From: Adam Dickmeiss <adam@indexdata.dk>
Date: Thu, 28 Sep 1995 14:22:56 +0000 (+0000)
Subject: Sort uses smaller temporary files.
X-Git-Tag: ZEBRA.1.0~739
X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=commitdiff_plain;h=3003d747565b03bac252b43eca48bb691d1adbfc

Sort uses smaller temporary files.
---

diff --git a/index/extract.c b/index/extract.c
index a7169f1..3560d13 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -4,7 +4,10 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: extract.c,v $
- * Revision 1.11  1995-09-28 12:10:31  adam
+ * Revision 1.12  1995-09-28 14:22:56  adam
+ * Sort uses smaller temporary files.
+ *
+ * Revision 1.11  1995/09/28  12:10:31  adam
  * Bug fixes. Field prefix used in queries.
  *
  * Revision 1.10  1995/09/28  09:19:41  adam
@@ -54,27 +57,44 @@
 
 static Dict file_idx;
 static SYSNO sysno_next;
-static int key_fd = -1;
 static int sys_idx_fd = -1;
-static char *key_buf;
-static int key_offset, key_buf_size;
+
 static int key_cmd;
 static int key_sysno;
+static char **key_buf;
+static size_t ptr_top;
+static size_t ptr_i;
+static size_t kused;
+static int key_file_no;
 
-void key_open (const char *fname)
+static int sort_compare (const void *p1, const void *p2)
+{
+    int r;
+    size_t l;
+    char *cp1 = *(char**) p1;
+    char *cp2 = *(char**) p2;
+
+    if ((r = strcmp (cp1, cp2)))
+        return r;
+    l = strlen(cp1); 
+    if ((r = key_compare (cp1+l, cp2+l)))
+        return r;
+    return cp1[l+sizeof(struct it_key)] -
+           cp2[l+sizeof(struct it_key)];
+}
+
+void key_open (int mem)
 {
     void *file_key;
-    if (key_fd != -1)
-        return;
-    if ((key_fd = open (fname, O_RDWR|O_CREAT, 0666)) == -1)
-    {
-        logf (LOG_FATAL|LOG_ERRNO, "open %s", fname);
-        exit (1);
-    }
-    logf (LOG_DEBUG, "key_open of %s", fname);
-    key_buf_size = 49100;
-    key_buf = xmalloc (key_buf_size);
-    key_offset = 0;
+
+    if (mem < 50000)
+        mem = 50000;
+    key_buf = xmalloc (mem);
+    ptr_top = mem/sizeof(char*);
+    ptr_i = 0;
+    kused = 0;
+    key_file_no = 0;
+
     if (!(file_idx = dict_open (FNAME_FILE_DICT, 40, 1)))
     {
         logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
@@ -91,41 +111,76 @@ void key_open (const char *fname)
         exit (1);
     }
 }
-
-int key_close (void)
+    
+void key_flush (void)
 {
-    if (key_fd == -1)
-    {
-        logf (LOG_DEBUG, "key_close - but no file");
-        return 0;
-    }
-    close (key_fd);
-    close (sys_idx_fd);
-    dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next);
-    dict_close (file_idx);
-    key_fd = -1;
-    xfree (key_buf);
-    return 1;
-}
+    FILE *outf;
+    char out_fname[200];
+    char *prevcp, *cp;
+    
+    if (ptr_i <= 0)
+        return;
+
+    key_file_no++;
+    logf (LOG_LOG, "sorting section %d", key_file_no);
+    qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), sort_compare);
+    sprintf (out_fname, TEMP_FNAME, key_file_no);
 
-void wordFlush (int sysno)
-{
-    size_t i = 0;
-    int w;
 
