X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=isamc%2Fisamd.c;h=8fb7dcaec219e5f243522bc3556fda3ee5421010;hb=e5e4f1de02396b2cfab488f8e5918853e96d6b53;hp=8e711aead10255a3638976d616ad3cc3032a0f8e;hpb=b918342ded2c57bcd57bf7af6748c17bce4d1cf6;p=idzebra-moved-to-github.git

diff --git a/isamc/isamd.c b/isamc/isamd.c
index 8e711ae..8fb7dca 100644
--- a/isamc/isamd.c
+++ b/isamc/isamd.c
@@ -1,15 +1,26 @@
-/*
- * Copyright (c) 1995-1998, Index Data.
- * See the file LICENSE for details.
- * $Id: isamd.c,v 1.12 1999-09-13 13:28:28 heikki Exp $ 
- *
- * Isamd - isam with diffs 
- * Programmed by: Heikki Levanto
- *
- * Todo
- *  - Statistics are missing and/or completely wrong
- *  - Lots of code stolen from isamc, not all needed any more
- */
+/* $Id: isamd.c,v 1.23 2002-08-02 19:26:56 adam Exp $
+   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
+   Index Data Aps
+
+This file is part of the Zebra server.
+
+Zebra is free software; you can redistribute it and/or modify it under
+the terms of the GNU General Public License as published by the Free
+Software Foundation; either version 2, or (at your option) any later
+version.
+
+Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
+WARRANTY; without even the implied warranty of MERCHANTABILITY or
+FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+for more details.
+
+You should have received a copy of the GNU General Public License
+along with Zebra; see the file LICENSE.zebra.  If not, write to the
+Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
+02111-1307, USA.
+*/
+
+
 
 
 #include <stdlib.h>
@@ -17,7 +28,7 @@
 #include <string.h>
 #include <stdio.h>
 
-#include <log.h>
+#include <yaz/log.h>
 #include "../index/index.h"  /* isamd uses the internal structure of it_key */
 #include "isamd-p.h"
 
@@ -38,22 +49,36 @@ ISAMD_M isamd_getmethod (ISAMD_M me)
 	{    64,    0 },
 #else
         {    32,    1 },
+	{   128,    1 },
+	{   256,    1 },
+	{   512,    1 },
+        {  1024,    1 },
+        {  2048,    1 },
+        {  4096,    1 },
+        {  8192,    0 },
+
+#endif
+#ifdef SKIPTHIS
+
+
+
+        {    32,    1 },
         {   128,    1 },
         {   512,    1 },
         {  2048,    1 },
         {  8192,    1 },
         { 32768,    1 },
         {131072,    0 },
-#endif 
 
-/* old values from isamc, long time ago...
-        {    24,   40 },
-	{   128,  256 },
-        {   512, 1024 },
-        {  2048, 4096 },
-        {  8192,16384 },
-        { 32768,   0  },
-*/
+        {    24,    1 }, /* Experimental sizes */
+        {    32,    1 },
+        {    64,    1 },
+        {   128,    1 },
+        {   256,    1 },
+        {   512,    1 },
+        {  1024,    1 },
+        {  2048,    0 },
+#endif 
 
     };
     ISAMD_M m = (ISAMD_M) xmalloc (sizeof(*m));  /* never released! */
@@ -136,21 +161,38 @@ ISAMD isamd_open (BFiles bfs, const char *name, int writeflag, ISAMD_M method)
 	is->files[i].sum_backward = 0;
 	is->files[i].no_next = 0;
 	is->files[i].no_prev = 0;
-        is->files[i].no_op_nodiff=0;
-        is->files[i].no_op_intdiff=0;
-        is->files[i].no_op_extdiff=0;
-        is->files[i].no_fbuilds=0;   
-        is->files[i].no_appds=0;     
-        is->files[i].no_merges=0;    
-        is->files[i].no_remerges=0;  
-
+        is->files[i].no_op_diffonly=0;
+        is->files[i].no_op_main=0;
         init_fc (is, i);
     }
+    is->last_pos=0;
+    is->last_cat=0;   
+    is->no_read=0;    
+    is->no_read_main=0;
+    is->no_write=0;   
+    is->no_op_single=0;
+    is->no_op_new=0;
+    is->no_read_keys=0;
+    is->no_read_eof=0;
+    is->no_seek_nxt=0;
+    is->no_seek_sam=0;
+    is->no_seek_fwd=0;
+    is->no_seek_prv=0;
+    is->no_seek_bak=0;
+    is->no_seek_cat=0;
+    is->no_fbuilds=0;
+    is->no_appds=0;
+    is->no_merges=0;
+    is->no_non=0;
+    is->no_singles=0;
+
     return is;
 }
 
