Changed extract code so that it optimizes updates of records where content
[idzebra-moved-to-github.git] / index / extract.c
1 /* $Id: extract.c,v 1.255 2007-04-07 22:26:27 adam Exp $
2    Copyright (C) 1995-2007
3    Index Data ApS
4
5 This file is part of the Zebra server.
6
7 Zebra is free software; you can redistribute it and/or modify it under
8 the terms of the GNU General Public License as published by the Free
9 Software Foundation; either version 2, or (at your option) any later
10 version.
11
12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
14 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
15 for more details.
16
17 You should have received a copy of the GNU General Public License
18 along with this program; if not, write to the Free Software
19 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
20
21 */
22
23 #include <stdio.h>
24 #include <assert.h>
25 #include <ctype.h>
26 #ifdef WIN32
27 #include <io.h>
28 #endif
29 #if HAVE_UNISTD_H
30 #include <unistd.h>
31 #endif
32 #include <fcntl.h>
33
34 #include "index.h"
35 #include "orddict.h"
36 #include <direntz.h>
37 #include <charmap.h>
38
39 static int log_level_extract = 0;
40 static int log_level_details = 0;
41 static int log_level_initialized = 0;
42
43 /* 1 if we use eliminitate identical delete/insert keys */
44 /* eventually this the 0-case code will be removed */
45 #define FLUSH2 1
46
47 void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
48                                 zebra_rec_keys_t ins_keys,
49                                 zint ins_rank,
50                                 zebra_rec_keys_t del_keys,
51                                 zint del_rank);
52
53 static void zebra_init_log_level(void)
54 {
55     if (!log_level_initialized)
56     {
57         log_level_initialized = 1;
58
59         log_level_extract = yaz_log_module_level("extract");
60         log_level_details = yaz_log_module_level("indexdetails");
61     }
62 }
63
64 static void extract_flush_record_keys(ZebraHandle zh, zint sysno,
65                                       int cmd, zebra_rec_keys_t reckeys,
66                                       zint staticrank);
67 static void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
68                                     int cmd, zebra_rec_keys_t skp);
69 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid);
70 static void extract_token_add (RecWord *p);
71
72 static void check_log_limit(ZebraHandle zh)
73 {
74     if (zh->records_processed + zh->records_skipped == zh->m_file_verbose_limit)
75     {
76         yaz_log(YLOG_LOG, "More than %d file log entries. Omitting rest",
77                 zh->m_file_verbose_limit);
78     }
79 }
80
81 static void logRecord (ZebraHandle zh)
82 {
83     check_log_limit(zh);
84     ++zh->records_processed;
85     if (!(zh->records_processed % 1000))
86     {
87         yaz_log(YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
88                 ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT, 
89                 zh->records_processed, zh->records_inserted, 
90                 zh->records_updated, zh->records_deleted);
91     }
92 }
93
94 static void extract_add_index_string (RecWord *p, 
95                                       zinfo_index_category_t cat,
96                                       const char *str, int length);
97
98 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
99
100 static void extract_init(struct recExtractCtrl *p, RecWord *w)
101 {
102     w->seqno = 1;
103     w->index_name = "any";
104     w->index_type = 'w';
105     w->extractCtrl = p;
106     w->record_id = 0;
107     w->section_id = 0;
108     w->segment = 0;
109 }
110
111 static void searchRecordKey(ZebraHandle zh,
112                             zebra_rec_keys_t reckeys,
113                             const char *index_name,
114                             const char **ws, int ws_length)
115 {
116     int i;
117     int ch = -1;
118     zinfo_index_category_t cat = zinfo_index_category_index;
119
120     for (i = 0; i<ws_length; i++)
121         ws[i] = NULL;
122
123     if (ch < 0)
124         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, '0', index_name);
125     if (ch < 0)
126         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'p', index_name);
127     if (ch < 0)
128         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'w', index_name);
129
130     if (ch < 0)
131         return ;
132
133     if (zebra_rec_keys_rewind(reckeys))
134     {
135         zint startSeq = -1;
136         const char *str;
137         size_t slen;
138         struct it_key key;
139         zint seqno;
140         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
141         {
142             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
143
144             seqno = key.mem[key.len-1];
145             
146             if (key.mem[0] == ch)
147             {
148                 zint woff;
149                 
150                 if (startSeq == -1)
151                     startSeq = seqno;
152                 woff = seqno - startSeq;
153                 if (woff >= 0 && woff < ws_length)
154                     ws[woff] = str;
155             }
156         }
157     }
158 }
159
160 #define FILE_MATCH_BLANK "\t "
161
162 static char *get_match_from_spec(ZebraHandle zh,
163                           zebra_rec_keys_t reckeys,
164                           const char *fname, const char *spec)
165 {
166     static char dstBuf[2048];      /* static here ??? */
167     char *dst = dstBuf;
168     const char *s = spec;
169
170     while (1)
171     {
172         for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
173             ;
174         if (!*s)
175             break;
176         if (*s == '(')
177         {
178             const char *ws[32];
179             char attset_str[64], attname_str[64];
180             int i;
181             int first = 1;
182             
183             for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
184                 ;
185             for (i = 0; *s && *s != ',' && *s != ')' && 
186                      !strchr(FILE_MATCH_BLANK, *s); s++)
187                 if (i+1 < sizeof(attset_str))
188                     attset_str[i++] = *s;
189             attset_str[i] = '\0';
190             
191             for (; strchr(FILE_MATCH_BLANK, *s); s++)
192                 ;
193             if (*s != ',')
194                 strcpy(attname_str, attset_str);
195             else
196             {
197                 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
198                     ;
199                 for (i = 0; *s && *s != ')' && 
200                          !strchr(FILE_MATCH_BLANK, *s); s++)
201                     if (i+1 < sizeof(attname_str))
202                         attname_str[i++] = *s;
203                 attname_str[i] = '\0';
204             }
205
206             searchRecordKey (zh, reckeys, attname_str, ws, 32);
207
208             if (*s != ')')
209             {
210                 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
211                       spec, zh->m_group ? zh->m_group : "none");
212                 return NULL;
213             }
214             s++;
215
216             for (i = 0; i<32; i++)
217                 if (ws[i])
218                 {
