X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=be70f55e7ed43cf86f521e63b25737a781097315;hb=3d4f0f1ddbc8b1ec6df244bb400184f086e211c8;hp=a726916ddfb773b07f7ea16359ebde1101b0545d;hpb=7e75317bed8eecabcb57e59b16093a32238738e2;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index a726916..be70f55 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,10 +1,26 @@
-/*
- * Copyright (C) 1994-2002, Index Data 
- * All rights reserved.
- * Sebastian Hammer, Adam Dickmeiss
- *
- * $Id: extract.c,v 1.110 2002-02-20 17:30:01 adam Exp $
- */
+/* $Id: extract.c,v 1.128 2002-10-24 13:07:02 heikki Exp $
+   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
+   Index Data Aps
+
+This file is part of the Zebra server.
+
+Zebra is free software; you can redistribute it and/or modify it under
+the terms of the GNU General Public License as published by the Free
+Software Foundation; either version 2, or (at your option) any later
+version.
+
+Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
+WARRANTY; without even the implied warranty of MERCHANTABILITY or
+FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+for more details.
+
+You should have received a copy of the GNU General Public License
+along with Zebra; see the file LICENSE.zebra.  If not, write to the
+Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
+02111-1307, USA.
+*/
+
+
 #include <stdio.h>
 #include <assert.h>
 #ifdef WIN32
@@ -14,12 +30,9 @@
 #endif
 #include <fcntl.h>
 
-#include <recctrl.h>
-#include <charmap.h>
-#include <sortidx.h>
 #include "index.h"
-#include "zserver.h"
-#include "zinfo.h"
+#include <direntz.h>
+#include <charmap.h>
 
 #if _FILE_OFFSET_BITS == 64
 #define PRINTF_OFF_T "%Ld"
@@ -27,35 +40,48 @@
 #define PRINTF_OFF_T "%ld"
 #endif
 
-#ifndef ZEBRASDR
-#define ZEBRASDR 0
-#endif
+#define USE_SHELLSORT 0
 
-#if ZEBRASDR
-#include "zebrasdr.h"
+#if USE_SHELLSORT
+static void shellsort(void *ar, int r, size_t s,
+                      int (*cmp)(const void *a, const void *b))
+{
+    char *a = ar;
+    char v[100];
+    int h, i, j, k;
+    static const int incs[16] = { 1391376, 463792, 198768, 86961, 33936,
+                                  13776, 4592, 1968, 861, 336, 
+                                  112, 48, 21, 7, 3, 1 };
+    for ( k = 0; k < 16; k++)
+        for (h = incs[k], i = h; i < r; i++)
+        { 
+            memcpy (v, a+s*i, s);
+            j = i;
+            while (j > h && (*cmp)(a + s*(j-h), v) > 0)
+            {
+                memcpy (a + s*j, a + s*(j-h), s);
+                j -= h;
+            }
+            memcpy (a+s*j, v, s);
+        } 
+}
 #endif
 
-static int records_inserted = 0;
-static int records_updated = 0;
-static int records_deleted = 0;
-static int records_processed = 0;
-
-static void logRecord (int showFlag)
+static void logRecord (ZebraHandle zh)
 {
-    if (!showFlag)
-        ++records_processed;
-    if (showFlag || !(records_processed % 1000))
+    ++zh->records_processed;
+    if (!(zh->records_processed % 1000))
     {
         logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d", 
-              records_processed, records_inserted, records_updated,
-              records_deleted);
+              zh->records_processed, zh->records_inserted, zh->records_updated,
+              zh->records_deleted);
     }
 }
 
 static void extract_init (struct recExtractCtrl *p, RecWord *w)
 {
     w->zebra_maps = p->zebra_maps;
-    w->seqnos = p->seqno;
+    w->seqno = 1;
     w->attrSet = VAL_BIB1;
     w->attrUse = 1016;
     w->reg_type = 'w';
@@ -82,7 +108,7 @@ static const char **searchRecordKey (ZebraHandle zh,
         ws[i] = NULL;
 
 #if SU_SCHEME
-    chS = zebraExplain_lookupSU (zh->service->zei, attrSetS, attrUseS);
+    chS = zebraExplain_lookupSU (zh->reg->zei, attrSetS, attrUseS);
     if (chS < 0)
 	return ws;
 #endif
@@ -232,7 +258,7 @@ static void file_end (void *handle, off_t offset)
     p->file_moffset = offset;
 }
 
-static char *fileMatchStr (ZebraHandle zh, 
+static char *fileMatchStr (ZebraHandle zh,
                            struct recKeys *reckeys, struct recordGroup *rGroup,
                            const char *fname, const char *spec)
 {
@@ -271,11 +297,11 @@ static char *fileMatchStr (ZebraHandle zh,
 		attname_str[i] = '\0';
 	    }
 	    
-	    if ((attset = data1_get_attset (zh->service->dh, attset_str)))
+	    if ((attset = data1_get_attset (zh->reg->dh, attset_str)))
 	    {
 		data1_att *att;
 		attSet = attset->reference;
-		att = data1_getattbyname(zh->service->dh, attset, attname_str);
+		att = data1_getattbyname(zh->reg->dh, attset, attname_str);
 		if (att)
 		    attUse = att->value;
 		else
@@ -388,19 +414,6 @@ struct recordLogInfo {
     struct recordGroup *rGroup;
 };
      
-static void recordLogPreamble (int level, const char *msg, void *info)
-{
-    struct recordLogInfo *p = (struct recordLogInfo *) info;
-    FILE *outf = yaz_log_file ();
-
-    if (level & LOG_LOG)
-        return ;
-    fprintf (outf, "File %s, offset %d, type %s\n",
-             p->rGroup->recordType, p->recordOffset, p->fname);
-    log_event_start (NULL, NULL);
-}
-
-
 static int recordExtract (ZebraHandle zh,
                           SYSNO *sysno, const char *fname,
                           struct recordGroup *rGroup, int deleteFlag,
@@ -412,7 +425,6 @@ static int recordExtract (ZebraHandle zh,
     char *matchStr;
     SYSNO sysnotmp;
     Record rec;
-    struct recordLogInfo logInfo;
     off_t recordOffset = 0;
 
     if (fi->fd != -1)
@@ -422,10 +434,13 @@ static int recordExtract (ZebraHandle zh,
         /* we are going to read from a file, so prepare the extraction */
 	int i;
 
