X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=isamb%2Fisamb.c;h=1a7b9bdfe673a5d3a12363fef7778d45f8a3fa36;hb=d587b2b301c06be60ad255eee6f7010cb2d56fe7;hp=e2042d188eb2c11c861114a1b6a5d33a3f1fc4a8;hpb=ae2d0647bd3a555df77397b6c2f9c129fda0733a;p=idzebra-moved-to-github.git

diff --git a/isamb/isamb.c b/isamb/isamb.c
index e2042d1..1a7b9bd 100644
--- a/isamb/isamb.c
+++ b/isamb/isamb.c
@@ -1,5 +1,5 @@
-/* $Id: isamb.c,v 1.27 2003-06-23 15:36:11 adam Exp $
-   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003
+/* $Id: isamb.c,v 1.44 2004-06-04 13:54:56 heikki Exp $
+   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
 This file is part of the Zebra server.
@@ -26,6 +26,10 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
 #include <isamb.h>
 #include <assert.h>
 
+#ifndef ISAMB_DEBUG
+#define ISAMB_DEBUG 0
+#endif
+
 struct ISAMB_head {
     int first_block;
     int last_block;
@@ -36,13 +40,24 @@ struct ISAMB_head {
 
 #define ISAMB_DATA_OFFSET 3
 
+/* maximum size of encoded buffer */
 #define DST_ITEM_MAX 256
 
-/* approx 2*4 K + max size of item */
-#define DST_BUF_SIZE 8448
+#define ISAMB_MAX_LEVEL 10
+/* approx 2*max page + max size of item */
+#define DST_BUF_SIZE 16840
 
 #define ISAMB_CACHE_ENTRY_SIZE 4096
 
+/* CAT_MAX: _must_ be power of 2 */
+#define CAT_MAX 4
+#define CAT_MASK (CAT_MAX-1)
+/* CAT_NO: <= CAT_MAX */
+#define CAT_NO 4
+
+/* ISAMB_PTR_CODEC=1 var, =0 fixed */
+#define ISAMB_PTR_CODEC  1
+
 struct ISAMB_cache_entry {
     ISAMB_P pos;
     unsigned char *buf;
@@ -51,7 +66,6 @@ struct ISAMB_cache_entry {
     struct ISAMB_cache_entry *next;
 };
 
-
 struct ISAMB_file {
     BFile bf;
     int head_dirty;
@@ -68,6 +82,10 @@ struct ISAMB_s {
     int cache; /* 0=no cache, 1=use cache, -1=dummy isam (for testing only) */
     int log_io;        /* log level for bf_read/bf_write calls */
     int log_freelist;  /* log level for freelist handling */
+    int skipped_numbers; /* on a leaf node */
+    int returned_numbers; 
+    int skipped_nodes[ISAMB_MAX_LEVEL]; /* [0]=skipped leaves, 1=higher etc */
+    int accessed_nodes[ISAMB_MAX_LEVEL]; /* nodes we did not skip */
 };
 
 struct ISAMB_block {
@@ -79,6 +97,7 @@ struct ISAMB_block {
     int deleted;
     int offset;
     char *bytes;
+    char *cbuf;
     unsigned char *buf;
     void *decodeClientData;
     int log_rw;
@@ -88,22 +107,63 @@ struct ISAMB_PP_s {
     ISAMB isamb;
     ISAMB_P pos;
     int level;
+    int maxlevel; /* total depth */
     int total_size;
     int no_blocks;
+    int skipped_numbers; /* on a leaf node */
+    int returned_numbers; 
+    int skipped_nodes[ISAMB_MAX_LEVEL]; /* [0]=skipped leaves, 1=higher etc */
+    int accessed_nodes[ISAMB_MAX_LEVEL]; /* nodes we did not skip */
     struct ISAMB_block **block;
 };
 
-void encode_ptr (char **dst, int pos)
+void isamb_pp_pos( ISAMB_PP pp, int *current, int *total );
+         /* FIXME - this should be in a header file */
+
+#if ISAMB_PTR_CODEC
+static void encode_ptr (char **dst, unsigned pos)
 {
-    memcpy (*dst, &pos, sizeof(pos));
+    unsigned char *bp = (unsigned char*) *dst;
+
+    while (pos > 127)
+    {
+         *bp++ = 128 | (pos & 127);
+         pos = pos >> 7;
+    }
+    *bp++ = pos;
+    *dst = (char *) bp;
+}
+#else
+static void encode_ptr (char **dst, unsigned pos)
+{
+    memcpy(*dst, &pos, sizeof(pos));
     (*dst) += sizeof(pos);
 }
+#endif
 
-void decode_ptr (char **src, int *pos)
+#if ISAMB_PTR_CODEC
+static void decode_ptr (char **src1, int *pos)
+{
+    unsigned char **src = (unsigned char **) src1;
+    unsigned d = 0;
+    unsigned char c;
+    unsigned r = 0;
+
+    while (((c = *(*src)++) & 128))
+    {
+        d += ((c & 127) << r);
+	r += 7;
+    }
+    d += (c << r);
+    *pos = d;
+}
+#else
+static void decode_ptr (char **src, int *pos)
 {
-    memcpy (pos, *src, sizeof(*pos));
-    (*src) += sizeof(*pos);
+     memcpy (pos, *src, sizeof(*pos));
+     (*src) += sizeof(*pos);
 }
+#endif
 
 ISAMB isamb_open (BFiles bfs, const char *name, int writeflag, ISAMC_M *method,
                   int cache)
@@ -114,10 +174,14 @@ ISAMB isamb_open (BFiles bfs, const char *name, int writeflag, ISAMC_M *method,
     isamb->bfs = bfs;
     isamb->method = (ISAMC_M *) xmalloc (sizeof(*method));
     memcpy (isamb->method, method, sizeof(*method));
-    isamb->no_cat = 4;
+    isamb->no_cat = CAT_NO;
     isamb->log_io = 0;
     isamb->log_freelist = 0;
     isamb->cache = cache;
+    isamb->skipped_numbers=0;
+    isamb->returned_numbers=0;
+    for (i=0;i<ISAMB_MAX_LEVEL;i++)
+      isamb->skipped_nodes[i]= isamb->accessed_nodes[i]=0;
 
     assert (cache == 0);
     isamb->file = xmalloc (sizeof(*isamb->file) * isamb->no_cat);
@@ -148,6 +212,9 @@ ISAMB isamb_open (BFiles bfs, const char *name, int writeflag, ISAMC_M *method,
         assert(isamb->file[i].head.block_size == b_size);
         b_size = b_size * 4;
     }
+#if ISAMB_DEBUG
+    logf(LOG_WARN, "isamb debug enabled. Things will be slower than usual");
+#endif
     return isamb;
 }
 
