X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=isamc%2Fisamc.c;h=4f5c8bc6df383f928053071a5d1c075a34ef97de;hp=6d64d7940e74c862eb7a80538c9ef24c6bee1f42;hb=ef696645cc3b7e0f4027008d1dc589c0f0f90c1f;hpb=b9093505b17a074e79137ed64595c8269f77d330

diff --git a/isamc/isamc.c b/isamc/isamc.c
index 6d64d79..4f5c8bc 100644
--- a/isamc/isamc.c
+++ b/isamc/isamc.c
@@ -1,10 +1,40 @@
 /*
- * Copyright (c) 1995-1996, Index Data.
+ * Copyright (c) 1995-1998, Index Data.
  * See the file LICENSE for details.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: isamc.c,v $
- * Revision 1.8  1997-09-17 12:19:20  adam
+ * Revision 1.17  1999-05-26 07:49:14  adam
+ * C++ compilation.
+ *
+ * Revision 1.16  1998/05/27 14:32:03  adam
+ * Changed default block category layout.
+ *
+ * Revision 1.15  1998/05/20 10:12:25  adam
+ * Implemented automatic EXPLAIN database maintenance.
+ * Modified Zebra to work with ASN.1 compiled version of YAZ.
+ *
+ * Revision 1.14  1998/03/19 10:04:35  adam
+ * Minor changes.
+ *
+ * Revision 1.13  1998/03/18 09:23:55  adam
+ * Blocks are stored in chunks on free list - up to factor 2 in speed.
+ * Fixed bug that could occur in block category rearrangemen.
+ *
+ * Revision 1.12  1998/03/16 10:37:24  adam
+ * Added more statistics.
+ *
+ * Revision 1.11  1998/03/13 15:30:50  adam
+ * New functions isc_block_used and isc_block_size. Fixed 'leak'
+ * in isc_alloc_block.
+ *
+ * Revision 1.10  1998/03/11 11:18:18  adam
+ * Changed the isc_merge to take into account the mfill (minimum-fill).
+ *
+ * Revision 1.9  1998/03/06 13:54:02  adam
+ * Fixed two nasty bugs in isc_merge.
+ *
+ * Revision 1.8  1997/09/17 12:19:20  adam
  * Zebra version corresponds to YAZ version 1.4.
  * Changed Zebra server so that it doesn't depend on global common_resource.
  *
@@ -50,18 +80,30 @@
 #include <log.h>
 #include "isamc-p.h"
 
+static void flush_block (ISAMC is, int cat);
 static void release_fc (ISAMC is, int cat);
 static void init_fc (ISAMC is, int cat);
 
+#define ISAMC_FREELIST_CHUNK 1
+
+#define SMALL_TEST 0
+
 ISAMC_M isc_getmethod (void)
 {
     static struct ISAMC_filecat_s def_cat[] = {
-        {   32,    28,     0,    20 },
-        {  512,   490,   100,    20 },
-        { 4096,  3950,  1000,    20 },
-        {32768, 32000, 10000,     0 },
+#if SMALL_TEST
+        {    32,     28,      0,  3 },
+	{    64,     54,     30,  0 },
+#else
+        {    24,     22,     18,  10 },
+	{   128,    120,    100,  10 },
+        {   512,    490,    350,  10 },
+        {  2048,   1900,   1700,  10 },
+        {  8192,   8000,   7900,  10 },
+        { 32768,  32000,  31000,  0 },
+#endif
     };
-    ISAMC_M m = xmalloc (sizeof(*m));
+    ISAMC_M m = (ISAMC_M) xmalloc (sizeof(*m));
     m->filecat = def_cat;
 
     m->code_start = NULL;
@@ -85,9 +127,9 @@ ISAMC isc_open (BFiles bfs, const char *name, int writeflag, ISAMC_M method)
     int i = 0;
     int max_buf_size = 0;
 
-    is = xmalloc (sizeof(*is));
+    is = (ISAMC) xmalloc (sizeof(*is));
 