-	zh->keys.buf_used = 0;
-	zh->keys.prevAttrUse = -1;
-	zh->keys.prevAttrSet = -1;
-	zh->keys.prevSeqNo = 0;
+	zh->reg->keys.buf_used = 0;
+	zh->reg->keys.prevAttrUse = -1;
+	zh->reg->keys.prevAttrSet = -1;
+	zh->reg->keys.prevSeqNo = 0;
+	zh->reg->sortKeys.buf_used = 0;
+	zh->reg->sortKeys.buf_max = 0;
+	zh->reg->sortKeys.buf = 0;
 	
 	recordOffset = fi->file_moffset;
 	extractCtrl.offset = fi->file_moffset;
@@ -438,44 +453,56 @@ static int recordExtract (ZebraHandle zh,
 	extractCtrl.init = extract_init;
 	extractCtrl.tokenAdd = extract_token_add;
 	extractCtrl.schemaAdd = extract_schema_add;
-	extractCtrl.dh = zh->service->dh;
+	extractCtrl.dh = zh->reg->dh;
         extractCtrl.handle = zh;
 	for (i = 0; i<256; i++)
 	{
-	    if (zebra_maps_is_positioned(zh->service->zebra_maps, i))
+	    if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
 		extractCtrl.seqno[i] = 1;
 	    else
 		extractCtrl.seqno[i] = 0;
 	}
-	extractCtrl.zebra_maps = zh->service->zebra_maps;
+	extractCtrl.zebra_maps = zh->reg->zebra_maps;
 	extractCtrl.flagShowRecords = !rGroup->flagRw;
 
         if (!rGroup->flagRw)
             printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset);
-
-        logInfo.fname = fname;
-        logInfo.recordOffset = recordOffset;
-        logInfo.rGroup = rGroup;
-        log_event_start (recordLogPreamble, &logInfo);
+        if (rGroup->flagRw)
+        {
+            char msg[512];
+            sprintf (msg, "%s:" PRINTF_OFF_T , fname, recordOffset);
+            yaz_log_init_prefix2 (msg);
+        }
 
         r = (*recType->extract)(clientData, &extractCtrl);
 
-        log_event_start (NULL, NULL);
-
+        yaz_log_init_prefix2 (0);
 	if (r == RECCTRL_EXTRACT_EOF)
 	    return 0;
-	else if (r == RECCTRL_EXTRACT_ERROR)
+	else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
 	{
             /* error occured during extraction ... */
             if (rGroup->flagRw &&
-		records_processed < rGroup->fileVerboseLimit)
+		zh->records_processed < rGroup->fileVerboseLimit)
             {
                 logf (LOG_WARN, "fail %s %s " PRINTF_OFF_T, rGroup->recordType,
                       fname, recordOffset);
             }
             return 0;
         }
-        if (zh->keys.buf_used == 0)
+	else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
+	{
+            /* error occured during extraction ... */
+            if (rGroup->flagRw &&
+		zh->records_processed < rGroup->fileVerboseLimit)
+            {
+                logf (LOG_WARN, "no filter for %s %s " 
+                      PRINTF_OFF_T, rGroup->recordType,
+                      fname, recordOffset);
+            }
+            return 0;
+        }
+        if (zh->reg->keys.buf_used == 0)
         {
             /* the extraction process returned no information - the record
                is probably empty - unless flagShowRecords is in use */
@@ -499,11 +526,11 @@ static int recordExtract (ZebraHandle zh,
         {
             char *rinfo;
         
-            matchStr = fileMatchStr (zh, &zh->keys, rGroup, fname, 
+            matchStr = fileMatchStr (zh, &zh->reg->keys, rGroup, fname, 
                                      rGroup->recordId);
             if (matchStr)
             {
-                rinfo = dict_lookup (zh->service->matchDict, matchStr);
+                rinfo = dict_lookup (zh->reg->matchDict, matchStr);
                 if (rinfo)
                     memcpy (sysno, rinfo+1, sizeof(*sysno));
             }
@@ -525,47 +552,53 @@ static int recordExtract (ZebraHandle zh,
             logf (LOG_WARN, "cannot delete record above (seems new)");
             return 1;
         }
-        if (records_processed < rGroup->fileVerboseLimit)
+        if (zh->records_processed < rGroup->fileVerboseLimit)
             logf (LOG_LOG, "add %s %s " PRINTF_OFF_T, rGroup->recordType,
                   fname, recordOffset);
-        rec = rec_new (zh->service->records);
+        rec = rec_new (zh->reg->records);
 
         *sysno = rec->sysno;
 
-	recordAttr = rec_init_attr (zh->service->zei, rec);
+	recordAttr = rec_init_attr (zh->reg->zei, rec);
 
         if (matchStr)
         {
-            dict_insert (zh->service->matchDict, matchStr, sizeof(*sysno), sysno);
+            dict_insert (zh->reg->matchDict, matchStr, sizeof(*sysno), sysno);
         }
-        extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
-	extract_flushSortKeys (zh, *sysno, 1, &zh->sortKeys);
+	extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
+        extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
 
-        records_inserted++;
+        zh->records_inserted++;
     }
     else
     {
         /* record already exists */
         struct recKeys delkeys;
+        struct sortKeys sortKeys;
 
-        rec = rec_get (zh->service->records, *sysno);
+        rec = rec_get (zh->reg->records, *sysno);
         assert (rec);
 	
