Implemented update action : adelete.
[idzebra-moved-to-github.git] / index / extract.c
index 526a05f..be000c2 100644 (file)
@@ -1,4 +1,4 @@
-/* $Id: extract.c,v 1.247 2007-01-15 15:10:16 adam Exp $
+/* $Id: extract.c,v 1.275 2007-12-20 11:15:42 adam Exp $
    Copyright (C) 1995-2007
    Index Data ApS
 
@@ -20,6 +20,10 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 
 */
 
+/** \file
+    \brief indexes records and extract tokens for indexing and sorting
+*/
+
 #include <stdio.h>
 #include <assert.h>
 #include <ctype.h>
@@ -31,15 +35,27 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 #endif
 #include <fcntl.h>
 
+
 #include "index.h"
 #include "orddict.h"
 #include <direntz.h>
 #include <charmap.h>
+#include <yaz/snprintf.h>
 
 static int log_level_extract = 0;
 static int log_level_details = 0;
 static int log_level_initialized = 0;
 
+/* 1 if we use eliminitate identical delete/insert keys */
+/* eventually this the 0-case code will be removed */
+#define FLUSH2 1
+
+void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
+                                zebra_rec_keys_t ins_keys,
+                                zint ins_rank,
+                                zebra_rec_keys_t del_keys,
+                                zint del_rank);
+
 static void zebra_init_log_level(void)
 {
     if (!log_level_initialized)
@@ -51,16 +67,23 @@ static void zebra_init_log_level(void)
     }
 }
 
-static void extract_flush_record_keys(ZebraHandle zh, zint sysno,
-                                      int cmd, zebra_rec_keys_t reckeys,
-                                      zint staticrank);
 static void extract_flush_sort_keys(ZebraHandle zh, zint 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 extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid);
+static void extract_token_add(RecWord *p);
 
-static void logRecord (ZebraHandle zh)
+static void check_log_limit(ZebraHandle zh)
 {
+    if (zh->records_processed + zh->records_skipped == zh->m_file_verbose_limit)
+    {
+        yaz_log(YLOG_LOG, "More than %d file log entries. Omitting rest",
+                zh->m_file_verbose_limit);
+    }
+}
+
+static void logRecord(ZebraHandle zh)
+{
+    check_log_limit(zh);
     ++zh->records_processed;
     if (!(zh->records_processed % 1000))
     {
@@ -71,7 +94,13 @@ static void logRecord (ZebraHandle zh)
     }
 }
 
-static void extract_add_index_string (RecWord *p, 
+static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
+{
+    ctrl->flagShowRecords = !zh->m_flag_rw;
+}
+
+
+static void extract_add_index_string(RecWord *p, 
                                       zinfo_index_category_t cat,
                                       const char *str, int length);
 
@@ -81,13 +110,244 @@ static void extract_init(struct recExtractCtrl *p, RecWord *w)
 {
     w->seqno = 1;
     w->index_name = "any";
-    w->index_type = 'w';
+    w->index_type = "w";
     w->extractCtrl = p;
     w->record_id = 0;
     w->section_id = 0;
     w->segment = 0;
 }
 
