The configuration, fileverboselimit, has a value of 1000. When
[idzebra-moved-to-github.git] / index / extract.c
1 /* $Id: extract.c,v 1.249 2007-02-06 09:34:56 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 static void zebra_init_log_level(void)
44 {
45     if (!log_level_initialized)
46     {
47         log_level_initialized = 1;
48
49         log_level_extract = yaz_log_module_level("extract");
50         log_level_details = yaz_log_module_level("indexdetails");
51     }
52 }
53
54 static void extract_flush_record_keys(ZebraHandle zh, zint sysno,
55                                       int cmd, zebra_rec_keys_t reckeys,
56                                       zint staticrank);
57 static void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
58                                     int cmd, zebra_rec_keys_t skp);
59 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid);
60 static void extract_token_add (RecWord *p);
61
62 static void check_log_limit(ZebraHandle zh)
63 {
64     if (zh->records_processed + zh->records_skipped == zh->m_file_verbose_limit)
65     {
66         yaz_log(YLOG_LOG, "More than %d file log entries. Omitting rest",
67                 zh->m_file_verbose_limit);
68     }
69 }
70
71 static void logRecord (ZebraHandle zh)
72 {
73     check_log_limit(zh);
74     ++zh->records_processed;
75     if (!(zh->records_processed % 1000))
76     {
77         yaz_log(YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
78                 ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT, 
79                 zh->records_processed, zh->records_inserted, 
80                 zh->records_updated, zh->records_deleted);
81     }
82 }
83
84 static void extract_add_index_string (RecWord *p, 
85                                       zinfo_index_category_t cat,
86                                       const char *str, int length);
87
88 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
89
90 static void extract_init(struct recExtractCtrl *p, RecWord *w)
91 {
92     w->seqno = 1;
93     w->index_name = "any";
94     w->index_type = 'w';
95     w->extractCtrl = p;
96     w->record_id = 0;
97     w->section_id = 0;
98     w->segment = 0;
99 }
100
101 static void searchRecordKey(ZebraHandle zh,
102                             zebra_rec_keys_t reckeys,
103                             const char *index_name,
104                             const char **ws, int ws_length)
105 {
106     int i;
107     int ch = -1;
108     zinfo_index_category_t cat = zinfo_index_category_index;
109
110     for (i = 0; i<ws_length; i++)
111         ws[i] = NULL;
112
113     if (ch < 0)
114         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, '0', index_name);
115     if (ch < 0)
116         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'p', index_name);
117     if (ch < 0)
118         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'w', index_name);
119
120     if (ch < 0)
121         return ;
122
123     if (zebra_rec_keys_rewind(reckeys))
124     {
125         zint startSeq = -1;
126         const char *str;
127         size_t slen;
128         struct it_key key;
129         zint seqno;
130         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
131         {
132             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
133
134             seqno = key.mem[key.len-1];
135             
136             if (key.mem[0] == ch)
137             {
138                 zint woff;
139                 
140                 if (startSeq == -1)
141                     startSeq = seqno;
142                 woff = seqno - startSeq;
143                 if (woff >= 0 && woff < ws_length)
144                     ws[woff] = str;
145             }
146         }
147     }
148 }
149
150 #define FILE_MATCH_BLANK "\t "
151
152 static char *get_match_from_spec(ZebraHandle zh,
153                           zebra_rec_keys_t reckeys,
154                           const char *fname, const char *spec)
155 {
156     static char dstBuf[2048];      /* static here ??? */
157     char *dst = dstBuf;
158     const char *s = spec;
159
160     while (1)
161     {
162         for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
163             ;
164         if (!*s)
165             break;
166         if (*s == '(')
167         {
168             const char *ws[32];
169             char attset_str[64], attname_str[64];
170             int i;
171             int first = 1;
172             
173             for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
174                 ;
175             for (i = 0; *s && *s != ',' && *s != ')' && 
176                      !strchr(FILE_MATCH_BLANK, *s); s++)
177                 if (i+1 < sizeof(attset_str))
178                     attset_str[i++] = *s;
179             attset_str[i] = '\0';
180             
181             for (; strchr(FILE_MATCH_BLANK, *s); s++)
182                 ;
183             if (*s != ',')
184                 strcpy(attname_str, attset_str);
185             else
186             {
187                 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
188                     ;
189                 for (i = 0; *s && *s != ')' && 
