X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=8358ade76267aafb7d93100dae38f678c74f3174;hb=b88909df16157ed1e7859bc3fad6b01520d4865e;hp=4484cdb9ae4d49b27529664b1b9d4692121007f2;hpb=6c9fcd3b5d3108702fa1ffc92dab4ab6060f9a19;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index 4484cdb..8358ade 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,5 +1,5 @@
-/* $Id: extract.c,v 1.171 2005-01-15 19:38:25 adam Exp $
-   Copyright (C) 1995-2005
+/* $Id: extract.c,v 1.220 2006-06-13 12:02:06 adam Exp $
+   Copyright (C) 1995-2006
    Index Data ApS
 
 This file is part of the Zebra server.
@@ -25,21 +25,60 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
 #include <ctype.h>
 #ifdef WIN32
 #include <io.h>
-#else
+#endif
+#if HAVE_UNISTD_H
 #include <unistd.h>
 #endif
 #include <fcntl.h>
 
 #include "index.h"
+#include "orddict.h"
 #include <direntz.h>
 #include <charmap.h>
 
-#if _FILE_OFFSET_BITS == 64
-#define PRINTF_OFF_T "%Ld"
+#ifdef WIN32
+#define PRINTF_OFF_T "%I64d"
+#else
+/* !WIN32 */
+#if SIZEOF_OFF_T == SIZEOF_LONG_LONG
+#define PRINTF_OFF_T "%lld"
 #else
 #define PRINTF_OFF_T "%ld"
 #endif
 
+#endif
+
+#define ENCODE_BUFLEN 768
+struct encode_info {
+    void *encode_handle;
+    void *decode_handle;
+    char buf[ENCODE_BUFLEN];
+};
+
+static int log_level = 0;
+static int log_level_initialized = 1;
+
+static void zebra_init_log_level()
+{
+    if (!log_level_initialized)
+    {
+        log_level = yaz_log_module_level("extract");
+        log_level_initialized = 1;
+    }
+}
+
+static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
+                                     int cmd, zebra_rec_keys_t reckeys,
+                                     zint staticrank);
+static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
+                                   int cmd, zebra_rec_keys_t skp);
+static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid);
+static void extract_token_add (RecWord *p);
+
+static void encode_key_init (struct encode_info *i);
+static void encode_key_write (char *k, struct encode_info *i, FILE *outf);
+static void encode_key_flush (struct encode_info *i, FILE *outf);
+
 #define USE_SHELLSORT 0
 
 #if USE_SHELLSORT
@@ -72,71 +111,74 @@ static void logRecord (ZebraHandle zh)
     ++zh->records_processed;
     if (!(zh->records_processed % 1000))
     {
-        yaz_log (YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
-			ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT, 
-              zh->records_processed, zh->records_inserted, zh->records_updated,
-              zh->records_deleted);
+        yaz_log(YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
+                ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT, 
+                zh->records_processed, zh->records_inserted, 
+                zh->records_updated, zh->records_deleted);
     }
 }
 
+static void extract_add_index_string (RecWord *p, const char *str, int length);
+
+static void extract_set_store_data_prepare(struct recExtractCtrl *p);
+
 static void extract_init (struct recExtractCtrl *p, RecWord *w)
 {
     w->zebra_maps = p->zebra_maps;
     w->seqno = 1;
-    w->attrSet = VAL_BIB1;
-    w->attrUse = 1016;
-    w->reg_type = 'w';
+    w->index_name = "any";
+    w->index_type = 'w';
     w->extractCtrl = p;
     w->record_id = 0;
     w->section_id = 0;
 }
 
-static const char **searchRecordKey (ZebraHandle zh,
-                                     struct recKeys *reckeys,
-				     int attrSetS, int attrUseS)
+static void searchRecordKey(ZebraHandle zh,
+			    zebra_rec_keys_t reckeys,
+                            const char *index_name,
+			    const char **ws, int ws_length)
 {
-    static const char *ws[32];
-    void *decode_handle = iscz1_start();
-    int off = 0;
-    int startSeq = -1;
-    int seqno = 0;
     int i;
+    int ch = -1;
 
-    for (i = 0; i<32; i++)
+    for (i = 0; i<ws_length; i++)
         ws[i] = NULL;
 
-    while (off < reckeys->buf_used)
-    {
-        const char *src = reckeys->buf + off;
-        struct it_key key;
-	char *dst = (char*) &key;
-	int attrSet, attrUse;
-
-	iscz1_decode(decode_handle, &dst, &src);
-	assert(key.len < 4 && key.len > 2);
+    if (ch < 0)
+        ch = zebraExplain_lookup_attr_str(zh->reg->zei, '0', index_name);
+    if (ch < 0)
+        ch = zebraExplain_lookup_attr_str(zh->reg->zei, 'p', index_name);
+    if (ch < 0)
+        ch = zebraExplain_lookup_attr_str(zh->reg->zei, 'w', index_name);
 
-	attrSet = (int) key.mem[0] >> 16;
-	attrUse = (int) key.mem[0] & 65535;
-	seqno = (int) key.mem[key.len-1];
-
-	if (attrUseS == attrUse && attrSetS == attrSet)
-        {
-            int woff;
+    if (ch < 0)
+	return ;
 
-            if (startSeq == -1)
-                startSeq = seqno;
-            woff = seqno - startSeq;
-            if (woff >= 0 && woff < 31)
-                ws[woff] = src;
-        }
+    if (zebra_rec_keys_rewind(reckeys))
+    {
+	zint startSeq = -1;
+	const char *str;
+	size_t slen;
+	struct it_key key;
+	zint seqno;
+	while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
+	{
+	    assert(key.len <= 4 && key.len > 2);
 
-        while (*src++)
-	    ;
-        off = src - reckeys->buf;
+	    seqno = key.mem[key.len-1];
+	    
+	    if (key.mem[0] == ch)
+	    {
+		zint woff;
+		
+		if (startSeq == -1)
+		    startSeq = seqno;
+		woff = seqno - startSeq;
+		if (woff >= 0 && woff < ws_length)
+		    ws[woff] = str;
+	    }
+	}
     }
-    iscz1_stop(decode_handle);
-    assert (off == reckeys->buf_used);
-    return ws;
 }
 
 struct file_read_info {
@@ -145,8 +187,6 @@ struct file_read_info {
     off_t file_moffset;     /* offset of rec/rec boundary */
     int file_more;
     int fd;
-    char *sdrbuf;
-    int sdrmax;
 };
 
 static struct file_read_info *file_read_start (int fd)
@@ -157,8 +197,8 @@ static struct file_read_info *file_read_start (int fd)
     fi->fd = fd;
     fi->file_max = 0;
     fi->file_moffset = 0;
-    fi->sdrbuf = 0;
-    fi->sdrmax = 0;
+    fi->file_offset = 0;
+    fi->file_more = 0;
     return fi;
 }
 
@@ -171,8 +211,6 @@ static off_t file_seek (void *handle, off_t offset)
 {
     struct file_read_info *p = (struct file_read_info *) handle;
     p->file_offset = offset;
-    if (p->sdrbuf)
-	return offset;
     return lseek (p->fd, offset, SEEK_SET);
 }
 
@@ -187,16 +225,7 @@ static int file_read (void *handle, char *buf, size_t count)
     struct file_read_info *p = (struct file_read_info *) handle;
     int fd = p->fd;
     int r;
-    if (p->sdrbuf)
-    {
-	r = count;
-	if (r > p->sdrmax - p->file_offset)
-	    r = p->sdrmax - p->file_offset;
-	if (r)
-	    memcpy (buf, p->sdrbuf + p->file_offset, r);
-    }
-    else
-	r = read (fd, buf, count);
+    r = read (fd, buf, count);
     if (r > 0)
     {
         p->file_offset += r;
@@ -206,83 +235,66 @@ static int file_read (void *handle, char *buf, size_t count)
     return r;
 }
 
-static void file_begin (void *handle)
-{
-    struct file_read_info *p = (struct file_read_info *) handle;
-
-    p->file_offset = p->file_moffset;
-    if (!p->sdrbuf && p->file_moffset)
-        lseek (p->fd, p->file_moffset, SEEK_SET);
-    p->file_more = 0;
-}
-
 static void file_end (void *handle, off_t offset)
 {
     struct file_read_info *p = (struct file_read_info *) handle;
 
-    assert (p->file_more == 0);
-    p->file_more = 1;
-    p->file_moffset = offset;
+    if (offset != p->file_moffset)
+    {
+	p->file_moffset = offset;
+	p->file_more = 1;
+    }
 }
 
