X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=acd63b277fe163c108c8e8c884ced982d351584c;hb=7a2d0f25682890bde5d8f2883d6020df2ed0b365;hp=5c17db509cdae90e723bcde6c98a4314dc063428;hpb=5437b50633032595afe6f87dc0f989bc92a5aea8;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index 5c17db5..acd63b2 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,6 +1,6 @@
-/* $Id: extract.c,v 1.166 2004-11-19 10:26:56 heikki Exp $
-   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
-   Index Data Aps
+/* $Id: extract.c,v 1.185 2005-06-07 14:53:39 adam Exp $
+   Copyright (C) 1995-2005
+   Index Data ApS
 
 This file is part of the Zebra server.
 
@@ -79,14 +79,19 @@ static void logRecord (ZebraHandle zh)
     }
 }
 
+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->attrStr = 0;
     w->reg_type = 'w';
     w->extractCtrl = p;
+    w->record_id = 0;
+    w->section_id = 0;
 }
 
 static const char **searchRecordKey (ZebraHandle zh,
@@ -111,11 +116,11 @@ static const char **searchRecordKey (ZebraHandle zh,
 	int attrSet, attrUse;
 
 	iscz1_decode(decode_handle, &dst, &src);
-	assert(key.len < 4 && key.len > 2);
+	assert(key.len <= 4 && key.len > 2);
 
-	attrSet = (int) key.mem[0];
-	attrUse = (int) key.mem[1];
-	seqno = (int) key.mem[2];
+	attrSet = (int) key.mem[0] >> 16;
+	attrUse = (int) key.mem[0] & 65535;
+	seqno = (int) key.mem[key.len-1];
 
 	if (attrUseS == attrUse && attrSetS == attrSet)
         {
@@ -143,8 +148,6 @@ 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)
@@ -155,8 +158,8 @@ 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;
 }
 
@@ -169,8 +172,6 @@ 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);
 }
 
@@ -185,16 +186,7 @@ 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;
@@ -204,23 +196,15 @@ 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;
+    }
 }
 
 static char *fileMatchStr (ZebraHandle zh,
@@ -385,30 +369,36 @@ void create_rec_keys_codec(struct recKeys *keys)
     keys->buf_used = 0;
     iscz1_reset(keys->codec_handle);
 }
-     
+
+static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
+{
+    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)
+			       int force_update,
+			       RecType recType,
+			       void *recTypeClientData)
 {
     RecordAttr *recordAttr;
     int r;
-    const char *matchStr;
+    const char *matchStr = 0;
     SYSNO sysnotmp;
     Record rec;
     off_t recordOffset = 0;
-    RecType recType;
-    void *clientData;
     
-    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]))
     {
@@ -422,13 +412,12 @@ static int file_extract_record(ZebraHandle zh,
 	struct recExtractCtrl extractCtrl;
 
         /* we are going to read from a file, so prepare the extraction */
-	int i;
-
 	create_rec_keys_codec(&zh->reg->keys);
 
 	zh->reg->sortKeys.buf_used = 0;
 	
 	recordOffset = fi->file_moffset;
+        extractCtrl.handle = zh;
 	extractCtrl.offset = fi->file_moffset;
 	extractCtrl.readf = file_read;
 	extractCtrl.seekf = file_seek;
@@ -439,16 +428,12 @@ static int file_extract_record(ZebraHandle zh,
 	extractCtrl.tokenAdd = extract_token_add;
 	extractCtrl.schemaAdd = extract_schema_add;
 	extractCtrl.dh = zh->reg->dh;
-        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.match_criteria[0] = '\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);
@@ -459,7 +444,7 @@ 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)
@@ -487,48 +472,48 @@ static int file_extract_record(ZebraHandle zh,
             }
             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;
-        }
+        if (extractCtrl.match_criteria[0])
+            matchStr = extractCtrl.match_criteria;
     }
 
     /* perform match if sysno not known and if match criteria is specified */
-       
-    matchStr = NULL;
     if (!sysno) 
     {
         sysnotmp = 0;
         sysno = &sysnotmp;
-        if (zh->m_record_id && *zh->m_record_id)
+
+        if (matchStr == 0 && zh->m_record_id && *zh->m_record_id)
         {
-            char *rinfo;
         