219                     if (first)
220                     {
221                         *dst++ = ' ';
222                         first = 0;
223                     }
224                     strcpy (dst, ws[i]);
225                     dst += strlen(ws[i]);
226                 }
227             if (first)
228             {
229                 yaz_log (YLOG_WARN, "Record didn't contain match"
230                       " fields in (%s,%s)", attset_str, attname_str);
231                 return NULL;
232             }
233         }
234         else if (*s == '$')
235         {
236             int spec_len;
237             char special[64];
238             const char *spec_src = NULL;
239             const char *s1 = ++s;
240             while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
241                 s1++;
242
243             spec_len = s1 - s;
244             if (spec_len > sizeof(special)-1)
245                 spec_len = sizeof(special)-1;
246             memcpy (special, s, spec_len);
247             special[spec_len] = '\0';
248             s = s1;
249
250             if (!strcmp (special, "group"))
251                 spec_src = zh->m_group;
252             else if (!strcmp (special, "database"))
253                 spec_src = zh->basenames[0];
254             else if (!strcmp (special, "filename")) {
255                 spec_src = fname;
256             }
257             else if (!strcmp (special, "type"))
258                 spec_src = zh->m_record_type;
259             else 
260                 spec_src = NULL;
261             if (spec_src)
262             {
263                 strcpy (dst, spec_src);
264                 dst += strlen (spec_src);
265             }
266         }
267         else if (*s == '\"' || *s == '\'')
268         {
269             int stopMarker = *s++;
270             char tmpString[64];
271             int i = 0;
272
273             while (*s && *s != stopMarker)
274             {
275                 if (i+1 < sizeof(tmpString))
276                     tmpString[i++] = *s++;
277             }
278             if (*s)
279                 s++;
280             tmpString[i] = '\0';
281             strcpy (dst, tmpString);
282             dst += strlen (tmpString);
283         }
284         else
285         {
286             yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
287                   spec, zh->m_group ? zh->m_group : "none");
288             return NULL;
289         }
290         *dst++ = 1;
291     }
292     if (dst == dstBuf)
293     {
294         yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
295               fname, zh->m_group ? zh->m_group : "none");
296         return NULL;
297     }
298     *dst = '\0';
299     return dstBuf;
300 }
301
302 struct recordLogInfo {
303     const char *fname;
304     int recordOffset;
305     struct recordGroup *rGroup;
306 };
307
308 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
309 {
310     int i;
311     for (i = 0; i<256; i++)
312     {
313         if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
314             ctrl->seqno[i] = 1;
315         else
316             ctrl->seqno[i] = 0;
317     }
318     ctrl->flagShowRecords = !zh->m_flag_rw;
319 }
320
321 static void all_matches_add(struct recExtractCtrl *ctrl)
322 {
323     RecWord word;
324     extract_init(ctrl, &word);
325     word.index_name = "_ALLRECORDS";
326     word.index_type = 'w';
327     word.seqno = 1;
328     extract_add_index_string (&word, zinfo_index_category_alwaysmatches,
329                               "", 0);
330 }
331
332 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh, 
333                                        struct ZebraRecStream *stream,
334                                        enum zebra_recctrl_action_t action,
335                                        int test_mode, 
336                                        const char *recordType,
337                                        zint *sysno,
338                                        const char *match_criteria,
339                                        const char *fname,
340                                        RecType recType,
341                                        void *recTypeClientData);
342
343
344 ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname, 
345                              int deleteFlag)
346 {
347     ZEBRA_RES r = ZEBRA_OK;
348     int i, fd;
349     char gprefix[128];
350     char ext[128];
351     char ext_res[128];
352     struct file_read_info *fi = 0;
353     const char *original_record_type = 0;
354     RecType recType;
355     void *recTypeClientData;
356     struct ZebraRecStream stream, *streamp;
357
358     zebra_init_log_level();
359
360     if (!zh->m_group || !*zh->m_group)
361         *gprefix = '\0';
362     else
363         sprintf (gprefix, "%s.", zh->m_group);
364     
365     yaz_log(log_level_extract, "zebra_extract_file %s", fname);
366
367     /* determine file extension */
368     *ext = '\0';
369     for (i = strlen(fname); --i >= 0; )
370         if (fname[i] == '/')
371             break;
372         else if (fname[i] == '.')
373         {
374             strcpy (ext, fname+i+1);
375             break;
376         }
377     /* determine file type - depending on extension */
378     original_record_type = zh->m_record_type;
379     if (!zh->m_record_type)
380     {
381         sprintf (ext_res, "%srecordType.%s", gprefix, ext);
382         zh->m_record_type = res_get (zh->res, ext_res);
383     }
384     if (!zh->m_record_type)
385     {
386         check_log_limit(zh);
387         if (zh->records_processed + zh->records_skipped
388             < zh->m_file_verbose_limit)
389             yaz_log (YLOG_LOG, "? %s", fname);
390         zh->records_skipped++;
391         return 0;
392     }
393     /* determine match criteria */
394     if (!zh->m_record_id)
395     {
396         sprintf (ext_res, "%srecordId.%s", gprefix, ext);
397         zh->m_record_id = res_get (zh->res, ext_res);
398     }
399
400     if (!(recType =
401           recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
402                           &recTypeClientData)))
403     {
404         yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
405         return ZEBRA_FAIL;
406     }
407
408     switch(recType->version)
409     {
410     case 0:
411         break;
412     default:
413         yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
414     }
415     if (sysno && deleteFlag)
416     {
417         streamp = 0;
418         fi = 0;
419     }
420     else
421     {
422         char full_rep[1024];
423
424         if (zh->path_reg && !yaz_is_abspath (fname))
425         {
426             strcpy (full_rep, zh->path_reg);
427             strcat (full_rep, "/");
428             strcat (full_rep, fname);
429         }
430         else
431             strcpy (full_rep, fname);
432         
433         if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
434         {
435             yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
436             zh->m_record_type = original_record_type;
437             return ZEBRA_FAIL;
438         }
439         streamp = &stream;
440         zebra_create_stream_fd(streamp, fd, 0);
441     }
442     r = zebra_extract_records_stream(zh, streamp,
443                                      deleteFlag ? 