+#define FILE_MATCH_BLANK "\t "
+
 static char *fileMatchStr (ZebraHandle zh,
-                           struct recKeys *reckeys,
+			   zebra_rec_keys_t reckeys,
                            const char *fname, const char *spec)
 {
     static char dstBuf[2048];      /* static here ??? */
     char *dst = dstBuf;
     const char *s = spec;
-    static const char **w;
 
     while (1)
     {
-        while (*s == ' ' || *s == '\t')
-            s++;
+	for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
+	    ;
         if (!*s)
             break;
         if (*s == '(')
         {
+	    const char *ws[32];
 	    char attset_str[64], attname_str[64];
-	    data1_attset *attset;
 	    int i;
-            char matchFlag[32];
-            int attSet = 1, attUse = 1;
             int first = 1;
-
-            s++;
-	    for (i = 0; *s && *s != ',' && *s != ')'; s++)
-		if (i < 63)
+	    
+	    for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
+		;
+	    for (i = 0; *s && *s != ',' && *s != ')' && 
+		     !strchr(FILE_MATCH_BLANK, *s); s++)
+		if (i+1 < sizeof(attset_str))
 		    attset_str[i++] = *s;
 	    attset_str[i] = '\0';
-
-	    if (*s == ',')
+	    
+	    for (; strchr(FILE_MATCH_BLANK, *s); s++)
+		;
+	    if (*s != ',')
+                strcpy(attname_str, attset_str);
+            else
 	    {
-		s++;
-		for (i = 0; *s && *s != ')'; s++)
-		    if (i < 63)
+		for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
+		    ;
+		for (i = 0; *s && *s != ')' && 
+			 !strchr(FILE_MATCH_BLANK, *s); s++)
+		    if (i+1 < sizeof(attname_str))
 			attname_str[i++] = *s;
 		attname_str[i] = '\0';
 	    }
-	    
-	    if ((attset = data1_get_attset (zh->reg->dh, attset_str)))
-	    {
-		data1_att *att;
-		attSet = attset->reference;
-		att = data1_getattbyname(zh->reg->dh, attset, attname_str);
-		if (att)
-		    attUse = att->value;
-		else
-		    attUse = atoi (attname_str);
-	    }
-            w = searchRecordKey (zh, reckeys, attSet, attUse);
-            assert (w);
 
-            if (*s == ')')
-            {
-                for (i = 0; i<32; i++)
-                    matchFlag[i] = 1;
-            }
-            else
+            searchRecordKey (zh, reckeys, attname_str, ws, 32);
+
+            if (*s != ')')
             {
                 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
                       spec, zh->m_group ? zh->m_group : "none");
@@ -291,15 +303,15 @@ static char *fileMatchStr (ZebraHandle zh,
             s++;
 
             for (i = 0; i<32; i++)
-                if (matchFlag[i] && w[i])
+                if (ws[i])
                 {
                     if (first)
                     {
                         *dst++ = ' ';
                         first = 0;
                     }
-                    strcpy (dst, w[i]);
-                    dst += strlen(w[i]);
+                    strcpy (dst, ws[i]);
+                    dst += strlen(ws[i]);
                 }
             if (first)
             {
@@ -314,12 +326,12 @@ static char *fileMatchStr (ZebraHandle zh,
             char special[64];
             const char *spec_src = NULL;
             const char *s1 = ++s;
-            while (*s1 && *s1 != ' ' && *s1 != '\t')
+            while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
                 s1++;
 
             spec_len = s1 - s;
-            if (spec_len > 63)
-                spec_len = 63;
+            if (spec_len > sizeof(special)-1)
+                spec_len = sizeof(special)-1;
             memcpy (special, s, spec_len);
             special[spec_len] = '\0';
             s = s1;
@@ -349,7 +361,7 @@ static char *fileMatchStr (ZebraHandle zh,
 
             while (*s && *s != stopMarker)
             {
-                if (i < 63)
+                if (i+1 < sizeof(tmpString))
                     tmpString[i++] = *s++;
             }
             if (*s)
@@ -382,55 +394,63 @@ struct recordLogInfo {
     struct recordGroup *rGroup;
 };
 
-void create_rec_keys_codec(struct recKeys *keys)
+static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
 {
-    keys->buf_used = 0;
-    iscz1_reset(keys->codec_handle);
+    int i;
+    for (i = 0; i<256; i++)
+    {
+	if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
+	    ctrl->seqno[i] = 1;
+	else
+	    ctrl->seqno[i] = 0;
+    }
+    ctrl->zebra_maps = zh->reg->zebra_maps;
+    ctrl->flagShowRecords = !zh->m_flag_rw;
 }
-     
-static int file_extract_record(ZebraHandle zh,
-			       SYSNO *sysno, const char *fname,
-			       int deleteFlag,
-			       struct file_read_info *fi,
-			       int force_update)
+
+static void all_matches_add(struct recExtractCtrl *ctrl)
+{
+    RecWord word;
+    extract_init(ctrl, &word);
+    word.index_name = "allrecords";
+    word.index_type = 'w';
+    word.seqno = 1;
+    extract_add_index_string (&word, "", 0);
+}
+
+static ZEBRA_RES file_extract_record(ZebraHandle zh,
+				     SYSNO *sysno, const char *fname,
+				     int deleteFlag,
+				     struct file_read_info *fi,
+				     int force_update,
+				     RecType recType,
+				     void *recTypeClientData)
 {
+    const char *match_str_to_print = "";
     RecordAttr *recordAttr;
     int r;
     const char *matchStr = 0;
     SYSNO sysnotmp;
     Record rec;
     off_t recordOffset = 0;
-    RecType recType;
-    void *clientData;
+    struct recExtractCtrl extractCtrl;
     
-    if (!(recType =
-	  recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
-			  &clientData)))
-    {
-        yaz_log (YLOG_WARN, "No such record type: %s", zh->m_record_type);
-        return 0;
-    }
-
     /* announce database */
     if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
     {
         if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
 				      zh->m_explain_database))
-	    return 0;
+	    return ZEBRA_FAIL;
     }
 
     if (fi->fd != -1)
     {
-	struct recExtractCtrl extractCtrl;
-
         /* we are going to read from a file, so prepare the extraction */
-	int i;
+	zebra_rec_keys_reset(zh->reg->keys);
 
-	create_rec_keys_codec(&zh->reg->keys);
-
-	zh->reg->sortKeys.buf_used = 0;
-	
+	zebra_rec_keys_reset(zh->reg->sortKeys);
 	recordOffset = fi->file_moffset;
+        extractCtrl.handle = zh;
 	extractCtrl.offset = fi->file_moffset;
 	extractCtrl.readf = file_read;
 	extractCtrl.seekf = file_seek;
@@ -442,16 +462,13 @@ static int file_extract_record(ZebraHandle zh,
 	extractCtrl.schemaAdd = extract_schema_add;
 	extractCtrl.dh = zh->reg->dh;
 	extractCtrl.match_criteria[0] = '\0';
-        extractCtrl.handle = zh;
-	for (i = 0; i<256; i++)
-	{
-	    if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
-		extractCtrl.seqno[i] = 1;
-	    else
-		extractCtrl.seqno[i] = 0;
-	}
-	extractCtrl.zebra_maps = zh->reg->zebra_maps;
-	extractCtrl.flagShowRecords = !zh->m_flag_rw;
+	extractCtrl.staticrank = 0;
+	
+	extractCtrl.first_record = fi->file_offset ? 0 : 1;
+
+	extract_set_store_data_prepare(&extractCtrl);
+
+	init_extractCtrl(zh, &extractCtrl);
 
         if (!zh->m_flag_rw)
             printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset);
@@ -462,11 +479,11 @@ static int file_extract_record(ZebraHandle zh,
             yaz_log_init_prefix2 (msg);
         }
 
-        r = (*recType->extract)(clientData, &extractCtrl);
+        r = (*recType->extract)(recTypeClientData, &extractCtrl);
 
         yaz_log_init_prefix2 (0);
 	if (r == RECCTRL_EXTRACT_EOF)
-	    return 0;
+	    return ZEBRA_FAIL;
 	else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
 	{
             /* error occured during extraction ... */
@@ -476,7 +493,7 @@ static int file_extract_record(ZebraHandle zh,
                 yaz_log (YLOG_WARN, "fail %s %s " PRINTF_OFF_T, zh->m_record_type,
                       fname, recordOffset);
             }
-            return 0;
+            return ZEBRA_FAIL;
         }
 	else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
 	{
@@ -488,25 +505,20 @@ static int file_extract_record(ZebraHandle zh,
                       PRINTF_OFF_T, zh->m_record_type,
                       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 */
-            if (!zh->m_flag_rw)
-                return 1;
-	    
-	    if (zh->records_processed < zh->m_file_verbose_limit)
-	        yaz_log (YLOG_WARN, "empty %s %s " PRINTF_OFF_T, zh->m_record_type,
-		    fname, recordOffset);
-            return 1;
+            return ZEBRA_FAIL;
         }
+        all_matches_add(&extractCtrl);
         if (extractCtrl.match_criteria[0])
             matchStr = extractCtrl.match_criteria;
     }
 