+struct snip_rec_info {
+    ZebraHandle zh;
+    zebra_snippets *snippets;
+};
+
+
+static void snippet_add_complete_field(RecWord *p, int ord,
+                                       zebra_map_t zm)
+{
+    struct snip_rec_info *h = p->extractCtrl->handle;
+
+    const char *b = p->term_buf;
+    char buf[IT_MAX_WORD+1];
+    const char **map = 0;
+    int i = 0, remain = p->term_len;
+    const char *start = b;
+    const char *last = 0;
+
+    if (remain > 0)
+       map = zebra_maps_input(zm, &b, remain, 1);
+
+    while (remain > 0 && i < IT_MAX_WORD)
+    {
+       while (map && *map && **map == *CHR_SPACE)
+       {
+           remain = p->term_len - (b - p->term_buf);
+
+            if (i == 0)
+                start = b;  /* set to first non-ws area */
+           if (remain > 0)
+           {
+               int first = i ? 0 : 1;  /* first position */
+
+               map = zebra_maps_input(zm, &b, remain, first);
+           }
+           else
+               map = 0;
+       }
+       if (!map)
+           break;
+
+       if (i && i < IT_MAX_WORD)
+           buf[i++] = *CHR_SPACE;
+       while (map && *map && **map != *CHR_SPACE)
+       {
+           const char *cp = *map;
+
+           if (**map == *CHR_CUT)
+           {
+               i = 0;
+           }
+           else
+           {
+               if (i >= IT_MAX_WORD)
+                   break;
+               while (i < IT_MAX_WORD && *cp)
+                   buf[i++] = *(cp++);
+           }
+            last = b;
+           remain = p->term_len  - (b - p->term_buf);
+           if (remain > 0)
+           {
+               map = zebra_maps_input(zm, &b, remain, 0);
+           }
+           else
+               map = 0;
+       }
+    }
+    if (!i)
+       return;
+    if (last && start != last)
+        zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
+                               start, last - start);
+}
+
+static void snippet_add_incomplete_field(RecWord *p, int ord, zebra_map_t zm)
+{
+    struct snip_rec_info *h = p->extractCtrl->handle;
+    const char *b = p->term_buf;
+    int remain = p->term_len;
+    int first = 1;
+    const char **map = 0;
+    const char *start = b;
+    const char *last = b;
+
+    if (remain > 0)
+       map = zebra_maps_input(zm, &b, remain, 0);
+
+    while (map)
+    {
+       char buf[IT_MAX_WORD+1];
+       int i, remain;
+
+       /* Skip spaces */
+       while (map && *map && **map == *CHR_SPACE)
+       {
+           remain = p->term_len - (b - p->term_buf);
+            last = b;
+           if (remain > 0)
+               map = zebra_maps_input(zm, &b, remain, 0);
+           else
+               map = 0;
+       }
+       if (!map)
+           break;
+        if (start != last)
+        {
+            zebra_snippets_appendn(h->snippets, p->seqno, 1, ord,
+                                   start, last - start);
+
+        }
+        start = last;
+
+       i = 0;
+       while (map && *map && **map != *CHR_SPACE)
+       {
+           const char *cp = *map;
+
+           while (i < IT_MAX_WORD && *cp)
+               buf[i++] = *(cp++);
+           remain = p->term_len - (b - p->term_buf);
+            last = b;
+           if (remain > 0)
+               map = zebra_maps_input(zm, &b, remain, 0);
+           else
+               map = 0;
+       }
+       if (!i)
+           return;
+
+        if (first)
+        {   
+            first = 0;
+            if (zebra_maps_is_first_in_field(zm))
+            {
+                /* first in field marker */
+                p->seqno++;
+            }
+        }
+        if (start != last)
+            zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
+                                   start, last - start);
+        start = last;
+        p->seqno++;
+    }
+
+}
+
+static void snippet_add_icu(RecWord *p, int ord, zebra_map_t zm)
+{
+    struct snip_rec_info *h = p->extractCtrl->handle;
+
+    const char *res_buf = 0;
+    size_t res_len = 0;
+
+    const char *display_buf = 0;
+    size_t display_len = 0;
+
+    zebra_map_tokenize_start(zm, p->term_buf, p->term_len);
+    while (zebra_map_tokenize_next(zm, &res_buf, &res_len,
+                                   &display_buf, &display_len))
+    {
+        zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
+                               display_buf, display_len);
+        p->seqno++;
+    }
+}
+
+static void snippet_token_add(RecWord *p)
+{
+    struct snip_rec_info *h = p->extractCtrl->handle;
+    ZebraHandle zh = h->zh;
+    zebra_map_t zm = zebra_map_get(zh->reg->zebra_maps, p->index_type);
+
+    if (zm && zebra_maps_is_index(zm))
+    {
+        ZebraExplainInfo zei = zh->reg->zei;
+        int ch = zebraExplain_lookup_attr_str(
+            zei, zinfo_index_category_index, p->index_type, p->index_name);
+
+        if (zebra_maps_is_icu(zm))
+            snippet_add_icu(p, ch, zm);
+        else
+        {
+            if (zebra_maps_is_complete(zm))
+                snippet_add_complete_field(p, ch, zm);
+            else
+                snippet_add_incomplete_field(p, ch, zm);
+        }
+    }
+}
+
+static void snippet_schema_add(
+    struct recExtractCtrl *p, Odr_oid *oid)
+{
+
+}
+
+void extract_snippet(ZebraHandle zh, zebra_snippets *sn,
+                     struct ZebraRecStream *stream,
+                     RecType rt, void *recTypeClientData)
+{
+    struct recExtractCtrl extractCtrl;
+    struct snip_rec_info info;
+    int r;
+
+    extractCtrl.stream = stream;
+    extractCtrl.first_record = 1;
+    extractCtrl.init = extract_init;
+    extractCtrl.tokenAdd = snippet_token_add;
+    extractCtrl.schemaAdd = snippet_schema_add;
+    assert(zh->reg);
+    assert(zh->reg->dh);
+
+    extractCtrl.dh = zh->reg->dh;
+    
+    info.zh = zh;
+    info.snippets = sn;
+    extractCtrl.handle = &info;
+    extractCtrl.match_criteria[0] = '\0';
+    extractCtrl.staticrank = 0;
+    extractCtrl.action = action_insert;
+    
+    init_extractCtrl(zh, &extractCtrl);
+
+    extractCtrl.setStoreData = 0;
+
+    r = (*rt->extract)(recTypeClientData, &extractCtrl);
+
+}
+
 static void searchRecordKey(ZebraHandle zh,
                            zebra_rec_keys_t reckeys,
                             const char *index_name,
@@ -101,11 +361,11 @@ static void searchRecordKey(ZebraHandle zh,
         ws[i] = NULL;
 
     if (ch < 0)
-        ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, '0', index_name);
+        ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "0", index_name);
     if (ch < 0)
-        ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'p', index_name);
+        ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "p", index_name);
     if (ch < 0)
-        ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'w', index_name);
+        ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, "w", index_name);
 
     if (ch < 0)
        return ;
@@ -183,11 +443,11 @@ static char *get_match_from_spec(ZebraHandle zh,
                attname_str[i] = '\0';
            }
 
-            searchRecordKey (zh, reckeys, attname_str, ws, 32);
+            searchRecordKey(zh, reckeys, attname_str, ws, 32);
 
             if (*s != ')')
             {
-                yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
+                yaz_log(YLOG_WARN, "Missing ) in match criteria %s in group %s",
                       spec, zh->m_group ? zh->m_group : "none");
                 return NULL;
             }
@@ -201,12 +461,12 @@ static char *get_match_from_spec(ZebraHandle zh,
                         *dst++ = ' ';
                         first = 0;
                     }
-                    strcpy (dst, ws[i]);
+                    strcpy(dst, ws[i]);
                     dst += strlen(ws[i]);
                 }
             if (first)
             {
-                yaz_log (YLOG_WARN, "Record didn't contain match"
+                yaz_log(YLOG_WARN, "Record didn't contain match"
                       " fields in (%s,%s)", attset_str, attname_str);
                 return NULL;
             }
@@ -223,25 +483,25 @@ static char *get_match_from_spec(ZebraHandle zh,
             spec_len = s1 - s;
             if (spec_len > sizeof(special)-1)
                 spec_len = sizeof(special)-1;
-            memcpy (special, s, spec_len);
+            memcpy(special, s, spec_len);
             special[spec_len] = '\0';
             s = s1;
 
