X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=rset%2Frset.c;h=38f02e8207065fb424727b6a297ecbc8336826bb;hp=dbf7b7bcc1f6283326aa77401e4467fdd66b0af2;hb=a876f6c2860bf13e36f47c8ce938d74b4ce98b8e;hpb=5660e317ea2972ea6bb6a4f8a415f71579f71103

diff --git a/rset/rset.c b/rset/rset.c
index dbf7b7b..38f02e8 100644
--- a/rset/rset.c
+++ b/rset/rset.c
@@ -1,6 +1,6 @@
-/* $Id: rset.c,v 1.23 2004-08-06 10:09:28 heikki Exp $
-   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
-   Index Data Aps
+/* $Id: rset.c,v 1.47 2005-06-02 11:59:54 adam Exp $
+   Copyright (C) 1995-2005
+   Index Data ApS
 
 This file is part of the Zebra server.
 
@@ -20,132 +20,347 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
 02111-1307, USA.
 */
 
-
-
 #include <stdio.h>
 #include <string.h>
-#include <zebrautl.h>
+#include <idzebra/util.h>
 #include <assert.h>
-
+#include <yaz/nmem.h>
 #include <rset.h>
-#include <../index/index.h> /* for log_keydump. Debugging only */
 
-RSET rset_create(const struct rset_control *sel, void *parms)
+static int log_level = 0;
+static int log_level_initialized = 0;
+
+/**
+   \brief Common constuctor for RFDs
+   \param rs Result set handle.
+   
+   Creates an rfd. Either allocates a new one, in which case the priv 
+   pointer is null, and will have to be filled in, or picks up one 
+   from the freelist, in which case the priv is already allocated,
+   and presumably everything that hangs from it as well 
+*/
+RSFD rfd_create_base(RSET rs)
 {
-    RSET rnew;
-    int i;
+    RSFD rnew = rs->free_list;
 
-    logf (LOG_DEBUG, "rs_create(%s)", sel->desc);
-    rnew = (RSET) xmalloc(sizeof(*rnew));
-    rnew->control = sel;
-    rnew->flags = 0;
-    rnew->count = 1;
-    rnew->rset_terms = NULL;
-    rnew->no_rset_terms = 0;
-    rnew->buf = (*sel->f_create)(rnew, sel, parms);
-    logf (LOG_DEBUG, "no_rset_terms: %d", rnew->no_rset_terms);
-    for (i = 0; i<rnew->no_rset_terms; i++)
-	logf (LOG_DEBUG, " %s", rnew->rset_terms[i]->name);
+    if (rnew) 
+    {
+        rs->free_list = rnew->next;
+        assert(rnew->rset==rs);
+        yaz_log(log_level, "rfd_create_base (fl): rfd=%p rs=%p fl=%p priv=%p", 
+		rnew, rs, rs->free_list, rnew->priv); 
+    } 
+    else
+    {
+        rnew = nmem_malloc(rs->nmem, sizeof(*rnew));
+	rnew->counted_buf = nmem_malloc(rs->nmem, rs->keycontrol->key_size);
+	rnew->priv = 0;
+        rnew->rset = rs;
+        yaz_log(log_level, "rfd_create_base (new): rfd=%p rs=%p fl=%p priv=%p", 
+		rnew, rs, rs->free_list, rnew->priv); 
+    }
+    rnew->next = rs->use_list;
+    rs->use_list = rnew;
+    rnew->counted_items = 0;
     return rnew;
 }
 