-    /* perform match if sysno not known and if match criteria is specified */
+    /* if matchStr is set now - we assume it's printable .
+       For internal matchStr (see below) we don't print */
+    if (matchStr)
+        match_str_to_print = matchStr;
+
+    /* perform internal match if sysno not known and if match criteria is
+       specified already */
     if (!sysno) 
     {
         sysnotmp = 0;
@@ -514,18 +526,24 @@ static int file_extract_record(ZebraHandle zh,
 
         if (matchStr == 0 && zh->m_record_id && *zh->m_record_id)
         {
-        
-            matchStr = fileMatchStr (zh, &zh->reg->keys, fname, 
+            matchStr = fileMatchStr (zh, zh->reg->keys, fname, 
 				     zh->m_record_id);
 	    if (!matchStr)
 	    {
 		yaz_log(YLOG_WARN, "Bad match criteria");
-		return 0;
+
+                if (zebra_rec_keys_empty(zh->reg->keys))
+                {
+                    yaz_log(YLOG_WARN, "And no index keys");
+                }
+		return ZEBRA_FAIL;
 	    }
 	}
 	if (matchStr)
 	{
-            char *rinfo = dict_lookup (zh->reg->matchDict, matchStr);
+	    int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
+            char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
+					  matchStr);
 	    if (rinfo)
 	    {
 		assert(*rinfo == sizeof(*sysno));
@@ -533,6 +551,18 @@ static int file_extract_record(ZebraHandle zh,
 	    }
 	}
     }
+    if (! *sysno && zebra_rec_keys_empty(zh->reg->keys)	)
+    {
+         /* the extraction process returned no information - the record
+            is probably empty - unless flagShowRecords is in use */
+         if (!zh->m_flag_rw)
+             return ZEBRA_OK;
+  
+         if (zh->records_processed < zh->m_file_verbose_limit)
+	     yaz_log (YLOG_WARN, "empty %s %s " PRINTF_OFF_T, zh->m_record_type,
+	    fname, recordOffset);
+         return ZEBRA_OK;
+    }
 
     if (! *sysno)
     {
@@ -542,98 +572,109 @@ static int file_extract_record(ZebraHandle zh,
 	    yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T, zh->m_record_type,
 		  fname, recordOffset);
             yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
-            return 1;
+            return ZEBRA_OK;
         }
-        if (zh->records_processed < zh->m_file_verbose_limit)
-            yaz_log (YLOG_LOG, "add %s %s " PRINTF_OFF_T, zh->m_record_type,
-                  fname, recordOffset);
-        rec = rec_new (zh->reg->records);
 
+        rec = rec_new (zh->reg->records);
+        
         *sysno = rec->sysno;
-
+        
+        if (zh->records_processed < zh->m_file_verbose_limit)
+        {
+            yaz_log(YLOG_LOG, "add %s %s " PRINTF_OFF_T 
+                    " " ZINT_FORMAT " %s" ,
+                    zh->m_record_type,
+                    fname, recordOffset, *sysno, match_str_to_print);
+        }
 	recordAttr = rec_init_attr (zh->reg->zei, rec);
+	recordAttr->staticrank = extractCtrl.staticrank;
 
         if (matchStr)
         {
-            dict_insert (zh->reg->matchDict, matchStr, sizeof(*sysno), sysno);
+	    int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
+            dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
+			    sizeof(*sysno), sysno);
         }
-	extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
-        extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
 
+
+	extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
+        extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
+				 recordAttr->staticrank);
         zh->records_inserted++;
     }
     else
     {
         /* record already exists */
-        struct recKeys delkeys;
-        struct sortKeys sortKeys;
+	zebra_rec_keys_t delkeys = zebra_rec_keys_open();
+
+	zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
 
         rec = rec_get (zh->reg->records, *sysno);
         assert (rec);
 	
 	recordAttr = rec_init_attr (zh->reg->zei, rec);
 
-	if (!force_update && recordAttr->runNumber ==
-            zebraExplain_runNumberIncrement (zh->reg->zei, 0))
-	{
-            yaz_log (YLOG_LOG, "run number = %d", recordAttr->runNumber);
-	    yaz_log (YLOG_LOG, "skipped %s %s " PRINTF_OFF_T,
-                     zh->m_record_type, fname, recordOffset);
-	    extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
-	    rec_rm (&rec);
-	    logRecord (zh);
-	    return 1;
-	}
-        delkeys.buf_used = rec->size[recInfo_delKeys];
-	delkeys.buf = rec->info[recInfo_delKeys];
-
-        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);
+	zebra_rec_keys_set_buf(delkeys,
+			       rec->info[recInfo_delKeys],
+			       rec->size[recInfo_delKeys],
+			       0);
+
+	zebra_rec_keys_set_buf(sortKeys,
+			       rec->info[recInfo_sortKeys],
+			       rec->size[recInfo_sortKeys],
+			       0);
+	extract_flushSortKeys (zh, *sysno, 0, sortKeys);
+        extract_flushRecordKeys (zh, *sysno, 0, delkeys,
+				 recordAttr->staticrank); /* old values */  
         if (deleteFlag)
         {
             /* record going to be deleted */
-            if (!delkeys.buf_used)
+            if (zebra_rec_keys_empty(delkeys))
             {
-                yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
-                      zh->m_record_type, fname, recordOffset);
-                yaz_log (YLOG_WARN, "cannot delete file above, storeKeys false");
+                yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T 
+                         " " ZINT_FORMAT,
+			 zh->m_record_type, fname, recordOffset, *sysno);
+                yaz_log (YLOG_WARN, "cannot delete file above, storeKeys false (1)");
             }
             else
             {
                 if (zh->records_processed < zh->m_file_verbose_limit)
-                    yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
-                         zh->m_record_type, fname, recordOffset);
+                {
+                    yaz_log(YLOG_LOG, "delete %s %s " PRINTF_OFF_T 
+                            " " ZINT_FORMAT " %s" ,
+                            zh->m_record_type,
+                            fname, recordOffset, *sysno, match_str_to_print);
+                }
                 zh->records_deleted++;
                 if (matchStr)
-                    dict_delete (zh->reg->matchDict, matchStr);
+		{
+		    int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
+                    dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
+		}
                 rec_del (zh->reg->records, &rec);
             }
 	    rec_rm (&rec);
             logRecord (zh);
-            return 1;
+            return ZEBRA_OK;
         }
         else
         {
-            /* record going to be updated */
-            if (!delkeys.buf_used)
+	    /* flush new keys for sort&search etc */
+            if (zh->records_processed < zh->m_file_verbose_limit)
             {
-                yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
-                      zh->m_record_type, fname, recordOffset);
-                yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
-            }
-            else
-            {
-                if (zh->records_processed < zh->m_file_verbose_limit)
-                    yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
-                        zh->m_record_type, fname, recordOffset);
-                extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
-                extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
-                zh->records_updated++;
+                yaz_log(YLOG_LOG, "update %s %s " PRINTF_OFF_T 
+                        " " ZINT_FORMAT " %s" ,
+                        zh->m_record_type,
+                        fname, recordOffset, *sysno, match_str_to_print);
             }
+	    recordAttr->staticrank = extractCtrl.staticrank;
+            extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
+            extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
+					 recordAttr->staticrank);
+            zh->records_updated++;
         }
+	zebra_rec_keys_close(delkeys);
+	zebra_rec_keys_close(sortKeys);
     }
     /* update file type */
     xfree (rec->info[recInfo_fileType]);
@@ -647,12 +688,11 @@ static int file_extract_record(ZebraHandle zh,
 
     /* update delete keys */
     xfree (rec->info[recInfo_delKeys]);
-    if (zh->reg->keys.buf_used > 0 && zh->m_store_keys == 1)
+    if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
     {
-        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;
+	zebra_rec_keys_get_buf(zh->reg->keys,
+			       &rec->info[recInfo_delKeys],
+			       &rec->size[recInfo_delKeys]);
     }
     else
     {
@@ -663,10 +703,9 @@ static int file_extract_record(ZebraHandle zh,
     /* 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;
+    zebra_rec_keys_get_buf(zh->reg->sortKeys,
+			   &rec->info[recInfo_sortKeys],
+			   &rec->size[recInfo_sortKeys]);
 