             matchStr = fileMatchStr (zh, &zh->reg->keys, fname, 
 				     zh->m_record_id);
-            if (matchStr)
-            {
-                rinfo = dict_lookup (zh->reg->matchDict, matchStr);
-                if (rinfo)
-		{
-		    assert(*rinfo == sizeof(*sysno));
-                    memcpy (sysno, rinfo+1, sizeof(*sysno));
-		}
-            }
-            else
-            {
-                yaz_log (YLOG_WARN, "Bad match criteria");
-                return 0;
-            }
-        }
+	    if (!matchStr)
+	    {
+		yaz_log(YLOG_WARN, "Bad match criteria");
+		return 0;
+	    }
+	}
+	if (matchStr)
+	{
+            char *rinfo = dict_lookup (zh->reg->matchDict, matchStr);
+	    if (rinfo)
+	    {
+		assert(*rinfo == sizeof(*sysno));
+		memcpy (sysno, rinfo+1, sizeof(*sysno));
+	    }
+	}
+    }
+    if (! *sysno && 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;
     }
 
     if (! *sysno)
@@ -573,7 +558,8 @@ static int file_extract_record(ZebraHandle zh,
 	if (!force_update && recordAttr->runNumber ==
             zebraExplain_runNumberIncrement (zh->reg->zei, 0))
 	{
-            yaz_log (YLOG_LOG, "run number = %d", recordAttr->runNumber);
+            yaz_log (YLOG_LOG, "run number = " ZINT_FORMAT,
+			    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);
@@ -680,7 +666,13 @@ 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 *)
@@ -727,6 +719,8 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
     char ext_res[128];
     struct file_read_info *fi;
     const char *original_record_type = 0;
+    RecType recType;
+    void *recTypeClientData;
 
     if (!zh->m_group || !*zh->m_group)
         *gprefix = '\0';
@@ -765,6 +759,21 @@ 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 0;
+    }
+
+    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
@@ -791,9 +800,17 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
     fi = file_read_start (fd);
     do
     {
-        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);
+	}
+    }
+    while (r && !sysno);
     file_read_stop (fi);
     if (fd != -1)
         close (fd);
@@ -807,20 +824,20 @@ 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 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)
 {
     RecordAttr *recordAttr;
     struct recExtractCtrl extractCtrl;
-    int i, r;
+    int r;
     const char *matchStr = 0;
     RecType recType = NULL;
     void *clientData;
@@ -828,6 +845,7 @@ 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;
 
     if (!pr_fname)
 	pr_fname = "<no file>";  /* make it printable if file is omitted */
@@ -844,6 +862,7 @@ 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);
@@ -854,17 +873,21 @@ int buffer_extract_record (ZebraHandle zh,
     {
         if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0], 
 				      zh->m_explain_database))
-            return 0;
+            return ZEBRA_FAIL;
     }
     
-    if (recordType && *recordType) {
+    if (recordType && *recordType)
+    {
         yaz_log (YLOG_DEBUG, "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);
         recType = recType_byName (zh->reg->recTypes, zh->res,
@@ -872,9 +895,10 @@ int buffer_extract_record (ZebraHandle zh,
         recordType = zh->m_record_type;
     }
     
-    if (!recType) {
+    if (!recType)
+    {
         yaz_log (YLOG_WARN, "No such record type: %s", zh->m_record_type);
-        return 0;
+        return ZEBRA_FAIL;
     }
     
     extractCtrl.init = extract_init;
@@ -882,45 +906,34 @@ 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;
-    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.match_criteria[0] = '\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;
 
+    if (extractCtrl.match_criteria[0])
+	match_criteria = extractCtrl.match_criteria;
+
     if (! *sysno) {
         char *rinfo;
         if (match_criteria && *match_criteria) {
@@ -932,7 +945,7 @@ int buffer_extract_record (ZebraHandle zh,
 		if (!matchStr)
                 {
                     yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
-		    return 1;
+		    return ZEBRA_FAIL;
                 }
             }
         }
@@ -945,19 +958,28 @@ int buffer_extract_record (ZebraHandle zh,
 	    }
         }
     }
+    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 ZEBRA_OK;
+    }
 
     if (! *sysno)
     {
         /* new record */
         if (delete_flag)
         {
-	    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);
             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;
@@ -980,11 +1002,13 @@ int buffer_extract_record (ZebraHandle zh,
         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;
+	if (!allow_update)
+	{
+	    if (show_progress)
+		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);
@@ -996,12 +1020,13 @@ int buffer_extract_record (ZebraHandle zh,
 	    if (recordAttr->runNumber ==
 		zebraExplain_runNumberIncrement (zh->reg->zei, 0))
 	    {
-		yaz_log (YLOG_LOG, "skipped %s %s %ld", recordType,
-		      pr_fname, (long) recordOffset);
+		if (show_progress)
+		    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;
+		return ZEBRA_FAIL;
 	    }
 	}
 