190                          !strchr(FILE_MATCH_BLANK, *s); s++)
191                     if (i+1 < sizeof(attname_str))
192                         attname_str[i++] = *s;
193                 attname_str[i] = '\0';
194             }
195
196             searchRecordKey (zh, reckeys, attname_str, ws, 32);
197
198             if (*s != ')')
199             {
200                 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
201                       spec, zh->m_group ? zh->m_group : "none");
202                 return NULL;
203             }
204             s++;
205
206             for (i = 0; i<32; i++)
207                 if (ws[i])
208                 {
209                     if (first)
210                     {
211                         *dst++ = ' ';
212                         first = 0;
213                     }
214                     strcpy (dst, ws[i]);
215                     dst += strlen(ws[i]);
216                 }
217             if (first)
218             {
219                 yaz_log (YLOG_WARN, "Record didn't contain match"
220                       " fields in (%s,%s)", attset_str, attname_str);
221                 return NULL;
222             }
223         }
224         else if (*s == '$')
225         {
226             int spec_len;
227             char special[64];
228             const char *spec_src = NULL;
229             const char *s1 = ++s;
230             while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
231                 s1++;
232
233             spec_len = s1 - s;
234             if (spec_len > sizeof(special)-1)
235                 spec_len = sizeof(special)-1;
236             memcpy (special, s, spec_len);
237             special[spec_len] = '\0';
238             s = s1;
239
240             if (!strcmp (special, "group"))
241                 spec_src = zh->m_group;
242             else if (!strcmp (special, "database"))
243                 spec_src = zh->basenames[0];
244             else if (!strcmp (special, "filename")) {
245                 spec_src = fname;
246             }
247             else if (!strcmp (special, "type"))
248                 spec_src = zh->m_record_type;
249             else 
250                 spec_src = NULL;
251             if (spec_src)
252             {
253                 strcpy (dst, spec_src);
254                 dst += strlen (spec_src);
255             }
256         }
257         else if (*s == '\"' || *s == '\'')
258         {
259             int stopMarker = *s++;
260             char tmpString[64];
261             int i = 0;
262
263             while (*s && *s != stopMarker)
264             {
265                 if (i+1 < sizeof(tmpString))
266                     tmpString[i++] = *s++;
267             }
268             if (*s)
269                 s++;
270             tmpString[i] = '\0';
271             strcpy (dst, tmpString);
272             dst += strlen (tmpString);
273         }
274         else
275         {
276             yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
277                   spec, zh->m_group ? zh->m_group : "none");
278             return NULL;
279         }
280         *dst++ = 1;
281     }
282     if (dst == dstBuf)
283     {
284         yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
285               fname, zh->m_group ? zh->m_group : "none");
286         return NULL;
287     }
288     *dst = '\0';
289     return dstBuf;
290 }
291
292 struct recordLogInfo {
293     const char *fname;
294     int recordOffset;
295     struct recordGroup *rGroup;
296 };
297
298 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
299 {
300     int i;
301     for (i = 0; i<256; i++)
302     {
303         if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
304             ctrl->seqno[i] = 1;
305         else
306             ctrl->seqno[i] = 0;
307     }
308     ctrl->flagShowRecords = !zh->m_flag_rw;
309 }
310
311 static void all_matches_add(struct recExtractCtrl *ctrl)
312 {
313     RecWord word;
314     extract_init(ctrl, &word);
315     word.index_name = "_ALLRECORDS";
316     word.index_type = 'w';
317     word.seqno = 1;
318     extract_add_index_string (&word, zinfo_index_category_alwaysmatches,
319                               "", 0);
320 }
321
322 ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname, 
323                              int deleteFlag)
324 {
325     ZEBRA_RES r = ZEBRA_OK;
326     int i, fd;
327     char gprefix[128];
328     char ext[128];
329     char ext_res[128];
330     struct file_read_info *fi = 0;
331     const char *original_record_type = 0;
332     RecType recType;
333     void *recTypeClientData;
334     struct ZebraRecStream stream, *streamp;
335
336     zebra_init_log_level();
337
338     if (!zh->m_group || !*zh->m_group)
339         *gprefix = '\0';
340     else
341         sprintf (gprefix, "%s.", zh->m_group);
342     
343     yaz_log(log_level_extract, "zebra_extract_file %s", fname);
344
345     /* determine file extension */
346     *ext = '\0';
347     for (i = strlen(fname); --i >= 0; )
348         if (fname[i] == '/')
349             break;
350         else if (fname[i] == '.')