@@ -170,11 +237,11 @@ static void flush_blocks (ISAMB b, int cat)
 
 static int get_block (ISAMB b, ISAMC_P pos, char *userbuf, int wr)
 {
-    int cat = pos&3;
-    int off = ((pos/4) & 
+    int cat = pos&CAT_MASK;
+    int off = ((pos/CAT_MAX) & 
                (ISAMB_CACHE_ENTRY_SIZE / b->file[cat].head.block_size - 1))
         * b->file[cat].head.block_size;
-    int norm = pos / (4*ISAMB_CACHE_ENTRY_SIZE / b->file[cat].head.block_size);
+    int norm = pos / (CAT_MASK*ISAMB_CACHE_ENTRY_SIZE / b->file[cat].head.block_size);
     int no = 0;
     struct ISAMB_cache_entry **ce, *ce_this = 0, **ce_last = 0;
 
@@ -244,6 +311,11 @@ static int get_block (ISAMB b, ISAMC_P pos, char *userbuf, int wr)
 void isamb_close (ISAMB isamb)
 {
     int i;
+    for (i=0;isamb->accessed_nodes[i];i++)
+        logf(LOG_DEBUG,"isamb_close  level leaf-%d: %d read, %d skipped",
+             i, isamb->accessed_nodes[i], isamb->skipped_nodes[i]);
+    logf(LOG_DEBUG,"isamb_close returned %d values, skipped %d",
+         isamb->skipped_numbers, isamb->returned_numbers);
     for (i = 0; i<isamb->no_cat; i++)
     {
         flush_blocks (isamb, i);
@@ -258,25 +330,25 @@ void isamb_close (ISAMB isamb)
     xfree (isamb);
 }
 
-
-struct ISAMB_block *open_block (ISAMB b, ISAMC_P pos)
+static struct ISAMB_block *open_block (ISAMB b, ISAMC_P pos)
 {
-    int cat = pos&3;
+    int cat = pos&CAT_MASK;
     struct ISAMB_block *p;
     if (!pos)
         return 0;
     p = xmalloc (sizeof(*p));
     p->pos = pos;
-    p->cat = pos & 3;
+    p->cat = pos & CAT_MASK;
     p->buf = xmalloc (b->file[cat].head.block_size);
+    p->cbuf = 0;
 
     if (!get_block (b, pos, p->buf, 0))
     {
         yaz_log (b->log_io, "bf_read: open_block");
-        if (!bf_read (b->file[cat].bf, pos/4, 0, 0, p->buf))
+        if (!bf_read (b->file[cat].bf, pos/CAT_MAX, 0, 0, p->buf))
         {
             yaz_log (LOG_FATAL, "isamb: read fail for pos=%ld block=%ld",
-                     (long) pos, (long) pos/4);
+                     (long) pos, (long) pos/CAT_MAX);
             abort();
         }
     }
@@ -285,13 +357,14 @@ struct ISAMB_block *open_block (ISAMB b, ISAMC_P pos)
     p->size = (p->buf[1] + 256 * p->buf[2]) - ISAMB_DATA_OFFSET;
     if (p->size < 0)
     {
-        fprintf (stderr, "pos=%d\n", pos);
+        yaz_log (LOG_FATAL, "Bad block size %d in pos=%d\n", p->size, pos);
     }
     assert (p->size >= 0);
     p->offset = 0;
     p->dirty = 0;
     p->deleted = 0;
     p->decodeClientData = (*b->method->code_start)(ISAMC_DECODE);
+    yaz_log (LOG_DEBUG, "isamb_open_block: Opened block %d ofs=%d",pos, p->offset);
     return p;
 }
 
@@ -306,24 +379,24 @@ struct ISAMB_block *new_block (ISAMB b, int leaf, int cat)
     {
         int block_no;
         block_no = b->file[cat].head.last_block++;
-        p->pos = block_no * 4 + cat;
+        p->pos = block_no * CAT_MAX + cat;
     }
     else
     {
         p->pos = b->file[cat].head.free_list;
-        assert((p->pos & 3) == cat);
+        assert((p->pos & CAT_MASK) == cat);
         if (!get_block (b, p->pos, p->buf, 0))
         {
             yaz_log (b->log_io, "bf_read: new_block");
-            if (!bf_read (b->file[cat].bf, p->pos/4, 0, 0, p->buf))
+            if (!bf_read (b->file[cat].bf, p->pos/CAT_MAX, 0, 0, p->buf))
             {
                 yaz_log (LOG_FATAL, "isamb: read fail for pos=%ld block=%ld",
-                         (long) p->pos/4, (long) p->pos/4);
+                         (long) p->pos/CAT_MAX, (long) p->pos/CAT_MAX);
                 abort ();
             }
         }
         yaz_log (b->log_freelist, "got block %d from freelist %d:%d", p->pos,
-                 cat, p->pos/4);
+                 cat, p->pos/CAT_MAX);
         memcpy (&b->file[cat].head.free_list, p->buf, sizeof(int));
     }
     p->cat = cat;
@@ -365,7 +438,7 @@ static void check_block (ISAMB b, struct ISAMB_block *p)
         int pos;
             