-	recordAttr = rec_init_attr (zh->service->zei, rec);
+	recordAttr = rec_init_attr (zh->reg->zei, rec);
 
 	if (recordAttr->runNumber ==
-            zebraExplain_runNumberIncrement (zh->service->zei, 0))
+            zebraExplain_runNumberIncrement (zh->reg->zei, 0))
 	{
-	    logf (LOG_LOG, "skipped %s %s " PRINTF_OFF_T, rGroup->recordType,
-		  fname, recordOffset);
-	    extract_flushSortKeys (zh, *sysno, -1, &zh->sortKeys);
+            yaz_log (LOG_LOG, "run number = %d", recordAttr->runNumber);
+	    yaz_log (LOG_LOG, "skipped %s %s " PRINTF_OFF_T,
+                     rGroup->recordType, fname, recordOffset);
+	    extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
 	    rec_rm (&rec);
-	    logRecord (0);
+	    logRecord (zh);
 	    return 1;
 	}
         delkeys.buf_used = rec->size[recInfo_delKeys];
 	delkeys.buf = rec->info[recInfo_delKeys];
-	extract_flushSortKeys (zh, *sysno, 0, &zh->sortKeys);
+
+        sortKeys.buf_used = rec->size[recInfo_sortKeys];
+        sortKeys.buf = rec->info[recInfo_sortKeys];
+
+	extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
         extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
         if (deleteFlag)
         {
@@ -578,16 +611,16 @@ static int recordExtract (ZebraHandle zh,
             }
             else
             {
-                if (records_processed < rGroup->fileVerboseLimit)
+                if (zh->records_processed < rGroup->fileVerboseLimit)
                     logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T,
                          rGroup->recordType, fname, recordOffset);
-                records_deleted++;
+                zh->records_deleted++;
                 if (matchStr)
-                    dict_delete (zh->service->matchDict, matchStr);
-                rec_del (zh->service->records, &rec);
+                    dict_delete (zh->reg->matchDict, matchStr);
+                rec_del (zh->reg->records, &rec);
             }
 	    rec_rm (&rec);
-            logRecord (0);
+            logRecord (zh);
             return 1;
         }
         else
@@ -601,11 +634,12 @@ static int recordExtract (ZebraHandle zh,
             }
             else
             {
-                if (records_processed < rGroup->fileVerboseLimit)
+                if (zh->records_processed < rGroup->fileVerboseLimit)
                     logf (LOG_LOG, "update %s %s " PRINTF_OFF_T,
                         rGroup->recordType, fname, recordOffset);
-                extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
-                records_updated++;
+                extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
+                extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
+                zh->records_updated++;
             }
         }
     }
@@ -621,19 +655,12 @@ static int recordExtract (ZebraHandle zh,
 
     /* update delete keys */
     xfree (rec->info[recInfo_delKeys]);
-    if (zh->keys.buf_used > 0 && rGroup->flagStoreKeys == 1)
+    if (zh->reg->keys.buf_used > 0 && rGroup->flagStoreKeys == 1)
     {
-#if 1
-        rec->size[recInfo_delKeys] = zh->keys.buf_used;
-        rec->info[recInfo_delKeys] = zh->keys.buf;
-        zh->keys.buf = NULL;
-        zh->keys.buf_max = 0;
-#else
-        rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
-        rec->size[recInfo_delKeys] = reckeys.buf_used;
-        memcpy (rec->info[recInfo_delKeys], reckeys.buf,
-                rec->size[recInfo_delKeys]);
-#endif
+        rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
+        rec->info[recInfo_delKeys] = zh->reg->keys.buf;
+        zh->reg->keys.buf = NULL;
+        zh->reg->keys.buf_max = 0;
     }
     else
     {
@@ -641,17 +668,25 @@ static int recordExtract (ZebraHandle zh,
         rec->size[recInfo_delKeys] = 0;
     }
 
+    /* update sort keys */
+    xfree (rec->info[recInfo_sortKeys]);
+
+    rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
+    rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
+    zh->reg->sortKeys.buf = NULL;
+    zh->reg->sortKeys.buf_max = 0;
+
     /* save file size of original record */
-    zebraExplain_recordBytesIncrement (zh->service->zei,
+    zebraExplain_recordBytesIncrement (zh->reg->zei,
                                        - recordAttr->recordSize);
     recordAttr->recordSize = fi->file_moffset - recordOffset;
     if (!recordAttr->recordSize)
 	recordAttr->recordSize = fi->file_max - recordOffset;
-    zebraExplain_recordBytesIncrement (zh->service->zei,
+    zebraExplain_recordBytesIncrement (zh->reg->zei,
                                        recordAttr->recordSize);
 
     /* set run-number for this record */
-    recordAttr->runNumber = zebraExplain_runNumberIncrement (zh->service->zei,
+    recordAttr->runNumber = zebraExplain_runNumberIncrement (zh->reg->zei,
                                                              0);
 
     /* update store data */
@@ -689,8 +724,8 @@ static int recordExtract (ZebraHandle zh,
     recordAttr->recordOffset = recordOffset;
     
     /* commit this record */
-    rec_put (zh->service->records, &rec);
-    logRecord (0);
+    rec_put (zh->reg->records, &rec);
+    logRecord (zh);
     return 1;
 }
 
@@ -731,22 +766,22 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
     if (!rGroup->recordType)
     {
         sprintf (ext_res, "%srecordType.%s", gprefix, ext);
-        if (!(rGroup->recordType = res_get (zh->service->res, ext_res)))
+        if (!(rGroup->recordType = res_get (zh->res, ext_res)))
         {
             sprintf (ext_res, "%srecordType", gprefix);
-            rGroup->recordType = res_get (zh->service->res, ext_res);
+            rGroup->recordType = res_get (zh->res, ext_res);
         }
     }
     if (!rGroup->recordType)
     {
-        if (records_processed < rGroup->fileVerboseLimit)
+        if (zh->records_processed < rGroup->fileVerboseLimit)
             logf (LOG_LOG, "? %s", fname);
         return 0;
     }
     if (!*rGroup->recordType)
 	return 0;
     if (!(recType =
-	  recType_byName (zh->service->recTypes, rGroup->recordType, subType,
+	  recType_byName (zh->reg->recTypes, rGroup->recordType, subType,
 			  &clientData)))
     {
         logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
@@ -757,17 +792,17 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
     if (!rGroup->recordId)
     {
         sprintf (ext_res, "%srecordId.%s", gprefix, ext);
-        rGroup->recordId = res_get (zh->service->res, ext_res);
+        rGroup->recordId = res_get (zh->res, ext_res);
     }
 