351         {
352             strcpy (ext, fname+i+1);
353             break;
354         }
355     /* determine file type - depending on extension */
356     original_record_type = zh->m_record_type;
357     if (!zh->m_record_type)
358     {
359         sprintf (ext_res, "%srecordType.%s", gprefix, ext);
360         zh->m_record_type = res_get (zh->res, ext_res);
361     }
362     if (!zh->m_record_type)
363     {
364         check_log_limit(zh);
365         if (zh->records_processed + zh->records_skipped
366             < zh->m_file_verbose_limit)
367             yaz_log (YLOG_LOG, "? %s", fname);
368         zh->records_skipped++;
369         return 0;
370     }
371     /* determine match criteria */
372     if (!zh->m_record_id)
373     {
374         sprintf (ext_res, "%srecordId.%s", gprefix, ext);
375         zh->m_record_id = res_get (zh->res, ext_res);
376     }
377
378     if (!(recType =
379           recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
380                           &recTypeClientData)))
381     {
382         yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
383         return ZEBRA_FAIL;
384     }
385
386     switch(recType->version)
387     {
388     case 0:
389         break;
390     default:
391         yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
392     }
393     if (sysno && deleteFlag)
394     {
395         streamp = 0;
396         fi = 0;
397     }
398     else
399     {
400         char full_rep[1024];
401
402         if (zh->path_reg && !yaz_is_abspath (fname))
403         {
404             strcpy (full_rep, zh->path_reg);
405             strcat (full_rep, "/");
406             strcat (full_rep, fname);
407         }
408         else
409             strcpy (full_rep, fname);
410         
411         if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
412         {
413             yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
414             zh->m_record_type = original_record_type;
415             return ZEBRA_FAIL;
416         }
417         streamp = &stream;
418         zebra_create_stream_fd(streamp, fd, 0);
419     }
420     while(1)
421     {
422         int more = 0;
423         r = zebra_extract_record_stream(zh, streamp,
424                                         deleteFlag,
425                                         0, /* tst_mode */
426                                         zh->m_record_type,
427                                         sysno,
428                                         0, /*match_criteria */
429                                         fname,
430                                         1, /* force_update */
431                                         1, /* allow_update */
432                                         recType, recTypeClientData, &more);
433         if (!more)
434             break;
435         if (sysno)
436         {
437             break;
438         }
439     }
440     if (streamp)
441         stream.destroy(streamp);
442     zh->m_record_type = original_record_type;
443     return r;
444 }
445
446 /*
447   If sysno is provided, then it's used to identify the reocord.
448   If not, and match_criteria is provided, then sysno is guessed
449   If not, and a record is provided, then sysno is got from there
450   
451  */
452
453 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh, 
454                                       const char *buf, size_t buf_size,
455                                       int delete_flag,
456                                       int test_mode, 
457                                       const char *recordType,
458                                       zint *sysno,
459                                       const char *match_criteria,
460                                       const char *fname,
461                                       int force_update,
462                                       int allow_update)
463 {
464     struct ZebraRecStream stream;
465     ZEBRA_RES res;
466     void *clientData;
467     RecType recType = 0;
468     int more = 0;
469
470     if (recordType && *recordType)
471     {
472         yaz_log(log_level_extract,
473                 "Record type explicitly specified: %s", recordType);
474         recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
475                                   &clientData);
476     } 
477     else
478     {
479         if (!(zh->m_record_type))
480         {
481             yaz_log (YLOG_WARN, "No such record type defined");
482             return ZEBRA_FAIL;
483         }
484         yaz_log(log_level_extract, "Get record type from rgroup: %s",
485                 zh->m_record_type);
486         recType = recType_byName (zh->reg->recTypes, zh->res,
487                                   zh->m_record_type, &clientData);
488         recordType = zh->m_record_type;
489     }
490     
491     if (!recType)
492     {
493         yaz_log (YLOG_WARN, "No such record type: %s", recordType);
494         return ZEBRA_FAIL;
495     }
496
497     zebra_create_stream_mem(&stream, buf, buf_size);
498
499     res = zebra_extract_record_stream(zh, &stream,
500                                       delete_flag,
501                                       test_mode, 
502                                       recordType,
503                                       sysno,
504                                       match_criteria,
505                                       fname,
506                                       force_update,
507                                       allow_update,
508                                       recType, clientData, &more);
509     stream.destroy(&stream);
510     return res;
511 }