-    is->method = xmalloc (sizeof(*is->method));
+    is->method = (ISAMC_M) xmalloc (sizeof(*is->method));
     memcpy (is->method, method, sizeof(*method));
     filecat = is->method->filecat;
     assert (filecat);
@@ -114,10 +156,10 @@ ISAMC isc_open (BFiles bfs, const char *name, int writeflag, ISAMC_M method)
         logf (LOG_LOG, "isc: max_buf_size %d", max_buf_size);
     
     assert (is->no_files > 0);
-    is->files = xmalloc (sizeof(*is->files)*is->no_files);
+    is->files = (ISAMC_file) xmalloc (sizeof(*is->files)*is->no_files);
     if (writeflag)
     {
-        is->merge_buf = xmalloc (max_buf_size+256);
+        is->merge_buf = (char *) xmalloc (max_buf_size+256);
 	memset (is->merge_buf, 0, max_buf_size+256);
     }
     else
@@ -136,23 +178,65 @@ ISAMC isc_open (BFiles bfs, const char *name, int writeflag, ISAMC_M method)
             is->files[i].head.lastblock = 1;
             is->files[i].head.freelist = 0;
         }
+	is->files[i].alloc_entries_num = 0;
+	is->files[i].alloc_entries_max =
+	    is->method->filecat[i].bsize / sizeof(int) - 1;
+	is->files[i].alloc_buf = (char *)
+	    xmalloc (is->method->filecat[i].bsize);
         is->files[i].no_writes = 0;
         is->files[i].no_reads = 0;
         is->files[i].no_skip_writes = 0;
         is->files[i].no_allocated = 0;
         is->files[i].no_released = 0;
         is->files[i].no_remap = 0;
+	is->files[i].no_forward = 0;
+	is->files[i].no_backward = 0;
+	is->files[i].sum_forward = 0;
+	is->files[i].sum_backward = 0;
+	is->files[i].no_next = 0;
+	is->files[i].no_prev = 0;
 
         init_fc (is, i);
     }
     return is;
 }
 
+int isc_block_used (ISAMC is, int type)
+{
+    if (type < 0 || type >= is->no_files)
+	return -1;
+    return is->files[type].head.lastblock-1;
+}
+
+int isc_block_size (ISAMC is, int type)
+{
+    ISAMC_filecat filecat = is->method->filecat;
+    if (type < 0 || type >= is->no_files)
+	return -1;
+    return filecat[type].bsize;
+}
+
 int isc_close (ISAMC is)
 {
     int i;
 
     if (is->method->debug)
+    {
+	logf (LOG_LOG, "isc:    next    forw   mid-f    prev   backw   mid-b");
+	for (i = 0; i<is->no_files; i++)
+	    logf (LOG_LOG, "isc:%8d%8d%8.1f%8d%8d%8.1f",
+		  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,
+		  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);
+    }
+    if (is->method->debug)
         logf (LOG_LOG, "isc:  writes   reads skipped   alloc released  remap");
     for (i = 0; i<is->no_files; i++)
     {
@@ -170,10 +254,12 @@ int isc_close (ISAMC is)
                   is->files[i].no_released,
                   is->files[i].no_remap);
         xfree (is->files[i].fc_list);
+	flush_block (is, i);
         bf_close (is->files[i].bf);
     }
     xfree (is->files);
     xfree (is->merge_buf);
+    xfree (is->method);
     xfree (is);
     return 0;
 }
@@ -195,7 +281,7 @@ int isc_write_block (ISAMC is, int cat, int pos, char *src)
 int isc_write_dblock (ISAMC is, int cat, int pos, char *src,
                       int nextpos, int offset)
 {
-    unsigned short size = offset + ISAMC_BLOCK_OFFSET_N;
+    ISAMC_BLOCK_SIZE size = offset + ISAMC_BLOCK_OFFSET_N;
     if (is->method->debug > 2)
         logf (LOG_LOG, "isc: write_dblock. size=%d nextpos=%d",
               (int) size, nextpos);
@@ -205,6 +291,109 @@ int isc_write_dblock (ISAMC is, int cat, int pos, char *src,
     return isc_write_block (is, cat, pos, src);
 }
 