444                                      action_delete : action_update,
445                                      0, /* tst_mode */
446                                      zh->m_record_type,
447                                      sysno,
448                                      0, /*match_criteria */
449                                      fname,
450                                      recType, recTypeClientData);
451     if (streamp)
452         stream.destroy(streamp);
453     zh->m_record_type = original_record_type;
454     return r;
455 }
456
457 /*
458   If sysno is provided, then it's used to identify the reocord.
459   If not, and match_criteria is provided, then sysno is guessed
460   If not, and a record is provided, then sysno is got from there
461   
462  */
463
464 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh, 
465                                       const char *buf, size_t buf_size,
466                                       enum zebra_recctrl_action_t action,
467                                       int test_mode, 
468                                       const char *recordType,
469                                       zint *sysno,
470                                       const char *match_criteria,
471                                       const char *fname)
472 {
473     struct ZebraRecStream stream;
474     ZEBRA_RES res;
475     void *clientData;
476     RecType recType = 0;
477
478     if (recordType && *recordType)
479     {
480         yaz_log(log_level_extract,
481                 "Record type explicitly specified: %s", recordType);
482         recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
483                                   &clientData);
484     } 
485     else
486     {
487         if (!(zh->m_record_type))
488         {
489             yaz_log (YLOG_WARN, "No such record type defined");
490             return ZEBRA_FAIL;
491         }
492         yaz_log(log_level_extract, "Get record type from rgroup: %s",
493                 zh->m_record_type);
494         recType = recType_byName (zh->reg->recTypes, zh->res,
495                                   zh->m_record_type, &clientData);
496         recordType = zh->m_record_type;
497     }
498     
499     if (!recType)
500     {
501         yaz_log (YLOG_WARN, "No such record type: %s", recordType);
502         return ZEBRA_FAIL;
503     }
504
505     zebra_create_stream_mem(&stream, buf, buf_size);
506
507     res = zebra_extract_records_stream(zh, &stream,
508                                        action,
509                                        test_mode, 
510                                        recordType,
511                                        sysno,
512                                        match_criteria,
513                                        fname,
514                                        recType, clientData);
515     stream.destroy(&stream);
516     return res;
517 }
518
519 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh, 
520                                        struct ZebraRecStream *stream,
521                                        enum zebra_recctrl_action_t action,
522                                        int test_mode, 
523                                        const char *recordType,
524                                        zint *sysno,
525                                        const char *match_criteria,
526                                        const char *fname,
527                                        RecType recType,
528                                        void *recTypeClientData)
529 {
530     ZEBRA_RES res = ZEBRA_OK;
531     while (1)
532     {
533         int more = 0;
534         res = zebra_extract_record_stream(zh, stream,
535                                           action,
536                                           test_mode, 
537                                           recordType,
538                                           sysno,
539                                           match_criteria,
540                                           fname,
541                                           recType, recTypeClientData, &more);
542         if (!more)
543         {
544             res = ZEBRA_OK;
545             break;
546         }
547         if (res != ZEBRA_OK)
548             break;
549         if (sysno)
550             break;
551     }
552     return res;
553 }
554
555
556 ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, 
557                                       struct ZebraRecStream *stream,
558                                       enum zebra_recctrl_action_t action,
559                                       int test_mode, 
560                                       const char *recordType,
561                                       zint *sysno,
562                                       const char *match_criteria,
563                                       const char *fname,
564                                       RecType recType,
565                                       void *recTypeClientData,
566                                       int *more)
567
568 {
569     zint sysno0 = 0;
570     RecordAttr *recordAttr;
571     struct recExtractCtrl extractCtrl;
572     int r;
573     const char *matchStr = 0;
574     Record rec;
575     off_t start_offset = 0, end_offset = 0;
576     const char *pr_fname = fname;  /* filename to print .. */
577     int show_progress = zh->records_processed + zh->records_skipped 
578         < zh->m_file_verbose_limit ? 1:0;
579
580     zebra_init_log_level();
581
582     if (!pr_fname)
583         pr_fname = "<no file>";  /* make it printable if file is omitted */
584
585     zebra_rec_keys_reset(zh->reg->keys);
586     zebra_rec_keys_reset(zh->reg->sortKeys);
587
588     if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
589     {
590         if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0], 
591                                       zh->m_explain_database))
592             return ZEBRA_FAIL;
593     }
594
595     if (stream)
596     {
597         off_t null_offset = 0;
598         extractCtrl.stream = stream;
599
600         start_offset = stream->tellf(stream);
601
602         extractCtrl.first_record = start_offset ? 0 : 1;
603         
604         stream->endf(stream, &null_offset);;
605
606         extractCtrl.init = extract_init;
607         extractCtrl.tokenAdd = extract_token_add;
608         extractCtrl.schemaAdd = extract_schema_add;
609         extractCtrl.dh = zh->reg->dh;
610         extractCtrl.handle = zh;
611         extractCtrl.match_criteria[0] = '\0';
612         extractCtrl.staticrank = 0;
613         extractCtrl.action = action;
614
615         init_extractCtrl(zh, &extractCtrl);
616
617         extract_set_store_data_prepare(&extractCtrl);
618         
619         r = (*recType->extract)(recTypeClientData, &extractCtrl);
620
621         if (action == action_update)
622         {
623             action = extractCtrl.action;