512
513
514 ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, 
515                                       struct ZebraRecStream *stream,
516                                       int delete_flag,
517                                       int test_mode, 
518                                       const char *recordType,
519                                       zint *sysno,
520                                       const char *match_criteria,
521                                       const char *fname,
522                                       int force_update,
523                                       int allow_update,
524                                       RecType recType,
525                                       void *recTypeClientData,
526                                       int *more)
527
528 {
529     zint sysno0 = 0;
530     RecordAttr *recordAttr;
531     struct recExtractCtrl extractCtrl;
532     int r;
533     const char *matchStr = 0;
534     Record rec;
535     off_t start_offset = 0, end_offset = 0;
536     const char *pr_fname = fname;  /* filename to print .. */
537     int show_progress = zh->records_processed + zh->records_skipped 
538         < zh->m_file_verbose_limit ? 1:0;
539
540     zebra_init_log_level();
541
542     if (!pr_fname)
543         pr_fname = "<no file>";  /* make it printable if file is omitted */
544
545     zebra_rec_keys_reset(zh->reg->keys);
546     zebra_rec_keys_reset(zh->reg->sortKeys);
547
548     if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
549     {
550         if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0], 
551                                       zh->m_explain_database))
552             return ZEBRA_FAIL;
553     }
554
555     if (stream)
556     {
557         off_t null_offset = 0;
558         extractCtrl.stream = stream;
559
560         start_offset = stream->tellf(stream);
561
562         extractCtrl.first_record = start_offset ? 0 : 1;
563         
564         stream->endf(stream, &null_offset);;
565
566         extractCtrl.init = extract_init;
567         extractCtrl.tokenAdd = extract_token_add;
568         extractCtrl.schemaAdd = extract_schema_add;
569         extractCtrl.dh = zh->reg->dh;
570         extractCtrl.handle = zh;
571         extractCtrl.match_criteria[0] = '\0';
572         extractCtrl.staticrank = 0;
573
574         init_extractCtrl(zh, &extractCtrl);
575         
576         extract_set_store_data_prepare(&extractCtrl);
577         
578         r = (*recType->extract)(recTypeClientData, &extractCtrl);
579         
580         if (r == RECCTRL_EXTRACT_EOF)
581             return ZEBRA_FAIL;
582         else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
583         {
584             /* error occured during extraction ... */
585             yaz_log (YLOG_WARN, "extract error: generic");
586             return ZEBRA_FAIL;
587         }
588         else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
589         {
590             /* error occured during extraction ... */
591             yaz_log (YLOG_WARN, "extract error: no such filter");
592             return ZEBRA_FAIL;
593         }
594         
595         all_matches_add(&extractCtrl);
596         
597         if (extractCtrl.match_criteria[0])
598             match_criteria = extractCtrl.match_criteria;
599
600
601         end_offset = stream->endf(stream, 0);
602
603         if (!end_offset)
604             end_offset = stream->tellf(stream);
605         else
606             stream->seekf(stream, end_offset);
607
608     }
609
610
611     *more = 1;
612     if (!sysno)
613     {
614         sysno = &sysno0;
615
616         if (match_criteria && *match_criteria) {
617             matchStr = match_criteria;
618         } else {
619             if (zh->m_record_id && *zh->m_record_id) {
620                 matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname, 
621                                                zh->m_record_id);
622                 if (!matchStr)
623                 {
624                     yaz_log (YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
625                              pr_fname, (zint) start_offset);
626                     return ZEBRA_FAIL;
627                 }
628             }
629         }
630         if (matchStr) 
631         {
632             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
633             char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
634                                           matchStr);
635             if (rinfo)
636             {
637                 assert(*rinfo == sizeof(*sysno));
638                 memcpy (sysno, rinfo+1, sizeof(*sysno));
639             }
640         }
641     }
642     if (zebra_rec_keys_empty(zh->reg->keys))
643     {
644         /* the extraction process returned no information - the record
645            is probably empty - unless flagShowRecords is in use */
646         if (test_mode)
647             return ZEBRA_OK;
648     }
649
650     if (! *sysno)
651     {
652         /* new record */
653         if (delete_flag)
654         {
655             yaz_log (YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
656                          pr_fname, (zint) start_offset);
657             yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
658             return ZEBRA_FAIL;
659         }
660         if (show_progress)
661             yaz_log (YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
662                      (zint) start_offset);
663         rec = rec_new (zh->reg->records);
664
665         *sysno = rec->sysno;
666
667         recordAttr = rec_init_attr (zh->reg->zei, rec);