+#if ISAMC_FREELIST_CHUNK
+static void flush_block (ISAMC is, int cat)
+{
+    char *abuf = is->files[cat].alloc_buf;
+    int block = is->files[cat].head.freelist;
+    if (block && is->files[cat].alloc_entries_num)
+    {
+	memcpy (abuf, &is->files[cat].alloc_entries_num, sizeof(int));
+	bf_write (is->files[cat].bf, block, 0, 0, abuf);
+	is->files[cat].alloc_entries_num = 0;
+    }
+    xfree (abuf);
+}
+
+static int alloc_block (ISAMC is, int cat)
+{
+    int block = is->files[cat].head.freelist;
+    char *abuf = is->files[cat].alloc_buf;
+
+    (is->files[cat].no_allocated)++;
+
+    if (!block)
+    {
+        block = (is->files[cat].head.lastblock)++;   /* no free list */
+	is->files[cat].head_is_dirty = 1;
+    }
+    else
+    {
+	if (!is->files[cat].alloc_entries_num) /* read first time */
+	{
+	    bf_read (is->files[cat].bf, block, 0, 0, abuf);
+	    memcpy (&is->files[cat].alloc_entries_num, abuf,
+		    sizeof(is->files[cat].alloc_entries_num));
+	    assert (is->files[cat].alloc_entries_num > 0);
+	}
+	/* have some free blocks now */
+	assert (is->files[cat].alloc_entries_num > 0);
+	is->files[cat].alloc_entries_num--;
+	if (!is->files[cat].alloc_entries_num)  /* last one in block? */
+	{
+	    memcpy (&is->files[cat].head.freelist, abuf + sizeof(int),
+		    sizeof(int));
+	    is->files[cat].head_is_dirty = 1;
+
+	    if (is->files[cat].head.freelist)
+	    {
+		bf_read (is->files[cat].bf, is->files[cat].head.freelist,
+			 0, 0, abuf);
+		memcpy (&is->files[cat].alloc_entries_num, abuf,
+			sizeof(is->files[cat].alloc_entries_num));
+		assert (is->files[cat].alloc_entries_num);
+	    }
+	}
+	else
+	    memcpy (&block, abuf + sizeof(int) + sizeof(int) *
+		    is->files[cat].alloc_entries_num, sizeof(int));
+    }
+    return block;
+}
+
+static void release_block (ISAMC is, int cat, int pos)
+{
+    char *abuf = is->files[cat].alloc_buf;
+    int block = is->files[cat].head.freelist;
+
+    (is->files[cat].no_released)++;
+
+    if (block && !is->files[cat].alloc_entries_num) /* must read block */
+    {
+	bf_read (is->files[cat].bf, block, 0, 0, abuf);
+	memcpy (&is->files[cat].alloc_entries_num, abuf,
+		sizeof(is->files[cat].alloc_entries_num));
+	assert (is->files[cat].alloc_entries_num > 0);
+    }
+    assert (is->files[cat].alloc_entries_num <= is->files[cat].alloc_entries_max);
+    if (is->files[cat].alloc_entries_num == is->files[cat].alloc_entries_max)
+    {
+	assert (block);
+	memcpy (abuf, &is->files[cat].alloc_entries_num, sizeof(int));
+	bf_write (is->files[cat].bf, block, 0, 0, abuf);
+	is->files[cat].alloc_entries_num = 0;
+    }
+    if (!is->files[cat].alloc_entries_num) /* make new buffer? */
+    {
+	memcpy (abuf + sizeof(int), &block, sizeof(int));
+	is->files[cat].head.freelist = pos;
+	is->files[cat].head_is_dirty = 1; 
+    }
+    else
+    {
+	memcpy (abuf + sizeof(int) +
+		is->files[cat].alloc_entries_num*sizeof(int),
+		&pos, sizeof(int));
+    }
+    is->files[cat].alloc_entries_num++;
+}
+#else
+static void flush_block (ISAMC is, int cat)
+{
+    char *abuf = is->files[cat].alloc_buf;
+    xfree (abuf);
+}
+
 static int alloc_block (ISAMC is, int cat)
 {
     int block;
@@ -222,6 +411,18 @@ static int alloc_block (ISAMC is, int cat)
     return block;
 }
 
+static void release_block (ISAMC is, int cat, int pos)
+{
+    char buf[sizeof(int)];
+   
+    (is->files[cat].no_released)++;
+    is->files[cat].head_is_dirty = 1; 
+    memcpy (buf, &is->files[cat].head.freelist, sizeof(int));
+    is->files[cat].head.freelist = pos;
+    bf_write (is->files[cat].bf, pos, 0, sizeof(int), buf);
+}
+#endif
+
 int isc_alloc_block (ISAMC is, int cat)
 {
     int block = 0;
@@ -233,6 +434,7 @@ int isc_alloc_block (ISAMC is, int cat)
             if ((nb = is->files[cat].fc_list[j]) && (!block || nb < block))
             {
                 is->files[cat].fc_list[j] = 0;
+		block = nb;
                 break;
             }
     }
@@ -243,17 +445,6 @@ int isc_alloc_block (ISAMC is, int cat)
     return block;
 }
 
