Option: only one word key per file.

diff --git a/index/Makefile b/index/Makefile
index a30b8be..ec6c61b 100644 (file)
--- a/index/Makefile
+++ b/index/Makefile
@@ -1,7 +1,7 @@
 # Copyright (C) 1995, Index Data I/S 
 # All rights reserved.
 # Sebastian Hammer, Adam Dickmeiss
-# $Id: Makefile,v 1.6 1995-09-05 15:28:39 adam Exp $
+# $Id: Makefile,v 1.7 1995-09-06 16:11:15 adam Exp $
 
 SHELL=/bin/sh
 RANLIB=ranlib
@@ -12,7 +12,7 @@ TPROG1=index
 TPROG2=kdump
 TPROG3=zserver
 DEFS=$(INCLUDE)
-O1 = main.o dir.o trav.o extract.o kinput.o kcompare.o ksort.o
+O1 = main.o dir.o trav.o extract.o kinput.o kcompare.o ksort.o symtab.o
 O2 = kdump.o
 O3 = zserver.o kcompare.o zrpn.o zsets.o
 CPP=cc -E

diff --git a/index/dir.c b/index/dir.c
index 6d0e11d..867f2b4 100644 (file)
--- a/index/dir.c
+++ b/index/dir.c
@@ -1,10 +1,13 @@
 /*
- * Copyright (C) 1995, Index Data I/S 
+ * Copyright (C) 1994-1995, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: dir.c,v $
- * Revision 1.4  1995-09-04 12:33:41  adam
+ * Revision 1.5  1995-09-06 16:11:16  adam
+ * Option: only one word key per file.
+ *
+ * Revision 1.4  1995/09/04  12:33:41  adam
  * Various cleanup. YAZ util used instead.
  *
  * Revision 1.3  1995/09/01  14:06:35  adam
@@ -37,7 +40,7 @@ struct dir_entry *dir_open (const char *rep)
     size_t idx = 0;
     struct dir_entry *entry;
 
-    logf (LOG_DEBUG, "dir_open %s", rep);
+    logf (LOG_LOG, "dir_open %s", rep);
     if (!(dir = opendir(rep)))
     {
         logf (LOG_WARN|LOG_ERRNO, "opendir %s", rep);

diff --git a/index/extract.c b/index/extract.c
index a763b7b..15e0873 100644 (file)
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,10 +1,13 @@
 /*
- * Copyright (C) 1995, Index Data I/S 
+ * Copyright (C) 1994-1995, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: extract.c,v $
- * Revision 1.4  1995-09-05 15:28:39  adam
+ * Revision 1.5  1995-09-06 16:11:16  adam
+ * Option: only one word key per file.
+ *
+ * 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
@@ -54,7 +57,7 @@ void key_open (const char *fname)
         exit (1);
     }
     key_offset = 0;
-    if (!(file_idx = dict_open (FNAME_FILE_DICT, 10, 1)))
+    if (!(file_idx = dict_open (FNAME_FILE_DICT, 40, 1)))
     {
         logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
         exit (1);
@@ -123,7 +126,27 @@ void key_write (int cmd, struct it_key *k, const char *str)
     key_offset += sizeof(*k);
 }
 
-void text_extract (SYSNO sysno, int cmd, const char *fname)
+void key_write_x (struct strtab *t, int cmd, struct it_key *k, const char *str)
+{
+    void **oldinfo;
+
+    if (strtab_src (t, str, &oldinfo))
+        ((struct it_key *) *oldinfo)->seqno++;
+    else
+    {
+        *oldinfo = xmalloc (sizeof(*k));
+        memcpy (*oldinfo, k, sizeof(*k));
+        ((struct it_key *) *oldinfo)->seqno = 1;
+    }
+}
+
+void key_rec_flush (const char *str, void *info, void *data)
+{
+    key_write (*((int*) data), (struct it_key *)info, str);
+    xfree (info);
+}
+
+void text_extract (struct strtab *t, SYSNO sysno, int cmd, const char *fname)
 {
     FILE *inf;
     struct it_key k;
@@ -152,8 +175,10 @@ void text_extract (SYSNO sysno, int cmd, const char *fname)
             w[i] = 0;
             
             k.seqno = seqno++;
+#if IT_KEY_HAVE_FIELD
             k.field = 0;
-            key_write (cmd, &k, w);
+#endif
+            key_write_x (t, cmd, &k, w);
         }
         if (c == EOF)
             break;
@@ -169,6 +194,7 @@ void file_extract (int cmd, const char *fname, const char *kname)
     char ext_res[128];
     const char *file_type;
     void *file_info;
+    struct strtab *t;
 
     logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
     for (i = strlen(fname); --i >= 0; )
@@ -196,8 +222,10 @@ void file_extract (int cmd, const char *fname, const char *kname)
     }
     else
         memcpy (&sysno, (char*) file_info+1, sizeof(sysno));
+    t = strtab_mk ();
     if (!strcmp (file_type, "text"))
-        text_extract (sysno, cmd, fname);
+        text_extract (t, sysno, cmd, fname);
+    strtab_del (t, key_rec_flush, &cmd);
 }
 
 

diff --git a/index/index.h b/index/index.h
index d5d626a..4a95800 100644 (file)
--- a/index/index.h
+++ b/index/index.h
@@ -1,10 +1,13 @@
 /*
- * Copyright (C) 1995, Index Data I/S 
+ * Copyright (C) 1994-1995, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: index.h,v $
- * Revision 1.6  1995-09-05 15:28:39  adam
+ * Revision 1.7  1995-09-06 16:11:16  adam
+ * Option: only one word key per file.
+ *
+ * Revision 1.6  1995/09/05  15:28:39  adam
  * More work on search engine.
  *
  * Revision 1.5  1995/09/04  12:33:42  adam
@@ -30,10 +33,14 @@
 #include <dict.h>
 #include <isam.h>
 
+#define IT_KEY_HAVE_FIELD 0
+
 struct it_key {
     int sysno;
     int seqno;
+#if IT_KEY_HAVE_FIELD
     int field;
+#endif
 };
 
 struct dir_entry {
@@ -62,3 +69,9 @@ int key_sort (const char *key_fname, size_t mem);
 #define FNAME_FILE_DICT "filedict"
 #define FNAME_SYS_IDX "sysidx"
 #define SYS_IDX_ENTRY_LEN 120
+
+struct strtab *strtab_mk (void);
+int strtab_src (struct strtab *t, const char *name, void ***infop);
+void strtab_del (struct strtab *t,
+                 void (*func)(const char *name, void *info, void *data),
+                 void *data);

diff --git a/index/kcompare.c b/index/kcompare.c
index 5217ed9..7ce3a24 100644 (file)
--- a/index/kcompare.c
+++ b/index/kcompare.c
@@ -1,10 +1,13 @@
 /*
- * Copyright (C) 1995, Index Data I/S 
+ * Copyright (C) 1994-1995, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: kcompare.c,v $
- * Revision 1.1  1995-09-04 09:10:36  adam
+ * Revision 1.2  1995-09-06 16:11:17  adam
+ * Option: only one word key per file.
+ *
+ * Revision 1.1  1995/09/04  09:10:36  adam
  * More work on index add/del/update.
  * Merge sort implemented.
  * Initial work on z39 server.
@@ -25,17 +28,25 @@ int key_compare (const void *p1, const void *p2)
     memcpy (&i2, p2, sizeof(i2));
     if ( i1.sysno != i2.sysno)
         return i1.sysno - i2.sysno;
+#if IT_KEY_HAVE_FIELD
     if ( i1.seqno != i2.seqno)
         return i1.seqno - i2.seqno;
     return i1.field - i2.field;
+#else
+    return i1.seqno - i2.seqno;
+#endif
 }
 
 int key_compare_x (const struct it_key *i1, const struct it_key *i2)
 {
     if ( i1->sysno != i2->sysno)
         return i1->sysno - i2->sysno;
+#if IT_KEY_HAVE_FIELD
     if ( i1->seqno != i2->seqno)
         return i1->seqno - i2->seqno;
     return i1->field - i2->field;
+#else
+    return i1->seqno - i2->seqno;
+#endif
 }
 

diff --git a/index/kdump.c b/index/kdump.c
index b9f9ae0..a37d379 100644 (file)
--- a/index/kdump.c
+++ b/index/kdump.c
@@ -1,10 +1,13 @@
 /*
- * Copyright (C) 1995, Index Data I/S 
+ * Copyright (C) 1994-1995, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: kdump.c,v $
- * Revision 1.2  1995-09-04 12:33:42  adam
+ * 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
  * Various cleanup. YAZ util used instead.
  *
  * Revision 1.1  1995/09/04  09:10:36  adam