         decode_ptr (&src, &pos);
-        assert ((pos&3) == p->cat);
+        assert ((pos&CAT_MASK) == p->cat);
         while (src != endp)
         {
             int item_len;
@@ -373,7 +446,7 @@ static void check_block (ISAMB b, struct ISAMB_block *p)
             assert (item_len > 0 && item_len < 30);
             src += item_len;
             decode_ptr (&src, &pos);
-            assert ((pos&3) == p->cat);
+            assert ((pos&CAT_MASK) == p->cat);
         }
     }
 }
@@ -385,13 +458,13 @@ void close_block (ISAMB b, struct ISAMB_block *p)
     if (p->deleted)
     {
         yaz_log (b->log_freelist, "release block %d from freelist %d:%d",
-                 p->pos, p->cat, p->pos/4);
+                 p->pos, p->cat, p->pos/CAT_MAX);
         memcpy (p->buf, &b->file[p->cat].head.free_list, sizeof(int));
         b->file[p->cat].head.free_list = p->pos;
         if (!get_block (b, p->pos, p->buf, 1))
         {
             yaz_log (b->log_io, "bf_write: close_block (deleted)");
-            bf_write (b->file[p->cat].bf, p->pos/4, 0, 0, p->buf);
+            bf_write (b->file[p->cat].bf, p->pos/CAT_MAX, 0, 0, p->buf);
         }
     }
     else if (p->dirty)
@@ -405,7 +478,7 @@ void close_block (ISAMB b, struct ISAMB_block *p)
         if (!get_block (b, p->pos, p->buf, 1))
         {
             yaz_log (b->log_io, "bf_write: close_block");
-            bf_write (b->file[p->cat].bf, p->pos/4, 0, 0, p->buf);
+            bf_write (b->file[p->cat].bf, p->pos/CAT_MAX, 0, 0, p->buf);
         }
     }
     (*b->method->code_stop)(ISAMC_DECODE, p->decodeClientData);
@@ -543,7 +616,7 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
     void *c1 = (*b->method->code_start)(ISAMC_DECODE);
     void *c2 = (*b->method->code_start)(ISAMC_ENCODE);
     int more = 1;
-    int quater = b->file[b->no_cat-1].head.block_max / 4;
+    int quater = b->file[b->no_cat-1].head.block_max / CAT_MAX;
     char *cut = dst_buf + quater * 2;
     char *maxp = dst_buf + b->file[b->no_cat-1].head.block_max;
     char *half1 = 0;
@@ -856,14 +929,20 @@ int isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I *stream)
 ISAMB_PP isamb_pp_open_x (ISAMB isamb, ISAMB_P pos, int *level)
 {
     ISAMB_PP pp = xmalloc (sizeof(*pp));
+    int i;
 
     pp->isamb = isamb;
-    pp->block = xmalloc (10 * sizeof(*pp->block));
+    pp->block = xmalloc (ISAMB_MAX_LEVEL * sizeof(*pp->block));
 
     pp->pos = pos;
     pp->level = 0;
+    pp->maxlevel=0;
     pp->total_size = 0;
     pp->no_blocks = 0;
+    pp->skipped_numbers=0;
+    pp->returned_numbers=0;
+    for (i=0;i<ISAMB_MAX_LEVEL;i++)
+        pp->skipped_nodes[i] = pp->accessed_nodes[i]=0;
     while (1)
     {
         struct ISAMB_block *p = open_block (isamb, pos);
@@ -872,15 +951,17 @@ ISAMB_PP isamb_pp_open_x (ISAMB isamb, ISAMB_P pos, int *level)
 
         pp->total_size += p->size;
         pp->no_blocks++;
-
         if (p->leaf)
             break;
 
+                                        
         decode_ptr (&src, &pos);
         p->offset = src - p->bytes;
         pp->level++;
+        pp->accessed_nodes[pp->level]++; 
     }
     pp->block[pp->level+1] = 0;
+    pp->maxlevel=pp->level;
     if (level)
         *level = pp->level;
     return pp;
@@ -896,6 +977,19 @@ void isamb_pp_close_x (ISAMB_PP pp, int *size, int *blocks)
     int i;
     if (!pp)
         return;
+    logf(LOG_DEBUG,"isamb_pp_close lev=%d returned %d values, skipped %d",
+        pp->maxlevel, pp->skipped_numbers, pp->returned_numbers);
+    for (i=pp->maxlevel;i>=0;i--)
+        if ( pp->skipped_nodes[i] || pp->accessed_nodes[i])
+            logf(LOG_DEBUG,"isamb_pp_close  level leaf-%d: %d read, %d skipped", i,
+                 pp->accessed_nodes[i], pp->skipped_nodes[i]);
+    pp->isamb->skipped_numbers += pp->skipped_numbers;
+    pp->isamb->returned_numbers += pp->returned_numbers;
+    for (i=pp->maxlevel;i>=0;i--)
+    {
+        pp->isamb->accessed_nodes[i] += pp->accessed_nodes[i];
+        pp->isamb->skipped_nodes[i] += pp->skipped_nodes[i];
+    }
     if (size)
         *size = pp->total_size;
     if (blocks)
@@ -918,6 +1012,70 @@ void isamb_pp_close (ISAMB_PP pp)
     isamb_pp_close_x (pp, 0, 0);
 }
 