624         }
625         
626         switch (r)
627         {
628         case RECCTRL_EXTRACT_EOF:
629             return ZEBRA_FAIL;
630         case RECCTRL_EXTRACT_ERROR_GENERIC:
631             /* error occured during extraction ... */
632             yaz_log (YLOG_WARN, "extract error: generic");
633             return ZEBRA_FAIL;
634         case RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER:
635             /* error occured during extraction ... */
636             yaz_log (YLOG_WARN, "extract error: no such filter");
637             return ZEBRA_FAIL;
638         case RECCTRL_EXTRACT_SKIP:
639             if (show_progress)
640                 yaz_log (YLOG_LOG, "skip %s %s " ZINT_FORMAT,
641                          recordType, pr_fname, (zint) start_offset);
642             *more = 1;
643             
644             end_offset = stream->endf(stream, 0);
645             if (end_offset)
646                 stream->seekf(stream, end_offset);
647
648             return ZEBRA_OK;
649         case RECCTRL_EXTRACT_OK:
650             break;
651         default:
652             yaz_log (YLOG_WARN, "extract error: unknown error: %d", r);
653             return ZEBRA_FAIL;
654         }
655         end_offset = stream->endf(stream, 0);
656         if (end_offset)
657             stream->seekf(stream, end_offset);
658         else
659             end_offset = stream->tellf(stream);
660
661         all_matches_add(&extractCtrl);
662         
663         if (extractCtrl.match_criteria[0])
664             match_criteria = extractCtrl.match_criteria;
665     }
666
667     *more = 1;
668     if (!sysno)
669     {
670         sysno = &sysno0;
671
672         if (match_criteria && *match_criteria) {
673             matchStr = match_criteria;
674         } else {
675             if (zh->m_record_id && *zh->m_record_id) {
676                 matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname, 
677                                                zh->m_record_id);
678                 if (!matchStr)
679                 {
680                     yaz_log (YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
681                              pr_fname, (zint) start_offset);
682                     return ZEBRA_FAIL;
683                 }
684             }
685         }
686         if (matchStr) 
687         {
688             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
689             char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
690                                           matchStr);
691             if (rinfo)
692             {
693                 assert(*rinfo == sizeof(*sysno));
694                 memcpy (sysno, rinfo+1, sizeof(*sysno));
695             }
696         }
697     }
698     if (zebra_rec_keys_empty(zh->reg->keys))
699     {
700         /* the extraction process returned no information - the record
701            is probably empty - unless flagShowRecords is in use */
702         if (test_mode)
703             return ZEBRA_OK;
704     }
705
706     if (! *sysno)
707     {
708         /* new record */
709         if (action == action_delete)
710         {
711             yaz_log (YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
712                          pr_fname, (zint) start_offset);
713             yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
714             return ZEBRA_FAIL;
715         }
716         else if (action == action_replace)
717         {
718             yaz_log (YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
719                          pr_fname, (zint) start_offset);
720             yaz_log (YLOG_WARN, "cannot update record above (seems new)");
721             return ZEBRA_FAIL;
722         }
723         if (show_progress)
724             yaz_log (YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
725                      (zint) start_offset);
726         rec = rec_new (zh->reg->records);
727
728         *sysno = rec->sysno;
729
730         recordAttr = rec_init_attr (zh->reg->zei, rec);
731         if (extractCtrl.staticrank < 0)
732         {
733             yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
734             extractCtrl.staticrank = 0;
735         }
736
737         if (matchStr)
738         {
739             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
740             dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
741                             sizeof(*sysno), sysno);
742         }
743
744         extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
745 #if FLUSH2
746         extract_flush_record_keys2(zh, *sysno,
747                                    zh->reg->keys, extractCtrl.staticrank,
748                                    0, recordAttr->staticrank);
749 #else
750         extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
751                                   extractCtrl.staticrank);
752 #endif
753         recordAttr->staticrank = extractCtrl.staticrank;
754         zh->records_inserted++;
755     } 
756     else
757     {
758         /* record already exists */
759         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
760         zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
761         if (action == action_insert)
762         {
763             yaz_log (YLOG_LOG, "skipped %s %s " ZINT_FORMAT, 
764                          recordType, pr_fname, (zint) start_offset);
765             logRecord(zh);
766             return ZEBRA_FAIL;
767         }
768
769         rec = rec_get (zh->reg->records, *sysno);
770         assert (rec);
771         
772         recordAttr = rec_init_attr (zh->reg->zei, rec);
773
774         /* decrease total size */
775         zebraExplain_recordBytesIncrement (zh->reg->zei,
776                                            - recordAttr->recordSize);
777
778         zebra_rec_keys_set_buf(delkeys,
779                                rec->info[recInfo_delKeys],
780                                rec->size[recInfo_delKeys],
781                                0);
782         zebra_rec_keys_set_buf(sortKeys,
783                                rec->info[recInfo_sortKeys],
784                                rec->size[recInfo_sortKeys],
785                                0);
786
787         extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
788 #if !FLUSH2
789         extract_flush_record_keys(zh, *sysno, 0, delkeys,
790                                   recordAttr->staticrank);
791 #endif
792         if (action == action_delete)
793         {
794             /* record going to be deleted */
795 #if FLUSH2
796             extract_flush_record_keys2(zh, *sysno, 0, recordAttr->staticrank,
797                                        delkeys, recordAttr->staticrank);
798 #endif       
799             if (zebra_rec_keys_empty(delkeys))
800             {
801                 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
802                         pr_fname, (zint) start_offset);
