X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=isamc%2Fisamc.c;h=6d64d7940e74c862eb7a80538c9ef24c6bee1f42;hb=b7942589b21cca1b6d44c7bf9446dcb6a5942953;hp=611f3faceb309952991a9e195d037dbe6cba36f9;hpb=932697e4b6d6dd00a210b84b285bd8c0000f379f;p=idzebra-moved-to-github.git

diff --git a/isamc/isamc.c b/isamc/isamc.c
index 611f3fa..6d64d79 100644
--- a/isamc/isamc.c
+++ b/isamc/isamc.c
@@ -4,11 +4,44 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: isamc.c,v $
- * Revision 1.1  1996-10-29 13:40:48  adam
+ * 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.
+ *
+ * Revision 1.7  1997/02/12 20:42:43  adam
+ * Bug fix: during isc_merge operations, some pages weren't marked dirty
+ * even though they should be. At this point the merge operation marks
+ * a page dirty if the previous page changed at all. A better approach is
+ * to mark it dirty if the last key written changed in previous page.
+ *
+ * Revision 1.6  1996/11/08 11:15:29  adam
+ * Number of keys in chain are stored in first block and the function
+ * to retrieve this information, isc_pp_num is implemented.
+ *
+ * Revision 1.5  1996/11/04 14:08:57  adam
+ * Optimized free block usage.
+ *
+ * Revision 1.4  1996/11/01 13:36:46  adam
+ * New element, max_blocks_mem, that control how many blocks of max size
+ * to store in memory during isc_merge.
+ * Function isc_merge now ignores delete/update of identical keys and
+ * the proper blocks are then non-dirty and not written in flush_blocks.
+ *
+ * Revision 1.3  1996/11/01  08:59:14  adam
+ * First version of isc_merge that supports update/delete.
+ *
+ * Revision 1.2  1996/10/29 16:44:56  adam
+ * Work on isc_merge.
+ *
+ * Revision 1.1  1996/10/29  13:40:48  adam
  * First work.
  *
  */
 
+/* 
+ * TODO:
+ *   Reduction to lower categories in isc_merge
+ */
 #include <stdlib.h>
 #include <assert.h>
 #include <string.h>
@@ -17,6 +50,9 @@
 #include <log.h>
 #include "isamc-p.h"
 
+static void release_fc (ISAMC is, int cat);
+static void init_fc (ISAMC is, int cat);
+
 ISAMC_M isc_getmethod (void)
 {
     static struct ISAMC_filecat_s def_cat[] = {
@@ -24,7 +60,6 @@ ISAMC_M isc_getmethod (void)
         {  512,   490,   100,    20 },
         { 4096,  3950,  1000,    20 },
         {32768, 32000, 10000,     0 },
-        {    0,     0,     0,     0 }
     };
     ISAMC_M m = xmalloc (sizeof(*m));
     m->filecat = def_cat;
@@ -35,17 +70,19 @@ ISAMC_M isc_getmethod (void)
 
     m->compare_item = NULL;
 
-    m->debug = 0;
+    m->debug = 1;
+
+    m->max_blocks_mem = 10;
 
     return m;
 }
 
 
-ISAMC isc_open (const char *name, int writeflag, ISAMC_M method)
+ISAMC isc_open (BFiles bfs, const char *name, int writeflag, ISAMC_M method)
 {
     ISAMC is;
     ISAMC_filecat filecat;
-    int i;
+    int i = 0;
     int max_buf_size = 0;
 
