X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=isamb%2Fisamb.c;h=343e5e3794e18de0d20a4e6edac4a01625db3c30;hb=1dec188ccbdafb0ec7b7c98857d701a7b77d46a1;hp=4f55b02d895408bc5b57fa7d7940098bb4c3672f;hpb=adfd6d1b6afa9964e4483a16dcf8c2f64d20989a;p=idzebra-moved-to-github.git

diff --git a/isamb/isamb.c b/isamb/isamb.c
index 4f55b02..343e5e3 100644
--- a/isamb/isamb.c
+++ b/isamb/isamb.c
@@ -1,8 +1,8 @@
 /*
- *  Copyright (c) 1995-1998, Index Data.
+ *  Copyright (c) 2000-2002, Index Data.
  *  See the file LICENSE for details.
  *
- *  $Id: isamb.c,v 1.8 2002-04-23 13:39:10 adam Exp $
+ *  $Id: isamb.c,v 1.17 2002-06-19 10:29:17 adam Exp $
  */
 #include <yaz/xmalloc.h>
 #include <yaz/log.h>
@@ -39,7 +39,7 @@ struct ISAMB_s {
 };
 
 struct ISAMB_block {
-    int pos;
+    ISAMB_P pos;
     int cat;
     int size;
     int leaf;
@@ -52,7 +52,10 @@ struct ISAMB_block {
 
 struct ISAMB_PP_s {
     ISAMB isamb;
+    ISAMB_P pos;
     int level;
+    int total_size;
+    int no_blocks;
     struct ISAMB_block **block;
 };
 
@@ -71,7 +74,7 @@ void decode_ptr (char **src, int *pos)
 ISAMB isamb_open (BFiles bfs, const char *name, int writeflag, ISAMC_M method)
 {
     ISAMB isamb = xmalloc (sizeof(*isamb));
-    int i, b_size = 64;
+    int i, b_size = 32;
 
     isamb->bfs = bfs;
     isamb->method = (ISAMC_M) xmalloc (sizeof(*method));
@@ -82,18 +85,21 @@ ISAMB isamb_open (BFiles bfs, const char *name, int writeflag, ISAMC_M method)
     for (i = 0; i<isamb->no_cat; i++)
     {
         char fname[DST_BUF_SIZE];
-        isamb->file[i].head.first_block = 1;
-        isamb->file[i].head.last_block = 1;
-        isamb->file[i].head.block_size = b_size;
-        isamb->file[i].head.block_max = b_size - ISAMB_DATA_OFFSET;
-        b_size = b_size * 4;
         isamb->file[i].head_dirty = 0;
-        sprintf (fname, "%s-%d", name, i);
-        isamb->file[i].bf =
-            bf_open (bfs, fname, isamb->file[i].head.block_size, writeflag);
+        sprintf (fname, "%s%c", name, i+'A');
+        isamb->file[i].bf = bf_open (bfs, fname, b_size, writeflag);
     
-        bf_read (isamb->file[i].bf, 0, 0, sizeof(struct ISAMB_head),
-                 &isamb->file[i].head);
+        if (!bf_read (isamb->file[i].bf, 0, 0, sizeof(struct ISAMB_head),
+                 &isamb->file[i].head))
+	{
+            isamb->file[i].head.first_block = 1;
+            isamb->file[i].head.last_block = 1;
+            isamb->file[i].head.block_size = b_size;
+            isamb->file[i].head.block_max = b_size - ISAMB_DATA_OFFSET;
+	}
+        assert (isamb->file[i].head.block_size >= ISAMB_DATA_OFFSET);
+        isamb->file[i].head_dirty = 0;
+        b_size = b_size * 4;
     }
     return isamb;
 }
@@ -122,7 +128,12 @@ struct ISAMB_block *open_block (ISAMB b, ISAMC_P pos)
     p->pos = pos;
     p->cat = pos & 3;
     p->buf = xmalloc (b->file[cat].head.block_size);
-    bf_read (b->file[cat].bf, pos/4, 0, 0, p->buf);
+    if (!bf_read (b->file[cat].bf, pos/4, 0, 0, p->buf))
+    {
+        yaz_log (LOG_FATAL, "read failure for pos=%ld block=%ld",
+                 (long) pos, (long) pos/4);
+        abort();
+    }
     p->bytes = p->buf + ISAMB_DATA_OFFSET;
     p->leaf = p->buf[0];
     p->size = p->buf[1] + 256 * p->buf[2] - ISAMB_DATA_OFFSET;
@@ -183,15 +194,16 @@ void close_block (ISAMB b, struct ISAMB_block *p)
 }
 
 int insert_sub (ISAMB b, struct ISAMB_block **p,
-                void *new_item,
+                void *new_item, int *mode,
                 ISAMC_I stream,
                 struct ISAMB_block **sp,
                 void *sub_item, int *sub_size,
                 void *max_item);
 