     /* save file size of original record */
     zebraExplain_recordBytesIncrement (zh->reg->zei,
@@ -683,7 +722,13 @@ static int file_extract_record(ZebraHandle zh,
 
     /* update store data */
     xfree (rec->info[recInfo_storeData]);
-    if (zh->m_store_data)
+    if (zh->store_data_buf)
+    {
+        rec->size[recInfo_storeData] = zh->store_data_size;
+        rec->info[recInfo_storeData] = zh->store_data_buf;
+	zh->store_data_buf = 0;
+    }
+    else if (zh->m_store_data)
     {
         rec->size[recInfo_storeData] = recordAttr->recordSize;
         rec->info[recInfo_storeData] = (char *)
@@ -718,25 +763,30 @@ static int file_extract_record(ZebraHandle zh,
     /* commit this record */
     rec_put (zh->reg->records, &rec);
     logRecord (zh);
-    return 1;
+    return ZEBRA_OK;
 }
 
-int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname, 
-		 int deleteFlag)
+ZEBRA_RES zebra_extract_file(ZebraHandle zh, SYSNO *sysno, const char *fname, 
+			     int deleteFlag)
 {
-    int r, i, fd;
+    ZEBRA_RES r = ZEBRA_OK;
+    int i, fd;
     char gprefix[128];
     char ext[128];
     char ext_res[128];
     struct file_read_info *fi;
     const char *original_record_type = 0;
+    RecType recType;
+    void *recTypeClientData;
+
+    zebra_init_log_level();
 
     if (!zh->m_group || !*zh->m_group)
         *gprefix = '\0';
     else
         sprintf (gprefix, "%s.", zh->m_group);
     
-    yaz_log (YLOG_DEBUG, "fileExtract %s", fname);
+    yaz_log(log_level, "zebra_extract_file %s", fname);
 
     /* determine file extension */
     *ext = '\0';
@@ -768,6 +818,21 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
         zh->m_record_id = res_get (zh->res, ext_res);
     }
 
+    if (!(recType =
+	  recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
+			  &recTypeClientData)))
+    {
+        yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
+        return ZEBRA_FAIL;
+    }
+
+    switch(recType->version)
+    {
+    case 0:
+	break;
+    default:
+	yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
+    }
     if (sysno && deleteFlag)
         fd = -1;
     else
@@ -783,20 +848,34 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
         else
             strcpy (full_rep, fname);
         
-
         if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
         {
             yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
 	    zh->m_record_type = original_record_type;
-            return 0;
+            return ZEBRA_FAIL;
         }
     }
     fi = file_read_start (fd);
-    do
+    while(1)
     {
-        file_begin (fi);
-        r = file_extract_record (zh, sysno, fname, deleteFlag, fi, 1);
-    } while (r && !sysno && fi->file_more);
+	fi->file_moffset = fi->file_offset;
+	fi->file_more = 0;  /* file_end not called (yet) */
+        r = file_extract_record (zh, sysno, fname, deleteFlag, fi, 1,
+				 recType, recTypeClientData);
+	if (fi->file_more)
+	{   /* file_end has been called so reset offset .. */
+	    fi->file_offset = fi->file_moffset;
+	    lseek(fi->fd, fi->file_moffset, SEEK_SET);
+	}
+	if (r != ZEBRA_OK)
+	{
+	    break;
+	}
+	if (sysno)
+	{
+	    break;
+	}
+    }
     file_read_stop (fi);
     if (fd != -1)
         close (fd);
@@ -810,20 +889,21 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
   If not, and a record is provided, then sysno is got from there
   
  */
-int buffer_extract_record (ZebraHandle zh, 
-			   const char *buf, size_t buf_size,
-			   int delete_flag,
-			   int test_mode, 
-			   const char *recordType,
-			   SYSNO *sysno,
-			   const char *match_criteria,
-			   const char *fname,
-			   int force_update,
-			   int allow_update)
+ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh, 
+                                      const char *buf, size_t buf_size,
+                                      int delete_flag,
+                                      int test_mode, 
+                                      const char *recordType,
+                                      SYSNO *sysno,
+                                      const char *match_criteria,
+                                      const char *fname,
+                                      int force_update,
+                                      int allow_update)
 {
+    SYSNO sysno0 = 0;
     RecordAttr *recordAttr;
     struct recExtractCtrl extractCtrl;
-    int i, r;
+    int r;
     const char *matchStr = 0;
     RecType recType = NULL;
     void *clientData;
@@ -831,6 +911,9 @@ int buffer_extract_record (ZebraHandle zh,
     long recordOffset = 0;
     struct zebra_fetch_control fc;
     const char *pr_fname = fname;  /* filename to print .. */
+    int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
+
+    zebra_init_log_level();
 
     if (!pr_fname)
 	pr_fname = "<no file>";  /* make it printable if file is omitted */
@@ -847,37 +930,43 @@ int buffer_extract_record (ZebraHandle zh,
     extractCtrl.seekf = zebra_record_int_seek;
     extractCtrl.tellf = zebra_record_int_tell;
     extractCtrl.endf = zebra_record_int_end;
+    extractCtrl.first_record = 1;
     extractCtrl.fh = &fc;
 
-    create_rec_keys_codec(&zh->reg->keys);
-
-    zh->reg->sortKeys.buf_used = 0;
+    zebra_rec_keys_reset(zh->reg->keys);
+    zebra_rec_keys_reset(zh->reg->sortKeys);
 
     if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
     {
         if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0], 
 				      zh->m_explain_database))
-            return 0;
+            return ZEBRA_FAIL;
     }
     
-    if (recordType && *recordType) {
-        yaz_log (YLOG_DEBUG, "Record type explicitly specified: %s", recordType);
+    if (recordType && *recordType)
+    {
+        yaz_log(log_level, "Record type explicitly specified: %s", recordType);
         recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
                                   &clientData);
-    } else {
-        if (!(zh->m_record_type)) {
+    } 
+    else
+    {
+        if (!(zh->m_record_type))
+	{
             yaz_log (YLOG_WARN, "No such record type defined");
-            return 0;
+            return ZEBRA_FAIL;
         }
-        yaz_log (YLOG_DEBUG, "Get record type from rgroup: %s",zh->m_record_type);
+        yaz_log(log_level, "Get record type from rgroup: %s",
+                zh->m_record_type);
         recType = recType_byName (zh->reg->recTypes, zh->res,
 				  zh->m_record_type, &clientData);
         recordType = zh->m_record_type;
     }
     
-    if (!recType) {
-        yaz_log (YLOG_WARN, "No such record type: %s", zh->m_record_type);
-        return 0;
+    if (!recType)
+    {
+        yaz_log (YLOG_WARN, "No such record type: %s", recordType);
+        return ZEBRA_FAIL;
     }
     
     extractCtrl.init = extract_init;
@@ -885,66 +974,57 @@ int buffer_extract_record (ZebraHandle zh,
     extractCtrl.schemaAdd = extract_schema_add;
     extractCtrl.dh = zh->reg->dh;
     extractCtrl.handle = zh;
-    extractCtrl.zebra_maps = zh->reg->zebra_maps;
-    extractCtrl.flagShowRecords = 0;
     extractCtrl.match_criteria[0] = '\0';
-    for (i = 0; i<256; i++)
-    {
-	if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
-	    extractCtrl.seqno[i] = 1;
-	else
-	    extractCtrl.seqno[i] = 0;
-    }
+    extractCtrl.staticrank = 0;
+    
+    init_extractCtrl(zh, &extractCtrl);
+
+    extract_set_store_data_prepare(&extractCtrl);
 
     r = (*recType->extract)(clientData, &extractCtrl);
 
     if (r == RECCTRL_EXTRACT_EOF)
-	return 0;
+	return ZEBRA_FAIL;
     else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
     {
 	/* error occured during extraction ... */
 	yaz_log (YLOG_WARN, "extract error: generic");
-	return 0;
+	return ZEBRA_FAIL;
     }
     else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
     {
 	/* error occured during extraction ... */
 	yaz_log (YLOG_WARN, "extract error: no such filter");
-	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 */
-	if (test_mode)
-	    return 1;
-	yaz_log (YLOG_WARN, "No keys generated for record");
-	yaz_log (YLOG_WARN, " The file is probably empty");
-	return 1;
+	return ZEBRA_FAIL;
     }
-    /* match criteria */
-    matchStr = NULL;
 
+    all_matches_add(&extractCtrl);
+        
     if (extractCtrl.match_criteria[0])
 	match_criteria = extractCtrl.match_criteria;
 