+/* simple recursive dumper .. */
+static void isamb_dump_r (ISAMB b, ISAMB_P pos, void (*pr)(const char *str),
+			  int level)
+{
+    char buf[1024];
+    char prefix_str[1024];
+    if (pos)
+    {
+	struct ISAMB_block *p = open_block (b, pos);
+	sprintf(prefix_str, "%*s %d cat=%d size=%d max=%d", level*2, "",
+		pos, p->cat, p->size, b->file[p->cat].head.block_max);
+	(*pr)(prefix_str);
+	sprintf(prefix_str, "%*s %d", level*2, "", pos);
+	if (p->leaf)
+	{
+	    while (p->offset < p->size)
+	    {
+		char *src = p->bytes + p->offset;
+		char *dst = buf;
+		(*b->method->code_item)(ISAMC_DECODE, p->decodeClientData,
+					&dst, &src);
+		(*b->method->log_item)(LOG_DEBUG, buf, prefix_str);
+		p->offset = src - (char*) p->bytes;
+	    }
+	    assert(p->offset == p->size);
+	}
+	else
+	{
+	    char *src = p->bytes + p->offset;
+	    int sub;
+	    int item_len;
+
+	    decode_ptr (&src, &sub);
+	    p->offset = src - (char*) p->bytes;
+
+	    isamb_dump_r(b, sub, pr, level+1);
+	    
+	    while (p->offset < p->size)
+	    {
+		decode_ptr (&src, &item_len);
+		(*b->method->log_item)(LOG_DEBUG, src, prefix_str);
+		src += item_len;
+		decode_ptr (&src, &sub);
+		
+		p->offset = src - (char*) p->bytes;
+		
+		isamb_dump_r(b, sub, pr, level+1);
+	    }		
+	}
+	close_block(b,p);
+    }
+}
+
+void isamb_dump (ISAMB b, ISAMB_P pos, void (*pr)(const char *str))
+{
+    return isamb_dump_r(b, pos, pr, 0);
+}
+
+#if 0
+/* Old isamb_pp_read that Adam wrote, kept as a reference in case we need to
+   debug the more complex pp_read that also forwards. May be deleted near end
+   of 2004, if it has not shown to be useful */
+
+
 int isamb_pp_read (ISAMB_PP pp, void *buf)
 {
     char *dst = buf;
@@ -937,7 +1095,7 @@ int isamb_pp_read (ISAMB_PP pp, void *buf)
             pp->block[pp->level] = 0;
             (pp->level)--;
             p = pp->block[pp->level];
-            assert (!p->leaf);  /* must be int */
+            assert (!p->leaf);  
         }
         src = p->bytes + p->offset;
         
@@ -956,7 +1114,7 @@ int isamb_pp_read (ISAMB_PP pp, void *buf)
             pp->total_size += p->size;
             pp->no_blocks++;
             