     /* determine database name */
     if (!rGroup->databaseName)
     {
         sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
-        if (!(rGroup->databaseName = res_get (zh->service->res, ext_res)))
+        if (!(rGroup->databaseName = res_get (zh->res, ext_res)))
         {
             sprintf (ext_res, "%sdatabase", gprefix);
-            rGroup->databaseName = res_get (zh->service->res, ext_res);
+            rGroup->databaseName = res_get (zh->res, ext_res);
         }
     }
     if (!rGroup->databaseName)
@@ -777,12 +812,12 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
     
     sprintf (ext_res, "%sexplainDatabase", gprefix);
     rGroup->explainDatabase =
-	atoi (res_get_def (zh->service->res, ext_res, "0"));
+	atoi (res_get_def (zh->res, ext_res, "0"));
 
     /* announce database */
-    if (zebraExplain_curDatabase (zh->service->zei, rGroup->databaseName))
+    if (zebraExplain_curDatabase (zh->reg->zei, rGroup->databaseName))
     {
-        if (zebraExplain_newDatabase (zh->service->zei, rGroup->databaseName,
+        if (zebraExplain_newDatabase (zh->reg->zei, rGroup->databaseName,
 				      rGroup->explainDatabase))
 	    return 0;
     }
@@ -791,10 +826,10 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
     {
         const char *sval;
         sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
-        if (!(sval = res_get (zh->service->res, ext_res)))
+        if (!(sval = res_get (zh->res, ext_res)))
         {
             sprintf (ext_res, "%sstoreData", gprefix);
-            sval = res_get (zh->service->res, ext_res);
+            sval = res_get (zh->res, ext_res);
         }
         if (sval)
             rGroup->flagStoreData = atoi (sval);
@@ -807,14 +842,14 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
         const char *sval;
 
         sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
-        sval = res_get (zh->service->res, ext_res);
+        sval = res_get (zh->res, ext_res);
 	if (!sval)
         {
             sprintf (ext_res, "%sstoreKeys", gprefix);
-            sval = res_get (zh->service->res, ext_res);
+            sval = res_get (zh->res, ext_res);
         }
 	if (!sval)
-	    sval = res_get (zh->service->res, "storeKeys");
+	    sval = res_get (zh->res, "storeKeys");
         if (sval)
             rGroup->flagStoreKeys = atoi (sval);
     }
@@ -825,9 +860,21 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
         fd = -1;
     else
     {
-        if ((fd = open (fname, O_BINARY|O_RDONLY)) == -1)
+        char full_rep[1024];
+
+        if (zh->path_reg && !yaz_is_abspath (fname))
         {
-            logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
+            strcpy (full_rep, zh->path_reg);
+            strcat (full_rep, "/");
+            strcat (full_rep, fname);
+        }
+        else
+            strcpy (full_rep, fname);
+        
+
+        if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
+        {
+            logf (LOG_WARN|LOG_ERRNO, "open %s", full_rep);
             return 0;
         }
     }
@@ -855,7 +902,7 @@ int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
     RecordAttr *recordAttr;
     struct recExtractCtrl extractCtrl;
     int i, r;
-    char *matchStr;
+    char *matchStr = 0;
     RecType recType;
     char subType[1024];
     void *clientData;
@@ -879,36 +926,36 @@ int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
     extractCtrl.fh = &fc;
 
     /* announce database */
-    if (zebraExplain_curDatabase (zh->service->zei, databaseName))
+    if (zebraExplain_curDatabase (zh->reg->zei, databaseName))
     {
-        if (zebraExplain_newDatabase (zh->service->zei, databaseName, 0))
+        if (zebraExplain_newDatabase (zh->reg->zei, databaseName, 0))
 	    return 0;
     }
     if (!(recType =
-	  recType_byName (zh->service->recTypes, recordType, subType,
+	  recType_byName (zh->reg->recTypes, recordType, subType,
 			  &clientData)))
     {
         logf (LOG_WARN, "No such record type: %s", recordType);
         return 0;
     }
 
-    zh->keys.buf_used = 0;
-    zh->keys.prevAttrUse = -1;
-    zh->keys.prevAttrSet = -1;
-    zh->keys.prevSeqNo = 0;
-    zh->sortKeys = 0;
+    zh->reg->keys.buf_used = 0;
+    zh->reg->keys.prevAttrUse = -1;
+    zh->reg->keys.prevAttrSet = -1;
+    zh->reg->keys.prevSeqNo = 0;
+    zh->reg->sortKeys.buf_used = 0;
 
     extractCtrl.subType = subType;
     extractCtrl.init = extract_init;
     extractCtrl.tokenAdd = extract_token_add;
     extractCtrl.schemaAdd = extract_schema_add;
-    extractCtrl.dh = zh->service->dh;
+    extractCtrl.dh = zh->reg->dh;
     extractCtrl.handle = zh;
-    extractCtrl.zebra_maps = zh->service->zebra_maps;
+    extractCtrl.zebra_maps = zh->reg->zebra_maps;
     extractCtrl.flagShowRecords = 0;
     for (i = 0; i<256; i++)
     {
-	if (zebra_maps_is_positioned(zh->service->zebra_maps, i))
+	if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
 	    extractCtrl.seqno[i] = 1;
 	else
 	    extractCtrl.seqno[i] = 0;
@@ -918,22 +965,19 @@ int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
 
     if (r == RECCTRL_EXTRACT_EOF)
 	return 0;
-    else if (r == RECCTRL_EXTRACT_ERROR)
+    else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
     {
 	/* error occured during extraction ... */
-#if 1
-	yaz_log (LOG_WARN, "extract error");
-#else
-	if (rGroup->flagRw &&
-	    records_processed < rGroup->fileVerboseLimit)
-	{
-	    logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
-		  fname, (long) recordOffset);
-	}
-#endif
+	yaz_log (LOG_WARN, "extract error: generic");
+	return 0;
+    }
+    else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
+    {
+	/* error occured during extraction ... */
+	yaz_log (LOG_WARN, "extract error: no such filter");
 	return 0;
     }
-    if (zh->keys.buf_used == 0)
+    if (zh->reg->keys.buf_used == 0)
     {
 	/* the extraction process returned no information - the record
 	   is probably empty - unless flagShowRecords is in use */
@@ -945,7 +989,6 @@ int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
     }
     /* match criteria */
 