     is = xmalloc (sizeof(*is));
@@ -58,7 +95,7 @@ ISAMC isc_open (const char *name, int writeflag, ISAMC_M method)
     /* determine number of block categories */
     if (is->method->debug)
         logf (LOG_LOG, "isc: bsize  ifill  mfill mblocks");
-    for (i = 0; filecat[i].bsize; i++)
+    do
     {
         if (is->method->debug)
             logf (LOG_LOG, "isc:%6d %6d %6d %6d",
@@ -66,23 +103,31 @@ ISAMC isc_open (const char *name, int writeflag, ISAMC_M method)
                   filecat[i].mfill, filecat[i].mblocks);
         if (max_buf_size < filecat[i].mblocks * filecat[i].bsize)
             max_buf_size = filecat[i].mblocks * filecat[i].bsize;
-    }
+    } while (filecat[i++].mblocks);
     is->no_files = i;
     is->max_cat = --i;
     /* max_buf_size is the larget buffer to be used during merge */
     max_buf_size = (1 + max_buf_size / filecat[i].bsize) * filecat[i].bsize;
+    if (max_buf_size < (1+is->method->max_blocks_mem) * filecat[i].bsize)
+        max_buf_size = (1+is->method->max_blocks_mem) * filecat[i].bsize;
     if (is->method->debug)
         logf (LOG_LOG, "isc: max_buf_size %d", max_buf_size);
     
     assert (is->no_files > 0);
-    is->files = xmalloc (sizeof(*is->files)*i);
-    is->r_buf = xmalloc (max_buf_size+128);
+    is->files = xmalloc (sizeof(*is->files)*is->no_files);
+    if (writeflag)
+    {
+        is->merge_buf = xmalloc (max_buf_size+256);
+	memset (is->merge_buf, 0, max_buf_size+256);
+    }
+    else
+        is->merge_buf = NULL;
     for (i = 0; i<is->no_files; i++)
     {
         char fname[512];
 
         sprintf (fname, "%s%c", name, i+'A');
-        is->files[i].bf = bf_open (fname, is->method->filecat[i].bsize,
+        is->files[i].bf = bf_open (bfs, fname, is->method->filecat[i].bsize,
                                    writeflag);
         is->files[i].head_is_dirty = 0;
         if (!bf_read (is->files[i].bf, 0, 0, sizeof(ISAMC_head),
@@ -91,6 +136,14 @@ ISAMC isc_open (const char *name, int writeflag, ISAMC_M method)
             is->files[i].head.lastblock = 1;
             is->files[i].head.freelist = 0;
         }
+        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;
+
+        init_fc (is, i);
     }
     return is;
 }
@@ -99,36 +152,42 @@ int isc_close (ISAMC is)
 {
     int i;
 
+    if (is->method->debug)
+        logf (LOG_LOG, "isc:  writes   reads skipped   alloc released  remap");
     for (i = 0; i<is->no_files; i++)
-        if (is->files[i].bf)
-        {
-            if (is->files[i].head_is_dirty)
-                bf_write (is->files[i].bf, 0, 0, sizeof(ISAMC_head),
-                     &is->files[i].head);
-            bf_close (is->files[i].bf);
-        }
+    {
+        release_fc (is, i);
+        assert (is->files[i].bf);
+        if (is->files[i].head_is_dirty)
+            bf_write (is->files[i].bf, 0, 0, sizeof(ISAMC_head),
+                 &is->files[i].head);
+        if (is->method->debug)
+            logf (LOG_LOG, "isc:%8d%8d%8d%8d%8d%8d",
+                  is->files[i].no_writes,
+                  is->files[i].no_reads,
+                  is->files[i].no_skip_writes,
+                  is->files[i].no_allocated,
+                  is->files[i].no_released,
+                  is->files[i].no_remap);
+        xfree (is->files[i].fc_list);
+        bf_close (is->files[i].bf);
+    }
     xfree (is->files);
-    xfree (is->r_buf);
+    xfree (is->merge_buf);
     xfree (is);
     return 0;
 }
 