-            if (p->leaf) /* leaf */
+            if (p->leaf) 
             {
                 break;
             }
@@ -972,10 +1130,746 @@ int isamb_pp_read (ISAMB_PP pp, void *buf)
     (*pp->isamb->method->code_item)(ISAMC_DECODE, p->decodeClientData,
                                     &dst, &src);
     p->offset = src - (char*) p->bytes;
+    /* key_logdump_txt(LOG_DEBUG,buf, "isamb_pp_read returning 1"); */
+    return 1;
+}
+
+#else
+int isamb_pp_read (ISAMB_PP pp, void *buf)
+{
+    return isamb_pp_forward(pp,buf,0);
+}
+#endif
+
+#define NEW_FORWARD 1
+
+#if NEW_FORWARD == 1
+
+/*
+#undef ISAMB_DEBUB
+#define ISAMB_DEBUG 1
+*/
+static int isamb_pp_on_right_node(ISAMB_PP pp, int level, const void *untilbuf)
+{ /* looks one node higher to see if we should be on this node at all */
+  /* useful in backing off quickly, and in avoiding tail descends */
+  /* call with pp->level to begin with */
+    struct ISAMB_block *p;
+    int cmp;
+    char *src;
+    int item_len;
+    assert(level>=0);
+    if ( level == 0) {
+#if ISAMB_DEBUG
+            logf(LOG_DEBUG,"isamb_pp_on_right returning true for root");
+#endif
+        return 1; /* we can never skip the root node */
+    }
+    level--;
+    p=pp->block[level];
+    assert(p->offset <= p->size);
+    if (p->offset < p->size )
+    {
+        assert(p->offset>0); 
+        src=p->bytes + p->offset;
+        decode_ptr(&src,&item_len);
+#if ISAMB_DEBUG
+        (*pp->isamb->method->log_item)(LOG_DEBUG,untilbuf,"on_leaf: until");
+        (*pp->isamb->method->log_item)(LOG_DEBUG,src,"on_leaf: value");
+#endif
+        cmp=(*pp->isamb->method->compare_item)(untilbuf,src);
+        if (cmp<2) {
+#if ISAMB_DEBUG
+            logf(LOG_DEBUG,"isamb_pp_on_right returning true "
+                            "cmp=%d lev=%d ofs=%d",cmp,level,p->offset);
+#endif
+            return 1; 
+        }
+        else {
+#if ISAMB_DEBUG
+            logf(LOG_DEBUG,"isamb_pp_on_right returning false "
+                            "cmp=%d lev=%d ofs=%d",cmp,level,p->offset);
+#endif
+            return 0; 
+        }
+    }
+    else {
+#if ISAMB_DEBUG
+        logf(LOG_DEBUG,"isamb_pp_on_right at tail, looking higher "
+                        "lev=%d",level);
+#endif
+        return isamb_pp_on_right_node(pp, level, untilbuf);
+    }
+} /* isamb_pp_on_right_node */
+
+static int isamb_pp_read_on_leaf(ISAMB_PP pp, void *buf)
+{ /* reads the next item on the current leaf, returns 0 if end of leaf*/
+    struct ISAMB_block *p = pp->block[pp->level];
+    char *dst;
+    char *src;
+    assert(pp);
+    assert(buf);
+    if (p->offset == p->size) {
+#if ISAMB_DEBUG
+        logf(LOG_DEBUG,"isamb_pp_read_on_leaf returning 0 on node %d",p->pos);
+#endif
+        return 0; /* at end of leaf */
+    }
+    src=p->bytes + p->offset;
+    dst=buf;
+    (*pp->isamb->method->code_item)
+           (ISAMC_DECODE, p->decodeClientData,&dst, &src);
+    p->offset = src - (char*) p->bytes;
+    /*
+#if ISAMB_DEBUG
+    (*pp->isamb->method->log_item)(LOG_DEBUG, buf, "read_on_leaf returning 1");
+#endif
+*/
+    return 1;
+} /* read_on_leaf */
+
+static int isamb_pp_forward_on_leaf(ISAMB_PP pp, void *buf, const void *untilbuf)
+{ /* forwards on the current leaf, returns 0 if not found */
+    int cmp;
+    int skips=0;
+    while (1){
+        if (!isamb_pp_read_on_leaf(pp,buf))
+            return 0;
+        /* FIXME - this is an extra function call, inline the read? */
+        cmp=(*pp->isamb->method->compare_item)(untilbuf,buf);
+        if (cmp <2){  /* found a good one */
+#if ISAMB_DEBUG
+            if (skips)
+                logf(LOG_DEBUG, "isam_pp_fwd_on_leaf skipped %d items",skips);
+#endif
+            pp->returned_numbers++;
+            return 1;
+        }
+        if (!skips)
+            if (!isamb_pp_on_right_node(pp, pp->level, untilbuf))
+                return 0; /* never mind the rest of this leaf */
+        pp->skipped_numbers++;
+        skips++;
+    }
+} /* forward_on_leaf */
+
+static int isamb_pp_climb_level(ISAMB_PP pp, int *pos)
+{ /* climbs higher in the tree, until finds a level with data left */
+  /* returns the node to (consider to) descend to in *pos) */
+    struct ISAMB_block *p = pp->block[pp->level];
+    char *src;
+    int item_len;
+#if ISAMB_DEBUG
+    logf(LOG_DEBUG,"isamb_pp_climb_level starting "
+                   "at level %d node %d ofs=%d sz=%d",
+                    pp->level, p->pos, p->offset, p->size);
+#endif
+    assert(pp->level >= 0);
+    assert(p->offset <= p->size);
+    if (pp->level==0)
+    {
+#if ISAMB_DEBUG
+        logf(LOG_DEBUG,"isamb_pp_climb_level returning 0 at root");
+#endif
+        return 0;
+    }
+    assert(pp->level>0); 