-    if (key_fd == -1)
-	return; 
-    while (i < key_offset)
+    if (!(outf = fopen (out_fname, "w")))
+    {
+        logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname);
+        exit (1);
+    }
+    logf (LOG_LOG, "writing section %d", key_file_no);
+    prevcp = cp = key_buf[ptr_top-ptr_i];
+    
+    if (fwrite (cp, strlen (cp)+2+sizeof(struct it_key), 1, outf) != 1)
     {
-        w = write (key_fd, key_buf + i, key_offset - i);
-        if (w == -1)
+        logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
+        exit (1);
+    }
+    while (--ptr_i > 0)
+    {
+        cp = key_buf[ptr_top-ptr_i];
+        if (strcmp (cp, prevcp))
+        {
+            if (fwrite (cp, strlen (cp)+2+sizeof(struct it_key), 1,
+                        outf) != 1)
+            {
+                logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
+                exit (1);
+            }
+            prevcp = cp;
+        }
+        else
         {
-            logf (LOG_FATAL|LOG_ERRNO, "Write key fail");
-            exit (1);
+            cp = strlen (cp) + cp;
+            if (fwrite (cp, 2+sizeof(struct it_key), 1, outf) != 1)
+            {
+                logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
+                exit (1);
+            }
         }
-        i += w;
     }
-    key_offset = 0;
+    if (fclose (outf))
+    {
+        logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
+        exit (1);
+    }
+    logf (LOG_LOG, "finished section %d", key_file_no);
+    ptr_i = 0;
+    kused = 0;
+}
+
+int key_close (void)
+{
+    key_flush ();
+    xfree (key_buf);
+    close (sys_idx_fd);
+    dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next);
+    dict_close (file_idx);
+    return key_file_no;
 }
 
 static void wordInit (RecWord *p)
@@ -138,39 +193,30 @@ static void wordInit (RecWord *p)
 static void wordAdd (const RecWord *p)
 {
     struct it_key key;
-    char x;
     size_t i;
 
-    if (key_offset + 1000 > key_buf_size)
-    {
-        char *new_key_buf;
-
-        key_buf_size *= 2;
-        new_key_buf = xmalloc (2*key_buf_size);
-        memcpy (new_key_buf, key_buf, key_offset);
-        xfree (key_buf);
-        key_buf = new_key_buf;
-    }
-    key_offset += index_word_prefix (key_buf + key_offset,
-                                     p->attrSet, p->attrUse);
+    if (kused + 1024 > (ptr_top-ptr_i)*sizeof(char*))
+        key_flush ();
+    ++ptr_i;
+    key_buf[ptr_top-ptr_i] = (char*)key_buf + kused;
+    kused += index_word_prefix ((char*)key_buf + kused,
+                                p->attrSet, p->attrUse);
     switch (p->which)
     {
     case Word_String:
         for (i = 0; p->u.string[i]; i++)
-            key_buf[key_offset++] = index_char_cvt (p->u.string[i]);
-        key_buf[key_offset++] = '\0';
+            ((char*)key_buf) [kused++] = index_char_cvt (p->u.string[i]);
+        ((char*)key_buf) [kused++] = '\0';
         break;
     default:
         return ;
     }
-    x = (key_cmd == 'a') ? 1 : 0;
-    memcpy (key_buf + key_offset, &x, 1);
-    key_offset++;
-
     key.sysno = key_sysno;
     key.seqno = p->seqno;
-    memcpy (key_buf + key_offset, &key, sizeof(key));
-    key_offset += sizeof(key);
+    memcpy ((char*)key_buf + kused, &key, sizeof(key));
+    kused += sizeof(key);
+
+    ((char*) key_buf)[kused++] = ((key_cmd == 'a') ? 1 : 0);
 }
 
 void file_extract (int cmd, const char *fname, const char *kname)
@@ -227,7 +273,4 @@ void file_extract (int cmd, const char *fname, const char *kname)
     key_cmd = cmd;
     (*rt->extract)(&extractCtrl);
     fclose (inf);
-    wordFlush (sysno);
 }
-
-
diff --git a/index/index.h b/index/index.h
index 5352205..1d8ea48 100644
--- a/index/index.h
+++ b/index/index.h
@@ -4,7 +4,10 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: index.h,v $
- * Revision 1.12  1995-09-28 12:10:32  adam
+ * Revision 1.13  1995-09-28 14:22:56  adam
+ * Sort uses smaller temporary files.
+ *
+ * Revision 1.12  1995/09/28  12:10:32  adam
  * Bug fixes. Field prefix used in queries.
  *
  * Revision 1.11  1995/09/27  12:22:28  adam
@@ -69,15 +72,16 @@ void repository (int cmd, const char *rep, const char *base_path);
 
 void file_extract (int cmd, const char *fname, const char *kname);
 