-void isc_catpos (ISAMC_P ipos, int *cat, int *pos)
-{
-    *pos = ipos >> 3;
-    *cat = ipos & 7;
-}
-
 int isc_read_block (ISAMC is, int cat, int pos, char *dst)
 {
-    if (is->method->debug > 1)
-        logf (LOG_LOG, "isc: read_block %d %d", cat, pos);
+    ++(is->files[cat].no_reads);
     return bf_read (is->files[cat].bf, pos, 0, 0, dst);
 }
 
 int isc_write_block (ISAMC is, int cat, int pos, char *src)
 {
-    if (is->method->debug > 1)
+    ++(is->files[cat].no_writes);
+    if (is->method->debug > 2)
         logf (LOG_LOG, "isc: write_block %d %d", cat, pos);
     return bf_write (is->files[cat].bf, pos, 0, 0, src);
 }
@@ -136,21 +195,23 @@ 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)
 {
-    int xoffset = offset + 2*sizeof(int);
+    unsigned short size = offset + ISAMC_BLOCK_OFFSET_N;
     if (is->method->debug > 2)
-        logf (LOG_LOG, "isc: write_dblock. offset=%d nextpos=%d",
-              offset, nextpos);
-    memcpy (src - sizeof(int)*2, &nextpos, sizeof(int));
-    memcpy (src - sizeof(int), &xoffset, sizeof(int));
-    return isc_write_block (is, cat, pos, src - sizeof(int)*2);
+        logf (LOG_LOG, "isc: write_dblock. size=%d nextpos=%d",
+              (int) size, nextpos);
+    src -= ISAMC_BLOCK_OFFSET_N;
+    memcpy (src, &nextpos, sizeof(int));
+    memcpy (src + sizeof(int), &size, sizeof(size));
+    return isc_write_block (is, cat, pos, src);
 }
 
-int isc_alloc_block (ISAMC is, int cat)
+static int alloc_block (ISAMC is, int cat)
 {
     int block;
     char buf[sizeof(int)];
 
     is->files[cat].head_is_dirty = 1;
+    (is->files[cat].no_allocated)++;
     if ((block = is->files[cat].head.freelist))
     {
         bf_read (is->files[cat].bf, block, 0, sizeof(int), buf);
@@ -158,268 +219,79 @@ int isc_alloc_block (ISAMC is, int cat)
     }
     else
         block = (is->files[cat].head.lastblock)++;
-    if (is->method->debug > 2)
-        logf (LOG_LOG, "isc: alloc_block in cat %d -> %d", cat, block);
     return block;
 }
 
-void isc_release_block (ISAMC is, int cat, int pos)
+int isc_alloc_block (ISAMC is, int cat)
+{
+    int block = 0;
+
+    if (is->files[cat].fc_list)
+    {
+        int j, nb;
+        for (j = 0; j < is->files[cat].fc_max; j++)
+            if ((nb = is->files[cat].fc_list[j]) && (!block || nb < block))
+            {
+                is->files[cat].fc_list[j] = 0;
+                break;
+            }
+    }
+    if (!block)
+        block = alloc_block (is, cat);
+    if (is->method->debug > 3)
+        logf (LOG_LOG, "isc: alloc_block in cat %d: %d", cat, block);
+    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);
 }
 