+    close_block(pp->isamb, pp->block[pp->level]);
+    pp->block[pp->level]=0;
+    (pp->level)--;
+    p=pp->block[pp->level];
+#if ISAMB_DEBUG
+    logf(LOG_DEBUG,"isamb_pp_climb_level climbed to level %d node %d ofs=%d",
+                    pp->level, p->pos, p->offset);
+#endif
+    assert(!p->leaf);
+    assert(p->offset <= p->size);
+    if (p->offset == p->size ) {
+        /* we came from the last pointer, climb on */
+        if (!isamb_pp_climb_level(pp,pos))
+            return 0;
+        p=pp->block[pp->level];
+    }
+    else
+    {
+        /* skip the child we just came from */
+#if ISAMB_DEBUG
+        logf(LOG_DEBUG,"isam_pp_climb_level: skipping lev=%d ofs=%d sz=%d", 
+                        pp->level, p->offset, p->size);
+#endif
+        assert (p->offset < p->size );
+        src=p->bytes + p->offset;
+        decode_ptr(&src, &item_len);
+        src += item_len;
+        decode_ptr(&src, pos);
+        p->offset=src - (char *)p->bytes;
+            
+    }
+    return 1;
+} /* climb_level */
+
+
+static int isamb_pp_forward_unode(ISAMB_PP pp, int pos, const void *untilbuf)
+{ /* scans a upper node until it finds a child <= untilbuf */
+  /* pp points to the key value, as always. pos is the child read from */
+  /* the buffer */
+  /* if all values are too small, returns the last child in the node */
+  /* FIXME - this can be detected, and avoided by looking at the */
+  /* parent node, but that gets messy. Presumably the cost is */
+  /* pretty low anyway */
+    struct ISAMB_block *p = pp->block[pp->level];
+    char *src=p->bytes + p->offset;
+    int item_len;
+    int cmp;
+    int nxtpos;
+#if ISAMB_DEBUG
+    int skips=0;
+    logf(LOG_DEBUG,"isamb_pp_forward_unode starting "
+                   "at level %d node %d ofs=%di sz=%d",
+                    pp->level, p->pos, p->offset, p->size);
+#endif
+    assert(!p->leaf);
+    assert(p->offset <= p->size);
+    if (p->offset == p->size) {
+#if ISAMB_DEBUG
+            logf(LOG_DEBUG,"isamb_pp_forward_unode returning at end "
+                   "at level %d node %d ofs=%di sz=%d",
+                    pp->level, p->pos, p->offset, p->size);
+#endif
+        return pos; /* already at the end of it */
+    }
+    while(p->offset < p->size) {
+        decode_ptr(&src,&item_len);
+        cmp=(*pp->isamb->method->compare_item)(untilbuf,src);
+        src+=item_len;
+        decode_ptr(&src,&nxtpos);
+        if (cmp<2)
+        {
+#if ISAMB_DEBUG
+            logf(LOG_DEBUG,"isamb_pp_forward_unode returning a hit "
+                   "at level %d node %d ofs=%d sz=%d",
+                    pp->level, p->pos, p->offset, p->size);
+#endif
+            return pos;
+        } /* found one */
+        pos=nxtpos;
+        p->offset=src-(char*)p->bytes;
+        (pp->skipped_nodes[pp->maxlevel - pp->level -1])++;
+#if ISAMB_DEBUG
+        skips++;
+#endif
+    }
+#if ISAMB_DEBUG
+            logf(LOG_DEBUG,"isamb_pp_forward_unode returning at tail "
+                   "at level %d node %d ofs=%d sz=%d skips=%d",
+                    pp->level, p->pos, p->offset, p->size, skips);
+#endif
+    return pos; /* that's the last one in the line */
+    
+} /* forward_unode */
+
+static void isamb_pp_descend_to_leaf(ISAMB_PP pp, int pos, const void *untilbuf)
+{ /* climbs down the tree, from pos, to the leftmost leaf */
+    struct ISAMB_block *p = pp->block[pp->level];
+    char *src;
+    assert(!p->leaf);
+#if ISAMB_DEBUG
+    logf(LOG_DEBUG,"isamb_pp_descend_to_leaf "
+                   "starting at lev %d node %d ofs=%d lf=%d u=%p", 
+                   pp->level, p->pos, p->offset, p->leaf, untilbuf);
+#endif
+    if (untilbuf)
+        pos=isamb_pp_forward_unode(pp,pos,untilbuf);
+    ++(pp->level);
+    assert(pos);
+    p=open_block(pp->isamb, pos);
+    pp->block[pp->level]=p;
+    ++(pp->accessed_nodes[pp->maxlevel-pp->level]);
+    ++(pp->no_blocks);
+#if ISAMB_DEBUG
+    logf(LOG_DEBUG,"isamb_pp_descend_to_leaf "
+                   "got lev %d node %d lf=%d", 
+                   pp->level, p->pos, p->leaf);
+#endif
+    if (p->leaf)
+        return;
+    assert (p->offset==0 );
+    src=p->bytes + p->offset;
+    decode_ptr(&src, &pos);
+    p->offset=src-(char*)p->bytes;
+    isamb_pp_descend_to_leaf(pp,pos,untilbuf);
+#if ISAMB_DEBUG
+    logf(LOG_DEBUG,"isamb_pp_descend_to_leaf "
+                   "returning at lev %d node %d ofs=%d lf=%d", 
+                   pp->level, p->pos, p->offset, p->leaf);
+#endif
+} /* descend_to_leaf */
+
+static int isamb_pp_find_next_leaf(ISAMB_PP pp)
+{ /* finds the next leaf by climbing up and down */
+    int pos;
+    if (!isamb_pp_climb_level(pp,&pos))
+        return 0;
+    isamb_pp_descend_to_leaf(pp, pos,0);
     return 1;
 }
 