-
     if (! *sysno)
     {
         /* new record */
@@ -958,32 +1001,32 @@ int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
         }
 	logf (LOG_LOG, "add %s %s %ld", recordType, fname,
 	      (long) recordOffset);
-        rec = rec_new (zh->service->records);
+        rec = rec_new (zh->reg->records);
 
         *sysno = rec->sysno;
 
-	recordAttr = rec_init_attr (zh->service->zei, rec);
+	recordAttr = rec_init_attr (zh->reg->zei, rec);
 
         if (matchStr)
         {
-            dict_insert (zh->service->matchDict, matchStr,
+            dict_insert (zh->reg->matchDict, matchStr,
                          sizeof(*sysno), sysno);
         }
-        extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
-	extract_flushSortKeys (zh, *sysno, 1, &zh->sortKeys);
+	extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
+        extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
     }
     else
     {
         /* record already exists */
         struct recKeys delkeys;
 
-        rec = rec_get (zh->service->records, *sysno);
+        rec = rec_get (zh->reg->records, *sysno);
         assert (rec);
 	
-	recordAttr = rec_init_attr (zh->service->zei, rec);
+	recordAttr = rec_init_attr (zh->reg->zei, rec);
 
 	if (recordAttr->runNumber ==
-	    zebraExplain_runNumberIncrement (zh->service->zei, 0))
+	    zebraExplain_runNumberIncrement (zh->reg->zei, 0))
 	{
 	    logf (LOG_LOG, "skipped %s %s %ld", recordType,
 		  fname, (long) recordOffset);
@@ -992,7 +1035,7 @@ int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
 	}
         delkeys.buf_used = rec->size[recInfo_delKeys];
 	delkeys.buf = rec->info[recInfo_delKeys];
-	extract_flushSortKeys (zh, *sysno, 0, &zh->sortKeys);
+	extract_flushSortKeys (zh, *sysno, 0, &zh->reg->sortKeys);
         extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
         if (delete_flag)
         {
@@ -1011,7 +1054,7 @@ int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
                 if (matchStr)
                     dict_delete (matchDict, matchStr);
 #endif
-                rec_del (zh->service->records, &rec);
+                rec_del (zh->reg->records, &rec);
             }
 	    rec_rm (&rec);
             return 1;
@@ -1029,7 +1072,8 @@ int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
             {
 		logf (LOG_LOG, "update %s %s %ld", recordType,
 		      fname, (long) recordOffset);
-                extract_flushRecordKeys (zh, *sysno, 1, &zh->keys);
+                extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
+                extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
             }
         }
     }
@@ -1045,12 +1089,12 @@ int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
 
     /* update delete keys */
     xfree (rec->info[recInfo_delKeys]);
-    if (zh->keys.buf_used > 0 && store_keys == 1)
+    if (zh->reg->keys.buf_used > 0 && store_keys == 1)
     {
-        rec->size[recInfo_delKeys] = zh->keys.buf_used;
-        rec->info[recInfo_delKeys] = zh->keys.buf;
-        zh->keys.buf = NULL;
-        zh->keys.buf_max = 0;
+        rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
+        rec->info[recInfo_delKeys] = zh->reg->keys.buf;
+        zh->reg->keys.buf = NULL;
+        zh->reg->keys.buf_max = 0;
     }
     else
     {
@@ -1059,7 +1103,7 @@ int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
     }
 
     /* save file size of original record */
-    zebraExplain_recordBytesIncrement (zh->service->zei,
+    zebraExplain_recordBytesIncrement (zh->reg->zei,
 				       - recordAttr->recordSize);
 #if 0
     recordAttr->recordSize = fi->file_moffset - recordOffset;
@@ -1068,12 +1112,12 @@ int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
 #else
     recordAttr->recordSize = buf_size;
 #endif
-    zebraExplain_recordBytesIncrement (zh->service->zei,
+    zebraExplain_recordBytesIncrement (zh->reg->zei,
 				       recordAttr->recordSize);
 
     /* set run-number for this record */
     recordAttr->runNumber =
-	zebraExplain_runNumberIncrement (zh->service->zei, 0);
+	zebraExplain_runNumberIncrement (zh->reg->zei, 0);
 
     /* update store data */
     xfree (rec->info[recInfo_storeData]);
@@ -1114,7 +1158,7 @@ int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
     recordAttr->recordOffset = recordOffset;
     
     /* commit this record */
-    rec_put (zh->service->records, &rec);
+    rec_put (zh->reg->records, &rec);
 
     return 0;
 }
@@ -1125,53 +1169,62 @@ int explain_extract (void *handle, Record rec, data1_node *n)
     struct recExtractCtrl extractCtrl;
     int i;
 
-    if (zebraExplain_curDatabase (zh->service->zei,
+    if (zebraExplain_curDatabase (zh->reg->zei,
 				  rec->info[recInfo_databaseName]))
     {
 	abort();
-        if (zebraExplain_newDatabase (zh->service->zei,
+        if (zebraExplain_newDatabase (zh->reg->zei,
 				      rec->info[recInfo_databaseName], 0))
             abort ();
     }
 
-    zh->keys.buf_used = 0;
-    zh->keys.prevAttrUse = -1;
-    zh->keys.prevAttrSet = -1;
-    zh->keys.prevSeqNo = 0;
-    zh->sortKeys = 0;
+    zh->reg->keys.buf_used = 0;
+    zh->reg->keys.prevAttrUse = -1;
+    zh->reg->keys.prevAttrSet = -1;
+    zh->reg->keys.prevSeqNo = 0;
+    zh->reg->sortKeys.buf_used = 0;
     
     extractCtrl.init = extract_init;
     extractCtrl.tokenAdd = extract_token_add;
     extractCtrl.schemaAdd = extract_schema_add;
-    extractCtrl.dh = zh->service->dh;
+    extractCtrl.dh = zh->reg->dh;
     for (i = 0; i<256; i++)
 	extractCtrl.seqno[i] = 0;
-    extractCtrl.zebra_maps = zh->service->zebra_maps;
+    extractCtrl.zebra_maps = zh->reg->zebra_maps;
     extractCtrl.flagShowRecords = 0;
     extractCtrl.handle = handle;
     
     grs_extract_tree(&extractCtrl, n);
 