-void rset_delete (RSET rs)
+/**
+   \brief Closes a result set RFD handle
+   \param rfd the RFD handle.
+*/
+void rset_close(RSFD rfd)
 {
-    (rs->count)--;
-    if (!rs->count)
+    RSFD *pfd;
+    RSET rs = rfd->rset;
+
+    if (rs->hits_count == 0)
     {
-	(*rs->control->f_delete)(rs);
-	xfree(rs);
+	TERMID termid;
+	char buf[100];
+	while(rfd->counted_items < rs->hits_limit
+	      && rset_default_read(rfd, buf, &termid))
+	    ;
+	
+	rs->hits_count = rfd->counted_items;
+	rs->hits_approx = 0;
+	if (rs->hits_count >= rs->hits_limit)
+	{
+	    double cur, tot;
+	    zint est;
+	    rset_pos(rfd, &cur, &tot); 
+	    if (tot > 0) {
+		int i;
+		double ratio = cur/tot;
+		est = (zint)(0.5 + rs->hits_count / ratio);
+		yaz_log(log_level, "Estimating hits (%s) "
+			"%0.1f->" ZINT_FORMAT
+			"; %0.1f->" ZINT_FORMAT,
+			rs->control->desc,
+			cur, rs->hits_count,
+			tot, est);
+		i = 0; /* round to significant digits */
+		while (est > rs->hits_round) {
+		    est /= 10;
+		    i++;
+		}
+		while (i--)
+		    est *= 10;
+		rs->hits_count = est;
+		rs->hits_approx = 1;
+	    }
+	}
+	yaz_log(log_level, "rset_close p=%p count=" ZINT_FORMAT, rs,
+		rs->hits_count);
     }
+    (*rs->control->f_close)(rfd);
+    
+    yaz_log(log_level, "rfd_delete_base: rfd=%p rs=%p priv=%p fl=%p",
+            rfd, rs, rfd->priv, rs->free_list); 
+    for (pfd = &rs->use_list; *pfd; pfd = &(*pfd)->next)
+	if (*pfd == rfd)
+	{
+	    *pfd = (*pfd)->next;
+	    rfd->next = rs->free_list;
+	    rs->free_list = rfd;
+	    return;
+	}
+    yaz_log(YLOG_WARN, "rset_close handle not found. type=%s",
+	    rs->control->desc);
 }
 
-RSET rset_dup (RSET rs)
+/**
+   \brief Common constuctor for RSETs
+   \param sel The interface control handle
+   \param nmem The memory handle for it.
+   \param kcontrol Key control info (decode, encode, comparison etc)
+   \param scope scope for set
+   \param term Information about term for it (NULL for none).
+   \param no_children number of child rsets (0 for none)
+   \param children child rsets (NULL for none).
+   
+   Creates an rfd. Either allocates a new one, in which case the priv 
+   pointer is null, and will have to be filled in, or picks up one 
+   from the freelist, in which case the priv is already allocated,
+   and presumably everything that hangs from it as well 
+*/
+RSET rset_create_base(const struct rset_control *sel, 
+                      NMEM nmem, struct rset_key_control *kcontrol,
+                      int scope, TERMID term,
+		      int no_children, RSET *children)
 {
-    (rs->count)++;
-    return rs;
+    RSET rset;
+    assert(nmem);
+    if (!log_level_initialized) 
+    {
+        log_level = yaz_log_module_level("rset");
+        log_level_initialized = 1;
+    }
+
+    rset = (RSET) nmem_malloc(nmem, sizeof(*rset));
+    yaz_log(log_level, "rs_create(%s) rs=%p (nm=%p)", sel->desc, rset, nmem); 
+    rset->nmem = nmem;
+    rset->control = sel;
+    rset->refcount = 1;
+    rset->priv = 0;
+    rset->free_list = NULL;
+    rset->use_list = NULL;
+    rset->hits_count = 0;
+    rset->hits_limit = 1000;
+    rset->hits_round = 1000;
+    rset->keycontrol = kcontrol;
+    (*kcontrol->inc)(kcontrol);
+    rset->scope = scope;
+    rset->term = term;
+    if (term)
+        term->rset = rset;
+
+    rset->no_children = no_children;
+    rset->children = 0;
+    if (no_children)
+    {
+	rset->children = (RSET*)
+	    nmem_malloc(rset->nmem, no_children*sizeof(RSET *));
+	memcpy(rset->children, children, no_children*sizeof(RSET *));
+    }
+    return rset;
 }
 