-    if (! *sysno) {
-        char *rinfo;
+    if (!sysno) {
+
+	sysno = &sysno0;
+
         if (match_criteria && *match_criteria) {
             matchStr = match_criteria;
         } else {
             if (zh->m_record_id && *zh->m_record_id) {
-                matchStr = fileMatchStr (zh, &zh->reg->keys, pr_fname, 
+                matchStr = fileMatchStr (zh, zh->reg->keys, pr_fname, 
                                          zh->m_record_id);
 		if (!matchStr)
                 {
                     yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
-		    return 1;
+		    return ZEBRA_FAIL;
                 }
             }
         }
-        if (matchStr) {
-            rinfo = dict_lookup (zh->reg->matchDict, matchStr);
+        if (matchStr) 
+	{
+	    int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
+	    char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
+					  matchStr);
             if (rinfo)
 	    {
 		assert(*rinfo == sizeof(*sysno));
@@ -952,6 +1032,13 @@ int buffer_extract_record (ZebraHandle zh,
 	    }
         }
     }
+    if (zebra_rec_keys_empty(zh->reg->keys))
+    {
+	/* the extraction process returned no information - the record
+	   is probably empty - unless flagShowRecords is in use */
+	if (test_mode)
+	    return ZEBRA_OK;
+    }
 
     if (! *sysno)
     {
@@ -959,107 +1046,103 @@ int buffer_extract_record (ZebraHandle zh,
         if (delete_flag)
         {
 	    yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
-		  pr_fname, (long) recordOffset);
+			 pr_fname, (long) recordOffset);
             yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
-            return 1;
+            return ZEBRA_FAIL;
         }
-	yaz_log (YLOG_LOG, "add %s %s %ld", recordType, pr_fname,
-	      (long) recordOffset);
+	if (show_progress)
+	    yaz_log (YLOG_LOG, "add %s %s %ld", recordType, pr_fname,
+		     (long) recordOffset);
         rec = rec_new (zh->reg->records);
 
         *sysno = rec->sysno;
 
 	recordAttr = rec_init_attr (zh->reg->zei, rec);
+	recordAttr->staticrank = extractCtrl.staticrank;
 
         if (matchStr)
         {
-            dict_insert (zh->reg->matchDict, matchStr,
-                         sizeof(*sysno), sysno);
+	    int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
+            dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
+			    sizeof(*sysno), sysno);
         }
-	extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
-        extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
 
+
+	extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
+        extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
+			 recordAttr->staticrank);
         zh->records_inserted++;
     } 
     else
     {
         /* record already exists */
-        struct recKeys delkeys;
-        struct sortKeys sortKeys;
-
-	if (!allow_update) {
-	      yaz_log (YLOG_LOG, "skipped %s %s %ld", 
-		    recordType, pr_fname, (long) recordOffset);
-	      logRecord(zh);
-	      return -1;
+	zebra_rec_keys_t delkeys = zebra_rec_keys_open();
+	zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
+	if (!allow_update)
+	{
+	    yaz_log (YLOG_LOG, "skipped %s %s %ld", 
+			 recordType, pr_fname, (long) recordOffset);
+	    logRecord(zh);
+	    return ZEBRA_FAIL;
 	}
 
         rec = rec_get (zh->reg->records, *sysno);
         assert (rec);
 	
 	recordAttr = rec_init_attr (zh->reg->zei, rec);
-	
-	if (!force_update) {
-	    if (recordAttr->runNumber ==
-		zebraExplain_runNumberIncrement (zh->reg->zei, 0))
-	    {
-		yaz_log (YLOG_LOG, "skipped %s %s %ld", recordType,
-		      pr_fname, (long) recordOffset);
-		extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
-		rec_rm (&rec);
-		logRecord(zh);
-		return -1;
-	    }
-	}
-
-        delkeys.buf_used = rec->size[recInfo_delKeys];
-	delkeys.buf = rec->info[recInfo_delKeys];
 
-        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);
+	zebra_rec_keys_set_buf(delkeys,
+			       rec->info[recInfo_delKeys],
+			       rec->size[recInfo_delKeys],
+			       0);
+	zebra_rec_keys_set_buf(sortKeys,
+			       rec->info[recInfo_sortKeys],
+			       rec->size[recInfo_sortKeys],
+			       0);
+
+	extract_flushSortKeys (zh, *sysno, 0, sortKeys);
+        extract_flushRecordKeys (zh, *sysno, 0, delkeys,
+				 recordAttr->staticrank);
         if (delete_flag)
         {
             /* record going to be deleted */
-            if (!delkeys.buf_used)
+            if (zebra_rec_keys_empty(delkeys))
             {
-                yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
-                      pr_fname, (long) recordOffset);
-                yaz_log (YLOG_WARN, "cannot delete file above, storeKeys false");
-            }
+		yaz_log(YLOG_LOG, "delete %s %s %ld", recordType,
+                        pr_fname, (long) recordOffset);
+		yaz_log(YLOG_WARN, "cannot delete file above, "
+                        "storeKeys false (3)");
+	    }
             else
             {
-		yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
-		      pr_fname, (long) recordOffset);
+		if (show_progress)
+		    yaz_log(YLOG_LOG, "delete %s %s %ld", recordType,
+                            pr_fname, (long) recordOffset);
                 zh->records_deleted++;
                 if (matchStr)
-                    dict_delete (zh->reg->matchDict, matchStr);
+		{
+		    int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
+                    dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
+		}
                 rec_del (zh->reg->records, &rec);
             }
 	    rec_rm (&rec);
             logRecord(zh);
-            return 0;
+            return ZEBRA_OK;
         }
         else
         {
-            /* record going to be updated */
-            if (!delkeys.buf_used)
-            {
-                yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
-                      pr_fname, (long) recordOffset);
-                yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
-            }
-            else
-            {
-		yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
-		      pr_fname, (long) recordOffset);
-                extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
-                extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
-                zh->records_updated++;
-            }
+	    if (show_progress)
+		    yaz_log(YLOG_LOG, "update %s %s %ld", recordType,
+                            pr_fname, (long) recordOffset);
+	    recordAttr->staticrank = extractCtrl.staticrank;
+            extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
+            extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys, 
+					 recordAttr->staticrank);
+            zh->records_updated++;
         }
+	zebra_rec_keys_close(delkeys);
+	zebra_rec_keys_close(sortKeys);
     }
     /* update file type */
     xfree (rec->info[recInfo_fileType]);
@@ -1073,26 +1156,23 @@ int buffer_extract_record (ZebraHandle zh,
 
     /* update delete keys */
     xfree (rec->info[recInfo_delKeys]);
-    if (zh->reg->keys.buf_used > 0 && zh->m_store_keys == 1)
+    if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
     {
-        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;
+	zebra_rec_keys_get_buf(zh->reg->keys,
+			       &rec->info[recInfo_delKeys],
+			       &rec->size[recInfo_delKeys]);
     }
     else
     {
         rec->info[recInfo_delKeys] = NULL;
         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;
+    zebra_rec_keys_get_buf(zh->reg->sortKeys,
+			   &rec->info[recInfo_sortKeys],
+			   &rec->size[recInfo_sortKeys]);
 
     /* save file size of original record */
     zebraExplain_recordBytesIncrement (zh->reg->zei,
@@ -1113,7 +1193,15 @@ int buffer_extract_record (ZebraHandle zh,
 
     /* update store data */
     xfree (rec->info[recInfo_storeData]);
-    if (zh->m_store_data)
+
+    /* update store data */
+    if (zh->store_data_buf)
+    {
+        rec->size[recInfo_storeData] = zh->store_data_size;
+        rec->info[recInfo_storeData] = zh->store_data_buf;
+	zh->store_data_buf = 0;
+    }
+    else if (zh->m_store_data)
     {
         rec->size[recInfo_storeData] = recordAttr->recordSize;
         rec->info[recInfo_storeData] = (char *)
@@ -1136,14 +1224,13 @@ int buffer_extract_record (ZebraHandle zh,
     /* commit this record */
     rec_put (zh->reg->records, &rec);
     logRecord(zh);
-    return 0;
+    return ZEBRA_OK;
 }
 
-int explain_extract (void *handle, Record rec, data1_node *n)
+ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
 {
     ZebraHandle zh = (ZebraHandle) handle;
     struct recExtractCtrl extractCtrl;
-    int i;
 
     if (zebraExplain_curDatabase (zh->reg->zei,
 				  rec->info[recInfo_databaseName]))
@@ -1154,64 +1241,122 @@ int explain_extract (void *handle, Record rec, data1_node *n)
             abort ();
     }
 
