rset_m, rfd->rfd_m, rfd->buf_m,
- &rfd->term_index_m);
- rfd->more_r = rset_read (info->rset_r, rfd->rfd_r, rfd->buf_r,
- &rfd->term_index_r);
- if (info->rset_attr)
- {
- int dummy;
- rfd->rfd_attr = rset_open (info->rset_attr, RSETF_READ);
- rfd->more_attr = rset_read (info->rset_attr, rfd->rfd_attr,
- rfd->buf_attr, &dummy);
+ rfd = rfd_create_base(ct);
+ if (rfd->priv)
+ p=(struct rset_between_rfd *)rfd->priv;
+ else {
+ p = (struct rset_between_rfd *) nmem_malloc(ct->nmem, (sizeof(*p)));
+ rfd->priv = p;
+ p->recbuf = nmem_malloc(ct->nmem, (ct->keycontrol->key_size));
+ p->startbuf = nmem_malloc(ct->nmem, (ct->keycontrol->key_size));
+ p->attrbuf = nmem_malloc(ct->nmem, (ct->keycontrol->key_size));
}
- rfd->level=0;
+ p->andrfd = rset_open (info->andset, RSETF_READ);
+ p->hits=-1;
+ p->depth = 0;
+ p->attrdepth = 0;
+ p->attrbufok = 0;
+ p->startbufok = 0;
+ yaz_log(log_level,"open rset=%p rfd=%p", ct, rfd);
return rfd;
}
-static void r_close_between (RSFD rfd)
+static void r_close (RSFD rfd)
{
- struct rset_between_info *info = ((struct rset_between_rfd*)rfd)->info;
- struct rset_between_rfd **rfdp;
-
- for (rfdp = &info->rfd_list; *rfdp; rfdp = &(*rfdp)->next)
- if (*rfdp == rfd)
- {
- xfree ((*rfdp)->buf_l);
- xfree ((*rfdp)->buf_m);
- xfree ((*rfdp)->buf_r);
- xfree ((*rfdp)->buf_attr);
- rset_close (info->rset_l, (*rfdp)->rfd_l);
- rset_close (info->rset_m, (*rfdp)->rfd_m);
- rset_close (info->rset_r, (*rfdp)->rfd_r);
- if (info->rset_attr)
- rset_close (info->rset_attr, (*rfdp)->rfd_attr);
-
- *rfdp = (*rfdp)->next;
- xfree (rfd);
- return;
- }
- logf (LOG_FATAL, "r_close_between but no rfd match!");
- assert (0);
+ struct rset_between_rfd *p=(struct rset_between_rfd *)rfd->priv;
+ yaz_log(log_level,"close rfd=%p", rfd);
+ rset_close (p->andrfd);
+ rfd_delete_base(rfd);
}
-static void r_delete_between (RSET ct)
-{
- struct rset_between_info *info = (struct rset_between_info *) ct->buf;
-
- assert (info->rfd_list == NULL);
- xfree (ct->rset_terms);
- rset_delete (info->rset_l);
- rset_delete (info->rset_m);
- rset_delete (info->rset_r);
- if (info->rset_attr)
- rset_delete (info->rset_attr);
- xfree (info);
-}
-static void r_rewind_between (RSFD rfd)
-{
- struct rset_between_info *info = ((struct rset_between_rfd*)rfd)->info;
- struct rset_between_rfd *p = (struct rset_between_rfd *) rfd;
-
- logf (LOG_DEBUG, "rsbetween_rewind");
- rset_rewind (info->rset_l, p->rfd_l);
- rset_rewind (info->rset_m, p->rfd_m);
- rset_rewind (info->rset_r, p->rfd_r);
- p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l, &p->term_index_l);
- p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m, &p->term_index_m);
- p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r, &p->term_index_r);
- if (info->rset_attr)
- {
- int dummy;
- rset_rewind (info->rset_attr, p->rfd_attr);
- p->more_attr = rset_read (info->rset_attr, p->rfd_attr, p->buf_attr,
- &dummy);
- }
- p->level=0;
-}
-static int r_count_between (RSET ct)
+static int r_forward(RSFD rfd, void *buf,
+ TERMID *term, const void *untilbuf)
{
- return 0;
+ struct rset_between_rfd *p=(struct rset_between_rfd *)rfd->priv;
+ int rc;
+ yaz_log(log_level, "forwarding ");
+ rc = rset_forward(p->andrfd,buf,term,untilbuf);