668         if (extractCtrl.staticrank < 0)
669         {
670             yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
671             extractCtrl.staticrank = 0;
672         }
673         recordAttr->staticrank = extractCtrl.staticrank;
674
675         if (matchStr)
676         {
677             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
678             dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
679                             sizeof(*sysno), sysno);
680         }
681
682         extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
683         extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
684                                   recordAttr->staticrank);
685         zh->records_inserted++;
686     } 
687     else
688     {
689         /* record already exists */
690         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
691         zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
692         if (!allow_update)
693         {
694             yaz_log (YLOG_LOG, "skipped %s %s " ZINT_FORMAT, 
695                          recordType, pr_fname, (zint) start_offset);
696             logRecord(zh);
697             return ZEBRA_FAIL;
698         }
699
700         rec = rec_get (zh->reg->records, *sysno);
701         assert (rec);
702         
703         recordAttr = rec_init_attr (zh->reg->zei, rec);
704
705         /* decrease total size */
706         zebraExplain_recordBytesIncrement (zh->reg->zei,
707                                            - recordAttr->recordSize);
708
709         zebra_rec_keys_set_buf(delkeys,
710                                rec->info[recInfo_delKeys],
711                                rec->size[recInfo_delKeys],
712                                0);
713         zebra_rec_keys_set_buf(sortKeys,
714                                rec->info[recInfo_sortKeys],
715                                rec->size[recInfo_sortKeys],
716                                0);
717
718         extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
719         extract_flush_record_keys(zh, *sysno, 0, delkeys,
720                                   recordAttr->staticrank);
721         if (delete_flag)
722         {
723             /* record going to be deleted */
724             if (zebra_rec_keys_empty(delkeys))
725             {
726                 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
727                         pr_fname, (zint) start_offset);
728                 yaz_log(YLOG_WARN, "cannot delete file above, "
729                         "storeKeys false (3)");
730             }
731             else
732             {
733                 if (show_progress)
734                     yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
735                             pr_fname, (zint) start_offset);
736                 zh->records_deleted++;
737                 if (matchStr)
738                 {
739                     int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
740                     dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
741                 }
742                 rec_del (zh->reg->records, &rec);
743             }
744             rec_free(&rec);
745             logRecord(zh);
746             return ZEBRA_OK;
747         }
748         else
749         {
750             if (show_progress)
751                 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
752                         pr_fname, (zint) start_offset);
753             recordAttr->staticrank = extractCtrl.staticrank;
754             extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
755             extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys, 
756                                       recordAttr->staticrank);
757             zh->records_updated++;
758         }
759         zebra_rec_keys_close(delkeys);
760         zebra_rec_keys_close(sortKeys);
761     }
762     /* update file type */
763     xfree (rec->info[recInfo_fileType]);
764     rec->info[recInfo_fileType] =
765         rec_strdup (recordType, &rec->size[recInfo_fileType]);
766
767     /* update filename */
768     xfree (rec->info[recInfo_filename]);
769     rec->info[recInfo_filename] =
770         rec_strdup (fname, &rec->size[recInfo_filename]);
771
772     /* update delete keys */
773     xfree (rec->info[recInfo_delKeys]);
774     if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
775     {
776         zebra_rec_keys_get_buf(zh->reg->keys,
777                                &rec->info[recInfo_delKeys],
778                                &rec->size[recInfo_delKeys]);
779     }
780     else
781     {
782         rec->info[recInfo_delKeys] = NULL;
783         rec->size[recInfo_delKeys] = 0;
784     }
785     /* update sort keys */
786     xfree (rec->info[recInfo_sortKeys]);
787
788     zebra_rec_keys_get_buf(zh->reg->sortKeys,
789                            &rec->info[recInfo_sortKeys],
790                            &rec->size[recInfo_sortKeys]);
791
792     if (stream)
793     {
794         recordAttr->recordSize = end_offset - start_offset;
795         zebraExplain_recordBytesIncrement(zh->reg->zei,
796                                           recordAttr->recordSize);
797     }
798
799     /* set run-number for this record */
800     recordAttr->runNumber =
801         zebraExplain_runNumberIncrement (zh->reg->zei, 0);
802
803     /* update store data */
804     xfree (rec->info[recInfo_storeData]);
805
806     /* update store data */
807     if (zh->store_data_buf)
808     {
809         rec->size[recInfo_storeData] = zh->store_data_size;
810         rec->info[recInfo_storeData] = zh->store_data_buf;
811         zh->store_data_buf = 0;