-            if (!strcmp (special, "group"))
+            if (!strcmp(special, "group"))
                 spec_src = zh->m_group;
-            else if (!strcmp (special, "database"))
+            else if (!strcmp(special, "database"))
                 spec_src = zh->basenames[0];
-            else if (!strcmp (special, "filename")) {
+            else if (!strcmp(special, "filename")) {
                 spec_src = fname;
            }
-            else if (!strcmp (special, "type"))
+            else if (!strcmp(special, "type"))
                 spec_src = zh->m_record_type;
             else 
                 spec_src = NULL;
             if (spec_src)
             {
-                strcpy (dst, spec_src);
-                dst += strlen (spec_src);
+                strcpy(dst, spec_src);
+                dst += strlen(spec_src);
             }
         }
         else if (*s == '\"' || *s == '\'')
@@ -258,12 +518,12 @@ static char *get_match_from_spec(ZebraHandle zh,
             if (*s)
                 s++;
             tmpString[i] = '\0';
-            strcpy (dst, tmpString);
-            dst += strlen (tmpString);
+            strcpy(dst, tmpString);
+            dst += strlen(tmpString);
         }
         else
         {
-            yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
+            yaz_log(YLOG_WARN, "Syntax error in match criteria %s in group %s",
                   spec, zh->m_group ? zh->m_group : "none");
             return NULL;
         }
@@ -271,7 +531,7 @@ static char *get_match_from_spec(ZebraHandle zh,
     }
     if (dst == dstBuf)
     {
-        yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
+        yaz_log(YLOG_WARN, "No match criteria for record %s in group %s",
               fname, zh->m_group ? zh->m_group : "none");
         return NULL;
     }
@@ -285,32 +545,47 @@ struct recordLogInfo {
     struct recordGroup *rGroup;
 };
 
-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->flagShowRecords = !zh->m_flag_rw;
-}
-
-static void all_matches_add(struct recExtractCtrl *ctrl)
+/** \brief add the always-matches index entry and map to real record ID
+    \param ctrl record control
+    \param record_id custom record ID
+    \param sysno system record ID
+    
+    This function serves two purposes.. It adds the always matches
+    entry and makes a pointer from the custom record ID (if defined)
+    back to the system record ID (sysno)
+    See zebra_recid_to_sysno .
+  */
+static void all_matches_add(struct recExtractCtrl *ctrl, zint record_id,
+                            zint sysno)
 {
     RecWord word;
     extract_init(ctrl, &word);
+    word.record_id = record_id;
+    /* we use the seqno as placeholder for a way to get back to
+       record database from _ALLRECORDS.. This is used if a custom
+       RECORD was defined */
+    word.seqno = sysno;
     word.index_name = "_ALLRECORDS";
-    word.index_type = 'w';
-    word.seqno = 1;
-    extract_add_index_string (&word, zinfo_index_category_alwaysmatches,
+    word.index_type = "w";
+
+    extract_add_index_string(&word, zinfo_index_category_alwaysmatches,
                               "", 0);
 }
 
+ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh, 
+                                       struct ZebraRecStream *stream,
+                                       enum zebra_recctrl_action_t action,
+                                       int test_mode, 
+                                       const char *recordType,
+                                       zint *sysno,
+                                       const char *match_criteria,
+                                       const char *fname,
+                                       RecType recType,
+                                       void *recTypeClientData);
+
+
 ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname, 
-                            int deleteFlag)
+                             enum zebra_recctrl_action_t action)
 {
     ZEBRA_RES r = ZEBRA_OK;
     int i, fd;
@@ -328,7 +603,7 @@ ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
     if (!zh->m_group || !*zh->m_group)
         *gprefix = '\0';
     else
-        sprintf (gprefix, "%s.", zh->m_group);
+        sprintf(gprefix, "%s.", zh->m_group);
     
     yaz_log(log_level_extract, "zebra_extract_file %s", fname);
 
@@ -339,31 +614,34 @@ ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
             break;
         else if (fname[i] == '.')
         {
-            strcpy (ext, fname+i+1);
+            strcpy(ext, fname+i+1);
             break;
         }
     /* determine file type - depending on extension */
     original_record_type = zh->m_record_type;
     if (!zh->m_record_type)
     {
-        sprintf (ext_res, "%srecordType.%s", gprefix, ext);
-        zh->m_record_type = res_get (zh->res, ext_res);
+        sprintf(ext_res, "%srecordType.%s", gprefix, ext);
+        zh->m_record_type = res_get(zh->res, ext_res);
     }
     if (!zh->m_record_type)
     {
-       if (zh->records_processed < zh->m_file_verbose_limit)
-            yaz_log (YLOG_LOG, "? %s", fname);
+        check_log_limit(zh);
+       if (zh->records_processed + zh->records_skipped
+            < zh->m_file_verbose_limit)
+            yaz_log(YLOG_LOG, "? %s", fname);
+        zh->records_skipped++;
         return 0;
     }
     /* determine match criteria */
     if (!zh->m_record_id)
     {
-        sprintf (ext_res, "%srecordId.%s", gprefix, ext);
-        zh->m_record_id = res_get (zh->res, ext_res);
+        sprintf(ext_res, "%srecordId.%s", gprefix, ext);
+        zh->m_record_id = res_get(zh->res, ext_res);
     }
 
     if (!(recType =
-         recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
+         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);
@@ -377,7 +655,7 @@ ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
     default:
        yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
     }
-    if (sysno && deleteFlag)
+    if (sysno && (action == action_delete || action == action_a_delete))
     {
         streamp = 0;
         fi = 0;
@@ -386,44 +664,32 @@ ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
     {
         char full_rep[1024];
 
-        if (zh->path_reg && !yaz_is_abspath (fname))
+        if (zh->path_reg && !yaz_is_abspath(fname))
         {
-            strcpy (full_rep, zh->path_reg);
-            strcat (full_rep, "/");
-            strcat (full_rep, fname);
+            strcpy(full_rep, zh->path_reg);
+            strcat(full_rep, "/");
+            strcat(full_rep, fname);
         }
         else
-            strcpy (full_rep, fname);
+            strcpy(full_rep, fname);
         
-        if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
+        if ((fd = open(full_rep, O_BINARY|O_RDONLY)) == -1)
         {
-            yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
+            yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
            zh->m_record_type = original_record_type;
             return ZEBRA_FAIL;
         }
         streamp = &stream;
         zebra_create_stream_fd(streamp, fd, 0);
     }
-    while(1)
-    {
-        int more = 0;
-        r = zebra_extract_record_stream(zh, streamp,
-                                        deleteFlag,
-                                        0, /* tst_mode */
-                                        zh->m_record_type,
-                                        sysno,
-                                        0, /*match_criteria */
-                                        fname,
-                                        1, /* force_update */
-                                        1, /* allow_update */
-                                        recType, recTypeClientData, &more);
-        if (!more)
-            break;
-       if (sysno)
-       {
-           break;
-       }
-    }
+    r = zebra_extract_records_stream(zh, streamp,
+                                     action,
+                                     0, /* tst_mode */
+                                     zh->m_record_type,
+                                     sysno,
+                                     0, /*match_criteria */
+                                     fname,
+                                     recType, recTypeClientData);
     if (streamp)
         stream.destroy(streamp);
     zh->m_record_type = original_record_type;
@@ -439,75 +705,118 @@ ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
 