+ return rc;
}
-static void log2 (struct rset_between_rfd *p, char *msg, int cmp_l, int cmp_r)
+
+static void checkattr(RSFD rfd)
{
- char buf_l[32];
- char buf_m[32];
- char buf_r[32];
- logf(LOG_DEBUG,"btw: %s l=%s(%d/%d) m=%s(%d) r=%s(%d/%d), lev=%d",
- msg,
- (*p->info->printer)(p->buf_l, buf_l), p->more_l, cmp_l,
- (*p->info->printer)(p->buf_m, buf_m), p->more_l,
- (*p->info->printer)(p->buf_r, buf_r), p->more_l, cmp_r,
- p->level);
+ struct rset_between_info *info =(struct rset_between_info *)rfd->rset->priv;
+ struct rset_between_rfd *p=(struct rset_between_rfd *)rfd->priv;
+ const struct key_control *kctrl = rfd->rset->keycontrol;
+ int cmp;
+ if (p->attrdepth)
+ return; /* already found one */
+ if (!info->attrterm)
+ {
+ p->attrdepth=-1; /* matches always */
+ return;
+ }
+ if ( p->startbufok && p->attrbufok )
+ { /* have buffers to compare */
+ cmp=(kctrl->cmp)(p->startbuf,p->attrbuf);
+ if (0==cmp) /* and the keys match */
+ {
+ p->attrdepth = p->depth;
+ yaz_log(log_level, "found attribute match at depth %d",p->attrdepth);
+ }
+ }
}
-static int r_read_between (RSFD rfd, void *buf, int *term_index)
+static int r_read (RSFD rfd, void *buf, TERMID *term)
{
- struct rset_between_rfd *p = (struct rset_between_rfd *) rfd;
- struct rset_between_info *info = p->info;
- int cmp_l=0;
- int cmp_r=0;
- int attr_match;
-
- while (p->more_m)
+ struct rset_between_info *info =(struct rset_between_info *)rfd->rset->priv;
+ struct rset_between_rfd *p=(struct rset_between_rfd *)rfd->priv;
+ const struct key_control *kctrl = rfd->rset->keycontrol;
+ int cmp;
+ TERMID dummyterm = 0;
+ yaz_log(log_level,"== read: term=%p",term);
+ if (!term)
+ term=&dummyterm;
+ while ( rset_read(p->andrfd,buf,term) )
{
- log2( p, "start of loop", cmp_l, cmp_r);
-
- /* forward L until past m, count levels, note rec boundaries */
- if (p->more_l)
- cmp_l= (*info->cmp)(p->buf_l, p->buf_m);
- else
- cmp_l=2; /* past this record */
- log2( p, "after first L", cmp_l, cmp_r);
-
- while (cmp_l < 0) /* l before m */
- {
- if (cmp_l == -2)
- p->level=0; /* earlier record */
- if (cmp_l == -1)
- {
- p->level++; /* relevant start tag */
-
- if (!info->rset_attr)
- attr_match = 1;
- else
- {
- int cmp_attr;
- int dummy_term;
- attr_match = 0;
- while (p->more_attr)
- {
- cmp_attr = (*info->cmp)(p->buf_attr, p->buf_l);
- if (cmp_attr == 0)
- {
- attr_match = 1;
- break;
- }
- else if (cmp_attr > 0)
- break;
- p->more_attr = rset_read (info->rset_attr, p->rfd_attr,
- p->buf_attr, &dummy_term);
- }
- }
- }
- p->more_l = rset_read (info->rset_l, p->rfd_l, p->buf_l,
- &p->term_index_l);
- if (p->more_l)
- {
- cmp_l= (*info->cmp)(p->buf_l, p->buf_m);
- }
- else
- cmp_l=2;
- log2( p, "end of L loop", cmp_l, cmp_r);
- } /* forward L */
-
-
- /* forward R until past m, count levels */
- log2( p, "Before moving R", cmp_l, cmp_r);
- if (p->more_r)
- cmp_r= (*info->cmp)(p->buf_r, p->buf_m);
- else
- cmp_r=2;
- log2( p, "after first R", cmp_l, cmp_r);
- while (cmp_r < 0) /* r before m */
- {
- /* -2, earlier record, doesn't matter */
- if (cmp_r == -1)
- p->level--; /* relevant end tag */
- if (p->more_r)
+ yaz_log(log_level,"read loop term=%p d=%d ad=%d",
+ *term,p->depth, p->attrdepth);
+ if (p->hits<0)