-    create_rec_keys_codec(&zh->reg->keys);
+    zebra_rec_keys_reset(zh->reg->keys);
+    zebra_rec_keys_reset(zh->reg->sortKeys);
 
-    zh->reg->sortKeys.buf_used = 0;
-    
     extractCtrl.init = extract_init;
     extractCtrl.tokenAdd = extract_token_add;
     extractCtrl.schemaAdd = extract_schema_add;
     extractCtrl.dh = zh->reg->dh;
-    for (i = 0; i<256; i++)
-	extractCtrl.seqno[i] = 0;
-    extractCtrl.zebra_maps = zh->reg->zebra_maps;
+
+    init_extractCtrl(zh, &extractCtrl);
+
     extractCtrl.flagShowRecords = 0;
     extractCtrl.match_criteria[0] = '\0';
+    extractCtrl.staticrank = 0;
     extractCtrl.handle = handle;
+    extractCtrl.first_record = 1;
+    
+    extract_set_store_data_prepare(&extractCtrl);
 
     if (n)
 	grs_extract_tree(&extractCtrl, n);
 
     if (rec->size[recInfo_delKeys])
     {
-	struct recKeys delkeys;
-	struct sortKeys sortkeys;
+	zebra_rec_keys_t delkeys = zebra_rec_keys_open();
+	
+	zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
 
-	delkeys.buf_used = rec->size[recInfo_delKeys];
-	delkeys.buf = rec->info[recInfo_delKeys];
+	zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
+			       rec->size[recInfo_delKeys],
+			       0);
+	extract_flushRecordKeys (zh, rec->sysno, 0, delkeys, 0);
+	zebra_rec_keys_close(delkeys);
 
-	sortkeys.buf_used = rec->size[recInfo_sortKeys];
-	sortkeys.buf = rec->info[recInfo_sortKeys];
+	zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
+			       rec->size[recInfo_sortKeys],
+			       0);
 
-	extract_flushSortKeys (zh, rec->sysno, 0, &sortkeys);
-	extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
+	extract_flushSortKeys (zh, rec->sysno, 0, sortkeys);
+	zebra_rec_keys_close(sortkeys);
     }
-    extract_flushRecordKeys (zh, rec->sysno, 1, &zh->reg->keys);
-    extract_flushSortKeys (zh, rec->sysno, 1, &zh->reg->sortKeys);
+    extract_flushRecordKeys (zh, rec->sysno, 1, zh->reg->keys, 0);
+    extract_flushSortKeys (zh, rec->sysno, 1, zh->reg->sortKeys);
 
     xfree (rec->info[recInfo_delKeys]);
-    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;
+    zebra_rec_keys_get_buf(zh->reg->keys,
+			   &rec->info[recInfo_delKeys],	
+			   &rec->size[recInfo_delKeys]);
 
     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;
+    zebra_rec_keys_get_buf(zh->reg->sortKeys,
+			   &rec->info[recInfo_sortKeys],
+			   &rec->size[recInfo_sortKeys]);
+    return ZEBRA_OK;
+}
+
+void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
+                             zebra_rec_keys_t reckeys)
+{
+    ZebraExplainInfo zei = zh->reg->zei;
+    struct ord_stat {
+        int no;
+        int ord;
+        struct ord_stat *next;
+    };
+
+    if (zebra_rec_keys_rewind(reckeys))
+    {
+        struct ord_stat *ord_list = 0;
+        struct ord_stat *p;
+	size_t slen;
+	const char *str;
+	struct it_key key_in;
+	while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
+        {
+            int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
 
-    return 0;
+            for (p = ord_list; p ; p = p->next)
+                if (p->ord == ord)
+                {
+                    p->no++;
+                    break;
+                }
+            if (!p)
+            {
+                p = xmalloc(sizeof(*p));
+                p->no = 1;
+                p->ord = ord;
+                p->next = ord_list;
+                ord_list = p;
+            }
+        }
+
+        p = ord_list;
+        while (p)
+        {
+            struct ord_stat *p1 = p;
+
+            if (is_insert)
+                zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
+            else
+                zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
+            p = p->next;
+            xfree(p1);
+        }
+    }
 }
 
 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
-                              int cmd, struct recKeys *reckeys)
+                              int cmd,
+			      zebra_rec_keys_t reckeys,
+			      zint staticrank)
 {
-    void *decode_handle = iscz1_start();
-    int off = 0;
-    int ch = 0;
     ZebraExplainInfo zei = zh->reg->zei;
 
+    extract_rec_keys_adjust(zh, cmd, reckeys);
+
     if (!zh->reg->key_buf)
     {
 	int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
@@ -1230,56 +1375,69 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
     }
     zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
 
-    while (off < reckeys->buf_used)
+    if (zebra_rec_keys_rewind(reckeys))
     {
-        const char *src = reckeys->buf + off;
-        struct it_key key;
-	char *dst = (char*) &key;
-	int attrSet, attrUse;
-
-	iscz1_decode(decode_handle, &dst, &src);
-	assert(key.len == 4);
-
-	attrSet = (int) key.mem[0] >> 16;
-	attrUse = (int) key.mem[0] & 65535;
-
-        if (zh->reg->key_buf_used + 1024 > 
-            (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
-            extract_flushWriteKeys (zh,0);
-        ++(zh->reg->ptr_i);
-        assert(zh->reg->ptr_i > 0);
-        (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
-	    (char*)zh->reg->key_buf + zh->reg->key_buf_used;
-
-        ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
-        if (ch < 0)
-            ch = zebraExplain_addSU (zei, attrSet, attrUse);
-
-        assert (ch > 0);
-	zh->reg->key_buf_used +=
-	    key_SU_encode (ch,((char*)zh->reg->key_buf) +
-                           zh->reg->key_buf_used);
-        while (*src)
-            ((char*)zh->reg->key_buf) [(zh->reg->key_buf_used)++] = *src++;
-        src++;
-        ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = '\0';
-        ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = cmd;
-
-        key.len = 3;
-	if (key.mem[1]) /* filter specify record ID */
-	    key.mem[0] = key.mem[1];
-	else
-	    key.mem[0] = sysno;
-	key.mem[1] = key.mem[2];  /* section_id */
-	key.mem[2] = key.mem[3];  /* sequence .. */
-
-        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;
+	size_t slen;
+	const char *str;
+	struct it_key key_in;
+	while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
+	{
+	    int ch = 0;
+	    struct it_key key_out;
+	    zint *keyp = key_out.mem;
+
+	    assert(key_in.len == 4);
+	    
+	    /* check for buffer overflow */
+	    if (zh->reg->key_buf_used + 1024 > 
+		(zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
+		extract_flushWriteKeys (zh, 0);
+	    
+	    ++(zh->reg->ptr_i);
+	    assert(zh->reg->ptr_i > 0);
+	    (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
+		(char*)zh->reg->key_buf + zh->reg->key_buf_used;
+
+	    /* encode the ordinal value (field/use/attribute) .. */
+	    ch = CAST_ZINT_TO_INT(key_in.mem[0]);
+	    zh->reg->key_buf_used +=
+		key_SU_encode(ch, (char*)zh->reg->key_buf +
+			      zh->reg->key_buf_used);
+
+	    /* copy the 0-terminated stuff from str to output */
+	    memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used, str, slen);
+	    zh->reg->key_buf_used += slen;
+	    ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = '\0';
+
+	    /* the delete/insert indicator */
+	    ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = cmd;
+
+	    if (zh->m_staticrank) /* rank config enabled ? */
+	    {
+		if (staticrank < 0)
+		{
+		    yaz_log(YLOG_WARN, "staticrank = %ld. Setting to 0",
+			    (long) staticrank);
+		    staticrank = 0;
+		}
+		*keyp++ = staticrank;
+		key_out.len = 4;
+	    }
+	    else
+		key_out.len = 3;
+	    
+	    if (key_in.mem[1]) /* filter specified record ID */
+		*keyp++ = key_in.mem[1];
+	    else
+		*keyp++ = sysno;
+	    *keyp++ = key_in.mem[2];  /* section_id */
+	    *keyp++ = key_in.mem[3];  /* sequence .. */
+	    
+	    memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used,
+		   &key_out, sizeof(key_out));
+	    (zh->reg->key_buf_used) += sizeof(key_out);
+	}
     }
-    assert (off == reckeys->buf_used);
-    iscz1_stop(decode_handle);
 }
 