 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh, 
                                       const char *buf, size_t buf_size,
-                                      int delete_flag,
+                                      enum zebra_recctrl_action_t action,
                                       int test_mode, 
                                       const char *recordType,
                                       zint *sysno,
                                       const char *match_criteria,
-                                      const char *fname,
-                                      int force_update,
-                                      int allow_update)
+                                      const char *fname)
 {
     struct ZebraRecStream stream;
     ZEBRA_RES res;
     void *clientData;
     RecType recType = 0;
-    int more = 0;
 
     if (recordType && *recordType)
     {
         yaz_log(log_level_extract,
                 "Record type explicitly specified: %s", recordType);
-        recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
+        recType = recType_byName(zh->reg->recTypes, zh->res, recordType,
                                   &clientData);
     } 
     else
     {
         if (!(zh->m_record_type))
        {
-            yaz_log (YLOG_WARN, "No such record type defined");
+            yaz_log(YLOG_WARN, "No such record type defined");
             return ZEBRA_FAIL;
         }
         yaz_log(log_level_extract, "Get record type from rgroup: %s",
                 zh->m_record_type);
-        recType = recType_byName (zh->reg->recTypes, zh->res,
+        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", recordType);
+        yaz_log(YLOG_WARN, "No such record type: %s", recordType);
         return ZEBRA_FAIL;
     }
 
     zebra_create_stream_mem(&stream, buf, buf_size);
 
-    res = zebra_extract_record_stream(zh, &stream,
-                                      delete_flag,
-                                      test_mode, 
-                                      recordType,
-                                      sysno,
-                                      match_criteria,
-                                      fname,
-                                      force_update,
-                                      allow_update,
-                                      recType, clientData, &more);
+    res = zebra_extract_records_stream(zh, &stream,
+                                       action,
+                                       test_mode, 
+                                       recordType,
+                                       sysno,
+                                       match_criteria,
+                                       fname,
+                                       recType, clientData);
     stream.destroy(&stream);
     return res;
 }
 
+ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh, 
+                                       struct ZebraRecStream *stream,
+                                       enum zebra_recctrl_action_t action,
+                                       int test_mode, 
+                                       const char *recordType,
+                                       zint *sysno,
+                                       const char *match_criteria,
+                                       const char *fname,
+                                       RecType recType,
+                                       void *recTypeClientData)
+{
+    ZEBRA_RES res = ZEBRA_OK;
+    while (1)
+    {
+        int more = 0;
+        res = zebra_extract_record_stream(zh, stream,
+                                          action,
+                                          test_mode, 
+                                          recordType,
+                                          sysno,
+                                          match_criteria,
+                                          fname,
+                                          recType, recTypeClientData, &more);
+        if (!more)
+        {
+            res = ZEBRA_OK;
+            break;
+        }
+        if (res != ZEBRA_OK)
+            break;
+        if (sysno)
+            break;
+    }
+    return res;
+}
+
+
+static WRBUF wrbuf_hex_str(const char *cstr)
+{
+    size_t i;
+    WRBUF w = wrbuf_alloc();
+    for (i = 0; cstr[i]; i++)
+    {
+        if (cstr[i] < ' ' || cstr[i] > 126)
+            wrbuf_printf(w, "\\%02X", cstr[i] & 0xff);
+        else
+            wrbuf_putc(w, cstr[i]);
+    }
+    return w;
+}
 
 ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, 
                                       struct ZebraRecStream *stream,
-                                      int delete_flag,
+                                      enum zebra_recctrl_action_t action,
                                       int test_mode, 
                                       const char *recordType,
                                       zint *sysno,
                                       const char *match_criteria,
                                       const char *fname,
-                                      int force_update,
-                                      int allow_update,
                                       RecType recType,
                                       void *recTypeClientData,
                                       int *more)
@@ -521,7 +830,8 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
     Record rec;
     off_t start_offset = 0, end_offset = 0;
     const char *pr_fname = fname;  /* filename to print .. */
-    int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
+    int show_progress = zh->records_processed + zh->records_skipped 
+        < zh->m_file_verbose_limit ? 1:0;
 
     zebra_init_log_level();
 
@@ -531,9 +841,9 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
     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_curDatabase(zh->reg->zei, zh->basenames[0]))
     {
-        if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0], 
+        if (zebraExplain_newDatabase(zh->reg->zei, zh->basenames[0], 
                                      zh->m_explain_database))
             return ZEBRA_FAIL;
     }
@@ -556,44 +866,58 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
         extractCtrl.handle = zh;
         extractCtrl.match_criteria[0] = '\0';
         extractCtrl.staticrank = 0;
+        extractCtrl.action = action;
 
         init_extractCtrl(zh, &extractCtrl);
-        
+
         extract_set_store_data_prepare(&extractCtrl);
         
         r = (*recType->extract)(recTypeClientData, &extractCtrl);
+
+        if (action == action_update)
+        {
+            action = extractCtrl.action;
+        }
         
-        if (r == RECCTRL_EXTRACT_EOF)
-            return ZEBRA_FAIL;
-        else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
+        switch (r)
         {
+        case RECCTRL_EXTRACT_EOF:
+            return ZEBRA_FAIL;
+        case RECCTRL_EXTRACT_ERROR_GENERIC:
             /* error occured during extraction ... */
-            yaz_log (YLOG_WARN, "extract error: generic");
+            yaz_log(YLOG_WARN, "extract error: generic");
             return ZEBRA_FAIL;
-        }
-        else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
-        {
+        case RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER:
             /* error occured during extraction ... */
-            yaz_log (YLOG_WARN, "extract error: no such filter");
+            yaz_log(YLOG_WARN, "extract error: no such filter");
             return ZEBRA_FAIL;
-        }
-        
-        all_matches_add(&extractCtrl);
-        
-        if (extractCtrl.match_criteria[0])
-            match_criteria = extractCtrl.match_criteria;
-
+        case RECCTRL_EXTRACT_SKIP:
+            if (show_progress)
+                yaz_log(YLOG_LOG, "skip %s %s " ZINT_FORMAT,
+                         recordType, pr_fname, (zint) start_offset);
+            *more = 1;
+            
+            end_offset = stream->endf(stream, 0);
+            if (end_offset)
+                stream->seekf(stream, end_offset);
 