-static void isc_flush_blocks (ISAMC is, int *r_ptr, int r_ptri, char *r_buf,
-                              int *nextpos, int *firstpos, int cat, int last)
-{
-    int i;
-
-    for (i = 1; i<r_ptri; i++)
-    {
-        int pos;
-        if (*nextpos)
-            pos = *nextpos;
-        else
-            pos = isc_alloc_block (is, cat);
-        if (!*firstpos)
-            *firstpos = pos;
-        if (last && i == r_ptri-1)
-            *nextpos = 0;
-        else
-            *nextpos = isc_alloc_block (is, cat);
-        isc_write_dblock (is, cat, pos, r_buf + r_ptr[i-1], *nextpos,
-                          r_ptr[i] - r_ptr[i-1]);
-    }
-}
-
-
-ISAMC_P isc_merge_first (ISAMC is, ISAMC_I data)
+void isc_release_block (ISAMC is, int cat, int pos)
 {
-    char i_item[128], *i_item_ptr;
-    int i_more, i_mode, i;
-
-    int firstpos = 0;
-    int nextpos = 0;    
-    int cat = 0;
-    char r_item_buf[128];
-    int r_offset = 0;
-    int r_ptr[100];
-    int r_ptri = 0;
-    void *r_clientData = (*is->method->code_start)(ISAMC_ENCODE);
-    char *r_buf = is->r_buf + ISAMC_BLOCK_OFFSET;
-
-    /* read first item from i */
-    i_item_ptr = i_item;
-    i_more = (*data->read_item)(data->clientData, &i_item_ptr, &i_mode);
-    if (i_more)
-        r_ptr[r_ptri++] = 0;
-    while (i_more)
+    if (is->method->debug > 3)
+        logf (LOG_LOG, "isc: release_block in cat %d: %d", cat, pos);
+    if (is->files[cat].fc_list)
     {
-        char *r_item = r_item_buf;
-
-        memcpy (r_item, i_item, i_item_ptr - i_item);
-        
-        if (r_item)  /* insert resulting item? */
-        {
-            char *r_out_ptr = r_buf + r_offset;
-            int new_offset;
-            int border = r_ptr[r_ptri-1] + is->method->filecat[cat].ifill
-                         -ISAMC_BLOCK_OFFSET;
-
-            (*is->method->code_item)(ISAMC_ENCODE, r_clientData,
-                                     &r_out_ptr, &r_item);
-            new_offset = r_out_ptr - r_buf; 
-
-            if (border >= r_offset && border < new_offset)
+        int j;
+        for (j = 0; j<is->files[cat].fc_max; j++)
+            if (!is->files[cat].fc_list[j])
             {
-                /* Initial fill of current block category reached... 
-                   Save offset in r_ptr 
-                 */
-                r_ptr[r_ptri++] = r_offset;
-                if (cat == is->max_cat)
-                {
-                    /* We are dealing with block of max size. Block(s)
-                       will be flushed. Note: the block(s) are surely not
-                       the last one(s).
-                     */
-                    if (is->method->debug > 1)
-                        logf (LOG_LOG, "isc: flush %d sections", r_ptri-1);
-                    isc_flush_blocks (is, r_ptr, r_ptri, r_buf,
-                                      &nextpos, &firstpos, cat, 0);
-                    r_ptri = 0;
-                    r_ptr[r_ptri++] = 0;
-                    memcpy (r_buf, r_buf + r_offset, new_offset - r_offset);
-                    new_offset = (new_offset - r_offset);
-                }
+                is->files[cat].fc_list[j] = pos;
+                return;
             }
-            r_offset = new_offset;
-            if (cat < is->max_cat &&
-                r_ptri>is->method->filecat[cat].mblocks)
-            {
-                /* Max number blocks in current category reached ->
-                   must switch to next category (with larger block size) 
-                */
-                int j = 1;
-                cat++;
-                /* r_ptr[0] = r_ptr[0] = 0 true anyway.. */
-                for (i = 2; i < r_ptri; i++)
-                {
-                    border = is->method->filecat[cat].ifill -
-                             ISAMC_BLOCK_OFFSET + r_ptr[j-1];
-                    if (r_ptr[i] > border && r_ptr[i-1] <= border)
-                        r_ptr[j++] = r_ptr[i-1];
-                }
-                r_ptri = j;
-            }
-        }
-        i_item_ptr = i_item;
-        i_more = (*data->read_item)(data->clientData, &i_item_ptr, &i_mode);
     }
-    r_ptr[r_ptri++] = r_offset;
-    /* flush rest of block(s) in r_buf */
-    if (is->method->debug > 1)
-        logf (LOG_LOG, "isc: flush rest, %d sections", r_ptri-1);
-    isc_flush_blocks (is, r_ptr, r_ptri, r_buf, &nextpos, &firstpos, cat, 1);
-    (*is->method->code_stop)(ISAMC_ENCODE, r_clientData);
-    return cat + firstpos * 8;
+    release_block (is, cat, pos);
 }
 