812         recordAttr->recordSize = zh->store_data_size;
813     }
814     else if (zh->m_store_data)
815     {
816         off_t cur_offset = stream->tellf(stream);
817
818         rec->size[recInfo_storeData] = recordAttr->recordSize;
819         rec->info[recInfo_storeData] = (char *)
820             xmalloc (recordAttr->recordSize);
821         stream->seekf(stream, start_offset);
822         stream->readf(stream, rec->info[recInfo_storeData],
823                       recordAttr->recordSize);
824         stream->seekf(stream, cur_offset);
825     }
826     else
827     {
828         rec->info[recInfo_storeData] = NULL;
829         rec->size[recInfo_storeData] = 0;
830     }
831     /* update database name */
832     xfree (rec->info[recInfo_databaseName]);
833     rec->info[recInfo_databaseName] =
834         rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]); 
835
836     /* update offset */
837     recordAttr->recordOffset = start_offset;
838     
839     /* commit this record */
840     rec_put (zh->reg->records, &rec);
841     logRecord(zh);
842     return ZEBRA_OK;
843 }
844
845 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
846 {
847     ZebraHandle zh = (ZebraHandle) handle;
848     struct recExtractCtrl extractCtrl;
849
850     if (zebraExplain_curDatabase (zh->reg->zei,
851                                   rec->info[recInfo_databaseName]))
852     {
853         abort();
854         if (zebraExplain_newDatabase (zh->reg->zei,
855                                       rec->info[recInfo_databaseName], 0))
856             abort ();
857     }
858
859     zebra_rec_keys_reset(zh->reg->keys);
860     zebra_rec_keys_reset(zh->reg->sortKeys);
861
862     extractCtrl.init = extract_init;
863     extractCtrl.tokenAdd = extract_token_add;
864     extractCtrl.schemaAdd = extract_schema_add;
865     extractCtrl.dh = zh->reg->dh;
866
867     init_extractCtrl(zh, &extractCtrl);
868
869     extractCtrl.flagShowRecords = 0;
870     extractCtrl.match_criteria[0] = '\0';
871     extractCtrl.staticrank = 0;
872     extractCtrl.handle = handle;
873     extractCtrl.first_record = 1;
874     
875     extract_set_store_data_prepare(&extractCtrl);
876
877     if (n)
878         grs_extract_tree(&extractCtrl, n);
879
880     if (rec->size[recInfo_delKeys])
881     {
882         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
883         
884         zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
885
886         zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
887                                rec->size[recInfo_delKeys],
888                                0);
889         extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
890         zebra_rec_keys_close(delkeys);
891
892         zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
893                                rec->size[recInfo_sortKeys],
894                                0);
895
896         extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
897         zebra_rec_keys_close(sortkeys);
898     }
899     extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
900     extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
901     
902     xfree (rec->info[recInfo_delKeys]);
903     zebra_rec_keys_get_buf(zh->reg->keys,
904                            &rec->info[recInfo_delKeys], 
905                            &rec->size[recInfo_delKeys]);
906
907     xfree (rec->info[recInfo_sortKeys]);
908     zebra_rec_keys_get_buf(zh->reg->sortKeys,
909                            &rec->info[recInfo_sortKeys],
910                            &rec->size[recInfo_sortKeys]);
911     return ZEBRA_OK;
912 }
913
914 void extract_rec_keys_log(ZebraHandle zh, int is_insert,
915                           zebra_rec_keys_t reckeys,
916                           int level)
917 {
918     if (zebra_rec_keys_rewind(reckeys))
919     {
920         size_t slen;
921         const char *str;
922         struct it_key key;
923         NMEM nmem = nmem_create();
924
925         while(zebra_rec_keys_read(reckeys, &str, &slen, &key))
926         {
927             char keystr[200]; /* room for zints to print */
928             char *dst_term = 0;
929             int ord = CAST_ZINT_TO_INT(key.mem[0]);
930             int index_type, i;
931             const char *string_index;
932             
933             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
934                                     0/* db */, &string_index);
935             assert(index_type);
936             zebra_term_untrans_iconv(zh, nmem, index_type,
937                                      &dst_term, str);
938             *keystr = '\0';
939             for (i = 0; i<key.len; i++)
940             {
941                 sprintf(keystr + strlen(keystr), ZINT_FORMAT " ", key.mem[i]);
942             }
943
944             if (*str < CHR_BASE_CHAR)
945             {
946                 int i;
947                 char dst_buf[200]; /* room for special chars */
948
949                 strcpy(dst_buf , "?");
950
951                 if (!strcmp(str, ""))
952                     strcpy(dst_buf, "alwaysmatches");
953                 if (!strcmp(str, FIRST_IN_FIELD_STR))
954                     strcpy(dst_buf, "firstinfield");
955                 else if (!strcmp(str, CHR_UNKNOWN))
956                     strcpy(dst_buf, "unknown");
957                 else if (!strcmp(str, CHR_SPACE))
958                     strcpy(dst_buf, "space");
959                 
960                 for (i = 0; i<slen; i++)
961                 {
962                     sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
963                 }
964                 yaz_log(level, "%s%c %s %s", keystr, index_type,