+            return ZEBRA_OK;
+        case RECCTRL_EXTRACT_OK:
+            break;
+        default:
+            yaz_log(YLOG_WARN, "extract error: unknown error: %d", r);
+            return ZEBRA_FAIL;
+        }
         end_offset = stream->endf(stream, 0);
-
-        if (!end_offset)
-            end_offset = stream->tellf(stream);
-        else
+        if (end_offset)
             stream->seekf(stream, end_offset);
+        else
+            end_offset = stream->tellf(stream);
 
+        if (extractCtrl.match_criteria[0])
+            match_criteria = extractCtrl.match_criteria;
     }
 
-
     *more = 1;
     if (!sysno)
     {
@@ -607,7 +931,7 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
                                                zh->m_record_id);
                if (!matchStr)
                 {
-                    yaz_log (YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
+                    yaz_log(YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
                              pr_fname, (zint) start_offset);
                    return ZEBRA_FAIL;
                 }
@@ -618,13 +942,22 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
            int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
            char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
                                          matchStr);
+
+            
+            if (log_level_extract)
+            {
+                WRBUF w = wrbuf_hex_str(matchStr);
+                yaz_log(log_level_extract, "matchStr: %s", wrbuf_cstr(w));
+                wrbuf_destroy(w);
+            }
             if (rinfo)
            {
                assert(*rinfo == sizeof(*sysno));
-                memcpy (sysno, rinfo+1, sizeof(*sysno));
+                memcpy(sysno, rinfo+1, sizeof(*sysno));
            }
-        }
+       }
     }
+
     if (zebra_rec_keys_empty(zh->reg->keys))
     {
        /* the extraction process returned no information - the record
@@ -635,28 +968,50 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
 
     if (! *sysno)
     {
-        /* new record */
-        if (delete_flag)
+        /* new record AKA does not exist already */
+        if (action == action_delete)
         {
-           yaz_log (YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
-                        pr_fname, (zint) start_offset);
-            yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
+                yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
+                        pr_fname, (zint) start_offset);
+            yaz_log(YLOG_WARN, "cannot delete record above (seems new)");
             return ZEBRA_FAIL;
         }
+        else if (action == action_a_delete)
+        {
+            if (show_progress)
+                yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
+                        pr_fname, (zint) start_offset);
+            return ZEBRA_OK;
+        }
+       else if (action == action_replace)
+       {
+           yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
+                        pr_fname, (zint) start_offset);
+            yaz_log(YLOG_WARN, "cannot update record above (seems new)");
+            return ZEBRA_FAIL;
+       }
        if (show_progress)
-           yaz_log (YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
+           yaz_log(YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
                     (zint) start_offset);
-        rec = rec_new (zh->reg->records);
+        rec = rec_new(zh->reg->records);
 
         *sysno = rec->sysno;
 
-       recordAttr = rec_init_attr (zh->reg->zei, rec);
+
+        if (stream)
+        {
+            all_matches_add(&extractCtrl,
+                            zebra_rec_keys_get_custom_record_id(zh->reg->keys),
+                            *sysno);
+        }
+
+
+       recordAttr = rec_init_attr(zh->reg->zei, rec);
        if (extractCtrl.staticrank < 0)
         {
             yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
            extractCtrl.staticrank = 0;
         }
-       recordAttr->staticrank = extractCtrl.staticrank;
 
         if (matchStr)
         {
@@ -666,8 +1021,15 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
         }
 
        extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
+#if FLUSH2
+        extract_flush_record_keys2(zh, *sysno,
+                                   zh->reg->keys, extractCtrl.staticrank,
+                                   0, recordAttr->staticrank);
+#else
         extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
-                                  recordAttr->staticrank);
+                                  extractCtrl.staticrank);
+#endif
+       recordAttr->staticrank = extractCtrl.staticrank;
         zh->records_inserted++;
     } 
     else
@@ -675,21 +1037,28 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
         /* record already exists */
        zebra_rec_keys_t delkeys = zebra_rec_keys_open();
        zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
-       if (!allow_update)
+       if (action == action_insert)
        {
-           yaz_log (YLOG_LOG, "skipped %s %s " ZINT_FORMAT, 
+           yaz_log(YLOG_LOG, "skipped %s %s " ZINT_FORMAT, 
                         recordType, pr_fname, (zint) start_offset);
            logRecord(zh);
            return ZEBRA_FAIL;
        }
 
-        rec = rec_get (zh->reg->records, *sysno);
-        assert (rec);
+        rec = rec_get(zh->reg->records, *sysno);
+        assert(rec);
+
+        if (stream)
+        {
+            all_matches_add(&extractCtrl,
+                            zebra_rec_keys_get_custom_record_id(zh->reg->keys),
+                            *sysno);
+        }
        
-       recordAttr = rec_init_attr (zh->reg->zei, rec);
+       recordAttr = rec_init_attr(zh->reg->zei, rec);
 
         /* decrease total size */
-        zebraExplain_recordBytesIncrement (zh->reg->zei,
+        zebraExplain_recordBytesIncrement(zh->reg->zei,
                                            - recordAttr->recordSize);
 
        zebra_rec_keys_set_buf(delkeys,
@@ -702,11 +1071,17 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
                               0);
 
        extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
+#if !FLUSH2
         extract_flush_record_keys(zh, *sysno, 0, delkeys,
                                   recordAttr->staticrank);
-        if (delete_flag)
+#endif
+        if (action == action_delete || action == action_a_delete)
         {
             /* record going to be deleted */
+#if FLUSH2
+            extract_flush_record_keys2(zh, *sysno, 0, recordAttr->staticrank,
+                                       delkeys, recordAttr->staticrank);
+#endif       
             if (zebra_rec_keys_empty(delkeys))
             {
                yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
@@ -725,38 +1100,47 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
                    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_del(zh->reg->records, &rec);
             }
+            zebra_rec_keys_close(delkeys);
+            zebra_rec_keys_close(sortKeys);
            rec_free(&rec);
             logRecord(zh);
             return ZEBRA_OK;
         }
         else
-        {
+        {   /* update or special_update */
            if (show_progress)
                 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
                         pr_fname, (zint) start_offset);
-           recordAttr->staticrank = extractCtrl.staticrank;
             extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
-            extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys, 
-                                      recordAttr->staticrank);
+
+#if FLUSH2
+            extract_flush_record_keys2(zh, *sysno,
+                                       zh->reg->keys, extractCtrl.staticrank,
+                                       delkeys, recordAttr->staticrank);
+#else
+            extract_flush_record_keys(zh, *sysno, 1, 
+                                      zh->reg->keys, extractCtrl.staticrank);
+#endif
+           recordAttr->staticrank = extractCtrl.staticrank;
             zh->records_updated++;
         }
        zebra_rec_keys_close(delkeys);
        zebra_rec_keys_close(sortKeys);
     }
     /* update file type */
-    xfree (rec->info[recInfo_fileType]);
+    xfree(rec->info[recInfo_fileType]);
     rec->info[recInfo_fileType] =
-        rec_strdup (recordType, &rec->size[recInfo_fileType]);
+        rec_strdup(recordType, &rec->size[recInfo_fileType]);
 
     /* update filename */
-    xfree (rec->info[recInfo_filename]);
+    xfree(rec->info[recInfo_filename]);
     rec->info[recInfo_filename] =
-        rec_strdup (fname, &rec->size[recInfo_filename]);
+        rec_strdup(fname, &rec->size[recInfo_filename]);
 
     /* update delete keys */
-    xfree (rec->info[recInfo_delKeys]);
+    xfree(rec->info[recInfo_delKeys]);
     if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
     {
        zebra_rec_keys_get_buf(zh->reg->keys,
@@ -769,7 +1153,7 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
         rec->size[recInfo_delKeys] = 0;
     }
     /* update sort keys */
-    xfree (rec->info[recInfo_sortKeys]);
+    xfree(rec->info[recInfo_sortKeys]);
 
     zebra_rec_keys_get_buf(zh->reg->sortKeys,
                           &rec->info[recInfo_sortKeys],
@@ -784,10 +1168,10 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
 
     /* set run-number for this record */
     recordAttr->runNumber =
-       zebraExplain_runNumberIncrement (zh->reg->zei, 0);
+       zebraExplain_runNumberIncrement(zh->reg->zei, 0);
 
     /* update store data */
-    xfree (rec->info[recInfo_storeData]);
+    xfree(rec->info[recInfo_storeData]);
 
     /* update store data */
     if (zh->store_data_buf)
@@ -803,7 +1187,7 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
 
         rec->size[recInfo_storeData] = recordAttr->recordSize;
         rec->info[recInfo_storeData] = (char *)
-           xmalloc (recordAttr->recordSize);
+           xmalloc(recordAttr->recordSize);
         stream->seekf(stream, start_offset);
         stream->readf(stream, rec->info[recInfo_storeData],
                       recordAttr->recordSize);
@@ -815,15 +1199,15 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
         rec->size[recInfo_storeData] = 0;
     }
     /* update database name */
-    xfree (rec->info[recInfo_databaseName]);
+    xfree(rec->info[recInfo_databaseName]);
     rec->info[recInfo_databaseName] =
-        rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]); 
+        rec_strdup(zh->basenames[0], &rec->size[recInfo_databaseName]); 
 
     /* update offset */
     recordAttr->recordOffset = start_offset;
     
     /* commit this record */
-    rec_put (zh->reg->records, &rec);
+    rec_put(zh->reg->records, &rec);
     logRecord(zh);
     return ZEBRA_OK;
 }
@@ -833,13 +1217,13 @@ ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
     ZebraHandle zh = (ZebraHandle) handle;
     struct recExtractCtrl extractCtrl;
 
-    if (zebraExplain_curDatabase (zh->reg->zei,
+    if (zebraExplain_curDatabase(zh->reg->zei,
                                  rec->info[recInfo_databaseName]))
     {
        abort();
-        if (zebraExplain_newDatabase (zh->reg->zei,
+        if (zebraExplain_newDatabase(zh->reg->zei,
                                      rec->info[recInfo_databaseName], 0))
-            abort ();
+            abort();
     }
 
     zebra_rec_keys_reset(zh->reg->keys);
@@ -855,6 +1239,8 @@ ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
     extractCtrl.flagShowRecords = 0;
     extractCtrl.match_criteria[0] = '\0';
     extractCtrl.staticrank = 0;
+    extractCtrl.action = action_update;
+
     extractCtrl.handle = handle;
     extractCtrl.first_record = 1;
     
@@ -872,7 +1258,13 @@ ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
        zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
                               rec->size[recInfo_delKeys],
                               0);
+#if FLUSH2
+       extract_flush_record_keys2(zh, rec->sysno, 
+                                   zh->reg->keys, 0, delkeys, 0);
+#else
        extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
+        extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
+#endif
        zebra_rec_keys_close(delkeys);
 
        zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
@@ -882,15 +1274,22 @@ ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
        extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
        zebra_rec_keys_close(sortkeys);
     }
-    extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
+    else
+    {
+#if FLUSH2
+       extract_flush_record_keys2(zh, rec->sysno, zh->reg->keys, 0, 0, 0);
+#else
+        extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);        
+#endif
+    }
     extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
     
-    xfree (rec->info[recInfo_delKeys]);
+    xfree(rec->info[recInfo_delKeys]);
     zebra_rec_keys_get_buf(zh->reg->keys,
                           &rec->info[recInfo_delKeys], 
                           &rec->size[recInfo_delKeys]);
 
-    xfree (rec->info[recInfo_sortKeys]);
+    xfree(rec->info[recInfo_sortKeys]);
     zebra_rec_keys_get_buf(zh->reg->sortKeys,
                           &rec->info[recInfo_sortKeys],
                           &rec->size[recInfo_sortKeys]);
@@ -913,7 +1312,8 @@ void extract_rec_keys_log(ZebraHandle zh, int is_insert,
             char keystr[200]; /* room for zints to print */
             char *dst_term = 0;
             int ord = CAST_ZINT_TO_INT(key.mem[0]);
-            int index_type, i;
+            const char *index_type;
+            int i;
             const char *string_index;
             