-ISAMC_P isc_merge (ISAMC is, ISAMC_P ipos, ISAMC_I data)
+static void init_fc (ISAMC is, int cat)
 {
-    assert (ipos == 0);
-    return isc_merge_first (is, data);
+    int j = 100;
+        
+    is->files[cat].fc_max = j;
+    is->files[cat].fc_list = xmalloc (sizeof(*is->files[0].fc_list) * j);
+    while (--j >= 0)
+        is->files[cat].fc_list[j] = 0;
 }
 
-
-#if 0
-ISAMC_P isc_merge (ISAMC is, ISAMC_P ipos, ISAMC_I data)
+static void release_fc (ISAMC is, int cat)
 {
-    ISAMC_PP pp; 
-    char f_item[128], *f_item_ptr;
-    int f_more;
-    int cat = 0;
-    int nextpos;
-
-    char i_item[128], *i_item_ptr;
-    int i_more, insertMode;
-
-    char r_item_buf[128];
-    int r_offset = ISAMC_BLOCK_OFFSET;
-    int r_dirty = 0;
-    char *r_ptr[100];
-    int r_ptri = 0;
-    int r_start = 0;
-    void *r_clientData = (*is->method->code_start)();
-
-    /* rewind and read first item from file */
-    pp = isc_position (is, ipos);
-    f_item_ptr = f_item;
-    f_more = isc_read_item (pp, &f_item_ptr);
-    cat = pp->cat;
-
-    /* read first item from i */
-    i_item_ptr = i_item;
-    i_more = (*data->read_item)(data->clientData, &i_item_ptr, &insertMode);
-   
-    while (f_more || i_more)
-    {
-        int cmp;
-        char *r_item = r_item_buf;
-
-        if (!f_more)
-            cmp = -1;
-        else if (!i_more)
-            cmp = 1;
-        else
-            cmp = (*is->method->compare_item)(i_item, f_item);
-        if (cmp == 0)                   /* insert i=f */
-        {
-            if (!insertMode) 
-            {
-                r_item = NULL;
-                r_dirty = 1;
-            }
-            else
-                memcpy (r_item, f_item, f_item_ptr - f_item);
-
-            /* move i */
-            i_item_ptr = i_item;
-            i_more = (*data->read_item)(data->clientData, &i_item_ptr,
-                                        &insertMode);
-            /* move f */
-            f_item_ptr = f_item;
-            f_more = isc_read_item (pp, &f_item_ptr);
-        }
-        else if (cmp > 0)               /* insert f */
-        {
-            memcpy (r_item, f_item, f_item_ptr - f_item);
-            /* move f */
-            f_item_ptr = f_item;
-            f_more = isc_read_item (pp, &f_item_ptr);
-        }
-        else                            /* insert i */
-        {
-            if (!insertMode)            /* delete item which isn't there? */
-            {
-                logf (LOG_FATAL, "Inconsistent register");
-                abort ();
-            }
-            memcpy (r_item, i_item, i_item_ptr - i_item);
-            r_dirty = 1;
-            /* move i */
-            i_item_ptr = i_item;
-            i_more = (*data->read_item)(data->clientData, &i_item_ptr,
-                                        &insertMode);
-        }
-        /* check for end of input block condition */
-
-        if (r_item)  /* insert resulting item? */
-        {
-            char *r_out_ptr = is->r_buf + r_offset;
-            int new_offset;
-            int border = is->method->filecat[cat].initsize - r_start;
-
-            (*is->method->code_item)(r_clientData, &r_out_ptr, &r_item);
-            new_offset = r_out_ptr - is->r_buf; 
+    int b, j = is->files[cat].fc_max;
 