965                         string_index, dst_buf);
966                 
967             }
968             else
969                 yaz_log(level, "%s%c %s \"%s\"", keystr, index_type,
970                         string_index, dst_term);
971
972             nmem_reset(nmem);
973         }
974         nmem_destroy(nmem);
975     }
976 }
977
978 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
979                              zebra_rec_keys_t reckeys)
980 {
981     ZebraExplainInfo zei = zh->reg->zei;
982     struct ord_stat {
983         int no;
984         int ord;
985         struct ord_stat *next;
986     };
987
988     if (zebra_rec_keys_rewind(reckeys))
989     {
990         struct ord_stat *ord_list = 0;
991         struct ord_stat *p;
992         size_t slen;
993         const char *str;
994         struct it_key key_in;
995         while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
996         {
997             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
998
999             for (p = ord_list; p ; p = p->next)
1000                 if (p->ord == ord)
1001                 {
1002                     p->no++;
1003                     break;
1004                 }
1005             if (!p)
1006             {
1007                 p = xmalloc(sizeof(*p));
1008                 p->no = 1;
1009                 p->ord = ord;
1010                 p->next = ord_list;
1011                 ord_list = p;
1012             }
1013         }
1014
1015         p = ord_list;
1016         while (p)
1017         {
1018             struct ord_stat *p1 = p;
1019
1020             if (is_insert)
1021                 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
1022             else
1023                 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
1024             p = p->next;
1025             xfree(p1);
1026         }
1027     }
1028 }
1029
1030 void extract_flush_record_keys(ZebraHandle zh, zint sysno, int cmd,
1031                                zebra_rec_keys_t reckeys,
1032                                zint staticrank)
1033 {
1034     ZebraExplainInfo zei = zh->reg->zei;
1035
1036     extract_rec_keys_adjust(zh, cmd, reckeys);
1037
1038     if (log_level_details)
1039     {
1040         yaz_log(log_level_details, "Keys for record " ZINT_FORMAT " %s",
1041                 sysno, cmd ? "insert" : "delete");
1042         extract_rec_keys_log(zh, cmd, reckeys, log_level_details);
1043     }
1044
1045     if (!zh->reg->key_block)
1046     {
1047         int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1048         const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", ".");
1049         int use_threads = atoi(res_get_def (zh->res, "threads", "1"));
1050         zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1051     }
1052     zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1053
1054     if (zebra_rec_keys_rewind(reckeys))
1055     {
1056         size_t slen;
1057         const char *str;
1058         struct it_key key_in;
1059         while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1060         {
1061             key_block_write(zh->reg->key_block, sysno, 
1062                             &key_in, cmd, str, slen,
1063                             staticrank, zh->m_staticrank);
1064         }
1065     }
1066 }
1067
1068
1069 ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
1070                                      zebra_rec_keys_t reckeys,
1071                                      zebra_snippets *snippets)
1072 {
1073     NMEM nmem = nmem_create();
1074     if (zebra_rec_keys_rewind(reckeys)) 
1075     {
1076         const char *str;
1077         size_t slen;
1078         struct it_key key;
1079         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1080         {
1081             char *dst_term = 0;
1082             int ord;
1083             zint seqno;
1084             int index_type;
1085
1086             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1087             seqno = key.mem[key.len-1];
1088             ord = CAST_ZINT_TO_INT(key.mem[0]);
1089             
1090             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1091                                     0/* db */, 0 /* string_index */);
1092             assert(index_type);
1093             zebra_term_untrans_iconv(zh, nmem, index_type,
1094                                      &dst_term, str);
1095             zebra_snippets_append(snippets, seqno, ord, dst_term);
1096             nmem_reset(nmem);
1097         }
1098     }
1099     nmem_destroy(nmem);
1100     return ZEBRA_OK;
1101 }
1102
1103 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1104 {
1105     yaz_log(YLOG_LOG, "print_rec_keys");
1106     if (zebra_rec_keys_rewind(reckeys))
1107     {
1108         const char *str;
1109         size_t slen;
1110         struct it_key key;
1111         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1112         {
1113             char dst_buf[IT_MAX_WORD];
1114             zint seqno;
1115             int index_type;
1116             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1117             const char *db = 0;
1118             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1119
1120             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1121             
1122             seqno = key.mem[key.len-1];
1123             
1124             zebra_term_untrans(zh, index_type, dst_buf, str);
1125             
1126             yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT 
1127                     " term=%s", ord, seqno, dst_buf); 
1128         }
1129     }
1130 }
1131