             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
@@ -947,12 +1347,12 @@ void extract_rec_keys_log(ZebraHandle zh, int is_insert,
                 {
                     sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
                 }
-                yaz_log(level, "%s%c %s %s", keystr, index_type,
+                yaz_log(level, "%s%s %s %s", keystr, index_type,
                         string_index, dst_buf);
                 
             }
             else
-                yaz_log(level, "%s%c %s \"%s\"", keystr, index_type,
+                yaz_log(level, "%s%s %s \"%s\"", keystr, index_type,
                         string_index, dst_term);
 
             nmem_reset(nmem);
@@ -1013,46 +1413,81 @@ void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
     }
 }
 
-void extract_flush_record_keys(ZebraHandle zh, zint sysno, int cmd,
-                               zebra_rec_keys_t reckeys,
-                               zint staticrank)
+void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
+                                zebra_rec_keys_t ins_keys, zint ins_rank,
+                                zebra_rec_keys_t del_keys, zint del_rank)
 {
     ZebraExplainInfo zei = zh->reg->zei;
-
-    extract_rec_keys_adjust(zh, cmd, reckeys);
-
-    if (log_level_details)
-    {
-        yaz_log(log_level_details, "Keys for record " ZINT_FORMAT " %s",
-                sysno, cmd ? "insert" : "delete");
-        extract_rec_keys_log(zh, cmd, reckeys, log_level_details);
-    }
+    int normal = 0;
+    int optimized = 0;
 
     if (!zh->reg->key_block)
     {
        int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
-        const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", ".");
-        int use_threads = atoi(res_get_def (zh->res, "threads", "1"));
+        const char *key_tmp_dir = res_get_def(zh->res, "keyTmpDir", ".");
+        int use_threads = atoi(res_get_def(zh->res, "threads", "1"));
         zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
     }
-    zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
 
-    if (zebra_rec_keys_rewind(reckeys))
+    if (ins_keys)
     {
-       size_t slen;
-       const char *str;
-       struct it_key key_in;
-       while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
-       {
+        extract_rec_keys_adjust(zh, 1, ins_keys);
+        if (!del_keys)
+            zebraExplain_recordCountIncrement(zei, 1);
+        zebra_rec_keys_rewind(ins_keys);
+    }
+    if (del_keys)
+    {
+        extract_rec_keys_adjust(zh, 0, del_keys);
+        if (!ins_keys)
+            zebraExplain_recordCountIncrement(zei, -1);
+        zebra_rec_keys_rewind(del_keys);
+    }
+
+    while (1)
+    {
+       size_t del_slen;
+       const char *del_str;
+       struct it_key del_key_in;
+        int del = 0;
+
+       size_t ins_slen;
+       const char *ins_str;
+       struct it_key ins_key_in;
+        int ins = 0;
+
+        if (del_keys)
+            del = zebra_rec_keys_read(del_keys, &del_str, &del_slen,
+                                      &del_key_in);
+        if (ins_keys)
+            ins = zebra_rec_keys_read(ins_keys, &ins_str, &ins_slen,
+                                      &ins_key_in);
+
+        if (del && ins && ins_rank == del_rank
+            && !key_compare(&del_key_in, &ins_key_in) 
+            && ins_slen == del_slen && !memcmp(del_str, ins_str, del_slen))
+        {
+            optimized++;
+            continue;
+        }
+        if (!del && !ins)
+            break;
+        
+        normal++;
+        if (del)
             key_block_write(zh->reg->key_block, sysno, 
-                            &key_in, cmd, str, slen,
-                            staticrank, zh->m_staticrank);
-       }
+                            &del_key_in, 0, del_str, del_slen,
+                            del_rank, zh->m_staticrank);
+        if (ins)
+            key_block_write(zh->reg->key_block, sysno, 
+                            &ins_key_in, 1, ins_str, ins_slen,
+                            ins_rank, zh->m_staticrank);
     }
+    yaz_log(log_level_extract, "normal=%d optimized=%d", normal, optimized);
 }
 
 
-ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
+ZEBRA_RES zebra_rec_keys_to_snippets1(ZebraHandle zh,
                                      zebra_rec_keys_t reckeys,
                                      zebra_snippets *snippets)
 {
@@ -1067,7 +1502,7 @@ ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
            char *dst_term = 0;
            int ord;
             zint seqno;
-           int index_type;
+           const char *index_type;
 
            assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
            seqno = key.mem[key.len-1];
@@ -1078,7 +1513,7 @@ ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
            assert(index_type);
            zebra_term_untrans_iconv(zh, nmem, index_type,
                                     &dst_term, str);
-           zebra_snippets_append(snippets, seqno, ord, dst_term);
+           zebra_snippets_append(snippets, seqno, 0, ord, dst_term);
            nmem_reset(nmem);
        }
     }
@@ -1098,7 +1533,7 @@ void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
        {
            char dst_buf[IT_MAX_WORD];
            zint seqno;
-           int index_type;
+           const char *index_type;
             int ord = CAST_ZINT_TO_INT(key.mem[0]);
            const char *db = 0;
            assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
@@ -1158,21 +1593,33 @@ static void extract_add_sort_string(RecWord *p, const char *str, int length)
     zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
 }
 
-static void extract_add_string(RecWord *p, const char *string, int length)
+static void extract_add_staticrank_string(RecWord *p,
+                                          const char *str, int length)
 {
-    ZebraHandle zh = p->extractCtrl->handle;
-    assert (length > 0);
+    char valz[40];
+    struct recExtractCtrl *ctrl = p->extractCtrl;
+
+    if (length > sizeof(valz)-1)
+        length = sizeof(valz)-1;
+
+    memcpy(valz, str, length);
+    valz[length] = '\0';
+    ctrl->staticrank = atozint(valz);
+}
+
+static void extract_add_string(RecWord *p, zebra_map_t zm,
+                               const char *string, int length)
+{
+    assert(length > 0);
 
     if (!p->index_name)
         return;
 