-void rset_default_pos (RSFD rfd, double *current, double *total)
-{ /* This should never really be needed, but it is still used in */
-  /* those rsets that we don't really plan to use, like isam-s */
-    assert(rfd);
-    assert(current);
-    assert(total);
-    *current=-1; /* signal that pos is not implemented */
-    *total=-1;
-} /* rset_default_pos */
-
-int rset_default_forward(RSET ct, RSFD rfd, void *buf, int *term_index, 
-                           int (*cmpfunc)(const void *p1, const void *p2), 
-                           const void *untilbuf)
+/**
+   \brief Destructor RSETs
+   \param rs Handle for result set.
+   
+   Destroys a result set and all its children.
+   The f_delete method of control is called for the result set.
+*/
+void rset_delete(RSET rs)
 {
-    int more=1;
-    int cmp=2;
-    logf (LOG_DEBUG, "rset_default_forward starting '%s' (ct=%p rfd=%p)",
-                    ct->control->desc, ct,rfd);
-    key_logdump(LOG_DEBUG, untilbuf);
-    while ( (cmp==2) && (more))
+    (rs->refcount)--;
+    yaz_log(log_level, "rs_delete(%s), rs=%p, refcount=%d",
+            rs->control->desc, rs, rs->refcount); 
+    if (!rs->refcount)
     {
-        logf (LOG_DEBUG, "rset_default_forward looping m=%d c=%d",more,cmp);
-        more=rset_read(ct, rfd, buf, term_index);
-        if (more)
-            cmp=(*cmpfunc)(untilbuf,buf);
-        if (more)
-            key_logdump(LOG_DEBUG,buf);
+	int i;
+	if (rs->use_list)
+	    yaz_log(YLOG_WARN, "rs_delete(%s) still has RFDs in use",
+		    rs->control->desc);
+	for (i = 0; i<rs->no_children; i++)
+	    rset_delete(rs->children[i]);
+        (*rs->control->f_delete)(rs);
+	(*rs->keycontrol->dec)(rs->keycontrol);
     }
-    logf (LOG_DEBUG, "rset_default_forward exiting m=%d c=%d",more,cmp);
+}
 
-    return more;
+/**
+   \brief Test for last use of RFD
+   \param rfd RFD handle.
+   
+   Returns 1 if this RFD is the last reference to it; 0 otherwise.
+*/
+int rfd_is_last(RSFD rfd)
+{
+    if (rfd->rset->use_list == rfd && rfd->next == 0)
+	return 1;
+    return 0;
+}
+
+/**
+   \brief Duplicate an RSET
+   \param rs Handle for result set.
+   
+   Duplicates a result set by incrementing the reference count to it.
+*/
+RSET rset_dup (RSET rs)
+{
+    (rs->refcount)++;
+    yaz_log(log_level, "rs_dup(%s), rs=%p, refcount=%d",
+            rs->control->desc, rs, rs->refcount); 
+    (*rs->keycontrol->inc)(rs->keycontrol);
+    return rs;
 }
 
-RSET_TERM *rset_terms(RSET rs, int *no)
+/** 
+    \brief Estimates hit count for result set.
+    \param rs Result Set.
+
+    rset_count uses rset_pos to get the total and returns that.
+    This is ok for rsisamb/c/s, and for some other rsets, but in case of
+    booleans etc it will give bad estimate, as nothing has been read
+    from that rset
+*/
+zint rset_count(RSET rs)
 {
-    *no = rs->no_rset_terms;
-    return rs->rset_terms;
+    double cur, tot;
+    RSFD rfd = rset_open(rs, 0);
+    rset_pos(rfd, &cur, &tot);
+    rset_close(rfd);
+    return (zint) tot;
 }
 