803                 yaz_log(YLOG_WARN, "cannot delete file above, "
804                         "storeKeys false (3)");
805             }
806             else
807             {
808                 if (show_progress)
809                     yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
810                             pr_fname, (zint) start_offset);
811                 zh->records_deleted++;
812                 if (matchStr)
813                 {
814                     int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
815                     dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
816                 }
817                 rec_del (zh->reg->records, &rec);
818             }
819             zebra_rec_keys_close(delkeys);
820             zebra_rec_keys_close(sortKeys);
821             rec_free(&rec);
822             logRecord(zh);
823             return ZEBRA_OK;
824         }
825         else
826         {   /* update or special_update */
827             if (show_progress)
828                 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
829                         pr_fname, (zint) start_offset);
830             extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
831
832 #if FLUSH2
833             extract_flush_record_keys2(zh, *sysno,
834                                        zh->reg->keys, extractCtrl.staticrank,
835                                        delkeys, recordAttr->staticrank);
836 #else
837             extract_flush_record_keys(zh, *sysno, 1, 
838                                       zh->reg->keys, extractCtrl.staticrank);
839 #endif
840             recordAttr->staticrank = extractCtrl.staticrank;
841             zh->records_updated++;
842         }
843         zebra_rec_keys_close(delkeys);
844         zebra_rec_keys_close(sortKeys);
845     }
846     /* update file type */
847     xfree (rec->info[recInfo_fileType]);
848     rec->info[recInfo_fileType] =
849         rec_strdup (recordType, &rec->size[recInfo_fileType]);
850
851     /* update filename */
852     xfree (rec->info[recInfo_filename]);
853     rec->info[recInfo_filename] =
854         rec_strdup (fname, &rec->size[recInfo_filename]);
855
856     /* update delete keys */
857     xfree (rec->info[recInfo_delKeys]);
858     if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
859     {
860         zebra_rec_keys_get_buf(zh->reg->keys,
861                                &rec->info[recInfo_delKeys],
862                                &rec->size[recInfo_delKeys]);
863     }
864     else
865     {
866         rec->info[recInfo_delKeys] = NULL;
867         rec->size[recInfo_delKeys] = 0;
868     }
869     /* update sort keys */
870     xfree (rec->info[recInfo_sortKeys]);
871
872     zebra_rec_keys_get_buf(zh->reg->sortKeys,
873                            &rec->info[recInfo_sortKeys],
874                            &rec->size[recInfo_sortKeys]);
875
876     if (stream)
877     {
878         recordAttr->recordSize = end_offset - start_offset;
879         zebraExplain_recordBytesIncrement(zh->reg->zei,
880                                           recordAttr->recordSize);
881     }
882
883     /* set run-number for this record */
884     recordAttr->runNumber =
885         zebraExplain_runNumberIncrement (zh->reg->zei, 0);
886
887     /* update store data */
888     xfree (rec->info[recInfo_storeData]);
889
890     /* update store data */
891     if (zh->store_data_buf)
892     {
893         rec->size[recInfo_storeData] = zh->store_data_size;
894         rec->info[recInfo_storeData] = zh->store_data_buf;
895         zh->store_data_buf = 0;
896         recordAttr->recordSize = zh->store_data_size;
897     }
898     else if (zh->m_store_data)
899     {
900         off_t cur_offset = stream->tellf(stream);
901
902         rec->size[recInfo_storeData] = recordAttr->recordSize;
903         rec->info[recInfo_storeData] = (char *)
904             xmalloc (recordAttr->recordSize);
905         stream->seekf(stream, start_offset);
906         stream->readf(stream, rec->info[recInfo_storeData],
907                       recordAttr->recordSize);
908         stream->seekf(stream, cur_offset);
909     }
910     else
911     {
912         rec->info[recInfo_storeData] = NULL;
913         rec->size[recInfo_storeData] = 0;
914     }
915     /* update database name */
916     xfree (rec->info[recInfo_databaseName]);
917     rec->info[recInfo_databaseName] =
918         rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]); 
919
920     /* update offset */
921     recordAttr->recordOffset = start_offset;
922     
923     /* commit this record */
924     rec_put (zh->reg->records, &rec);
925     logRecord(zh);
926     return ZEBRA_OK;
927 }
928
929 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
930 {
931     ZebraHandle zh = (ZebraHandle) handle;
932     struct recExtractCtrl extractCtrl;
933
934     if (zebraExplain_curDatabase (zh->reg->zei,
935                                   rec->info[recInfo_databaseName]))
936     {
937         abort();
938         if (zebraExplain_newDatabase (zh->reg->zei,
939                                       rec->info[recInfo_databaseName], 0))
940             abort ();
941     }
942
943     zebra_rec_keys_reset(zh->reg->keys);
944     zebra_rec_keys_reset(zh->reg->sortKeys);
945
946     extractCtrl.init = extract_init;
947     extractCtrl.tokenAdd = extract_token_add;
948     extractCtrl.schemaAdd = extract_schema_add;
949     extractCtrl.dh = zh->reg->dh;
950
951     init_extractCtrl(zh, &extractCtrl);
952
953     extractCtrl.flagShowRecords = 0;
954     extractCtrl.match_criteria[0] = '\0';
955     extractCtrl.staticrank = 0;
956     extractCtrl.action = action_update;
957
958     extractCtrl.handle = handle;
959     extractCtrl.first_record = 1;
960     
961     extract_set_store_data_prepare(&extractCtrl);
962
963     if (n)
964         grs_extract_tree(&extractCtrl, n);
965
966     if (rec->size[recInfo_delKeys])
967     {
968         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
969         
970         zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
971
972         zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
973                                rec->size[recInfo_delKeys],
974                                0);
975 #if FLUSH2
976         extract_flush_record_keys2(zh, rec->sysno, 
977                                    zh->reg->keys, 0, delkeys, 0);
978 #else
979         extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
980         extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
981 #endif
982         zebra_rec_keys_close(delkeys);
983
984         zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
985                                rec->size[recInfo_sortKeys],
986                                0);
987
988         extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
989         zebra_rec_keys_close(sortkeys);
990     }
991     else
992     {