-    if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
-       extract_add_sort_string(p, string, length);
-    else
+    if (zebra_maps_is_index(zm))
     {
        extract_add_index_string(p, zinfo_index_category_index,
                                  string, length);
-        if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type))
+        if (zebra_maps_is_alwaysmatches(zm))
         {
             RecWord word;
             memcpy(&word, p, sizeof(word));
@@ -1182,18 +1629,25 @@ static void extract_add_string(RecWord *p, const char *string, int length)
                 &word, zinfo_index_category_alwaysmatches, "", 0);
         }
     }
+    else if (zebra_maps_is_sort(zm))
+    {
+       extract_add_sort_string(p, string, length);
+    }
+    else if (zebra_maps_is_staticrank(zm))
+    {
+       extract_add_staticrank_string(p, string, length);
+    }
 }
 
-static void extract_add_incomplete_field(RecWord *p)
+static void extract_add_incomplete_field(RecWord *p, zebra_map_t zm)
 {
-    ZebraHandle zh = p->extractCtrl->handle;
     const char *b = p->term_buf;
     int remain = p->term_len;
     int first = 1;
     const char **map = 0;
     
     if (remain > 0)
-       map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
+       map = zebra_maps_input(zm, &b, remain, 0);
 
     while (map)
     {
@@ -1205,8 +1659,7 @@ static void extract_add_incomplete_field(RecWord *p)
        {
            remain = p->term_len - (b - p->term_buf);
            if (remain > 0)
-               map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
-                                      remain, 0);
+               map = zebra_maps_input(zm, &b, remain, 0);
            else
                map = 0;
        }
@@ -1221,7 +1674,7 @@ static void extract_add_incomplete_field(RecWord *p)
                buf[i++] = *(cp++);
            remain = p->term_len - (b - p->term_buf);
            if (remain > 0)
-               map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
+               map = zebra_maps_input(zm, &b, remain, 0);
            else
                map = 0;
        }
@@ -1231,28 +1684,27 @@ static void extract_add_incomplete_field(RecWord *p)
         if (first)
         {   
             first = 0;
-            if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
+            if (zebra_maps_is_first_in_field(zm))
             {
                 /* first in field marker */
-                extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
+                extract_add_string(p, zm, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
                 p->seqno++;
             }
         }
-       extract_add_string (p, buf, i);
+       extract_add_string(p, zm, buf, i);
         p->seqno++;
     }
 }
 
-static void extract_add_complete_field (RecWord *p)
+static void extract_add_complete_field(RecWord *p, zebra_map_t zm)
 {
-    ZebraHandle zh = p->extractCtrl->handle;
     const char *b = p->term_buf;
     char buf[IT_MAX_WORD+1];
     const char **map = 0;
     int i = 0, remain = p->term_len;
 
     if (remain > 0)
-       map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
+       map = zebra_maps_input(zm, &b, remain, 1);
 
     while (remain > 0 && i < IT_MAX_WORD)
     {
@@ -1263,7 +1715,7 @@ static void extract_add_complete_field (RecWord *p)
            if (remain > 0)
            {
                int first = i ? 0 : 1;  /* first position */
-               map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, first);
+               map = zebra_maps_input(zm, &b, remain, first);
            }
            else
                map = 0;
@@ -1291,8 +1743,7 @@ static void extract_add_complete_field (RecWord *p)
            remain = p->term_len  - (b - p->term_buf);
            if (remain > 0)
            {
-               map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
-                                       remain, 0);
+               map = zebra_maps_input(zm, &b, remain, 0);
            }
            else
                map = 0;
@@ -1300,31 +1751,67 @@ static void extract_add_complete_field (RecWord *p)
     }
     if (!i)
        return;
-    extract_add_string (p, buf, i);
+    extract_add_string(p, zm, buf, i);
 }
 
+static void extract_add_icu(RecWord *p, zebra_map_t zm)
+{
+    const char *res_buf = 0;
+    size_t res_len = 0;
+
+    zebra_map_tokenize_start(zm, p->term_buf, p->term_len);
+    while (zebra_map_tokenize_next(zm, &res_buf, &res_len, 0, 0))
+    {
+        extract_add_string(p, zm, res_buf, res_len);
+        p->seqno++;
+    }
+}
+
+
+/** \brief top-level indexing handler for recctrl system
+    \param p token data to be indexed
+
+    Call sequence:
+    extract_token_add
+    extract_add_{in}_complete / extract_add_icu
+    extract_add_string
+    
+    extract_add_index_string
+    or
+    extract_add_sort_string
+    or
+    extract_add_staticrank_string
+    
+*/
 static void extract_token_add(RecWord *p)
 {
     ZebraHandle zh = p->extractCtrl->handle;
+    zebra_map_t zm = zebra_map_get_or_add(zh->reg->zebra_maps, p->index_type);
     WRBUF wrbuf;
 
-    if (log_level_extract)
+    if (log_level_details)
     {
-        yaz_log(log_level_extract, "extract_token_add "
-                "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
+        yaz_log(log_level_details, "extract_token_add "
+                "type=%s 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(zh->reg->zebra_maps, p->index_type, 0,
-                              p->term_buf, p->term_len)))
+    if ((wrbuf = zebra_replace(zm, 0, p->term_buf, p->term_len)))
     {
-       p->term_buf = wrbuf_buf(wrbuf);
-       p->term_len = wrbuf_len(wrbuf);
+        p->term_buf = wrbuf_buf(wrbuf);
+        p->term_len = wrbuf_len(wrbuf);
+    }
+    if (zebra_maps_is_icu(zm))
+    {
+        extract_add_icu(p, zm);
     }
-    if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type))
-       extract_add_complete_field (p);
     else
-       extract_add_incomplete_field(p);
+    {
+        if (zebra_maps_is_complete(zm))
+            extract_add_complete_field(p, zm);
+        else
+            extract_add_incomplete_field(p, zm);
+    }
 }
 
 static void extract_set_store_data_cb(struct recExtractCtrl *p,
@@ -1355,7 +1842,7 @@ static void extract_set_store_data_prepare(struct recExtractCtrl *p)
 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
 {
     ZebraHandle zh = (ZebraHandle) p->handle;
-    zebraExplain_addSchema (zh->reg->zei, oid);
+    zebraExplain_addSchema(zh->reg->zei, oid);
 }
 
 void extract_flush_sort_keys(ZebraHandle zh, zint sysno,