diff --git a/index/kinput.c b/index/kinput.c
index 9e6a773..9f834a9 100644 (file)
--- a/index/kinput.c
+++ b/index/kinput.c
@@ -1,10 +1,13 @@
 /*
- * Copyright (C) 1995, Index Data I/S 
+ * Copyright (C) 1994-1995, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: kinput.c,v $
- * Revision 1.2  1995-09-04 12:33:42  adam
+ * 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
  * Various cleanup. YAZ util used instead.
  *
  * Revision 1.1  1995/09/04  09:10:37  adam
@@ -130,7 +133,7 @@ void key_input (const char *dict_fname, const char *isam_fname,
         logf (LOG_FATAL, "dict_open fail of `%s'", dict_fname);
         exit (1);
     }
-    isam = is_open (isam_fname, key_compare, 1);
+    isam = is_open (isam_fname, key_compare, 1, sizeof(struct it_key));
     if (!isam)
     {
         logf (LOG_FATAL, "is_open fail of `%s'", isam_fname);

diff --git a/index/main.c b/index/main.c
index 71e8325..f8e1362 100644 (file)
--- a/index/main.c
+++ b/index/main.c
@@ -1,10 +1,13 @@
 /*
- * 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 1.7  1995-09-05 15:28:39  adam
+ * 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
@@ -108,11 +111,11 @@ int main (int argc, char **argv)
     key_flush ();
     if (!key_close ())
         exit (0);
-    logf (LOG_DEBUG, "Sorting");
-    if (!key_sort ("keys.tmp", 1000000))
+    logf (LOG_LOG, "Sorting");
+    if (!key_sort ("keys.tmp", 3000000))
         exit (0);
-    logf (LOG_DEBUG, "Input");
-    key_input (FNAME_WORD_DICT, FNAME_WORD_ISAM, "keys.tmp", 50);
+    logf (LOG_LOG, "Input");
+    key_input (FNAME_WORD_DICT, FNAME_WORD_ISAM, "keys.tmp", 60);
     exit (0);
 }
 

diff --git a/index/symtab.c b/index/symtab.c
new file mode 100644 (file)
index 0000000..d656621
--- /dev/null
+++ b/index/symtab.c
@@ -0,0 +1,79 @@
+/*
+ * Copyright (C) 1994-1995, Index Data I/S 
+ * All rights reserved.
+ * Sebastian Hammer, Adam Dickmeiss
+ *
+ * $Log: symtab.c,v $
+ * Revision 1.1  1995-09-06 16:11:18  adam
+ * Option: only one word key per file.
+ *
+ */
+#include <stdio.h>
+#include <stdlib.h>
+#include <assert.h>
+
+#include <alexutil.h>
+#include "index.h"
+
+struct strentry {
+    char *name;
+    void *info;
+    struct strentry *next;
+};
+
+#define STR_HASH 401
+
+struct strtab {
+    struct strentry *ar[STR_HASH];
+};
+
+struct strtab *strtab_mk (void)
+{
+    int i;
+    struct strtab *p = xmalloc (sizeof (*p));
+    for (i=0; i<STR_HASH; i++)
+        p->ar[i] = NULL;
+    return p;
+}
+
+int strtab_src (struct strtab *t, const char *name, void ***infop)
+{
+    unsigned hash = 0;
+    int i;
+    struct strentry *e;
+
+    for (i=0; name[i]; i++)
+        hash += hash*65519 + name[i];
+    hash = hash % STR_HASH;
+    for (e = t->ar[hash]; e; e = e->next)
+        if (!strcmp(e->name, name))
+        {
+            *infop = &e->info;
+            return 1;
+        }
+    e = xmalloc (sizeof(*e));
+    e->name = xmalloc (strlen(name)+1);
+    strcpy (e->name, name);
+    e->next = t->ar[hash];
+    t->ar[hash] = e;
+    *infop = &e->info;
+    return 0;
+}
+
+void strtab_del (struct strtab *t,
+                 void (*func)(const char *name, void *info, void *data),
+                 void *data)
+{
+    int i;
+    struct strentry *e, *e1;
+
+    for (i = 0; i<STR_HASH; i++)
+        for (e = t->ar[i]; e; e = e1)
+        {
+            e1 = e->next;
+            (*func)(e->name, e->info, data);
+            free (e->name);
+            free (e);
+        }
+    free (t);
+}

