X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=f32e57536b0b19ceea6a471e5a05c2decbb86217;hb=0f3b8bcc6fe2e3beeec7c834d9a64dca48a4f1b7;hp=4484cdb9ae4d49b27529664b1b9d4692121007f2;hpb=6c9fcd3b5d3108702fa1ffc92dab4ab6060f9a19;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index 4484cdb..f32e575 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,4 +1,4 @@
-/* $Id: extract.c,v 1.171 2005-01-15 19:38:25 adam Exp $
+/* $Id: extract.c,v 1.179 2005-04-28 08:20:39 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -79,12 +79,15 @@ 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;
@@ -392,7 +395,9 @@ 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;
@@ -400,17 +405,7 @@ static int file_extract_record(ZebraHandle zh,
     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]))
     {
@@ -429,8 +424,10 @@ static int file_extract_record(ZebraHandle zh,
 	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;
@@ -442,7 +439,10 @@ 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;
+	extractCtrl.first_record = fi->file_offset ? 0 : 1;
+
+	extract_set_store_data_prepare(&extractCtrl);
+
 	for (i = 0; i<256; i++)
 	{
 	    if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
@@ -462,7 +462,7 @@ 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)
@@ -490,18 +490,6 @@ 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;
     }
@@ -533,6 +521,18 @@ static int file_extract_record(ZebraHandle zh,
 	    }
 	}
     }
+    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)
     {
@@ -576,7 +576,8 @@ 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);
@@ -683,7 +684,14 @@ 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;
+	file_end(fi, fi->file_offset);
+    }
+    else if (zh->m_store_data)
     {
         rec->size[recInfo_storeData] = recordAttr->recordSize;
         rec->info[recInfo_storeData] = (char *)
@@ -730,6 +738,8 @@ 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';
@@ -768,6 +778,21 @@ 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
@@ -795,7 +820,8 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
     do
     {
         file_begin (fi);
-        r = file_extract_record (zh, sysno, fname, deleteFlag, fi, 1);
+        r = file_extract_record (zh, sysno, fname, deleteFlag, fi, 1,
+				 recType, recTypeClientData);
     } while (r && !sysno && fi->file_more);
     file_read_stop (fi);
     if (fd != -1)
@@ -810,16 +836,16 @@ 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;
@@ -831,6 +857,7 @@ 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 */
@@ -847,6 +874,7 @@ 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);
@@ -857,17 +885,21 @@ 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,
@@ -875,9 +907,10 @@ 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;
@@ -895,32 +928,23 @@ int buffer_extract_record (ZebraHandle zh,
 	else
 	    extractCtrl.seqno[i] = 0;
     }
+    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;
@@ -939,7 +963,7 @@ int buffer_extract_record (ZebraHandle zh,
 		if (!matchStr)
                 {
                     yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
-		    return 1;
+		    return ZEBRA_FAIL;
                 }
             }
         }
@@ -952,19 +976,28 @@ 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;
@@ -987,11 +1020,13 @@ 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);
@@ -1003,12 +1038,13 @@ 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;
 	    }
 	}
 
@@ -1025,14 +1061,19 @@ 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);
@@ -1040,21 +1081,25 @@ 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++;
@@ -1136,7 +1181,7 @@ 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)
@@ -1168,6 +1213,9 @@ int explain_extract (void *handle, Record rec, data1_node *n)
     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);
@@ -1235,14 +1283,10 @@ 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);
 
-	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);
@@ -1251,11 +1295,8 @@ 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);
@@ -1266,7 +1307,7 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
         ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = cmd;
 
         key.len = 3;
-	if (key.mem[1]) /* filter specify record ID */
+	if (key.mem[1]) /* filter specified record ID */
 	    key.mem[0] = key.mem[1];
 	else
 	    key.mem[0] = sysno;
@@ -1426,9 +1467,7 @@ void extract_add_it_key (ZebraHandle zh,
     
     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);
@@ -1448,8 +1487,25 @@ void extract_add_it_key (ZebraHandle zh,
 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] = 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;
@@ -1513,11 +1569,11 @@ 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->length, p->string);
+    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);
@@ -1530,7 +1586,7 @@ 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
@@ -1545,7 +1601,7 @@ 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
@@ -1560,13 +1616,13 @@ 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);
@@ -1575,11 +1631,11 @@ static void extract_add_complete_field (RecWord *p)
     {
 	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
@@ -1606,7 +1662,7 @@ static void extract_add_complete_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,
@@ -1631,10 +1687,10 @@ 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);
@@ -1642,9 +1698,34 @@ 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);
 }