 void extract_flushWriteKeys (ZebraHandle zh, int final)
@@ -1298,19 +1456,19 @@ void extract_flushWriteKeys (ZebraHandle zh, int final)
 #endif
     if (!zh->reg->key_buf || ptr_i <= 0)
     {
-        yaz_log (YLOG_DEBUG, "  nothing to flush section=%d buf=%p i=%d",
+        yaz_log(log_level, "  nothing to flush section=%d buf=%p i=%d",
                zh->reg->key_file_no, zh->reg->key_buf, ptr_i);
-        yaz_log (YLOG_DEBUG, "  buf=%p ",
+        yaz_log(log_level, "  buf=%p ",
                zh->reg->key_buf);
-        yaz_log (YLOG_DEBUG, "  ptr=%d ",zh->reg->ptr_i);
-        yaz_log (YLOG_DEBUG, "  reg=%p ",zh->reg);
+        yaz_log(log_level, "  ptr=%d ",zh->reg->ptr_i);
+        yaz_log(log_level, "  reg=%p ",zh->reg);
                
         return;
     }
 
     (zh->reg->key_file_no)++;
     yaz_log (YLOG_LOG, "sorting section %d", (zh->reg->key_file_no));
-    yaz_log (YLOG_DEBUG, "  sort_buff at %p n=%d",
+    yaz_log(log_level, "  sort_buff at %p n=%d",
                     zh->reg->key_buf + zh->reg->ptr_top - ptr_i,ptr_i);
 #if !SORT_EXTRA
     qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
@@ -1416,96 +1574,154 @@ void extract_flushWriteKeys (ZebraHandle zh, int final)
     zh->reg->key_buf_used = 0;
 }
 
-void extract_add_it_key (ZebraHandle zh,
-			 int reg_type,
-			 const char *str, int slen, struct it_key *key)
+ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh,
+				  zebra_rec_keys_t reckeys,
+				  zebra_snippets *snippets)
 {
-    char *dst;
-    struct recKeys *keys = &zh->reg->keys;
-    const char *src = (char*) key;
-    
-    if (keys->buf_used+1024 > keys->buf_max)
+    NMEM nmem = nmem_create();
+    if (zebra_rec_keys_rewind(reckeys)) 
     {
-        char *b;
-
-        b = (char *) xmalloc (keys->buf_max += 128000);
-        if (keys->buf_used > 0)
-            memcpy (b, keys->buf, keys->buf_used);
-        xfree (keys->buf);
-        keys->buf = b;
+	const char *str;
+	size_t slen;
+	struct it_key key;
+	while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
+	{
+	    char dst_buf[IT_MAX_WORD];
+	    char *dst_term = dst_buf;
+	    int ord;
+            zint seqno;
+	    int index_type;
+
+	    assert(key.len <= 4 && key.len > 2);
+	    seqno = key.mem[key.len-1];
+	    ord = CAST_ZINT_TO_INT(key.mem[0]);
+	    
+	    zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
+				    0/* db */, 0 /* string_index */);
+	    assert(index_type);
+	    zebra_term_untrans_iconv(zh, nmem, index_type,
+				     &dst_term, str);
+	    zebra_snippets_append(snippets, seqno, ord, dst_term);
+	    nmem_reset(nmem);
+	}
     }
-    dst = keys->buf + keys->buf_used;
-
-    iscz1_encode(keys->codec_handle, &dst, &src);
+    nmem_destroy(nmem);
+    return ZEBRA_OK;
+}
 
-    *dst++ = reg_type;
-    memcpy (dst, str, slen);
-    dst += slen;
-    *dst++ = '\0';
-    keys->buf_used = dst - keys->buf;
+void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
+{
+    yaz_log(YLOG_LOG, "print_rec_keys");
+    if (zebra_rec_keys_rewind(reckeys))
+    {
+	const char *str;
+	size_t slen;
+	struct it_key key;
+	while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
+	{
+	    char dst_buf[IT_MAX_WORD];
+	    zint seqno;
+	    int index_type;
+            int ord = CAST_ZINT_TO_INT(key.mem[0]);
+	    const char *db = 0;
+	    assert(key.len <= 4 && key.len > 2);
+
+	    zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
+	    
+	    seqno = key.mem[key.len-1];
+	    
+	    zebra_term_untrans(zh, index_type, dst_buf, str);
+	    
+	    yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT 
+                    " term=%s", ord, seqno, dst_buf); 
+	}
+    }
 }
 
-void extract_add_index_string (RecWord *p, const char *str, int length)
+static void extract_add_index_string(RecWord *p, const char *str, int length)
 {
     struct it_key key;
+
+    ZebraHandle zh = p->extractCtrl->handle;
+    ZebraExplainInfo zei = zh->reg->zei;
+    int ch;
+
+    if (!p->index_name)
+        return;
+
+    ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
+    if (ch < 0)
+        ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
+
     key.len = 4;
-    key.mem[0] = p->attrSet * 65536 + p->attrUse;
+    key.mem[0] = ch;
     key.mem[1] = p->record_id;
     key.mem[2] = p->section_id;
     key.mem[3] = p->seqno;
 
 #if 0
-    /* just for debugging .. */
-    yaz_log(YLOG_LOG, "add: set=%d use=%d "
-	    "record_id=%lld section_id=%lld seqno=%lld",
-	    p->attrSet, p->attrUse, p->record_id, p->section_id, p->seqno);
+    if (1)
+    {
+	char strz[80];
+	int i;
+
+	strz[0] = 0;
+	for (i = 0; i<length && i < 20; i++)
+	    sprintf(strz+strlen(strz), "%02X", str[i] & 0xff);
+	/* just for debugging .. */
+	yaz_log(YLOG_LOG, "add: set=%d use=%d "
+		"record_id=%lld section_id=%lld seqno=%lld %s",
+		p->attrSet, p->attrUse, p->record_id, p->section_id, p->seqno,
+		strz);
+    }
 #endif
 
-    extract_add_it_key(p->extractCtrl->handle,  p->reg_type, str,
-		       length, &key);
+    zebra_rec_keys_write(zh->reg->keys, str, length, &key);
 }
 
-static void extract_add_sort_string (RecWord *p, const char *str,
-				     int length)
+static void extract_add_sort_string(RecWord *p, const char *str, int length)
 {
+    struct it_key key;
+
     ZebraHandle zh = p->extractCtrl->handle;
-    struct sortKeys *sk = &zh->reg->sortKeys;
-    int off = 0;
+    ZebraExplainInfo zei = zh->reg->zei;
+    int ch;
 
-    while (off < sk->buf_used)
-    {
-        int set, use, slen;
-
-        off += key_SU_decode(&set, sk->buf + off);
-        off += key_SU_decode(&use, sk->buf + off);
-        off += key_SU_decode(&slen, sk->buf + off);
-        off += slen;
-        if (p->attrSet == set && p->attrUse == use)
-            return;
-    }
-    assert (off == sk->buf_used);
-    
-    if (sk->buf_used + IT_MAX_WORD > sk->buf_max)
+    if (!p->index_name)
+        return;
+
+    ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
+    if (ch < 0)
+        ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
+    key.len = 4;
+    key.mem[0] = ch;
+    key.mem[1] = p->record_id;
+    key.mem[2] = p->section_id;
+    key.mem[3] = p->seqno;
+
+#if 0
+    if (1)
     {
-        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;
+	char strz[80];
+	int i;
+
+	strz[0] = 0;
+	for (i = 0; i<length && i < 20; i++)
+	    sprintf(strz+strlen(strz), "%02X", str[i] & 0xff);
+	/* just for debugging .. */
+	yaz_log(YLOG_LOG, "add: set=%d use=%d "
+		"record_id=%lld section_id=%lld seqno=%lld %s",
+		p->attrSet, p->attrUse, p->record_id, p->section_id, p->seqno,
+		strz);
     }
-    off += key_SU_encode(p->attrSet, sk->buf + off);
-    off += key_SU_encode(p->attrUse, sk->buf + off);
-    off += key_SU_encode(length, sk->buf + off);
-    memcpy (sk->buf + off, str, length);
-    sk->buf_used = off + length;
+#endif
+    zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
 }
 
-void extract_add_string (RecWord *p, const char *string, int length)
+static void extract_add_string (RecWord *p, const char *string, int length)
 {
     assert (length > 0);
-    if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
+    if (zebra_maps_is_sort (p->zebra_maps, p->index_type))
 	extract_add_sort_string (p, string, length);
     else
 	extract_add_index_string (p, string, length);
@@ -1513,14 +1729,12 @@ void extract_add_string (RecWord *p, const char *string, int length)
 
 static void extract_add_incomplete_field (RecWord *p)
 {
-    const char *b = p->string;
-    int remain = p->length;
+    const char *b = p->term_buf;
+    int remain = p->term_len;
     const char **map = 0;
     