+ {/* first time? */
+ memcpy(p->recbuf,buf,kctrl->key_size);
+ p->hits = 0;
+ cmp = rfd->rset->scope; /* force newrecord */
+ }
+ else {
+ cmp=(kctrl->cmp)(buf,p->recbuf);
+ yaz_log(log_level, "cmp=%d",cmp);
+ }
+
+ if (cmp>=rfd->rset->scope)
+ {
+ yaz_log(log_level,"new record");
+ p->depth = 0;
+ p->attrdepth = 0;
+ memcpy(p->recbuf,buf,kctrl->key_size);
+ }
+
+ if (*term)
+ yaz_log(log_level," term: '%s'", (*term)->name);
+ if (*term==info->startterm)
+ {
+ p->depth++;
+ yaz_log(log_level,"read start tag. d=%d",p->depth);
+ memcpy(p->startbuf,buf,kctrl->key_size);
+ p->startbufok = 1;
+ checkattr(rfd); /* in case we already saw the attr here */
+ }
+ else if (*term==info->stopterm)
+ {
+ if (p->depth == p->attrdepth)
+ p->attrdepth = 0; /* ending the tag with attr match */
+ p->depth--;
+ yaz_log(log_level,"read end tag. d=%d ad=%d",p->depth, p->attrdepth);
+ }
+ else if (*term==info->attrterm)
+ {
+ yaz_log(log_level,"read attr");
+ memcpy(p->attrbuf,buf,kctrl->key_size);
+ p->attrbufok = 1;
+ checkattr(rfd); /* in case the start tag came first */
+ }
+ else
+ { /* mut be a real hit */
+ if (p->depth && p->attrdepth)
{
- p->more_r = rset_read (info->rset_r, p->rfd_r, p->buf_r,
- &p->term_index_r);
- cmp_r= (*info->cmp)(p->buf_r, p->buf_m);
- }
- else
- cmp_r=2;
- log2( p, "End of R loop", cmp_l, cmp_r);
- } /* forward R */
-
- if ( ( p->level <= 0 ) && ! p->more_l)
- return 0; /* no more start tags, nothing more to find */
-
- if ( attr_match && p->level > 0) /* within a tag pair (or deeper) */
- {
- memcpy (buf, p->buf_m, info->key_size);
- *term_index = p->term_index_m;
- log2( p, "Returning a hit (and forwarding m)", cmp_l, cmp_r);
- p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m,
- &p->term_index_m);
- return 1;
- }
- else
- if ( ! p->more_l ) /* not in data, no more starts */
- {
- log2( p, "no more starts, exiting without a hit", cmp_l, cmp_r);
- return 0; /* ergo, nothing can be found. stop scanning */
- }
-
- p->more_m = rset_read (info->rset_m, p->rfd_m, p->buf_m,
- &p->term_index_m);
- log2( p, "End of M loop", cmp_l, cmp_r);
- } /* while more_m */
-
- log2( p, "Exiting, nothing more in m", cmp_l, cmp_r);
- return 0; /* no more data possible */
+ p->hits++;
+ yaz_log(log_level,"got a hit h="ZINT_FORMAT" d=%d ad=%d",
+ p->hits,p->depth,p->attrdepth);
+ return 1; /* we have everything in place already! */
+ } else
+ yaz_log(log_level, "Ignoring hit. h="ZINT_FORMAT" d=%d ad=%d",
+ p->hits,p->depth,p->attrdepth);
+ }
+ } /* while read */
+ return 0;
} /* r_read */
-static int r_write_between (RSFD rfd, const void *buf)
+static int r_write (RSFD rfd, const void *buf)
{
- logf (LOG_FATAL, "between set type is read-only");
+ yaz_log (YLOG_FATAL, "between set type is read-only");
return -1;
}
+
+static void r_pos (RSFD rfd, double *current, double *total)
+{
+ struct rset_between_rfd *p=(struct rset_between_rfd *)rfd->priv;
+ rset_pos(p->andrfd,current, total);
+ yaz_log(log_level,"pos: %0.1f/%0.1f ", *current, *total);
+}
+
+static void r_get_terms(RSET ct, TERMID *terms, int maxterms, int *curterm)
+{
+ struct rset_between_info *info = (struct rset_between_info *) ct->priv;
+ rset_getterms(info->andset, terms, maxterms, curterm);
+}
+
+