993 #if FLUSH2
994         extract_flush_record_keys2(zh, rec->sysno, zh->reg->keys, 0, 0, 0);
995 #else
996         extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);        
997 #endif
998     }
999     extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
1000     
1001     xfree (rec->info[recInfo_delKeys]);
1002     zebra_rec_keys_get_buf(zh->reg->keys,
1003                            &rec->info[recInfo_delKeys], 
1004                            &rec->size[recInfo_delKeys]);
1005
1006     xfree (rec->info[recInfo_sortKeys]);
1007     zebra_rec_keys_get_buf(zh->reg->sortKeys,
1008                            &rec->info[recInfo_sortKeys],
1009                            &rec->size[recInfo_sortKeys]);
1010     return ZEBRA_OK;
1011 }
1012
1013 void extract_rec_keys_log(ZebraHandle zh, int is_insert,
1014                           zebra_rec_keys_t reckeys,
1015                           int level)
1016 {
1017     if (zebra_rec_keys_rewind(reckeys))
1018     {
1019         size_t slen;
1020         const char *str;
1021         struct it_key key;
1022         NMEM nmem = nmem_create();
1023
1024         while(zebra_rec_keys_read(reckeys, &str, &slen, &key))
1025         {
1026             char keystr[200]; /* room for zints to print */
1027             char *dst_term = 0;
1028             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1029             int index_type, i;
1030             const char *string_index;
1031             
1032             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1033                                     0/* db */, &string_index);
1034             assert(index_type);
1035             zebra_term_untrans_iconv(zh, nmem, index_type,
1036                                      &dst_term, str);
1037             *keystr = '\0';
1038             for (i = 0; i<key.len; i++)
1039             {
1040                 sprintf(keystr + strlen(keystr), ZINT_FORMAT " ", key.mem[i]);
1041             }
1042
1043             if (*str < CHR_BASE_CHAR)
1044             {
1045                 int i;
1046                 char dst_buf[200]; /* room for special chars */
1047
1048                 strcpy(dst_buf , "?");
1049
1050                 if (!strcmp(str, ""))
1051                     strcpy(dst_buf, "alwaysmatches");
1052                 if (!strcmp(str, FIRST_IN_FIELD_STR))
1053                     strcpy(dst_buf, "firstinfield");
1054                 else if (!strcmp(str, CHR_UNKNOWN))
1055                     strcpy(dst_buf, "unknown");
1056                 else if (!strcmp(str, CHR_SPACE))
1057                     strcpy(dst_buf, "space");
1058                 
1059                 for (i = 0; i<slen; i++)
1060                 {
1061                     sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
1062                 }
1063                 yaz_log(level, "%s%c %s %s", keystr, index_type,
1064                         string_index, dst_buf);
1065                 
1066             }
1067             else
1068                 yaz_log(level, "%s%c %s \"%s\"", keystr, index_type,
1069                         string_index, dst_term);
1070
1071             nmem_reset(nmem);
1072         }
1073         nmem_destroy(nmem);
1074     }
1075 }
1076
1077 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
1078                              zebra_rec_keys_t reckeys)
1079 {
1080     ZebraExplainInfo zei = zh->reg->zei;
1081     struct ord_stat {
1082         int no;
1083         int ord;
1084         struct ord_stat *next;
1085     };
1086
1087     if (zebra_rec_keys_rewind(reckeys))
1088     {
1089         struct ord_stat *ord_list = 0;
1090         struct ord_stat *p;
1091         size_t slen;
1092         const char *str;
1093         struct it_key key_in;
1094         while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1095         {
1096             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1097
1098             for (p = ord_list; p ; p = p->next)
1099                 if (p->ord == ord)
1100                 {
1101                     p->no++;
1102                     break;
1103                 }
1104             if (!p)
1105             {
1106                 p = xmalloc(sizeof(*p));
1107                 p->no = 1;
1108                 p->ord = ord;
1109                 p->next = ord_list;
1110                 ord_list = p;
1111             }
1112         }
1113
1114         p = ord_list;
1115         while (p)
1116         {
1117             struct ord_stat *p1 = p;
1118
1119             if (is_insert)
1120                 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
1121             else
1122                 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
1123             p = p->next;
1124             xfree(p1);
1125         }
1126     }
1127 }
1128
1129 void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
1130                                 zebra_rec_keys_t ins_keys, zint ins_rank,
1131                                 zebra_rec_keys_t del_keys, zint del_rank)
1132 {
1133     ZebraExplainInfo zei = zh->reg->zei;
1134     int normal = 0;
1135     int optimized = 0;
1136
1137     if (!zh->reg->key_block)
1138     {
1139         int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1140         const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", ".");
1141         int use_threads = atoi(res_get_def (zh->res, "threads", "1"));
1142         zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1143     }
1144
1145     if (ins_keys)
1146     {
1147         extract_rec_keys_adjust(zh, 1, ins_keys);
1148         if (!del_keys)
1149             zebraExplain_recordCountIncrement (zei, 1);
1150         zebra_rec_keys_rewind(ins_keys);
1151     }
1152     if (del_keys)
1153     {
1154         extract_rec_keys_adjust(zh, 0, del_keys);
1155         if (!ins_keys)
1156             zebraExplain_recordCountIncrement (zei, -1);
1157         zebra_rec_keys_rewind(del_keys);
1158     }
1159
1160     while (1)
1161     {
1162         size_t del_slen;
1163         const char *del_str;
1164         struct it_key del_key_in;
1165         int del = 0;
1166
1167         size_t ins_slen;
1168         const char *ins_str;
1169         struct it_key ins_key_in;
1170         int ins = 0;
1171
1172         if (del_keys)
1173             del = zebra_rec_keys_read(del_keys, &del_str, &del_slen,
1174                                       &del_key_in);
1175         if (ins_keys)
1176             ins = zebra_rec_keys_read(ins_keys, &ins_str, &ins_slen,
1177                                       &ins_key_in);
1178
1179         if (del && ins && ins_rank == del_rank
1180             && !key_compare(&del_key_in, &ins_key_in) 
1181             && ins_slen == del_slen && !memcmp(del_str, ins_str, del_slen))
1182         {
1183             optimized++;
1184             continue;
1185         }
1186         if (!del && !ins)
1187             break;
1188         
1189         normal++;
1190         if (del)