@@ -1018,14 +1043,19 @@ int buffer_extract_record (ZebraHandle zh,
             /* record going to be deleted */
             if (!delkeys.buf_used)
             {
-                yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
-                      pr_fname, (long) recordOffset);
-                yaz_log (YLOG_WARN, "cannot delete file above, storeKeys false");
-            }
+		if (show_progress)
+		{
+		    yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
+			     pr_fname, (long) recordOffset);
+		    yaz_log (YLOG_WARN, "cannot delete file above, "
+			     "storeKeys false");
+		}
+	    }
             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);
@@ -1033,21 +1063,25 @@ int buffer_extract_record (ZebraHandle zh,
             }
 	    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");
-            }
+		if (show_progress)
+		{
+		    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);
+		if (show_progress)
+		    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++;
@@ -1106,7 +1140,15 @@ 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 *)
@@ -1129,14 +1171,13 @@ 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)
 {
     ZebraHandle zh = (ZebraHandle) handle;
     struct recExtractCtrl extractCtrl;
-    int i;
 
     if (zebraExplain_curDatabase (zh->reg->zei,
 				  rec->info[recInfo_databaseName]))
@@ -1155,11 +1196,15 @@ int explain_extract (void *handle, Record rec, data1_node *n)
     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.handle = handle;
+    extractCtrl.first_record = 1;
+    
+    extract_set_store_data_prepare(&extractCtrl);
 
     if (n)
 	grs_extract_tree(&extractCtrl, n);
@@ -1227,13 +1272,9 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
         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);
-
-	attrSet = (int) key.mem[0];
-	attrUse = (int) key.mem[1]; /* sequence in mem[2] */
+	assert(key.len == 4);
 
         if (zh->reg->key_buf_used + 1024 > 
             (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
@@ -1243,11 +1284,8 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
         (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);
+	ch = (int) key.mem[0];  /* ordinal for field/use/attribute */
 
-        assert (ch > 0);
 	zh->reg->key_buf_used +=
 	    key_SU_encode (ch,((char*)zh->reg->key_buf) +
                            zh->reg->key_buf_used);
@@ -1257,10 +1295,14 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
         ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = '\0';
         ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = cmd;
 
-        key.len = 2;
-	key.mem[0] = sysno;
-	key.mem[1] = key.mem[2];  /* sequence .. */
-	
+        key.len = 3;
+	if (key.mem[1]) /* filter specified 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);
@@ -1404,19 +1446,17 @@ void extract_flushWriteKeys (ZebraHandle zh, int final)
     zh->reg->key_buf_used = 0;
 }
 
-void extract_add_index_string (RecWord *p, const char *str, int length)
+void extract_add_it_key (ZebraHandle zh,
+			 int reg_type,
+			 const char *str, int slen, struct it_key *key)
 {
     char *dst;
-    ZebraHandle zh = p->extractCtrl->handle;
     struct recKeys *keys = &zh->reg->keys;
-    struct it_key key;
-    const char *src = (char*) &key;
+    const char *src = (char*) key;
     
     if (keys->buf_used+1024 > keys->buf_max)
     {
-        char *b;
-
-        b = (char *) xmalloc (keys->buf_max += 128000);
+        char *b = (char *) xmalloc (keys->buf_max += 128000);
         if (keys->buf_used > 0)
             memcpy (b, keys->buf, keys->buf_used);
         xfree (keys->buf);
@@ -1424,24 +1464,125 @@ void extract_add_index_string (RecWord *p, const char *str, int length)
     }
     dst = keys->buf + keys->buf_used;
 