 int insert_int (ISAMB b, struct ISAMB_block *p, void *lookahead_item,
+                int *mode,
                 ISAMC_I stream, struct ISAMB_block **sp,
-                void *split_item, int *split_size)
+                void *split_item, int *split_size, void *last_max_item)
 {
     char *startp = p->bytes;
     char *src = startp;
@@ -215,7 +227,8 @@ int insert_int (ISAMB b, struct ISAMB_block *p, void *lookahead_item,
         {
             sub_p1 = open_block (b, pos);
             assert (sub_p1);
-            more = insert_sub (b, &sub_p1, lookahead_item, stream, &sub_p2, 
+            more = insert_sub (b, &sub_p1, lookahead_item, mode,
+                               stream, &sub_p2, 
                                sub_item, &sub_size, src);
             break;
         }
@@ -226,8 +239,8 @@ int insert_int (ISAMB b, struct ISAMB_block *p, void *lookahead_item,
     {
         sub_p1 = open_block (b, pos);
         assert (sub_p1);
-        more = insert_sub (b, &sub_p1, lookahead_item, stream, &sub_p2, 
-                           sub_item, &sub_size, 0);
+        more = insert_sub (b, &sub_p1, lookahead_item, mode, stream, &sub_p2, 
+                           sub_item, &sub_size, last_max_item);
     }
     if (sub_p2)
     {
@@ -283,8 +296,6 @@ int insert_int (ISAMB b, struct ISAMB_block *p, void *lookahead_item,
             (*sp)->size = endp - src;
             memcpy ((*sp)->bytes, src, (*sp)->size);
 
-            yaz_log (LOG_LOG, "i split %d -> %d %d",
-                     p->size, p_new_size, (*sp)->size);
             p->size = p_new_size;
         }
         p->dirty = 1;
@@ -296,7 +307,7 @@ int insert_int (ISAMB b, struct ISAMB_block *p, void *lookahead_item,
 
 
 int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
-                 ISAMC_I stream, struct ISAMB_block **sp2,
+                 int *lookahead_mode, ISAMC_I stream, struct ISAMB_block **sp2,
                  void *sub_item, int *sub_size,
                  void *max_item)
 {
@@ -328,17 +339,23 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
             char *dst_item = 0;
             char *dst_0 = dst;
             char *lookahead_next;
-            int lookahead_mode;
             int d = -1;
             
             if (lookahead_item)
                 d = (*b->method->compare_item)(file_item_buf, lookahead_item);
             
-            if (d > 0)  
+            if (d > 0)
+            {
                 dst_item = lookahead_item;
+                assert (*lookahead_mode);
+            }
             else
                 dst_item = file_item_buf;
-            if (!half1 && dst > cut)   
+            if (!*lookahead_mode && d == 0)
+            {
+                p->dirty = 1;
+            }
+            else if (!half1 && dst > cut)
             {
                 char *dst_item_0 = dst_item;
                 half1 = dst; /* candidate for splitting */
@@ -364,15 +381,15 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
                     lookahead_next = lookahead_item;
                     if (!(*stream->read_item)(stream->clientData,
                                               &lookahead_next,
-                                              &lookahead_mode))
+                                              lookahead_mode))
                     {
                         lookahead_item = 0;
                         more = 0;
                     }
-                    if (max_item &&
+                    if (lookahead_item && max_item &&
                         (*b->method->compare_item)(max_item, lookahead_item) <= 0)
                     {
-                        assert (0);
+                        /* max_item 1 */
                         lookahead_item = 0;
                     }
                     
@@ -383,7 +400,7 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
             {
                 lookahead_next = lookahead_item;
                 if (!(*stream->read_item)(stream->clientData,
-                                          &lookahead_next, &lookahead_mode))
+                                          &lookahead_next, lookahead_mode))
                 {
                     lookahead_item = 0;
                     more = 0;
@@ -405,17 +422,21 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
     maxp = dst_buf + b->file[b->no_cat-1].head.block_max + quater;
     while (lookahead_item)
     {
-        int lookahead_mode;
         char *dst_item = lookahead_item;
         char *dst_0 = dst;
         
         if (max_item &&
             (*b->method->compare_item)(max_item, lookahead_item) <= 0)
         {
-            assert (0);
+            /* max_item 2 */
             break;
         }
-        if (!half1 && dst > cut)   
+        if (!*lookahead_mode)
+        {
+            yaz_log (LOG_WARN, "Inconsistent register (2)");
+            abort();
+        }
+        else if (!half1 && dst > cut)   
         {
             char *dst_item_0 = dst_item;
             half1 = dst; /* candidate for splitting */
@@ -439,7 +460,7 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
             p->dirty = 1;
         dst_item = lookahead_item;
         if (!(*stream->read_item)(stream->clientData, &dst_item,
-                                  &lookahead_mode))
+                                  lookahead_mode))
         {
             lookahead_item = 0;
             more = 0;
@@ -492,8 +513,6 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
         p->dirty = 1;
         memcpy (sub_item, cut_item_buf, cut_item_size);
         *sub_size = cut_item_size;
-
-        yaz_log (LOG_LOG, "l split %d / %d", p->size, (*sp2)->size);
     }
     else
     {
@@ -507,16 +526,18 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
 }
 