 int isamd_block_used (ISAMD is, int type)
 {
+    if ( type==-1) /* singleton */
+      return 0; 
     if (type < 0 || type >= is->no_files)
 	return -1;
     return is->files[type].head.lastblock-1;
@@ -159,6 +201,8 @@ int isamd_block_used (ISAMD is, int type)
 int isamd_block_size (ISAMD is, int type)
 {
     ISAMD_filecat filecat = is->method->filecat;
+    if ( type==-1) /* singleton */
+      return 0; /* no bytes used */ 
     if (type < 0 || type >= is->no_files)
 	return -1;
     return filecat[type].bsize;
@@ -167,23 +211,24 @@ int isamd_block_size (ISAMD is, int type)
 int isamd_close (ISAMD is)
 {
     int i;
+    int s;
 
     if (is->method->debug>0)
     {
         logf (LOG_LOG, "isamd statistics");
-	logf (LOG_LOG, "f      nxt    forw   mid-f    prev   backw   mid-b");
+	logf (LOG_LOG, "f    nxt   forw  mid-f   prev  backw  mid-b");
 	for (i = 0; i<is->no_files; i++)
-	    logf (LOG_LOG, "%d%8d%8d%8.1f%8d%8d%8.1f",i,
+	    logf (LOG_LOG, "%d%7d%7d%7.1f%7d%7d%7.1f",i,
 		  is->files[i].no_next,
 		  is->files[i].no_forward,
 		  is->files[i].no_forward ?
-		  (double) is->files[i].sum_forward/is->files[i].no_forward
-		  : 0.0,
+		    (double) is->files[i].sum_forward/is->files[i].no_forward
+		    : 0.0,
 		  is->files[i].no_prev,
 		  is->files[i].no_backward,
 		  is->files[i].no_backward ?
-		  (double) is->files[i].sum_backward/is->files[i].no_backward
-		  : 0.0);
+		    (double) is->files[i].sum_backward/is->files[i].no_backward
+		    : 0.0);
     }
     if (is->method->debug>0)
         logf (LOG_LOG, "f  writes   reads skipped   alloc released ");
@@ -208,23 +253,50 @@ int isamd_close (ISAMD is)
     
