Charmap work

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

diff --git a/index/extract.c b/index/extract.c
index 0662621..c6897dc 100644 (file)
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,10 +1,27 @@
 /*
- * Copyright (C) 1994-1995, Index Data I/S 
+ * Copyright (C) 1994-1996, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: extract.c,v $
- * Revision 1.55  1996-05-09 07:28:55  quinn
+ * Revision 1.60  1996-06-04 10:18:12  adam
+ * Search/scan uses character mapping module.
+ *
+ * Revision 1.59  1996/05/14  15:47:07  adam
+ * Cleanup of various buffer size entities.
+ *
+ * Revision 1.58  1996/05/14  06:16:38  adam
+ * Compact use/set bytes used in search service.
+ *
+ * Revision 1.57  1996/05/13 14:23:04  adam
+ * Work on compaction of set/use bytes in dictionary.
+ *
+ * Revision 1.56  1996/05/09  09:54:42  adam
+ * Server supports maps from one logical attributes to a list of physical
+ * attributes.
+ * The extraction process doesn't make space consuming 'any' keys.
+ *
+ * Revision 1.55  1996/05/09  07:28:55  quinn
  * Work towards phrases and multiple registers
  *
  * Revision 1.54  1996/05/01  13:46:35  adam
@@ -202,13 +219,12 @@
 #include <assert.h>
 #include <unistd.h>
 #include <fcntl.h>
-#include <ctype.h>
 
 #include <alexutil.h>
 #include <recctrl.h>
 #include "index.h"
 
-#include "recindex.h"
+#include "zinfo.h"
 
 static Dict matchDict;
 
@@ -225,6 +241,8 @@ static int records_updated = 0;
 static int records_deleted = 0;
 static int records_processed = 0;
 
+static ZebTargetInfo *zti = NULL;
+
 static void logRecord (int showFlag)
 {
     if (!showFlag)
@@ -255,12 +273,15 @@ void key_open (int mem)
     }
     assert (!records);
     records = rec_open (1);
+#if 1
+    zti = zebTargetInfo_open (records, 1);
+#endif
 }
 
 struct encode_info {
     int  sysno;
     int  seqno;
-    char buf[512];
+    char buf[768];
 };
 
 void encode_key_init (struct encode_info *i)
@@ -368,6 +389,9 @@ int key_close (void)
 {
     key_flush ();
     xfree (key_buf);
+#if 1
+    zebTargetInfo_close (zti, 1);
+#endif
     rec_close (&records);
     dict_close (matchDict);
 
@@ -421,32 +445,35 @@ static void addRecordKey (const RecWord *p)
     else
         reckeys.prevAttrUse = attrUse;
 
+    *dst++ = lead;
+
+    if (!(lead & 1))
+    {
+        memcpy (dst, &attrSet, sizeof(attrSet));
+        dst += sizeof(attrSet);
+    }
+    if (!(lead & 2))
+    {
+        memcpy (dst, &attrUse, sizeof(attrUse));
+        dst += sizeof(attrUse);
+    }
     switch (p->which)
     {
-    case Word_String: case Word_Phrase:
-        *dst++ = lead;
-
-        if (!(lead & 1))
-        {
-            memcpy (dst, &attrSet, sizeof(attrSet));
-            dst += sizeof(attrSet);
-        }
-        if (!(lead & 2))
-        {
-            memcpy (dst, &attrUse, sizeof(attrUse));
-            dst += sizeof(attrUse);
-        }
-        for (i = 0; p->u.string[i]; i++)
-            *dst++ = p->u.string[i];
-        *dst++ = '\0';
-
-        memcpy (dst, &p->seqno, sizeof(p->seqno));
-        dst += sizeof(p->seqno);
-
-        break;
-    default:
-        return;
+        case Word_String:
+            *dst++ = 'w';
+            break;
+        case Word_Phrase:
+            *dst++ = 'p';
+            break;
+        case Word_Numeric:
+            *dst++ = 'n';
     }
+    for (i = 0; p->u.string[i]; i++)
+        *dst++ = p->u.string[i];
+    *dst++ = '\0';
+
+    memcpy (dst, &p->seqno, sizeof(p->seqno));
+    dst += sizeof(p->seqno);
     reckeys.buf_used = dst - reckeys.buf;
 }
 
@@ -456,6 +483,12 @@ static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys,
     char attrSet = -1;
     short attrUse = -1;
     int off = 0;
+
+    if (zebTargetInfo_curDatabase (zti, databaseName))
+    {
+        if (zebTargetInfo_newDatabase (zti, databaseName))
+            abort ();
+    }
     while (off < reckeys->buf_used)
     {
         const char *src = reckeys->buf + off;
@@ -478,10 +511,14 @@ static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys,
             key_flush ();
         ++ptr_i;
         key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
-        key_buf_used += index_word_prefix ((char*)key_buf + key_buf_used,
-                                           attrSet, attrUse, databaseName);
+
+        lead = zebTargetInfo_lookupSU (zti, attrSet, attrUse);
+        if (lead < 0)
+            lead = zebTargetInfo_addSU (zti, attrSet, attrUse);
+        assert (lead > 0);
+        ((char*) key_buf) [key_buf_used++] = lead;
         while (*src)
-            ((char*)key_buf) [key_buf_used++] = index_char_cvt (*src++);
+            ((char*)key_buf) [key_buf_used++] = *src++;
         src++;
         ((char*)key_buf) [key_buf_used++] = '\0';
         
@@ -550,21 +587,6 @@ static const char **searchRecordKey (struct recKeys *reckeys,
     return ws;
 }
 
-static void addRecordKeyAny (const RecWord *p)
-{
-    if (p->attrSet != 1 || p->attrUse != 1016)
-    {
-        RecWord w;
-
-        memcpy (&w, p, sizeof(w));
-        w.attrSet = 1;
-        w.attrUse = 1016;
-        addRecordKey (&w);
-    }
-    if (p->attrSet != -1)
-        addRecordKey (p);
-}
-
 struct file_read_info {
     off_t file_max;
     off_t file_offset;
@@ -816,7 +838,7 @@ static int recordExtract (SYSNO *sysno, const char *fname,
         extractCtrl.fh = fi;
         extractCtrl.subType = subType;
         extractCtrl.init = wordInit;
-        extractCtrl.add = addRecordKeyAny;
+        extractCtrl.add = addRecordKey;
 
         reckeys.buf_used = 0;
         reckeys.prevAttrUse = -1;