-    key.len = 3;
-    key.mem[0] = p->attrSet;
-    key.mem[1] = p->attrUse;
-    key.mem[2] = p->seqno;
+    iscz1_encode(keys->codec_handle, &dst, &src);
+
+    *dst++ = reg_type;
+    memcpy (dst, str, slen);
+    dst += slen;
+    *dst++ = '\0';
+    keys->buf_used = dst - keys->buf;
+}
+
+ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh, struct recKeys *reckeys,
+				  zebra_snippets *snippets)
+{
+   void *decode_handle = iscz1_start();
+    int off = 0;
+    int seqno = 0;
+    NMEM nmem = nmem_create();
+
+    yaz_log(YLOG_LOG, "zebra_rec_keys_snippets buf=%p sz=%d", reckeys->buf,
+	    reckeys->buf_used);
+    assert(reckeys->buf);
+    while (off < reckeys->buf_used)
+    {
+        const char *src = reckeys->buf + off;
+        struct it_key key;
+	char *dst = (char*) &key;
+	char dst_buf[IT_MAX_WORD];
+	char *dst_term = dst_buf;
+
+	iscz1_decode(decode_handle, &dst, &src);
+	assert(key.len <= 4 && key.len > 2);
+
+	seqno = (int) key.mem[key.len-1];
+
+	zebra_term_untrans_iconv(zh, nmem, src[0], &dst_term, src+1);
+	zebra_snippets_append(snippets, seqno, src[0], key.mem[0], dst_term);
+        while (*src++)
+	    ;
+        off = src - reckeys->buf;
+	nmem_reset(nmem);
+    }
+    nmem_destroy(nmem);
+    iscz1_stop(decode_handle);
+    return ZEBRA_OK;
+}
+
+void print_rec_keys(ZebraHandle zh, struct recKeys *reckeys)
+{
+    void *decode_handle = iscz1_start();
+    int off = 0;
+    int seqno = 0;
+    NMEM nmem = nmem_create();
+
+    yaz_log(YLOG_LOG, "print_rec_keys buf=%p sz=%d", reckeys->buf,
+	    reckeys->buf_used);
+    assert(reckeys->buf);
+    while (off < reckeys->buf_used)
+    {
+        const char *src = reckeys->buf + off;
+        struct it_key key;
+	char *dst = (char*) &key;
+	int attrSet, attrUse;
+	char dst_buf[IT_MAX_WORD];
+	char *dst_term = dst_buf;
+
+	iscz1_decode(decode_handle, &dst, &src);
+	assert(key.len <= 4 && key.len > 2);
+
+	attrSet = (int) key.mem[0] >> 16;
+	attrUse = (int) key.mem[0] & 65535;
+	seqno = (int) key.mem[key.len-1];
+
+	zebra_term_untrans_iconv(zh, nmem, src[0], &dst_term, src+1);
+	
+	yaz_log(YLOG_LOG, "ord=" ZINT_FORMAT " seqno=%d term=%s",
+		key.mem[0], seqno, dst_term); 
+        while (*src++)
+	    ;
+        off = src - reckeys->buf;
+	nmem_reset(nmem);
+    }
+    nmem_destroy(nmem);
+    iscz1_stop(decode_handle);
+}
+
+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->attrStr)
+    {
+	ch = zebraExplain_lookup_attr_str(zei, p->attrStr);
+	if (ch < 0)
+	    ch = zebraExplain_add_attr_str(zei, p->attrStr);
+    }
+    else
+    {
+	ch = zebraExplain_lookup_attr_su(zei, p->attrSet, p->attrUse);
+	if (ch < 0)
+	    ch = zebraExplain_add_attr_su(zei, p->attrSet, p->attrUse);
+    }
+    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
     /* just for debugging .. */
-    yaz_log(YLOG_LOG, "set=%d use=%d seqno=%d", p->attrSet, p->attrUse,
-	    p->seqno);
+    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);
 #endif
 
-    iscz1_encode(keys->codec_handle, &dst, &src);
-
-    *dst++ = p->reg_type;
-    memcpy (dst, str, length);
-    dst += length;
-    *dst++ = '\0';
-    keys->buf_used = dst - keys->buf;
+    extract_add_it_key(p->extractCtrl->handle,  p->reg_type, str,
+		       length, &key);
 }
 
 static void extract_add_sort_string (RecWord *p, const char *str,
@@ -1492,9 +1633,11 @@ 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->term_len, p->term_buf);
 
     if (remain > 0)
 	map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
@@ -1507,7 +1650,7 @@ 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);
 	    else
@@ -1522,7 +1665,7 @@ 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);
 	    else
@@ -1537,13 +1680,13 @@ 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 */
+    int i = 0, remain = p->term_len;
 
-yaz_log(YLOG_DEBUG, "Complete field, w='%.*s'", p->length, p->string);
+    yaz_log(YLOG_DEBUG, "Complete field, w='%.*s'",
+	    p->term_len, p->term_buf);
 
     if (remain > 0)
 	map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain, 1);
@@ -1552,11 +1695,11 @@ yaz_log(YLOG_DEBUG, "Complete field, w='%.*s'", p->length, p->string);
     {
 	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;
+		int first = i ? 0 : 1;  /* first position */
 		map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, first);
 	    }
 	    else
@@ -1583,7 +1726,7 @@ yaz_log(YLOG_DEBUG, "Complete field, w='%.*s'", p->length, p->string);
 		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,
@@ -1608,10 +1751,10 @@ void extract_token_add (RecWord *p)
              p->string);
 #endif
     if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
-			       p->string, p->length)))
+			       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))
 	extract_add_complete_field (p);
@@ -1619,9 +1762,34 @@ void extract_token_add (RecWord *p)
 	extract_add_incomplete_field(p);
 }
 
+static void extract_set_store_data_cb(struct recExtractCtrl *p,
+				      void *buf, size_t sz)
+{
+    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;
+}
+
 void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
 {
-    ZebraHandle zh = (ZebraHandle) (p->handle);
+    ZebraHandle zh = (ZebraHandle) p->handle;
     zebraExplain_addSchema (zh->reg->zei, oid);
 }