 int insert_sub (ISAMB b, struct ISAMB_block **p, void *new_item,
+                int *mode,
                 ISAMC_I stream,
                 struct ISAMB_block **sp,
                 void *sub_item, int *sub_size,
                 void *max_item)
 {
     if (!*p || (*p)->leaf)
-        return insert_leaf (b, p, new_item, stream, sp, sub_item, sub_size, 
-                            max_item);
+        return insert_leaf (b, p, new_item, mode, stream, sp, sub_item, 
+                            sub_size, max_item);
     else
-        return insert_int (b, *p, new_item, stream, sp, sub_item, sub_size);
+        return insert_int (b, *p, new_item, mode, stream, sp, sub_item,
+                           sub_size, max_item);
 }
 
 int isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I stream)
@@ -536,7 +557,7 @@ int isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I stream)
         
         if (pos)
             p = open_block (b, pos);
-        more = insert_sub (b, &p, item_buf, stream, &sp,
+        more = insert_sub (b, &p, item_buf, &i_mode, stream, &sp,
                             sub_item, &sub_size, 0);
         if (sp)
         {    /* increase level of tree by one */
@@ -562,21 +583,27 @@ int isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I stream)
     return pos;
 }
 
-ISAMB_PP isamb_pp_open (ISAMB isamb, ISAMB_P pos)
+ISAMB_PP isamb_pp_open_x (ISAMB isamb, ISAMB_P pos, int *level)
 {
     ISAMB_PP pp = xmalloc (sizeof(*pp));
 
     pp->isamb = isamb;
     pp->block = xmalloc (10 * sizeof(*pp->block));
 
+    pp->pos = pos;
     pp->level = 0;
+    pp->total_size = 0;
+    pp->no_blocks = 0;
     while (1)
     {
         struct ISAMB_block *p = open_block (isamb, pos);
         char *src = p->bytes + p->offset;
         pp->block[pp->level] = p;
 
-        if (p->bytes[0]) /* leaf */
+        pp->total_size += p->size;
+        pp->no_blocks++;
+
+        if (p->leaf)
             break;
 
         decode_ptr (&src, &pos);
@@ -584,20 +611,43 @@ ISAMB_PP isamb_pp_open (ISAMB isamb, ISAMB_P pos)
         pp->level++;
     }
     pp->block[pp->level+1] = 0;
+    if (level)
+        *level = pp->level;
     return pp;
 }
 
-void isamb_pp_close (ISAMB_PP pp)
+ISAMB_PP isamb_pp_open (ISAMB isamb, ISAMB_P pos)
+{
+    return isamb_pp_open_x (isamb, pos, 0);
+}
+
+void isamb_pp_close_x (ISAMB_PP pp, int *size, int *blocks)
 {
     int i;
     if (!pp)
         return;
+    if (size)
+        *size = pp->total_size;
+    if (blocks)
+        *blocks = pp->no_blocks;
     for (i = 0; i <= pp->level; i++)
         close_block (pp->isamb, pp->block[i]);
     xfree (pp->block);
     xfree (pp);
 }
 
+int isamb_block_info (ISAMB isamb, int cat)
+{
+    if (cat >= 0 && cat < isamb->no_cat)
+        return isamb->file[cat].head.block_size;
+    return -1;
+}
+
+void isamb_pp_close (ISAMB_PP pp)
+{
+    return isamb_pp_close_x (pp, 0, 0);
+}
+
 int isamb_pp_read (ISAMB_PP pp, void *buf)
 {
     char *dst = buf;
@@ -617,7 +667,7 @@ int isamb_pp_read (ISAMB_PP pp, void *buf)
             pp->block[pp->level] = 0;
             (pp->level)--;
             p = pp->block[pp->level];
-            assert (p->bytes[0] == 0);  /* must be int */
+            assert (!p->leaf);  /* must be int */
         }
         src = p->bytes + p->offset;
         
@@ -632,6 +682,9 @@ int isamb_pp_read (ISAMB_PP pp, void *buf)
         while (1)
         {
             pp->block[pp->level] = p = open_block (pp->isamb, pos);
+
+            pp->total_size += p->size;
+            pp->no_blocks++;
             
             if (p->leaf) /* leaf */
             {
@@ -644,7 +697,7 @@ int isamb_pp_read (ISAMB_PP pp, void *buf)
         }
     }
     assert (p->offset < p->size);
-    assert (p->bytes[0]);
+    assert (p->leaf);
     src = p->bytes + p->offset;
     (*pp->isamb->method->code_item)(ISAMC_DECODE, p->decodeClientData,
                                     &dst, &src);