-static void release_block (ISAMC is, int cat, int pos)
-{
-    char buf[sizeof(int)];
-   
-    (is->files[cat].no_released)++;
-    is->files[cat].head_is_dirty = 1; 
-    memcpy (buf, &is->files[cat].head.freelist, sizeof(int));
-    is->files[cat].head.freelist = pos;
-    bf_write (is->files[cat].bf, pos, 0, sizeof(int), buf);
-}
-
 void isc_release_block (ISAMC is, int cat, int pos)
 {
     if (is->method->debug > 3)
@@ -276,7 +467,8 @@ static void init_fc (ISAMC is, int cat)
     int j = 100;
         
     is->files[cat].fc_max = j;
-    is->files[cat].fc_list = xmalloc (sizeof(*is->files[0].fc_list) * j);
+    is->files[cat].fc_list = (int *)
+	xmalloc (sizeof(*is->files[0].fc_list) * j);
     while (--j >= 0)
         is->files[cat].fc_list[j] = 0;
 }
@@ -304,13 +496,13 @@ void isc_pp_close (ISAMC_PP pp)
 
 ISAMC_PP isc_pp_open (ISAMC is, ISAMC_P ipos)
 {
-    ISAMC_PP pp = xmalloc (sizeof(*pp));
+    ISAMC_PP pp = (ISAMC_PP) xmalloc (sizeof(*pp));
     char *src;
    
     pp->cat = isc_type(ipos);
     pp->pos = isc_block(ipos); 
 
-    src = pp->buf = xmalloc (is->method->filecat[pp->cat].bsize);
+    src = pp->buf = (char *) xmalloc (is->method->filecat[pp->cat].bsize);
 
     pp->next = 0;
     pp->size = 0;
@@ -334,8 +526,8 @@ ISAMC_PP isc_pp_open (ISAMC is, ISAMC_P ipos)
         pp->offset = src - pp->buf; 
         assert (pp->offset == ISAMC_BLOCK_OFFSET_1);
         if (is->method->debug > 2)
-            logf (LOG_LOG, "isc: read_block size=%d %d %d",
-                 pp->size, pp->cat, pp->pos);
+            logf (LOG_LOG, "isc: read_block size=%d %d %d next=%d",
+                 pp->size, pp->cat, pp->pos, pp->next);
     }
     return pp;
 }
@@ -346,7 +538,11 @@ int isc_pp_read (ISAMC_PP pp, void *buf)
     return isc_read_item (pp, (char **) &buf);
 }
 
-/* returns non-zero if item could be read; 0 otherwise */
+/* read one item from file - decode and store it in *dst.
+   Returns
+     0 if end-of-file
+     1 if item could be read ok and NO boundary
+     2 if item could be read ok and boundary */
 int isc_read_item (ISAMC_PP pp, char **dst)
 {
     ISAMC is = pp->is;
@@ -354,10 +550,35 @@ int isc_read_item (ISAMC_PP pp, char **dst)
 
     if (pp->offset >= pp->size)
     {
+	if (!pp->next)
+	{
+	    pp->pos = 0;
+	    return 0; /* end of file */
+	}
+	if (pp->next > pp->pos)
+	{
+	    if (pp->next == pp->pos + 1)
+		is->files[pp->cat].no_next++;
+	    else
+	    {
+		is->files[pp->cat].no_forward++;
+		is->files[pp->cat].sum_forward += pp->next - pp->pos;
+	    }
+	}
+	else
+	{
+	    if (pp->next + 1 == pp->pos)
+		is->files[pp->cat].no_prev++;
+	    else
+	    {
+		is->files[pp->cat].no_backward++;
+		is->files[pp->cat].sum_backward += pp->pos - pp->next;
+	    }
+	}
+	/* out new block position */
         pp->pos = pp->next;
-        if (!pp->pos)
-            return 0;
         src = pp->buf;
+	/* read block and save 'next' and 'size' entry */
         isc_read_block (is, pp->cat, pp->pos, src);
         memcpy (&pp->next, src, sizeof(pp->next));
         src += sizeof(pp->next);
@@ -371,8 +592,8 @@ int isc_read_item (ISAMC_PP pp, char **dst)
         (*is->method->code_item)(ISAMC_DECODE, pp->decodeClientData, dst, &src);
         pp->offset = src - pp->buf; 
         if (is->method->debug > 2)
-            logf (LOG_LOG, "isc: read_block size=%d %d %d",
-                 pp->size, pp->cat, pp->pos);
+            logf (LOG_LOG, "isc: read_block size=%d %d %d next=%d",
+                 pp->size, pp->cat, pp->pos, pp->next);
         return 2;
     }
     (*is->method->code_item)(ISAMC_DECODE, pp->decodeClientData, dst, &src);