1191             key_block_write(zh->reg->key_block, sysno, 
1192                             &del_key_in, 0, del_str, del_slen,
1193                             del_rank, zh->m_staticrank);
1194         if (ins)
1195             key_block_write(zh->reg->key_block, sysno, 
1196                             &ins_key_in, 1, ins_str, ins_slen,
1197                             ins_rank, zh->m_staticrank);
1198     }
1199     yaz_log(YLOG_LOG, "normal=%d optimized=%d", normal, optimized);
1200 }
1201
1202 void extract_flush_record_keys(ZebraHandle zh, zint sysno, int cmd,
1203                                zebra_rec_keys_t reckeys,
1204                                zint staticrank)
1205 {
1206     ZebraExplainInfo zei = zh->reg->zei;
1207
1208     extract_rec_keys_adjust(zh, cmd, reckeys);
1209
1210     if (log_level_details)
1211     {
1212         yaz_log(log_level_details, "Keys for record " ZINT_FORMAT " %s",
1213                 sysno, cmd ? "insert" : "delete");
1214         extract_rec_keys_log(zh, cmd, reckeys, log_level_details);
1215     }
1216
1217     if (!zh->reg->key_block)
1218     {
1219         int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1220         const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", ".");
1221         int use_threads = atoi(res_get_def (zh->res, "threads", "1"));
1222         zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1223     }
1224     zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1225
1226 #if 0
1227     yaz_log(YLOG_LOG, "sysno=" ZINT_FORMAT " cmd=%d", sysno, cmd);
1228     print_rec_keys(zh, reckeys);
1229 #endif
1230     if (zebra_rec_keys_rewind(reckeys))
1231     {
1232         size_t slen;
1233         const char *str;
1234         struct it_key key_in;
1235         while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1236         {
1237             key_block_write(zh->reg->key_block, sysno, 
1238                             &key_in, cmd, str, slen,
1239                             staticrank, zh->m_staticrank);
1240         }
1241     }
1242 }
1243
1244 ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
1245                                      zebra_rec_keys_t reckeys,
1246                                      zebra_snippets *snippets)
1247 {
1248     NMEM nmem = nmem_create();
1249     if (zebra_rec_keys_rewind(reckeys)) 
1250     {
1251         const char *str;
1252         size_t slen;
1253         struct it_key key;
1254         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1255         {
1256             char *dst_term = 0;
1257             int ord;
1258             zint seqno;
1259             int index_type;
1260
1261             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1262             seqno = key.mem[key.len-1];
1263             ord = CAST_ZINT_TO_INT(key.mem[0]);
1264             
1265             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1266                                     0/* db */, 0 /* string_index */);
1267             assert(index_type);
1268             zebra_term_untrans_iconv(zh, nmem, index_type,
1269                                      &dst_term, str);
1270             zebra_snippets_append(snippets, seqno, ord, dst_term);
1271             nmem_reset(nmem);
1272         }
1273     }
1274     nmem_destroy(nmem);
1275     return ZEBRA_OK;
1276 }
1277
1278 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1279 {
1280     yaz_log(YLOG_LOG, "print_rec_keys");
1281     if (zebra_rec_keys_rewind(reckeys))
1282     {
1283         const char *str;
1284         size_t slen;
1285         struct it_key key;
1286         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1287         {
1288             char dst_buf[IT_MAX_WORD];
1289             zint seqno;
1290             int index_type;
1291             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1292             const char *db = 0;
1293             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1294
1295             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1296             
1297             seqno = key.mem[key.len-1];
1298             
1299             zebra_term_untrans(zh, index_type, dst_buf, str);
1300             
1301             yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT 
1302                     " term=%s", ord, seqno, dst_buf); 
1303         }
1304     }
1305 }
1306
1307 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1308                                      const char *str, int length)
1309 {
1310     struct it_key key;
1311     ZebraHandle zh = p->extractCtrl->handle;
1312     ZebraExplainInfo zei = zh->reg->zei;
1313     int ch, i;
1314
1315     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1316     if (ch < 0)
1317         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1318
1319     i = 0;
1320     key.mem[i++] = ch;
1321     key.mem[i++] = p->record_id;
1322     key.mem[i++] = p->section_id;
1323
1324     if (zh->m_segment_indexing)
1325         key.mem[i++] = p->segment;
1326     key.mem[i++] = p->seqno;
1327     key.len = i;
1328
1329     zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1330 }
1331
1332 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1333 {
1334     struct it_key key;
1335     ZebraHandle zh = p->extractCtrl->handle;
1336     ZebraExplainInfo zei = zh->reg->zei;
1337     int ch;
1338     zinfo_index_category_t cat = zinfo_index_category_sort;
1339
1340     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1341     if (ch < 0)
1342         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1343     key.len = 2;
1344     key.mem[0] = ch;
1345     key.mem[1] = p->record_id;
1346
1347     zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1348 }
1349
1350 static void extract_add_staticrank_string(RecWord *p,
1351                                           const char *str, int length)
1352 {
1353     char valz[40];
1354     struct recExtractCtrl *ctrl = p->extractCtrl;
1355
1356     if (length > sizeof(valz)-1)
1357         length = sizeof(valz)-1;
1358
1359     memcpy(valz, str, length);
1360     valz[length] = '\0';
1361     ctrl->staticrank = atozint(valz);
1362 }
1363
1364 static void extract_add_string(RecWord *p, const char *string, int length)
1365 {
1366     ZebraHandle zh = p->extractCtrl->handle;
1367     assert (length > 0);
1368
1369     if (!p->index_name)
1370         return;
1371
1372     if (zebra_maps_is_index(zh->reg->zebra_maps, p->index_type))
1373     {
1374         extract_add_index_string(p, zinfo_index_category_index,
1375                                  string, length);
1376         if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type))
1377         {
1378             RecWord word;
1379             memcpy(&word, p, sizeof(word));
1380
1381             word.seqno = 1;
1382             extract_add_index_string(
1383                 &word, zinfo_index_category_alwaysmatches, "", 0);
1384         }
1385     }