+static int isamb_pp_climb_desc(ISAMB_PP pp, void *buf, const void *untilbuf)
+{ /* climbs up and descends to a leaf where values >= *untilbuf are found */
+    int pos;
+#if ISAMB_DEBUG
+    struct ISAMB_block *p = pp->block[pp->level];
+    logf(LOG_DEBUG,"isamb_pp_climb_desc starting "
+                   "at level %d node %d ofs=%d sz=%d",
+                    pp->level, p->pos, p->offset, p->size);
+#endif
+    if (!isamb_pp_climb_level(pp,&pos))
+        return 0;
+    /* see if it would pay to climb one higher */
+    if (!isamb_pp_on_right_node(pp, pp->level, untilbuf))
+        if (!isamb_pp_climb_level(pp,&pos))
+            return 0;
+    isamb_pp_descend_to_leaf(pp, pos,untilbuf);
+#if ISAMB_DEBUG
+    p = pp->block[pp->level];
+    logf(LOG_DEBUG,"isamb_pp_climb_desc done "
+                   "at level %d node %d ofs=%d sz=%d",
+                    pp->level, p->pos, p->offset, p->size);
+#endif
+    return 1;
+} /* climb_desc */
+
+int isamb_pp_forward (ISAMB_PP pp, void *buf, const void *untilbuf)
+{
+#if ISAMB_DEBUG
+    struct ISAMB_block *p = pp->block[pp->level];
+    assert(p->leaf);
+    logf(LOG_DEBUG,"isamb_pp_forward starting "
+                   "at level %d node %d ofs=%d sz=%d u=%p",
+                    pp->level, p->pos, p->offset, p->size,untilbuf);
+#endif
+    if (untilbuf) {
+        if (isamb_pp_forward_on_leaf( pp, buf, untilbuf)) {
+#if ISAMB_DEBUG
+            logf(LOG_DEBUG,"isamb_pp_forward (f) returning (A) "
+                   "at level %d node %d ofs=%d sz=%d",
+                    pp->level, p->pos, p->offset, p->size);
+#endif
+            return 1;
+        }
+        if (! isamb_pp_climb_desc( pp, buf, untilbuf)) {
+#if ISAMB_DEBUG
+            logf(LOG_DEBUG,"isamb_pp_forward (f) returning notfound (B) "
+                   "at level %d node %d ofs=%d sz=%d",
+                    pp->level, p->pos, p->offset, p->size);
+#endif
+            return 0; /* could not find a leaf */
+        }
+        do{
+            if (isamb_pp_forward_on_leaf( pp, buf, untilbuf)) {
+#if ISAMB_DEBUG
+            logf(LOG_DEBUG,"isamb_pp_forward (f) returning (C) "
+                   "at level %d node %d ofs=%d sz=%d",
+                    pp->level, p->pos, p->offset, p->size);
+#endif
+                return 1;
+            }
+        }while ( isamb_pp_find_next_leaf(pp));
+        return 0; /* could not find at all */
+    }
+    else { /* no untilbuf, a straight read */
+        /* FIXME - this should be moved
+         * directly into the pp_read */
+        /* keeping here now, to keep same
+         * interface as the old fwd */
+        if (isamb_pp_read_on_leaf( pp, buf)) {
+#if ISAMB_DEBUG
+            logf(LOG_DEBUG,"isamb_pp_forward (read) returning (D) "
+                   "at level %d node %d ofs=%d sz=%d",
+                    pp->level, p->pos, p->offset, p->size);
+#endif
+            return 1;
+        }
+        if (isamb_pp_find_next_leaf(pp)) {
+#if ISAMB_DEBUG
+            logf(LOG_DEBUG,"isamb_pp_forward (read) returning (E) "
+                   "at level %d node %d ofs=%d sz=%d",
+                    pp->level, p->pos, p->offset, p->size);
+#endif
+            return isamb_pp_read_on_leaf(pp, buf);
+        }
+        else
+            return 0;
+    }
+} /* isam_pp_forward (new version) */
+
+#elif NEW_FORWARD == 0
+
+int isamb_pp_forward (ISAMB_PP pp, void *buf, const void *untilbuf)
+{
+    /* pseudocode:
+     *   while 1
+     *     while at end of node
+     *       climb higher. If out, return 0
+     *     while not on a leaf (and not at its end)
+     *       decode next
+     *       if cmp
+     *         descend to node
+     *     decode next
+     *     if cmp
+     *       return 1
+     */
+     /* 
+      * The upper nodes consist of a sequence of nodenumbers and keys
+      * When opening a block,  the first node number is read in, and
+      * offset points to the first key, which is the upper limit of keys
+      * in the node just read.
+      */
+    char *dst = buf;
+    char *src;
+    struct ISAMB_block *p = pp->block[pp->level];
+    int cmp;
+    int item_len;
+    int pos;
+    int nxtpos;
+    int descending=0; /* used to prevent a border condition error */
+    if (!p)
+        return 0;
+#if ISAMB_DEBUG
+    logf(LOG_DEBUG,"isamb_pp_forward starting [%p] p=%d",pp,p->pos);
+    
+    (*pp->isamb->method->log_item)(LOG_DEBUG, untilbuf, "until");
+    (*pp->isamb->method->log_item)(LOG_DEBUG, buf, "buf");
+#endif
+
+    while (1)
+    {
+        while ( (p->offset == p->size) && !descending )
+        {  /* end of this block - climb higher */
+	    assert (p->offset <= p->size);
+#if ISAMB_DEBUG
+            logf(LOG_DEBUG,"isamb_pp_forward climbing from l=%d",
+                            pp->level);
+#endif
+            if (pp->level == 0)
+            {
+#if ISAMB_DEBUG
+                logf(LOG_DEBUG,"isamb_pp_forward returning 0 at root");
+#endif
+                return 0; /* at end of the root, nothing left */
+            }
+            close_block(pp->isamb, pp->block[pp->level]);
+            pp->block[pp->level]=0;
+            (pp->level)--;
+            p=pp->block[pp->level];
+#if ISAMB_DEBUG
+            logf(LOG_DEBUG,"isamb_pp_forward climbed to node %d off=%d",
+                            p->pos, p->offset);
+#endif
+            assert(!p->leaf);
+            assert(p->offset <= p->size);
+            /* skip the child we have handled */
+            if (p->offset != p->size)
+            { 
+                src = p->bytes + p->offset;
+                decode_ptr(&src, &item_len);
+#if ISAMB_DEBUG		
+		(*pp->isamb->method->log_item)(LOG_DEBUG, src,
+					       " isamb_pp_forward "
+					       "climb skipping old key");
+#endif
+                src += item_len;
+                decode_ptr(&src,&pos);
+                p->offset = src - (char*) p->bytes;
+                break; /* even if this puts us at the end of the block, we
+			  need to descend to the last pos. UGLY coding,
+			  clean up some day */
+            }
+        }
+        if (!p->leaf)
+        { 
+            src = p->bytes + p->offset;
+            if (p->offset == p->size)
+                cmp=-2 ; /* descend to the last node, as we have
+			    no value to cmp */
+            else
+            {
+                decode_ptr(&src, &item_len);
+#if ISAMB_DEBUG
+                logf(LOG_DEBUG,"isamb_pp_forward (B) on a high node. "
+		     "ofs=%d sz=%d nxtpos=%d ",
+                        p->offset,p->size,pos);
+		(*pp->isamb->method->log_item)(LOG_DEBUG, src, "");
+#endif
+                if (untilbuf)
+                    cmp=(*pp->isamb->method->compare_item)(untilbuf,src);
+                else
+                    cmp=-2;
+                src += item_len;
+                decode_ptr(&src,&nxtpos);
+            }
+            if (cmp<2)
+            { 
+#if ISAMB_DEBUG
+                logf(LOG_DEBUG,"isambb_pp_forward descending l=%d p=%d ",
+                            pp->level, pos);
+#endif
+                descending=1; /* prevent climbing for a while */
+                ++(pp->level);
+                p = open_block(pp->isamb,pos);
+                pp->block[pp->level] = p ;
+                pp->total_size += p->size;
+                (pp->accessed_nodes[pp->maxlevel - pp->level])++;