     if (is->method->debug>0) 
     {
-        logf (LOG_LOG, "f   opens  simple     int     ext");
+        logf (LOG_LOG, "f   opens    main  diffonly");
         for (i = 0; i<is->no_files; i++)
         {
-            logf (LOG_LOG, "%d%8d%8d%8d%8d",i,
-                  is->files[i].no_op_nodiff+
-                  is->files[i].no_op_intdiff+
-                  is->files[i].no_op_extdiff,
-                  is->files[i].no_op_nodiff,
-                  is->files[i].no_op_intdiff,
-                  is->files[i].no_op_extdiff);
+            logf (LOG_LOG, "%d%8d%8d%8d",i,
+                  is->files[i].no_op_main+
+                  is->files[i].no_op_diffonly,
+                  is->files[i].no_op_main,
+                  is->files[i].no_op_diffonly);
         }
-        logf (LOG_LOG, "    build  append   merge   remrg");
-        logf (LOG_LOG, "=%8d%8d%8d%8d",
-                  is->files[0].no_fbuilds,
-                  is->files[0].no_appds,
-                  is->files[0].no_merges,
-                  is->files[0].no_remerges);
+        logf(LOG_LOG,"open single  %8d", is->no_op_single);
+        logf(LOG_LOG,"open new     %8d", is->no_op_new);
+
+        logf(LOG_LOG, "new build   %8d", is->no_fbuilds);
+        logf(LOG_LOG, "append      %8d", is->no_appds);
+        logf(LOG_LOG, "  merges    %8d", is->no_merges);
+        logf(LOG_LOG, "  singles   %8d", is->no_singles);
+        logf(LOG_LOG, "  no-ops    %8d", is->no_non);
+
+        logf(LOG_LOG, "read blocks %8d", is->no_read);
+        logf(LOG_LOG, "read keys:  %8d %8.1f k/bl", 
+                  is->no_read_keys, 
+                  1.0*(is->no_read_keys+1)/(is->no_read+1) );
+        logf(LOG_LOG, "read main-k %8d %8.1f %% of keys",
+                  is->no_read_main,
+                  100.0*(is->no_read_main+1)/(is->no_read_keys+1) );
+        logf(LOG_LOG, "read ends:  %8d %8.1f k/e",
+                  is->no_read_eof,
+                  1.0*(is->no_read_keys+1)/(is->no_read_eof+1) );
+        s= is->no_seek_nxt+ is->no_seek_sam+ is->no_seek_fwd +
+           is->no_seek_prv+ is->no_seek_bak+ is->no_seek_cat;
+        if (s==0) 
+          s++;
+        logf(LOG_LOG, "seek same   %8d %8.1f%%",
+            is->no_seek_sam, 100.0*is->no_seek_sam/s );
+        logf(LOG_LOG, "seek next   %8d %8.1f%%",
+            is->no_seek_nxt, 100.0*is->no_seek_nxt/s );
+        logf(LOG_LOG, "seek prev   %8d %8.1f%%",
+            is->no_seek_prv, 100.0*is->no_seek_prv/s );
+        logf(LOG_LOG, "seek forw   %8d %8.1f%%",
+            is->no_seek_fwd, 100.0*is->no_seek_fwd/s );
+        logf(LOG_LOG, "seek back   %8d %8.1f%%",
+            is->no_seek_bak, 100.0*is->no_seek_bak/s );
+        logf(LOG_LOG, "seek cat    %8d %8.1f%%",
+            is->no_seek_cat, 100.0*is->no_seek_cat/s );
     }
     xfree (is->files);
     xfree (is->method);
@@ -232,9 +304,29 @@ int isamd_close (ISAMD is)
     return 0;
 }
 
+static void isamd_seek_stat(ISAMD is, int cat, int pos)
+{
+  if (cat != is->last_cat)
+     is->no_seek_cat++;
+  else if ( pos == is->last_pos)
+     is->no_seek_sam++;
+  else if ( pos == is->last_pos+1)
+     is->no_seek_nxt++;
+  else if ( pos == is->last_pos-1)
+     is->no_seek_prv++;
+  else if ( pos > is->last_pos)
+     is->no_seek_fwd++;
+  else if ( pos < is->last_pos)
+     is->no_seek_bak++;
+  is->last_cat = cat;
+  is->last_pos = pos;
+} /* seek_stat */
+
 int isamd_read_block (ISAMD is, int cat, int pos, char *dst)
 {
+    isamd_seek_stat(is,cat,pos);
     ++(is->files[cat].no_reads);
+    ++(is->no_read);
     if (is->method->debug > 6)
         logf (LOG_LOG, "isamd: read_block %d:%d",cat, pos);
     return bf_read (is->files[cat].bf, pos, 0, 0, dst);
@@ -242,7 +334,9 @@ int isamd_read_block (ISAMD is, int cat, int pos, char *dst)
 
 int isamd_write_block (ISAMD is, int cat, int pos, char *src)
 {
+    isamd_seek_stat(is,cat,pos);
     ++(is->files[cat].no_writes);
+    ++(is->no_write);
     if (is->method->debug > 6)
         logf (LOG_LOG, "isamd: write_block %d:%d", cat, pos);
     return bf_write (is->files[cat].bf, pos, 0, 0, src);
@@ -464,26 +558,56 @@ void isamd_pp_close (ISAMD_PP pp)
 