1386     else if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
1387     {
1388         extract_add_sort_string(p, string, length);
1389     }
1390     else if (zebra_maps_is_staticrank(zh->reg->zebra_maps, p->index_type))
1391     {
1392         extract_add_staticrank_string(p, string, length);
1393     }
1394 }
1395
1396 static void extract_add_incomplete_field(RecWord *p)
1397 {
1398     ZebraHandle zh = p->extractCtrl->handle;
1399     const char *b = p->term_buf;
1400     int remain = p->term_len;
1401     int first = 1;
1402     const char **map = 0;
1403     
1404     if (remain > 0)
1405         map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1406
1407     while (map)
1408     {
1409         char buf[IT_MAX_WORD+1];
1410         int i, remain;
1411
1412         /* Skip spaces */
1413         while (map && *map && **map == *CHR_SPACE)
1414         {
1415             remain = p->term_len - (b - p->term_buf);
1416             if (remain > 0)
1417                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
1418                                        remain, 0);
1419             else
1420                 map = 0;
1421         }
1422         if (!map)
1423             break;
1424         i = 0;
1425         while (map && *map && **map != *CHR_SPACE)
1426         {
1427             const char *cp = *map;
1428
1429             while (i < IT_MAX_WORD && *cp)
1430                 buf[i++] = *(cp++);
1431             remain = p->term_len - (b - p->term_buf);
1432             if (remain > 0)
1433                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1434             else
1435                 map = 0;
1436         }
1437         if (!i)
1438             return;
1439
1440         if (first)
1441         {   
1442             first = 0;
1443             if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
1444             {
1445                 /* first in field marker */
1446                 extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1447                 p->seqno++;
1448             }
1449         }
1450         extract_add_string (p, buf, i);
1451         p->seqno++;
1452     }
1453 }
1454
1455 static void extract_add_complete_field (RecWord *p)
1456 {
1457     ZebraHandle zh = p->extractCtrl->handle;
1458     const char *b = p->term_buf;
1459     char buf[IT_MAX_WORD+1];
1460     const char **map = 0;
1461     int i = 0, remain = p->term_len;
1462
1463     if (remain > 0)
1464         map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
1465
1466     while (remain > 0 && i < IT_MAX_WORD)
1467     {
1468         while (map && *map && **map == *CHR_SPACE)
1469         {
1470             remain = p->term_len - (b - p->term_buf);
1471
1472             if (remain > 0)
1473             {
1474                 int first = i ? 0 : 1;  /* first position */
1475                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, first);
1476             }
1477             else
1478                 map = 0;
1479         }
1480         if (!map)
1481             break;
1482
1483         if (i && i < IT_MAX_WORD)
1484             buf[i++] = *CHR_SPACE;
1485         while (map && *map && **map != *CHR_SPACE)
1486         {
1487             const char *cp = *map;
1488
1489             if (**map == *CHR_CUT)
1490             {
1491                 i = 0;
1492             }
1493             else
1494             {
1495                 if (i >= IT_MAX_WORD)
1496                     break;
1497                 while (i < IT_MAX_WORD && *cp)
1498                     buf[i++] = *(cp++);
1499             }
1500             remain = p->term_len  - (b - p->term_buf);
1501             if (remain > 0)
1502             {
1503                 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
1504                                         remain, 0);
1505             }
1506             else
1507                 map = 0;
1508         }
1509     }
1510     if (!i)
1511         return;
1512     extract_add_string (p, buf, i);
1513 }
1514
1515 static void extract_token_add(RecWord *p)
1516 {
1517     ZebraHandle zh = p->extractCtrl->handle;
1518     WRBUF wrbuf;
1519
1520     if (log_level_extract)
1521     {
1522         yaz_log(log_level_extract, "extract_token_add "
1523                 "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
1524                 p->index_type, p->index_name, 
1525                 p->seqno, p->term_len, p->term_buf);
1526     }
1527     if ((wrbuf = zebra_replace(zh->reg->zebra_maps, p->index_type, 0,
1528                                p->term_buf, p->term_len)))
1529     {
1530         p->term_buf = wrbuf_buf(wrbuf);
1531         p->term_len = wrbuf_len(wrbuf);
1532     }
1533     if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type))
1534         extract_add_complete_field (p);
1535     else
1536         extract_add_incomplete_field(p);
1537 }
1538
1539 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1540                                       void *buf, size_t sz)
1541 {
1542     ZebraHandle zh = (ZebraHandle) p->handle;
1543
1544     xfree(zh->store_data_buf);
1545     zh->store_data_buf = 0;
1546     zh->store_data_size = 0;
1547     if (buf && sz)
1548     {
1549         zh->store_data_buf = xmalloc(sz);
1550         zh->store_data_size = sz;
1551         memcpy(zh->store_data_buf, buf, sz);
1552     }
1553 }
1554
1555 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1556 {
1557     ZebraHandle zh = (ZebraHandle) p->handle;
1558     xfree(zh->store_data_buf);
1559     zh->store_data_buf = 0;
1560     zh->store_data_size = 0;
1561     p->setStoreData = extract_set_store_data_cb;
1562 }
1563
1564 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
1565 {
1566     ZebraHandle zh = (ZebraHandle) p->handle;
1567     zebraExplain_addSchema (zh->reg->zei, oid);
1568 }
1569
1570 void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
1571                              int cmd, zebra_rec_keys_t reckeys)
1572 {
1573 #if 0
1574     yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT,
1575             cmd, sysno);
1576     extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG);
1577 #endif
1578
1579     if (zebra_rec_keys_rewind(reckeys))
1580     {
1581         zebra_sort_index_t si = zh->reg->sort_index;
1582         size_t slen;
1583         const char *str;
1584         struct it_key key_in;
1585
1586         zebra_sort_sysno(si, sysno);
1587
1588         while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1589         {
1590             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1591             
1592             zebra_sort_type(si, ord);
1593             if (cmd == 1)
1594                 zebra_sort_add(si, str, slen);
1595             else
1596                 zebra_sort_delete(si);
1597         }
1598     }
1599 }
1600
1601 /*
1602  * Local variables:
1603  * c-basic-offset: 4
1604  * indent-tabs-mode: nil
1605  * End:
1606  * vim: shiftwidth=4 tabstop=8 expandtab
1607  */
1608