-    logf (LOG_LOG, "flush explain record, sysno=%d", rec->sysno);
-
     if (rec->size[recInfo_delKeys])
     {
 	struct recKeys delkeys;
-	struct sortKey *sortKeys = 0;
+	struct sortKeys sortkeys;
 
 	delkeys.buf_used = rec->size[recInfo_delKeys];
 	delkeys.buf = rec->info[recInfo_delKeys];
-	extract_flushSortKeys (zh, rec->sysno, 0, &sortKeys);
+
+	sortkeys.buf_used = rec->size[recInfo_sortKeys];
+	sortkeys.buf = rec->info[recInfo_sortKeys];
+
+	extract_flushSortKeys (zh, rec->sysno, 0, &sortkeys);
 	extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
     }
-    extract_flushRecordKeys (zh, rec->sysno, 1, &zh->keys);
-    extract_flushSortKeys (zh, rec->sysno, 1, &zh->sortKeys);
+    extract_flushRecordKeys (zh, rec->sysno, 1, &zh->reg->keys);
+    extract_flushSortKeys (zh, rec->sysno, 1, &zh->reg->sortKeys);
 
     xfree (rec->info[recInfo_delKeys]);
-    rec->size[recInfo_delKeys] = zh->keys.buf_used;
-    rec->info[recInfo_delKeys] = zh->keys.buf;
-    zh->keys.buf = NULL;
-    zh->keys.buf_max = 0;
+    rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
+    rec->info[recInfo_delKeys] = zh->reg->keys.buf;
+    zh->reg->keys.buf = NULL;
+    zh->reg->keys.buf_max = 0;
+
+    xfree (rec->info[recInfo_sortKeys]);
+    rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
+    rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
+    zh->reg->sortKeys.buf = NULL;
+    zh->reg->sortKeys.buf_max = 0;
+
     return 0;
 }
 
@@ -1186,16 +1239,23 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
     int seqno = 0;
     int off = 0;
     int ch = 0;
-    ZebraExplainInfo zei = zh->service->zei;
+    ZebraExplainInfo zei = zh->reg->zei;
 
-    if (!zh->key_buf)
+    if (!zh->reg->key_buf)
     {
-	int mem = 8*1024*1024;
-	zh->key_buf = (char**) xmalloc (mem);
-	zh->ptr_top = mem/sizeof(char*);
-	zh->ptr_i = 0;
-	zh->key_buf_used = 0;
-	zh->key_file_no = 0;
+	int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
+	if (mem <= 0)
+	{
+	    logf(LOG_WARN, "Invalid memory setting, using default 8 MB");
+	    mem= 1024*1024*8;
+	}
+	/* FIXME: That "8" should be in a default settings include */
+	/* not hard-coded here! -H */
+	zh->reg->key_buf = (char**) xmalloc (mem);
+	zh->reg->ptr_top = mem/sizeof(char*);
+	zh->reg->ptr_i = 0;
+	zh->reg->key_buf_used = 0;
+	zh->reg->key_file_no = 0;
     }
     zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
     while (off < reckeys->buf_used)
@@ -1224,11 +1284,12 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
             src += sizeof(attrUse);
         }
 #endif
-        if (zh->key_buf_used + 1024 > (zh->ptr_top-zh->ptr_i)*sizeof(char*))
+        if (zh->reg->key_buf_used + 1024 > 
+            (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
             extract_flushWriteKeys (zh);
-        ++(zh->ptr_i);
-        (zh->key_buf)[zh->ptr_top - zh->ptr_i] =
-	    (char*)zh->key_buf + zh->key_buf_used;
+        ++(zh->reg->ptr_i);
+        (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
+	    (char*)zh->reg->key_buf + zh->reg->key_buf_used;
 #if SU_SCHEME
 #else
         ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
@@ -1236,14 +1297,15 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
             ch = zebraExplain_addSU (zei, attrSet, attrUse);
 #endif
         assert (ch > 0);
-	zh->key_buf_used +=
-	    key_SU_encode (ch,((char*)zh->key_buf) + zh->key_buf_used);
+	zh->reg->key_buf_used +=
+	    key_SU_encode (ch,((char*)zh->reg->key_buf) +
+                           zh->reg->key_buf_used);
 
         while (*src)
-            ((char*)zh->key_buf) [(zh->key_buf_used)++] = *src++;
+            ((char*)zh->reg->key_buf) [(zh->reg->key_buf_used)++] = *src++;
         src++;
-        ((char*)(zh->key_buf))[(zh->key_buf_used)++] = '\0';
-        ((char*)(zh->key_buf))[(zh->key_buf_used)++] = cmd;
+        ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = '\0';
+        ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = cmd;
 
         if (lead & 60)
             seqno += ((lead>>2) & 15)-1;
@@ -1254,8 +1316,8 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
         }
         key.seqno = seqno;
         key.sysno = sysno;
-        memcpy ((char*)zh->key_buf + zh->key_buf_used, &key, sizeof(key));
-        (zh->key_buf_used) += sizeof(key);
+        memcpy ((char*)zh->reg->key_buf + zh->reg->key_buf_used, &key, sizeof(key));
+        (zh->reg->key_buf_used) += sizeof(key);
         off = src - reckeys->buf;
     }
     assert (off == reckeys->buf_used);
@@ -1267,36 +1329,37 @@ void extract_flushWriteKeys (ZebraHandle zh)
     char out_fname[200];
     char *prevcp, *cp;
     struct encode_info encode_info;
-    int ptr_i = zh->ptr_i;
+    int ptr_i = zh->reg->ptr_i;
 #if SORT_EXTRA
     int i;
 #endif
-    if (!zh->key_buf || ptr_i <= 0)
+    if (!zh->reg->key_buf || ptr_i <= 0)
         return;
 
