More work on search engine.

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

diff --git a/index/extract.c b/index/extract.c
index 34cb68b..a763b7b 100644 (file)
--- a/index/extract.c
+++ b/index/extract.c
@@ -4,7 +4,18 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: extract.c,v $
- * Revision 1.1  1995-09-01 14:06:35  adam
+ * Revision 1.4  1995-09-05 15:28:39  adam
+ * More work on search engine.
+ *
+ * Revision 1.3  1995/09/04  12:33:41  adam
+ * Various cleanup. YAZ util used instead.
+ *
+ * Revision 1.2  1995/09/04  09:10:34  adam
+ * More work on index add/del/update.
+ * Merge sort implemented.
+ * Initial work on z39 server.
+ *
+ * Revision 1.1  1995/09/01  14:06:35  adam
  * Split of work into more files.
  *
  */
@@ -14,7 +25,7 @@
 #include <fcntl.h>
 #include <ctype.h>
 
-#include <util.h>
+#include <alexutil.h>
 #include "index.h"
 
 #define KEY_BUF_SIZE 100000
@@ -22,6 +33,7 @@
 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;
 
@@ -32,18 +44,19 @@ void key_open (const char *fname)
         return;
     if ((key_fd = open (fname, O_RDWR|O_CREAT, 0666)) == -1)
     {
-        log (LOG_FATAL|LOG_ERRNO, "Creat %s", fname);
+        logf (LOG_FATAL|LOG_ERRNO, "open %s", fname);
         exit (1);
     }
+    logf (LOG_DEBUG, "key_open of %s", fname);
     if (!(key_buf = malloc (KEY_BUF_SIZE)))
     {
-        log (LOG_FATAL|LOG_ERRNO, "malloc");
+        logf (LOG_FATAL|LOG_ERRNO, "malloc");
         exit (1);
     }
     key_offset = 0;
-    if (!(file_idx = dict_open ("fileidx", 10, 1)))
+    if (!(file_idx = dict_open (FNAME_FILE_DICT, 10, 1)))
     {
-        log (LOG_FATAL, "dict_open fail of %s", "fileidx");
+        logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
         exit (1);
     }
     file_key = dict_lookup (file_idx, ".");
@@ -51,29 +64,42 @@ void key_open (const char *fname)
         memcpy (&sysno_next, (char*)file_key+1, sizeof(sysno_next));
     else
         sysno_next = 1;
+    if ((sys_idx_fd = open (FNAME_SYS_IDX, O_RDWR|O_CREAT, 0666)) == -1)
+    {
+        logf (LOG_FATAL|LOG_ERRNO, "open %s", FNAME_SYS_IDX);
+        exit (1);
+    }
 }
 
-void key_close (void)
+int key_close (void)
 {
     if (key_fd == -1)
-        return;
+    {
+        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;
+    logf (LOG_DEBUG, "key close - key file exist");
+    return 1;
 }
 
 void key_flush (void)
 {
     size_t i = 0;
     int w;
-    
+
+    if (key_fd == -1)
+       return; 
     while (i < key_offset)
     {
         w = write (key_fd, key_buf + i, key_offset - i);
         if (w == -1)
         {
-            log (LOG_FATAL|LOG_ERRNO, "Write key fail");
+            logf (LOG_FATAL|LOG_ERRNO, "Write key fail");
             exit (1);
         }
         i += w;
@@ -83,17 +109,18 @@ void key_flush (void)
 
 void key_write (int cmd, struct it_key *k, const char *str)
 {
-    char x = cmd;
+    char x;
     size_t slen = strlen(str);
 
     if (key_offset + sizeof(*k) + slen >= KEY_BUF_SIZE - 2)
         key_flush ();
+    x = (cmd == 'a') ? 1 : 0;
+    memcpy (key_buf + key_offset, str, slen+1);
+    key_offset += slen+1;
     memcpy (key_buf + key_offset, &x, 1);
     key_offset++;
     memcpy (key_buf + key_offset, k, sizeof(*k));
     key_offset += sizeof(*k);
-    memcpy (key_buf + key_offset, str, slen+1);
-    key_offset += slen+1;
 }
 
 void text_extract (SYSNO sysno, int cmd, const char *fname)
@@ -104,12 +131,12 @@ void text_extract (SYSNO sysno, int cmd, const char *fname)
     int c;
     char w[256];
 
-    log (LOG_DEBUG, "Text extract of %d", sysno);
+    logf (LOG_DEBUG, "Text extract of %d", sysno);
     k.sysno = sysno;
     inf = fopen (fname, "r");
     if (!inf)
     {
-        log (LOG_WARN|LOG_ERRNO, "open %s", fname);
+        logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
         return;
     }
     while ((c=getc (inf)) != EOF)
@@ -143,7 +170,7 @@ void file_extract (int cmd, const char *fname, const char *kname)
     const char *file_type;
     void *file_info;
 
-    log (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
+    logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
     for (i = strlen(fname); --i >= 0; )
         if (fname[i] == '/')
         {
@@ -164,6 +191,8 @@ void file_extract (int cmd, const char *fname, const char *kname)
     {
         sysno = sysno_next++;
         dict_insert (file_idx, kname, sizeof(sysno), &sysno);
+        lseek (sys_idx_fd, sysno * SYS_IDX_ENTRY_LEN, SEEK_SET);
+        write (sys_idx_fd, kname, strlen(kname)+1);
     }
     else
         memcpy (&sysno, (char*) file_info+1, sizeof(sysno));