-void key_open (const char *fname);
+void key_open (int mem);
 int key_close (void);
 void key_write (int cmd, struct it_key *k, const char *str);
 int key_compare (const void *p1, const void *p2);
 void key_logdump (int mask, const void *p);
 void key_input (const char *dict_fname, const char *isam_fname, 
                 const char *key_fname, int cache);
-int key_sort (const char *key_fname, size_t mem);
+int merge_sort (char **buf, int from, int to);
 
+#define TEMP_FNAME  "keys%d.tmp"
 #define FNAME_WORD_DICT "worddict"
 #define FNAME_WORD_ISAM "wordisam"
 #define FNAME_FILE_DICT "filedict"
diff --git a/index/kinput.c b/index/kinput.c
index 9f834a9..889a641 100644
--- a/index/kinput.c
+++ b/index/kinput.c
@@ -4,7 +4,10 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: kinput.c,v $
- * Revision 1.3  1995-09-06 16:11:17  adam
+ * Revision 1.4  1995-09-28 14:22:57  adam
+ * Sort uses smaller temporary files.
+ *
+ * Revision 1.3  1995/09/06  16:11:17  adam
  * Option: only one word key per file.
  *
  * Revision 1.2  1995/09/04  12:33:42  adam
@@ -79,7 +82,7 @@ static int inp (Dict dict, ISAM isam, const char *name)
         {
             if (!read_one (inf, next_name, next_key))
                 break;
-            if (strcmp (next_name, cur_name))
+            if (*next_name && strcmp (next_name, cur_name))
                 break;
             memcpy (key_buf + key_buf_ptr, next_key, KEY_SIZE);
             key_buf_ptr += KEY_SIZE;
diff --git a/index/main.c b/index/main.c
index a270a89..0616499 100644
--- a/index/main.c
+++ b/index/main.c
@@ -4,7 +4,10 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: main.c,v $
- * Revision 1.9  1995-09-14 07:48:24  adam
+ * Revision 1.10  1995-09-28 14:22:57  adam
+ * Sort uses smaller temporary files.
+ *
+ * Revision 1.9  1995/09/14  07:48:24  adam
  * Record control management.
  *
  * Revision 1.8  1995/09/06  16:11:18  adam
@@ -50,6 +53,8 @@ int main (int argc, char **argv)
     char *arg;
     char *base_name = NULL;
     char *base_path = NULL;
+    int nsections;
+    char **mbuf;
 
     prog = *argv;
     while ((ret = options ("r:v:", argv, argc, &arg)) != -2)
@@ -86,7 +91,7 @@ int main (int argc, char **argv)
             else
             {
                 unlink ("keys.tmp");
-                key_open ("keys.tmp");
+                key_open (3000000);
                 repository (cmd, arg, base_path);
                 cmd = 0;
             }
@@ -111,13 +116,14 @@ int main (int argc, char **argv)
                  "base cmd1 dir1 cmd2 dir2 ...\n");
         exit (1);
     }
-    if (!key_close ())
-        exit (0);
-    logf (LOG_LOG, "Sorting");
-    if (!key_sort ("keys.tmp", 3000000))
+    nsections = key_close ();
+    if (!nsections)
         exit (0);
+    logf (LOG_LOG, "Merge sorting");
+    mbuf = xmalloc (100000);
+    merge_sort (mbuf, 1, nsections+1);
     logf (LOG_LOG, "Input");
-    key_input (FNAME_WORD_DICT, FNAME_WORD_ISAM, "keys.tmp", 60);
+    key_input (FNAME_WORD_DICT, FNAME_WORD_ISAM, "keys1.tmp", 60);
     exit (0);
 }