-    (zh->key_file_no)++;
-    logf (LOG_LOG, "sorting section %d", (zh->key_file_no));
+    (zh->reg->key_file_no)++;
+    logf (LOG_LOG, "sorting section %d", (zh->reg->key_file_no));
 #if !SORT_EXTRA
-    qsort (zh->key_buf + zh->ptr_top - ptr_i, ptr_i, sizeof(char*),
-	    key_qsort_compare);
-    extract_get_fname_tmp (zh, out_fname, zh->key_file_no);
+    qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
+               sizeof(char*), key_qsort_compare);
+    extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
 
     if (!(outf = fopen (out_fname, "wb")))
     {
         logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
         exit (1);
     }
-    logf (LOG_LOG, "writing section %d", zh->key_file_no);
-    prevcp = cp = (zh->key_buf)[zh->ptr_top - ptr_i];
+    logf (LOG_LOG, "writing section %d", zh->reg->key_file_no);
+    prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
     
     encode_key_init (&encode_info);
     encode_key_write (cp, &encode_info, outf);
     
     while (--ptr_i > 0)
     {
-        cp = (zh->key_buf)[zh->ptr_top - ptr_i];
+        cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
         if (strcmp (cp, prevcp))
         {
+            encode_key_flush ( &encode_info, outf);
             encode_key_init (&encode_info);
             encode_key_write (cp, &encode_info, outf);
             prevcp = cp;
@@ -1304,6 +1367,7 @@ void extract_flushWriteKeys (ZebraHandle zh)
         else
             encode_key_write (cp + strlen(cp), &encode_info, outf);
     }
+    encode_key_flush ( &encode_info, outf);
 #else
     qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
     extract_get_fname_tmp (out_fname, key_file_no);
@@ -1335,6 +1399,7 @@ void extract_flushWriteKeys (ZebraHandle zh)
                 cp = key_buf[ptr_top-ptr_i];
                 encode_key_write (cp+key_y_len, &encode_info, outf);
             }
+            encode_key_flush ( &encode_info, outf);
             if (!i)
                 break;
             prevcp = key_buf[ptr_top-ptr_i];
@@ -1345,9 +1410,9 @@ void extract_flushWriteKeys (ZebraHandle zh)
         logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
         exit (1);
     }
-    logf (LOG_LOG, "finished section %d", zh->key_file_no);
-    zh->ptr_i = 0;
-    zh->key_buf_used = 0;
+    logf (LOG_LOG, "finished section %d", zh->reg->key_file_no);
+    zh->reg->ptr_i = 0;
+    zh->reg->key_buf_used = 0;
 }
 
 void extract_add_index_string (RecWord *p, const char *string,
@@ -1358,11 +1423,11 @@ void extract_add_index_string (RecWord *p, const char *string,
     unsigned short attrUse;
     int lead = 0;
     int diff = 0;
-    int *pseqno = &p->seqnos[p->reg_type];
+    int *pseqno = &p->seqno;
     ZebraHandle zh = p->extractCtrl->handle;
-    ZebraExplainInfo zei = zh->service->zei;
-    struct recKeys *keys = &zh->keys;
-
+    ZebraExplainInfo zei = zh->reg->zei;
+    struct recKeys *keys = &zh->reg->keys;
+    
     if (keys->buf_used+1024 > keys->buf_max)
     {
         char *b;
@@ -1403,7 +1468,7 @@ void extract_add_index_string (RecWord *p, const char *string,
         if (ch < 0)
         {
             ch = zebraExplain_addSU (zei, attrSet, attrUse);
-            yaz_log (LOG_LOG, "addSU set=%d use=%d SU=%d",
+            yaz_log (LOG_DEBUG, "addSU set=%d use=%d SU=%d",
                      attrSet, attrUse, ch);
         }
 	assert (ch > 0);
@@ -1433,24 +1498,60 @@ void extract_add_index_string (RecWord *p, const char *string,
         dst += sizeof(*pseqno);
     }
     keys->buf_used = dst - keys->buf;
-    if (*pseqno)
-	(*pseqno)++;
 }
 
 static void extract_add_sort_string (RecWord *p, const char *string,
 				     int length)
 {
+#if 1
+    ZebraHandle zh = p->extractCtrl->handle;
+    struct sortKeys *sk = &zh->reg->sortKeys;
+    size_t off = 0;
+    int slen;
+
+    while (off < sk->buf_used)
+    {
+        int set, use, l;
+
+        l = key_SU_decode(&set, sk->buf + off);
+        off += l;
+        l = key_SU_decode(&use, sk->buf + off);
+        off += l;
+        l = key_SU_decode(&slen, sk->buf + off);
+        off += l + slen;
+        if (p->attrSet == set && p->attrUse == use)
+            return;
+    }
+    assert (off == sk->buf_used);
+    
+    if (sk->buf_used + IT_MAX_WORD > sk->buf_max)
+    {
+        char *b;
+        
+        b = (char *) xmalloc (sk->buf_max += 128000);
+        if (sk->buf_used > 0)
+            memcpy (b, sk->buf, sk->buf_used);
+        xfree (sk->buf);
+        sk->buf = b;
+    }
+    off += key_SU_encode(p->attrSet, sk->buf + off);
+    off += key_SU_encode(p->attrUse, sk->buf + off);
+    slen = strlen(string);
+    off += key_SU_encode(slen, sk->buf + off);
+    memcpy (sk->buf + off, string, slen);
+    sk->buf_used = off + slen;
+#else
     struct sortKey *sk;
     ZebraHandle zh = p->extractCtrl->handle;
-    struct sortKey *sortKeys = zh->sortKeys;
 