     (*is->method->code_stop)(ISAMD_DECODE, pp->decodeClientData);
     isamd_free_diffs(pp);  /* see merge-d.h */
-    xfree (pp->buf);
-    xfree (pp);
     if (is->method->debug > 5)
-       logf (LOG_LOG, "isamd_pp_close %p %d=%d:%d  sz=%d n=%d=%d:%d",
+       logf (LOG_LOG, "isamd_pp_close %p %d=%d:%d  sz=%d n=%d=%d:%d nk=%d",
              pp, isamd_addr(pp->pos, pp->cat), pp->cat, pp->pos, pp->size, 
-             pp->next, isamd_type(pp->next), isamd_block(pp->next) );
+             pp->next, isamd_type(pp->next), isamd_block(pp->next), 
+             pp->numKeys );
+    xfree (pp->buf);
+    xfree (pp);
 }
 
 
+ISAMD_PP isamd_pp_create (ISAMD is, int cat)
+/* creates a pp_buff without data in it. pos=0, cat as given */
+{
+    ISAMD_PP pp = (ISAMD_PP) xmalloc (sizeof(*pp));
+    int sz = is->method->filecat[is->max_cat].bsize;
+
+    pp->numKeys = 0;
+    pp->buf = (char *) xmalloc (sz);
+    memset(pp->buf,'\0',sz); /* clear the buffer, for new blocks */
+    
+    pp->next = 0;
+    pp->size = 0;
+    pp->offset = 0;
+    pp->is = is;
+    pp->diffs=0;
+    pp->diffbuf=0;
+    pp->diffinfo=0;
+    pp->decodeClientData = (*is->method->code_start)(ISAMD_DECODE);
+    pp->cat = cat;
+    pp->pos = 0;
+    is->no_op_new++; 
+    return pp;
+      
+}
+
 
-ISAMD_PP isamd_pp_open (ISAMD is, ISAMD_P ipos)
+ISAMD_PP isamd_pp_open (ISAMD is, const char *dictbuf, int dictlen)
 {
+    ISAMD_P ipos;
     ISAMD_PP pp = (ISAMD_PP) xmalloc (sizeof(*pp));
     char *src;
     int sz = is->method->filecat[is->max_cat].bsize;
                  /* always allocate for the largest blocks, saves trouble */
-   
-    pp->cat = isamd_type(ipos);
-    pp->pos = isamd_block(ipos); 
-
+    struct it_key singlekey;
+    char *c_ptr; /* for fake encoding the singlekey */
+    char *i_ptr;
+    int ofs;
+    int dictnum;
+    
+    pp->numKeys = 0;
     src = pp->buf = (char *) xmalloc (sz);
     memset(src,'\0',sz); /* clear the buffer, for new blocks */
     
@@ -491,12 +615,36 @@ ISAMD_PP isamd_pp_open (ISAMD is, ISAMD_P ipos)
     pp->size = 0;
     pp->offset = 0;
     pp->is = is;
-    pp->decodeClientData = (*is->method->code_start)(ISAMD_DECODE);
-    pp->numKeys = 0;
     pp->diffs=0;
     pp->diffbuf=0;
     pp->diffinfo=0;
+    pp->decodeClientData = (*is->method->code_start)(ISAMD_DECODE);
     