-            if (border >= r_offset && border < r_newoffset)
-            {
-                r_ptr[r_ptri++] = r_offset;
-                if (!is->method->filecat[cat].mblocks)
-                {
-                    assert (r_ptri == 1);
-                    /* dump block from 0 -> r_offset in max cat */
-                    r_ptri = 0;
-                    r_offset = ISAMC_BLOCK_OFFSET;
-                }
-            }
-            r_offset = new_offset;
-        }
-        if (r_ptri && r_ptri == is->method->filecat[cat].mblocks)
+    while (--j >= 0)
+        if ((b = is->files[cat].fc_list[j]))
         {
-            int i, j = 0;
-
-            /* dump previous blocks in chain */
-
-            /* recalc r_ptr's */
-            cat++;
-            for (i = 1; i<r_ptr; i++)
-            {
-                if (r_ptr[i] > is->method->filecat[cat].ifill &&
-                    r_ptr[i-1] <= is->method->filecat[cat].ifill)
-                    r_ptr[j++] = r_ptr[i-1];
-            }
-            r_ptri = j;
+            release_block (is, cat, b);
+            is->files[cat].fc_list[j] = 0;
         }
-    }
-    (*is->method->code_stop)(r_clientData);
-    return ipos;
 }
-#endif
 
 void isc_pp_close (ISAMC_PP pp)
 {
@@ -434,26 +306,47 @@ ISAMC_PP isc_pp_open (ISAMC is, ISAMC_P ipos)
 {
     ISAMC_PP pp = xmalloc (sizeof(*pp));
     char *src;
-    
-    isc_catpos (ipos, &pp->cat, &pp->next);
+   
+    pp->cat = isc_type(ipos);
+    pp->pos = isc_block(ipos); 
 
     src = pp->buf = xmalloc (is->method->filecat[pp->cat].bsize);
 
-    pp->pos = 0;    
+    pp->next = 0;
     pp->size = 0;
     pp->offset = 0;
     pp->is = is;
     pp->decodeClientData = (*is->method->code_start)(ISAMC_DECODE);
+    pp->deleteFlag = 0;
+    pp->numKeys = 0;
+
+    if (pp->pos)
+    {
+        src = pp->buf;
+        isc_read_block (is, pp->cat, pp->pos, src);
+        memcpy (&pp->next, src, sizeof(pp->next));
+        src += sizeof(pp->next);
+        memcpy (&pp->size, src, sizeof(pp->size));
+        src += sizeof(pp->size);
+        memcpy (&pp->numKeys, src, sizeof(pp->numKeys));
+        src += sizeof(pp->numKeys);
+        assert (pp->next != pp->pos);
+        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);
+    }
     return pp;
 }
 
-/* returns 1 if item could be read; 0 otherwise */
-int isc_read_key (ISAMC_PP pp, void *buf)
+/* returns non-zero if item could be read; 0 otherwise */
+int isc_pp_read (ISAMC_PP pp, void *buf)
 {
     return isc_read_item (pp, (char **) &buf);
 }
 
-/* returns 1 if item could be read; 0 otherwise */
+/* returns non-zero if item could be read; 0 otherwise */
 int isc_read_item (ISAMC_PP pp, char **dst)
 {
     ISAMC is = pp->is;
@@ -466,26 +359,29 @@ int isc_read_item (ISAMC_PP pp, char **dst)
             return 0;
         src = pp->buf;
         isc_read_block (is, pp->cat, pp->pos, src);
-        
         memcpy (&pp->next, src, sizeof(pp->next));
         src += sizeof(pp->next);
         memcpy (&pp->size, src, sizeof(pp->size));
         src += sizeof(pp->size);
         /* assume block is non-empty */
+        assert (src - pp->buf == ISAMC_BLOCK_OFFSET_N);
         assert (pp->next != pp->pos);
+        if (pp->deleteFlag)
+            isc_release_block (is, pp->cat, pp->pos);
+        (*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);
+        return 2;
     }
     (*is->method->code_item)(ISAMC_DECODE, pp->decodeClientData, dst, &src);
     pp->offset = src - pp->buf; 
     return 1;
 }
 
-int isc_read_islast (ISAMC_PP pp)
-{
-    return pp->offset >= pp->size;
-}
-
-int isc_numkeys (ISAMC_PP pp)
+int isc_pp_num (ISAMC_PP pp)
 {
-    return 1;
+    return pp->numKeys;
 }