-    for (sk = sortKeys; sk; sk = sk->next)
+
+    for (sk = zh->reg->sortKeys; sk; sk = sk->next)
 	if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
 	    return;
 
     sk = (struct sortKey *) xmalloc (sizeof(*sk));
-    sk->next = sortKeys;
-    sortKeys = sk;
+    sk->next = zh->reg->sortKeys;
+    zh->reg->sortKeys = sk;
 
     sk->string = (char *) xmalloc (length);
     sk->length = length;
@@ -1458,6 +1559,7 @@ static void extract_add_sort_string (RecWord *p, const char *string,
 
     sk->attrSet = p->attrSet;
     sk->attrUse = p->attrUse;
+#endif
 }
 
 void extract_add_string (RecWord *p, const char *string, int length)
@@ -1510,8 +1612,8 @@ static void extract_add_incomplete_field (RecWord *p)
 	if (!i)
 	    return;
 	extract_add_string (p, buf, i);
+        p->seqno++;
     }
-    (p->seqnos[p->reg_type])++; /* to separate this from next one  */
 }
 
 static void extract_add_complete_field (RecWord *p)
@@ -1578,26 +1680,32 @@ void extract_token_add (RecWord *p)
 void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
 {
     ZebraHandle zh = (ZebraHandle) (p->handle);
-    zebraExplain_addSchema (zh->service->zei, oid);
+    zebraExplain_addSchema (zh->reg->zei, oid);
 }
 
 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
-                            int cmd, struct sortKey **skp)
+                            int cmd, struct sortKeys *sk)
 {
-    struct sortKey *sk = *skp;
-    SortIdx sortIdx = zh->service->sortIdx;
+    SortIdx sortIdx = zh->reg->sortIdx;
+    size_t off = 0;
 
     sortIdx_sysno (sortIdx, sysno);
-    while (sk)
+
+    while (off < sk->buf_used)
     {
-	struct sortKey *sk_next = sk->next;
-	sortIdx_type (sortIdx, sk->attrUse);
-	sortIdx_add (sortIdx, sk->string, sk->length);
-	xfree (sk->string);
-	xfree (sk);
-	sk = sk_next;
+        int set, use, slen, l;
+        
+        off += key_SU_decode(&set, sk->buf + off);
+        off += key_SU_decode(&use, sk->buf + off);
+        off += key_SU_decode(&slen, sk->buf + off);
+        
+        sortIdx_type(sortIdx, use);
+        if (cmd == 1)
+            sortIdx_add(sortIdx, sk->buf + off, slen);
+        else
+            sortIdx_add(sortIdx, "", 1);
+        off += slen;
     }
-    *skp = 0;
 }
 
 void encode_key_init (struct encode_info *i)
@@ -1605,6 +1713,10 @@ void encode_key_init (struct encode_info *i)
     i->sysno = 0;
     i->seqno = 0;
     i->cmd = -1;
+    i->prevsys=0;
+    i->prevseq=0;
+    i->prevcmd=-1;
+    i->keylen=0;
 }
 
 char *encode_key_int (int d, char *bp)
@@ -1632,6 +1744,11 @@ char *encode_key_int (int d, char *bp)
     return bp;
 }
 
+#ifdef OLDENCODE
+/* this is the old encode_key_write 
+ * may be deleted once we are confident that the new works
+ * HL 15-oct-2002
+ */
 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
 {
     struct it_key key;
@@ -1658,3 +1775,108 @@ void encode_key_write (char *k, struct encode_info *i, FILE *outf)
     }
 }
 
+void encode_key_flush (struct encode_info *i, FILE *outf)
+{ /* dummy routine */
+}
+
+#else
+
+/* new encode_key_write
+ * The idea is to buffer one more key, and compare them
+ * If we are going to delete and insert the same key, 
+ * we may as well not bother. Should make a difference in 
+ * updates with small modifications (appending to a mbox)
+ */
+void encode_key_write (char *k, struct encode_info *i, FILE *outf)
+{
+    struct it_key key;
+    char *bp; 
+
+    if (*k)  /* first time for new key */
+    {
+        bp = i->buf;
+        while ((*bp++ = *k++))
+            ;
+	i->keylen= bp - i->buf -1;    
+	assert(i->keylen+1+sizeof(struct it_key) < ENCODE_BUFLEN);
+    }
+    else
+    {
+	bp=i->buf + i->keylen;
+	*bp++=0;
+	k++;
+    }
+
+    memcpy (&key, k+1, sizeof(struct it_key));
+    if (0==i->prevsys) /* no previous filter, fill up */
+    {
+        i->prevsys=key.sysno;
+	i->prevseq=key.seqno;
+	i->prevcmd=*k;
+    }
+    else if ( (i->prevsys==key.sysno) &&
+              (i->prevseq==key.seqno) &&
+	      (i->prevcmd!=*k) )
+    { /* same numbers, diff cmd, they cancel out */
+        i->prevsys=0;
+    }
+    else 
+    { /* different stuff, write previous, move buf */
+        bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
+	if (i->sysno != i->prevsys)
+	{
+	    i->sysno = i->prevsys;
+	    i->seqno = 0;
+        }
+        else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
+	{
+	    return; /* ??? Filters some sort of duplicates away */
+	            /* ??? Can this ever happen   -H 15oct02 */
+	}
+        bp = encode_key_int (i->prevseq - i->seqno, bp);
+        i->seqno = i->prevseq;
+        i->cmd = i->prevcmd;
+        if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
+        {
+            logf (LOG_FATAL|LOG_ERRNO, "fwrite");
+            exit (1);
+        }
+        i->keylen=0; /* ok, it's written, forget it */
+	i->prevsys=key.sysno;
+	i->prevseq=key.seqno;
+	i->prevcmd=*k;
+    }
+}
+
+void encode_key_flush (struct encode_info *i, FILE *outf)
+{ /* flush the last key from i */
+    char *bp =i->buf + i->keylen;
+    if (0==i->prevsys)
+    {
+        return; /* nothing to flush */
+    }
+    *bp++=0;
+    bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
+    if (i->sysno != i->prevsys)
+    {
+        i->sysno = i->prevsys;
+        i->seqno = 0;
+    }
+    else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
+    {
+        return; /* ??? Filters some sort of duplicates away */
+                /* ??? Can this ever happen   -H 15oct02 */
+    }
+    bp = encode_key_int (i->prevseq - i->seqno, bp);
+    i->seqno = i->prevseq;
+    i->cmd = i->prevcmd;
+    if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
+    {
+        logf (LOG_FATAL|LOG_ERRNO, "fwrite");
+        exit (1);
+    }
+    i->keylen=0; /* ok, it's written, forget it */
+    i->prevsys=0; /* forget the values too */
+    i->prevseq=0;
+}
+#endif