+    dictnum=*dictbuf;  // numkeys for internals, 0 for externals
+
+    if (0==dictnum)
+    {
+        memcpy(&ipos, dictbuf+1, sizeof(ISAMD_P) );
+    }
+    else /* dictionary block, fake a real one */
+    {
+       pp->cat=0; 
+       pp->pos=0;
+       if (is->method->debug > 5)
+          logf (LOG_LOG, "isamd_pp_open dict");
+       pp->numKeys=(unsigned char) dictbuf[0];
+       memcpy(pp->buf+ISAMD_BLOCK_OFFSET_1, dictbuf+1,dictlen-1);
+       pp->size=pp->offset=dictlen+ISAMD_BLOCK_OFFSET_1-1;
+       is->no_op_single++;
+       return pp;
+    } /* dict block */
+   
+    pp->cat = isamd_type(ipos);
+    pp->pos = isamd_block(ipos); 
+    
+    if (0==pp->pos)
+      is->no_op_new++; 
+      
     if (pp->pos)
     {
         src = pp->buf;
@@ -507,27 +655,20 @@ ISAMD_PP isamd_pp_open (ISAMD is, ISAMD_P ipos)
         src += sizeof(pp->size);
         memcpy (&pp->numKeys, src, sizeof(pp->numKeys));
         src += sizeof(pp->numKeys);
-        memcpy (&pp->diffs, src, sizeof(pp->diffs));
-        src += sizeof(pp->diffs);
-        assert (pp->next != pp->pos);
+        assert (pp->next != isamd_addr(pp->pos,pp->cat));
         pp->offset = src - pp->buf; 
         assert (pp->offset == ISAMD_BLOCK_OFFSET_1);
-        if (0==pp->diffs)
-           ++(is->files[pp->cat].no_op_nodiff);
+        assert(pp->size>=ISAMD_BLOCK_OFFSET_1); /*??*/
+        if (pp->next)
+          is->files[pp->cat].no_op_main++;
         else
-           if(pp->diffs&1)
-               ++(is->files[pp->cat].no_op_extdiff);
-           else
-               ++(is->files[pp->cat].no_op_intdiff);
-      //  if (!pp->diffbuf)
-      //    pp->diffbuf=pp->buf;
+          is->files[pp->cat].no_op_diffonly++;
     }
     if (is->method->debug > 5)
        logf (LOG_LOG, "isamd_pp_open  %p %d=%d:%d  sz=%d n=%d=%d:%d",
              pp, isamd_addr(pp->pos, pp->cat), pp->cat, pp->pos, pp->size, 
              pp->next, isamd_type(pp->next), isamd_block(pp->next) );
-    
-      
+
     return pp;
 }
 
@@ -536,22 +677,19 @@ ISAMD_PP isamd_pp_open (ISAMD is, ISAMD_P ipos)
 void isamd_buildfirstblock(ISAMD_PP pp){
   char *dst=pp->buf;
   assert(pp->buf);
-  assert(pp->next != pp->pos); 
+  assert(pp->next != isamd_addr(pp->pos,pp->cat)); 
   memcpy(dst, &pp->next, sizeof(pp->next) );
   dst += sizeof(pp->next);
   memcpy(dst, &pp->size,sizeof(pp->size));
   dst += sizeof(pp->size);
   memcpy(dst, &pp->numKeys, sizeof(pp->numKeys));
   dst += sizeof(pp->numKeys);
-  memcpy(dst, &pp->diffs, sizeof(pp->diffs));
-  dst += sizeof(pp->diffs);  
   assert (dst - pp->buf  == ISAMD_BLOCK_OFFSET_1);
   if (pp->is->method->debug > 5)
-     logf (LOG_LOG, "isamd: first: sz=%d  p=%d/%d>%d/%d nk=%d d=%d",
-           pp->size, 
-           pp->cat, pp->pos, 
-           isamd_type(pp->next), isamd_block(pp->next),
-           pp->numKeys, pp->diffs);
+     logf (LOG_LOG, "isamd: bldfirst:  p=%d=%d:%d n=%d:%d:%d sz=%d nk=%d ",
+           isamd_addr(pp->pos,pp->cat),pp->cat, pp->pos, 
+           pp->next, isamd_type(pp->next), isamd_block(pp->next),
+           pp->size, pp->numKeys);
 }
 
 void isamd_buildlaterblock(ISAMD_PP pp){
@@ -575,9 +713,10 @@ void isamd_buildlaterblock(ISAMD_PP pp){
 /* returns non-zero if item could be read; 0 otherwise */
 int isamd_pp_read (ISAMD_PP pp, void *buf)
 {
+
     return isamd_read_item (pp, (char **) &buf);
-    /* note: isamd_read_item is in merge-d.c, because it is so */
-    /* convoluted with the merge process */
+       /* note: isamd_read_item is in merge-d.c, because it is so */
+       /* convoluted with the merge process */
 }
 