-    yaz_log(YLOG_DEBUG, "Incomplete field, w='%.*s'", p->length, p->string);
-
     if (remain > 0)
-	map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
+	map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, 0);
 
     while (map)
     {
@@ -1530,9 +1744,10 @@ static void extract_add_incomplete_field (RecWord *p)
 	/* Skip spaces */
 	while (map && *map && **map == *CHR_SPACE)
 	{
-	    remain = p->length - (b - p->string);
+	    remain = p->term_len - (b - p->term_buf);
 	    if (remain > 0)
-		map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
+		map = zebra_maps_input(p->zebra_maps, p->index_type, &b,
+				       remain, 0);
 	    else
 		map = 0;
 	}
@@ -1545,9 +1760,9 @@ static void extract_add_incomplete_field (RecWord *p)
 
 	    while (i < IT_MAX_WORD && *cp)
 		buf[i++] = *(cp++);
-	    remain = p->length - (b - p->string);
+	    remain = p->term_len - (b - p->term_buf);
 	    if (remain > 0)
-		map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
+		map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, 0);
 	    else
 		map = 0;
 	}
@@ -1560,27 +1775,24 @@ static void extract_add_incomplete_field (RecWord *p)
 
 static void extract_add_complete_field (RecWord *p)
 {
-    const char *b = p->string;
+    const char *b = p->term_buf;
     char buf[IT_MAX_WORD+1];
     const char **map = 0;
-    int i = 0, remain = p->length;
-    int first; /* first position */
-
-    yaz_log(YLOG_DEBUG, "Complete field, w='%.*s'", p->length, p->string);
+    int i = 0, remain = p->term_len;
 
     if (remain > 0)
-	map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain, 1);
+	map = zebra_maps_input (p->zebra_maps, p->index_type, &b, remain, 1);
 
     while (remain > 0 && i < IT_MAX_WORD)
     {
 	while (map && *map && **map == *CHR_SPACE)
 	{
-	    remain = p->length - (b - p->string);
+	    remain = p->term_len - (b - p->term_buf);
 
 	    if (remain > 0)
 	    {
-		first = i ? 0 : 1;
-		map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, first);
+		int first = i ? 0 : 1;  /* first position */
+		map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, first);
 	    }
 	    else
 		map = 0;
@@ -1602,14 +1814,13 @@ static void extract_add_complete_field (RecWord *p)
 	    {
 		if (i >= IT_MAX_WORD)
 		    break;
-		yaz_log(YLOG_DEBUG, "Adding string to index '%d'", **map);
 		while (i < IT_MAX_WORD && *cp)
 		    buf[i++] = *(cp++);
 	    }
-	    remain = p->length  - (b - p->string);
+	    remain = p->term_len  - (b - p->term_buf);
 	    if (remain > 0)
 	    {
-		map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
+		map = zebra_maps_input (p->zebra_maps, p->index_type, &b,
 					remain, 0);
 	    }
 	    else
@@ -1621,78 +1832,89 @@ static void extract_add_complete_field (RecWord *p)
     extract_add_string (p, buf, i);
 }
 
-void extract_token_add (RecWord *p)
+static void extract_token_add(RecWord *p)
 {
     WRBUF wrbuf;
-#if 0
-    yaz_log (YLOG_LOG, "token_add "
-	     "reg_type=%c attrSet=%d attrUse=%d seqno=%d s=%.*s",
-             p->reg_type, p->attrSet, p->attrUse, p->seqno, p->length,
-             p->string);
-#endif
-    if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
-			       p->string, p->length)))
+    if (log_level)
+        yaz_log(log_level, "extract_token_add "
+                "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
+                p->index_type, p->index_name, 
+                p->seqno, p->term_len, p->term_buf);
+    if ((wrbuf = zebra_replace(p->zebra_maps, p->index_type, 0,
+			       p->term_buf, p->term_len)))
     {
-	p->string = wrbuf_buf(wrbuf);
-	p->length = wrbuf_len(wrbuf);
+	p->term_buf = wrbuf_buf(wrbuf);
+	p->term_len = wrbuf_len(wrbuf);
     }
-    if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
+    if (zebra_maps_is_complete (p->zebra_maps, p->index_type))
 	extract_add_complete_field (p);
     else
 	extract_add_incomplete_field(p);
 }
 
-void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
+static void extract_set_store_data_cb(struct recExtractCtrl *p,
+				      void *buf, size_t sz)
 {
-    ZebraHandle zh = (ZebraHandle) (p->handle);
+    ZebraHandle zh = (ZebraHandle) p->handle;
+
+    xfree(zh->store_data_buf);
+    zh->store_data_buf = 0;
+    zh->store_data_size = 0;
+    if (buf && sz)
+    {
+	zh->store_data_buf = xmalloc(sz);
+	zh->store_data_size = sz;
+	memcpy(zh->store_data_buf, buf, sz);
+    }
+}
+
+static void extract_set_store_data_prepare(struct recExtractCtrl *p)
+{
+    ZebraHandle zh = (ZebraHandle) p->handle;
+    xfree(zh->store_data_buf);
+    zh->store_data_buf = 0;
+    zh->store_data_size = 0;
+    p->setStoreData = extract_set_store_data_cb;
+}
+
+static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
+{
+    ZebraHandle zh = (ZebraHandle) p->handle;
     zebraExplain_addSchema (zh->reg->zei, oid);
 }
 
 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
-                            int cmd, struct sortKeys *sk)
+                            int cmd, zebra_rec_keys_t reckeys)
 {
-    SortIdx sortIdx = zh->reg->sortIdx;
-    int off = 0;
+    if (zebra_rec_keys_rewind(reckeys))
+    {
+        SortIdx sortIdx = zh->reg->sortIdx;
+	size_t slen;
+	const char *str;
+	struct it_key key_in;
 
-    sortIdx_sysno (sortIdx, sysno);
+        sortIdx_sysno (sortIdx, sysno);
 
-    while (off < sk->buf_used)
-    {
-        int set, use, slen;
-        
-        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;
+	while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
+        {
+            int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
+            
+            sortIdx_type(sortIdx, ord);
+            if (cmd == 1)
+                sortIdx_add(sortIdx, str, slen);
+            else
+                sortIdx_add(sortIdx, "", 1);
+        }
     }
 }
 
-void encode_key_init (struct encode_info *i)
+static 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;
     i->encode_handle = iscz1_start();
+    i->decode_handle = iscz1_start();
 }
 
-#define OLDENCODE 1
-
-#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)
+static void encode_key_write (char *k, struct encode_info *i, FILE *outf)
 {
     struct it_key key;
     char *bp = i->buf, *bp0;
@@ -1704,119 +1926,49 @@ void encode_key_write (char *k, struct encode_info *i, FILE *outf)
     /* and copy & align key so we can mangle */
     memcpy (&key, k+1, sizeof(struct it_key));  /* *k is insert/delete */
 
+#if 0
+    /* debugging */
+    key_logdump_txt(YLOG_LOG, &key, *k ? "i" : "d");
+#endif
+    assert(key.mem[0] >= 0);
+
     bp0 = bp++;
     iscz1_encode(i->encode_handle, &bp, &src);
+
     *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */
     if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
     {
         yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
         exit (1);
     }
-}
 
-void encode_key_flush (struct encode_info *i, FILE *outf)
-{ /* dummy routine */
-    iscz1_stop(i->encode_handle);
-}
-
-#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
+#if 0
+    /* debugging */
+    if (1)
     {
-	bp=i->buf + i->keylen;
-	*bp++=0;
-	k++;
-    }
+	struct it_key key2;
+	const char *src = bp0+1;
+	char *dst = (char*) &key2;
+	iscz1_decode(i->decode_handle, &dst, &src);
 
-    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)
-        {
-            yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
-            exit (1);
-        }
-        i->keylen=0; /* ok, it's written, forget it */
-	i->prevsys=key.sysno;
-	i->prevseq=key.seqno;
-	i->prevcmd=*k;
+	key_logdump_txt(YLOG_LOG, &key2, *k ? "i" : "d");
+
+	assert(key2.mem[1]);
     }
+#endif
 }
 
-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)
-    {
-        yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
-        exit (1);
-    }
-    i->keylen=0; /* ok, it's written, forget it */
-    i->prevsys=0; /* forget the values too */
-    i->prevseq=0;
+static void encode_key_flush (struct encode_info *i, FILE *outf)
+{ 
+    iscz1_stop(i->encode_handle);
+    iscz1_stop(i->decode_handle);
 }
-#endif
+
+/*
+ * Local variables:
+ * c-basic-offset: 4
+ * indent-tabs-mode: nil
+ * End:
+ * vim: shiftwidth=4 tabstop=8 expandtab
+ */
+