-RSET_TERM rset_term_create (const char *name, int length, const char *flags,
-                            int type)
+/**
+   \brief is a getterms function for those that don't have any
+   \param ct result set handle
+   \param terms array of terms (0..maxterms-1)
+   \param maxterms length of terms array
+   \param curterm current size of terms array
+
+   If there is a term associated with rset the term is appeneded; otherwise
+   the terms array is untouched but curterm is incremented anyway.
+*/
+void rset_get_one_term(RSET ct, TERMID *terms, int maxterms, int *curterm)
 {
-    RSET_TERM t = (RSET_TERM) xmalloc (sizeof(*t));
+    if (ct->term)
+    {
+        if (*curterm < maxterms)
+            terms[*curterm] = ct->term;
+	(*curterm)++;
+    }
+}
+
+/**
+   \brief Creates a TERMID entry.
+   \param name Term/Name buffer with given length
+   \param length of term
+   \param flags for term
+   \param type Term Type, Z_Term_general, Z_Term_characterString,..
+   \param nmem memory for term.
+*/
+TERMID rset_term_create(const char *name, int length, const char *flags,
+			int type, NMEM nmem)
+
+{
+    TERMID t;
+    yaz_log (log_level, "term_create '%s' %d f=%s type=%d nmem=%p",
+            name, length, flags, type, nmem);
+    t= (TERMID) nmem_malloc(nmem, sizeof(*t));
     if (!name)
-	t->name = NULL;
+        t->name = NULL;
     else if (length == -1)
-	t->name = xstrdup (name);
+        t->name = nmem_strdup(nmem, name);
     else
     {
-	t->name = (char*) xmalloc (length+1);
-	memcpy (t->name, name, length);
-	t->name[length] = '\0';
+        t->name = (char*) nmem_malloc(nmem, length+1);
+        memcpy (t->name, name, length);
+        t->name[length] = '\0';
     }
     if (!flags)
-	t->flags = NULL;
+        t->flags = NULL;
     else
-	t->flags = xstrdup (flags);
-    t->nn = -1;
-    t->count = 0;
+        t->flags = nmem_strdup(nmem, flags);
     t->type = type;
+    t->rankpriv = 0;
+    t->rset = 0;
     return t;
 }
 
-void rset_term_destroy (RSET_TERM t)
+int rset_default_read(RSFD rfd, void *buf, TERMID *term)
 {
-    xfree (t->name);
-    xfree (t->flags);
-    xfree (t);
+    RSET rset = rfd->rset;
+    int rc = (*rset->control->f_read)(rfd, buf, term);
+    if (rc > 0)
+    {
+	if (rfd->counted_items == 0 ||
+	    (rset->keycontrol->cmp)(buf, rfd->counted_buf) >= rset->scope)
+	{
+	    memcpy(rfd->counted_buf, buf, rset->keycontrol->key_size);
+	    rfd->counted_items++;
+	}
+    }
+    return rc;
 }
 
-RSET_TERM rset_term_dup (RSET_TERM t)
+int rset_default_forward(RSFD rfd, void *buf, TERMID *term,
+			 const void *untilbuf)
 {
-    RSET_TERM nt = (RSET_TERM) xmalloc (sizeof(*nt));
-    if (t->name)
-	nt->name = xstrdup (t->name);
-    else
-	nt->name = NULL;
-    if (t->flags)
-	nt->flags = xstrdup (t->flags);
-    else
-	nt->flags = NULL;
-    nt->nn = t->nn;
-    return nt;
+    RSET rset = rfd->rset;
+    int more;
+
+    if (rset->control->f_forward &&
+	rfd->counted_items >= rset->hits_limit)
+    {
+	assert (rset->control->f_forward != rset_default_forward);
+	return rset->control->f_forward(rfd, buf, term, untilbuf);
+    }
+    
+    while ((more = rset_read(rfd, buf, term)) > 0)
+    {
+	if ((rfd->rset->keycontrol->cmp)(untilbuf, buf) <= 1)
+	    break;
+    }
+    if (log_level)
+	yaz_log (log_level, "rset_default_forward exiting m=%d c=%d",
+		 more, rset->scope);
+    
+    return more;
 }
+
+void rset_visit(RSET rset, int level)
+{
+    int i;
+    yaz_log(YLOG_LOG, "%*s%c " ZINT_FORMAT, level, "",
+	    rset->hits_approx ? '~' : '=',
+	    rset->hits_count);
+    for (i = 0; i<rset->no_children; i++)
+	rset_visit(rset->children[i], level+1);
+}
+