Key input and merge sort in one pass.

[idzebra-moved-to-github.git] / index / main.c

diff --git a/index/main.c b/index/main.c
index 389585d..7c56296 100644 (file)
--- a/index/main.c
+++ b/index/main.c
@@ -1,10 +1,31 @@
 /*
- * Copyright (C) 1995, Index Data I/S 
+ * Copyright (C) 1994-1995, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: main.c,v $
- * Revision 1.5  1995-09-04 09:10:39  adam
+ * Revision 1.12  1995-10-04 16:57:20  adam
+ * Key input and merge sort in one pass.
+ *
+ * Revision 1.11  1995/09/29  14:01:45  adam
+ * Bug fixes.
+ *
+ * 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
+ * Option: only one word key per file.
+ *
+ * Revision 1.7  1995/09/05  15:28:39  adam
+ * More work on search engine.
+ *
+ * Revision 1.6  1995/09/04  12:33:43  adam
+ * Various cleanup. YAZ util used instead.
+ *
+ * Revision 1.5  1995/09/04  09:10:39  adam
  * More work on index add/del/update.
  * Merge sort implemented.
  * Initial work on z39 server.
@@ -26,10 +47,11 @@
 #include <assert.h>
 #include <unistd.h>
 
-#include <util.h>
+#include <alexutil.h>
 #include "index.h"
 
 char *prog;
+size_t mem_max = 4*1024*1024;
 
 int main (int argc, char **argv)
 {
@@ -38,9 +60,11 @@ 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)
+    while ((ret = options ("r:v:m:", argv, argc, &arg)) != -2)
     {
         if (ret == 0)
         {
@@ -51,7 +75,7 @@ int main (int argc, char **argv)
                 common_resource = res_open (base_name);
                 if (!common_resource)
                 {
-                    log (LOG_FATAL, "Cannot open resource `%s'", base_name);
+                    logf (LOG_FATAL, "Cannot open resource `%s'", base_name);
                     exit (1);
                 }
             }
@@ -67,14 +91,14 @@ int main (int argc, char **argv)
                 }
                 else
                 {
-                    log (LOG_FATAL, "Unknown command: %s", arg);
+                    logf (LOG_FATAL, "Unknown command: %s", arg);
                     exit (1);
                 }
             }
             else
             {
                 unlink ("keys.tmp");
-                key_open ("keys.tmp");
+                key_open (mem_max);
                 repository (cmd, arg, base_path);
                 cmd = 0;
             }
@@ -87,9 +111,13 @@ int main (int argc, char **argv)
         {
             base_path = arg;
         }
+        else if (ret == 'm')
+        {
+            mem_max = 1024*1024*atoi(arg);
+        }
         else
         {
-            log (LOG_FATAL, "Unknown option '-%s'", arg);
+            logf (LOG_FATAL, "Unknown option '-%s'", arg);
             exit (1);
         }
     }
@@ -99,13 +127,20 @@ int main (int argc, char **argv)
                  "base cmd1 dir1 cmd2 dir2 ...\n");
         exit (1);
     }
-    key_flush ();
-    if (!key_close ())
+    nsections = key_close ();
+    if (!nsections)
         exit (0);
-    log (LOG_DEBUG, "Sorting");
-    if (!key_sort ("keys.tmp", 1000000))
-        exit (0);
-    log (LOG_DEBUG, "Input");
-    key_input ("dictinv", "isaminv", "keys.tmp", 50);
+#if 0
+    logf (LOG_LOG, "Merge sorting");
+    mbuf = xmalloc (100000);
+    merge_sort (mbuf, 1, nsections+1);
+    xfree (mbuf);
+    logf (LOG_LOG, "Input");
+    key_input (FNAME_WORD_DICT, FNAME_WORD_ISAM, "keys1.tmp", 60);
+#else
+    logf (LOG_LOG, "Input");
+    key_input2 (FNAME_WORD_DICT, FNAME_WORD_ISAM, nsections, 60);
+#endif
     exit (0);
 }
+