 /* read one main item from file - decode and store it in *dst.
@@ -624,7 +763,7 @@ int isamd_read_main_item (ISAMD_PP pp, char **dst)
 	newcat = isamd_type(pp->next);
         pp->pos = isamd_block(pp->next);
         pp->cat = isamd_type(pp->next);
-        
+        pp->is->no_read_main++;
         src = pp->buf;
 	/* read block and save 'next' and 'size' entry */
         isamd_read_block (is, pp->cat, pp->pos, src);
@@ -674,6 +813,8 @@ static char *hexdump(unsigned char *p, int len, char *buff) {
 }
 
 
+#ifdef SKIPTHIS
+  /* needs different arguments, or something */
 void isamd_pp_dump (ISAMD is, ISAMD_P ipos)
 {
   ISAMD_PP pp;
@@ -685,12 +826,14 @@ void isamd_pp_dump (ISAMD is, ISAMD_P ipos)
   int diffmax=1;
   int diffidx;
   char hexbuff[64];
+  int olddebug= is->method->debug;
+  is->method->debug=0; /* no debug logs while reading for dump */
   
   logf(LOG_LOG,"dumping isamd block %d (%d:%d)",
                   (int)ipos, isamd_type(ipos), isamd_block(ipos) );
   pp=isamd_pp_open(is,ipos);
-  logf(LOG_LOG,"numKeys=%d,  ofs=%d sz=%d d=%d",
-       pp->numKeys, pp->offset, pp->size, pp->diffs);
+  logf(LOG_LOG,"numKeys=%d,  ofs=%d sz=%d",
+       pp->numKeys, pp->offset, pp->size );
   diffidx=oldoffs= pp->offset;
   while ((diffidx < is->method->filecat[pp->cat].bsize) && (diffmax>0))
   {
@@ -731,11 +874,50 @@ void isamd_pp_dump (ISAMD is, ISAMD_P ipos)
   }
   /*!*/ /*TODO: dump diffs too!!! */
   isamd_pp_close(pp);
+  is->method->debug=olddebug;
 } /* dump */
 
+#endif
+
 /*
  * $Log: isamd.c,v $
- * Revision 1.12  1999-09-13 13:28:28  heikki
+ * Revision 1.23  2002-08-02 19:26:56  adam
+ * Towards GPL
+ *
+ * Revision 1.22  2002/07/12 18:12:21  heikki
+ * Isam-D now stores small entries directly in the dictionary.
+ * Needs more tuning and cleaning...
+ *
+ * Revision 1.21  2002/07/11 16:16:00  heikki
+ * Fixed a bug in isamd, failed to store a single key when its bits
+ * did not fit into a singleton.
+ *
+ * Revision 1.20  2002/06/19 10:29:18  adam
+ * align block sizes for isam sys. Better plot for test
+ *
+ * Revision 1.19  1999/11/30 13:48:04  adam
+ * Improved installation. Updated for inclusion of YAZ header files.
+ *
+ * Revision 1.18  1999/10/06 15:18:13  heikki
+ *
+ * Improving block sizes again
+ *
+ * Revision 1.17  1999/10/06 11:46:36  heikki
+ * mproved statistics on isam-d
+ *
+ * Revision 1.16  1999/10/05 09:57:40  heikki
+ * Tuning the isam-d (and fixed a small "detail")
+ *
+ * Revision 1.15  1999/09/27 14:36:36  heikki
+ * singletons
+ *
+ * Revision 1.14  1999/09/23 18:01:18  heikki
+ * singleton optimising
+ *
+ * Revision 1.13  1999/09/20 15:48:06  heikki
+ * Small changes
+ *
+ * Revision 1.12  1999/09/13 13:28:28  heikki
  * isam-d optimizing: merging input data in the same go
  *
  * Revision 1.11  1999/08/25 18:09:24  heikki
@@ -764,4 +946,4 @@ void isamd_pp_dump (ISAMD is, ISAMD_P ipos)
  * Copied from isamh, starting to change things...
  *
  *
- */
\ No newline at end of file
+ */