+                pp->no_blocks++;
+                if ( !p->leaf)
+                { /* block starts with a pos */
+                    src = p->bytes + p->offset;
+                    decode_ptr(&src,&pos);
+                    p->offset=src-(char*) p->bytes;
+#if ISAMB_DEBUG
+                    logf(LOG_DEBUG,"isamb_pp_forward: block %d starts with %d",
+                                    p->pos, pos);
+#endif
+                } 
+            } /* descend to the node */
+            else
+            { /* skip the node */
+                p->offset = src - (char*) p->bytes;
+                pos=nxtpos;
+                (pp->skipped_nodes[pp->maxlevel - pp->level -1])++;
+#if ISAMB_DEBUG
+                logf(LOG_DEBUG,
+                    "isamb_pp_forward: skipping block on level %d, noting "
+		     "on %d (%d)",
+                    pp->level, pp->maxlevel - pp->level-1 , 
+                    pp->skipped_nodes[pp->maxlevel - pp->level-1 ]);
+#endif
+                /* 0 is always leafs, 1 is one level above leafs etc, no
+                 * matter how high tree */
+            }
+        } /* not on a leaf */
+        else
+        { /* on a leaf */
+	    if (p->offset == p->size) { 
+		descending = 0;
+	    }
+	    else
+	    {
+		assert (p->offset < p->size);
+		src = p->bytes + p->offset;
+		dst=buf;
+		(*pp->isamb->method->code_item)(ISAMC_DECODE, p->decodeClientData,
+						&dst, &src);
+		p->offset = src - (char*) p->bytes;
+		if (untilbuf)
+		    cmp=(*pp->isamb->method->compare_item)(untilbuf,buf);
+		else
+		    cmp=-2;
+#if ISAMB_DEBUG
+		logf(LOG_DEBUG,"isamb_pp_forward on a leaf. cmp=%d", 
+		     cmp);
+		(*pp->isamb->method->log_item)(LOG_DEBUG, buf, "");
+#endif
+		if (cmp <2)
+		{
+#if ISAMB_DEBUG
+		    if (untilbuf)
+		    {
+			(*pp->isamb->method->log_item)(
+			    LOG_DEBUG, buf,  "isamb_pp_forward returning 1");
+		    }
+		    else
+		    {
+			(*pp->isamb->method->log_item)(
+			    LOG_DEBUG, buf, "isamb_pp_read returning 1 (fwd)");
+		    }
+#endif
+		    pp->returned_numbers++;
+		    return 1;
+		}
+		else
+		    pp->skipped_numbers++;
+	    }
+        } /* leaf */
+    } /* main loop */
+}
+
+#elif NEW_FORWARD == 2
+
+int isamb_pp_forward (ISAMB_PP pp, void *buf, const void *untilb)
+{
+    char *dst = buf;
+    char *src;
+    struct ISAMB_block *p = pp->block[pp->level];
+    if (!p)
+        return 0;
+
+again:
+    while (p->offset == p->size)
+    {
+        int pos, item_len;
+        while (p->offset == p->size)
+        {
+            if (pp->level == 0)
+                return 0;
+            close_block (pp->isamb, pp->block[pp->level]);
+            pp->block[pp->level] = 0;
+            (pp->level)--;
+            p = pp->block[pp->level];
+            assert (!p->leaf);  
+        }
+
+	assert(!p->leaf);
+        src = p->bytes + p->offset;
+        
+        decode_ptr (&src, &item_len);
+        src += item_len;
+        decode_ptr (&src, &pos);
+        
+        p->offset = src - (char*) p->bytes;
+
+	src = p->bytes + p->offset;
+
+	while(1)
+	{
+	    if (!untilb || p->offset == p->size)
+		break;
+	    assert(p->offset < p->size);
+	    decode_ptr (&src, &item_len);
+	    if ((*pp->isamb->method->compare_item)(untilb, src) <= 1)
+		break;
+	    src += item_len;
+	    decode_ptr (&src, &pos);
+	    p->offset = src - (char*) p->bytes;
+	}
+
+	pp->level++;
+
+        while (1)
+        {
+            pp->block[pp->level] = p = open_block (pp->isamb, pos);
+
+            pp->total_size += p->size;
+            pp->no_blocks++;
+            
+            if (p->leaf) 
+            {
+                break;
+            }
+	    
+            src = p->bytes + p->offset;
+	    while(1)
+	    {
+		decode_ptr (&src, &pos);
+		p->offset = src - (char*) p->bytes;
+		
+		if (!untilb || p->offset == p->size)
+		    break;
+		assert(p->offset < p->size);
+		decode_ptr (&src, &item_len);
+		if ((*pp->isamb->method->compare_item)(untilb, src) <= 1)
+		    break;
+		src += item_len;
+	    }
+            pp->level++;
+        }
+    }
+    assert (p->offset < p->size);
+    assert (p->leaf);
+    while(1)
+    {
+	char *dst0 = dst;
+        src = p->bytes + p->offset;
+        (*pp->isamb->method->code_item)(ISAMC_DECODE, p->decodeClientData,
+                                    &dst, &src);
+        p->offset = src - (char*) p->bytes;
+        if (!untilb || (*pp->isamb->method->compare_item)(untilb, dst0) <= 1)
+	    break;
+	dst = dst0;
+	if (p->offset == p->size) goto again;
+    }
+    /* key_logdump_txt(LOG_DEBUG,buf, "isamb_pp_read returning 1"); */
+    return 1;
+}
+
+#endif
+
 int isamb_pp_num (ISAMB_PP pp)
 {
     return 1;
 }
+
+static void isamb_pp_leaf_pos( ISAMB_PP pp, 
+                               int *current, int *total, void *dummybuf )
+{
+    struct ISAMB_block *p = pp->block[pp->level];
+    char *src=p->bytes;
+    char *end=p->bytes+p->size;
+    char *cur=p->bytes+p->offset;
+    char *dst;
+    assert(p->offset <= p->size);
+    assert(cur <= end);
+    assert(p->leaf);
+    *current=0;
+    *total=0;
+
+    while(src < end) {
+        dst=dummybuf;
+        (*pp->isamb->method->code_item)
+           (ISAMC_DECODE, p->decodeClientData,&dst, &src);
+        assert(dst<dummybuf+100); /*FIXME */
+        (*total)++;
+        if (src<=cur)
+             (*current)++;
+    }
+    logf(LOG_DEBUG, "isamb_pp_leaf_pos: cur=%d tot=%d ofs=%d sz=%d lev=%d",
+                    *current, *total, p->offset, p->size, pp->level);
+    assert(src==end);
+}
+
+static void isamb_pp_upper_pos( ISAMB_PP pp, int *current, int *total, 
+                                int size, int level )
+{ /* estimates total/current occurrences from here up, excl leaf */
+    struct ISAMB_block *p = pp->block[level];
+    char *src=p->bytes;
+    char *end=p->bytes+p->size;
+    char *cur=p->bytes+p->offset;
+    int item_size;
+    int child;
+    assert(level>=0);
+    assert(!p->leaf);
+    logf(LOG_DEBUG,"isamb_pp_upper_pos at beginning     l=%d "
+                   "cur=%d tot=%d ofs=%d sz=%d pos=%d", 
+                   level, *current, *total, p->offset, p->size, p->pos);
+    assert (p->offset <= p->size);
+	decode_ptr (&src, &child ); /* first child */
+    while(src < end) {
+        if (src!=cur) {
+            *total += size;
+            if (src < cur)
+                *current +=size;
+        }
+	    decode_ptr (&src, &item_size ); 
+        assert(src+item_size<=end);
+        src += item_size;
+	    decode_ptr (&src, &child );
+    }
+    if (level>0)
+        isamb_pp_upper_pos(pp, current, total, *total, level-1);
+} /* upper_pos */
+
+void isamb_pp_pos( ISAMB_PP pp, int *current, int *total )
+{ /* return an estimate of the current position and of the total number of */
+  /* occureences in the isam tree, based on the current leaf */
+    struct ISAMB_block *p = pp->block[pp->level];
+    char dummy[100]; /* 100 bytes/entry must be enough */
+    assert(total);
+    assert(current);
+    assert(p->leaf);
+    isamb_pp_leaf_pos(pp,current, total, dummy);
+    if (pp->level>0)
+        isamb_pp_upper_pos(pp, current, total, *total, pp->level-1);
+    /*
+    logf(LOG_DEBUG,"isamb_pp_pos: C=%d T=%d =%6.2f%%",
+                    *current, *total, 100.0*(*current)/(*total));
+    */
+}