diff --git a/index/trav.c b/index/trav.c
index 4accc88..b87dbab 100644 (file)
--- a/index/trav.c
+++ b/index/trav.c
@@ -1,10 +1,13 @@
 /*
- * Copyright (C) 1995, Index Data I/S 
+ * Copyright (C) 1994-1995, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: trav.c,v $
- * Revision 1.2  1995-09-04 12:33:43  adam
+ * Revision 1.3  1995-09-06 16:11:18  adam
+ * Option: only one word key per file.
+ *
+ * Revision 1.2  1995/09/04  12:33:43  adam
  * Various cleanup. YAZ util used instead.
  *
  * Revision 1.1  1995/09/01  14:06:36  adam

diff --git a/index/zrpn.c b/index/zrpn.c
index 46d7afd..23b8243 100644 (file)
--- a/index/zrpn.c
+++ b/index/zrpn.c
@@ -1,10 +1,13 @@
 /*
- * Copyright (C) 1995, Index Data I/S 
+ * Copyright (C) 1994-1995, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: zrpn.c,v $
- * Revision 1.5  1995-09-06 10:33:04  adam
+ * Revision 1.6  1995-09-06 16:11:18  adam
+ * Option: only one word key per file.
+ *
+ * Revision 1.5  1995/09/06  10:33:04  adam
  * More work on present. Some log messages removed.
  *
  * Revision 1.4  1995/09/05  15:28:40  adam

diff --git a/index/zserver.c b/index/zserver.c
index 8fa470f..204f5e4 100644 (file)
--- a/index/zserver.c
+++ b/index/zserver.c
@@ -1,10 +1,13 @@
 /*
- * Copyright (C) 1995, Index Data I/S 
+ * Copyright (C) 1994-1995, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: zserver.c,v $
- * Revision 1.4  1995-09-06 10:33:04  adam
+ * Revision 1.5  1995-09-06 16:11:18  adam
+ * Option: only one word key per file.
+ *
+ * Revision 1.4  1995/09/06  10:33:04  adam
  * More work on present. Some log messages removed.
  *
  * Revision 1.3  1995/09/05  15:28:40  adam
@@ -47,20 +50,21 @@ bend_initresult *bend_init (bend_initrequest *q)
         r.errstring = "dict_open fail: filedict";
         return &r;
     }
-    if (!(server_info.fileDict = dict_open (FNAME_FILE_DICT, 5, 0)))
+    if (!(server_info.fileDict = dict_open (FNAME_FILE_DICT, 10, 0)))
     {
         r.errcode = 1;
         r.errstring = "dict_open fail: filedict";
         return &r;
     }    
-    if (!(server_info.wordDict = dict_open (FNAME_WORD_DICT, 20, 0)))
+    if (!(server_info.wordDict = dict_open (FNAME_WORD_DICT, 40, 0)))
     {
         dict_close (server_info.fileDict);
         r.errcode = 1;
         r.errstring = "dict_open fail: worddict";
         return &r;
     }    
-    if (!(server_info.wordIsam = is_open (FNAME_WORD_ISAM, key_compare, 0)))
+    if (!(server_info.wordIsam = is_open (FNAME_WORD_ISAM, key_compare, 0,
+                                          sizeof (struct it_key))))
     {
         dict_close (server_info.wordDict);
         dict_close (server_info.fileDict);

diff --git a/index/zserver.h b/index/zserver.h
index 33e566b..5190026 100644 (file)
--- a/index/zserver.h
+++ b/index/zserver.h
@@ -1,10 +1,13 @@
 /*
- * Copyright (C) 1995, Index Data I/S 
+ * Copyright (C) 1994-1995, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: zserver.h,v $
- * Revision 1.1  1995-09-05 15:28:40  adam
+ * Revision 1.2  1995-09-06 16:11:19  adam
+ * Option: only one word key per file.
+ *
+ * Revision 1.1  1995/09/05  15:28:40  adam
  * More work on search engine.
  *
  */

diff --git a/index/zsets.c b/index/zsets.c
index 2963e4f..cc1c544 100644 (file)
--- a/index/zsets.c
+++ b/index/zsets.c
@@ -1,10 +1,13 @@
 /*
- * Copyright (C) 1995, Index Data I/S 
+ * Copyright (C) 1994-1995, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: zsets.c,v $
- * Revision 1.2  1995-09-06 10:33:04  adam
+ * Revision 1.3  1995-09-06 16:11:19  adam
+ * Option: only one word key per file.
+ *
+ * Revision 1.2  1995/09/06  10:33:04  adam
  * More work on present. Some log messages removed.
  *
  * Revision 1.1  1995/09/05  15:28:40  adam