1132 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1133                                      const char *str, int length)
1134 {
1135     struct it_key key;
1136     ZebraHandle zh = p->extractCtrl->handle;
1137     ZebraExplainInfo zei = zh->reg->zei;
1138     int ch, i;
1139
1140     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1141     if (ch < 0)
1142         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1143
1144     i = 0;
1145     key.mem[i++] = ch;
1146     key.mem[i++] = p->record_id;
1147     key.mem[i++] = p->section_id;
1148
1149     if (zh->m_segment_indexing)
1150         key.mem[i++] = p->segment;
1151     key.mem[i++] = p->seqno;
1152     key.len = i;
1153
1154     zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1155 }
1156
1157 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1158 {
1159     struct it_key key;
1160     ZebraHandle zh = p->extractCtrl->handle;
1161     ZebraExplainInfo zei = zh->reg->zei;
1162     int ch;
1163     zinfo_index_category_t cat = zinfo_index_category_sort;
1164
1165     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1166     if (ch < 0)
1167         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1168     key.len = 2;
1169     key.mem[0] = ch;
1170     key.mem[1] = p->record_id;
1171
1172     zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1173 }
1174
1175 static void extract_add_staticrank_string(RecWord *p,
1176                                           const char *str, int length)
1177 {
1178     char valz[40];
1179     struct recExtractCtrl *ctrl = p->extractCtrl;
1180
1181     if (length > sizeof(valz)-1)
1182         length = sizeof(valz)-1;
1183
1184     memcpy(valz, str, length);
1185     valz[length] = '\0';
1186     ctrl->staticrank = atozint(valz);
1187 }
1188
1189 static void extract_add_string(RecWord *p, const char *string, int length)
1190 {
1191     ZebraHandle zh = p->extractCtrl->handle;
1192     assert (length > 0);
1193
1194     if (!p->index_name)
1195         return;
1196
1197     if (zebra_maps_is_index(zh->reg->zebra_maps, p->index_type))
1198     {
1199         extract_add_index_string(p, zinfo_index_category_index,
1200                                  string, length);
1201         if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type))
1202         {
1203             RecWord word;
1204             memcpy(&word, p, sizeof(word));
1205
1206             word.seqno = 1;
1207             extract_add_index_string(
1208                 &word, zinfo_index_category_alwaysmatches, "", 0);
1209         }
1210     }
1211     else if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
1212     {
1213         extract_add_sort_string(p, string, length);
1214     }
1215     else if (zebra_maps_is_staticrank(zh->reg->zebra_maps, p->index_type))
1216     {
1217         extract_add_staticrank_string(p, string, length);
1218     }
1219 }
1220
1221 static void extract_add_incomplete_field(RecWord *p)
1222 {
1223     ZebraHandle zh = p->extractCtrl->handle;
1224     const char *b = p->term_buf;
1225     int remain = p->term_len;
1226     int first = 1;
1227     const char **map = 0;
1228     
1229     if (remain > 0)
1230         map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1231
1232     while (map)
1233     {
1234         char buf[IT_MAX_WORD+1];
1235         int i, remain;
1236
1237         /* Skip spaces */
1238         while (map && *map && **map == *CHR_SPACE)
1239         {
1240             remain = p->term_len - (b - p->term_buf);
1241             if (remain > 0)
1242                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
1243                                        remain, 0);
1244             else
1245                 map = 0;
1246         }
1247         if (!map)
1248             break;
1249         i = 0;
1250         while (map && *map && **map != *CHR_SPACE)
1251         {
1252             const char *cp = *map;
1253
1254             while (i < IT_MAX_WORD && *cp)
1255                 buf[i++] = *(cp++);
1256             remain = p->term_len - (b - p->term_buf);
1257             if (remain > 0)
1258                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1259             else
1260                 map = 0;
1261         }
1262         if (!i)
1263             return;
1264
1265         if (first)
1266         {   
1267             first = 0;
1268             if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
1269             {
1270                 /* first in field marker */
1271                 extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1272                 p->seqno++;
1273             }
1274         }
1275         extract_add_string (p, buf, i);
1276         p->seqno++;
1277     }
1278 }
1279
1280 static void extract_add_complete_field (RecWord *p)
1281 {
1282     ZebraHandle zh = p->extractCtrl->handle;
1283     const char *b = p->term_buf;
1284     char buf[IT_MAX_WORD+1];
1285     const char **map = 0;
1286     int i = 0, remain = p->term_len;
1287
1288     if (remain > 0)
1289         map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
1290
1291     while (remain > 0 && i < IT_MAX_WORD)
1292     {
1293         while (map && *map && **map == *CHR_SPACE)
1294         {
1295             remain = p->term_len - (b - p->term_buf);
1296
1297             if (remain > 0)
1298             {
1299                 int first = i ? 0 : 1;  /* first position */
1300                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, first);
1301             }
1302             else
1303                 map = 0;
1304         }
1305         if (!map)
1306             break;
1307
1308         if (i && i < IT_MAX_WORD)
1309             buf[i++] = *CHR_SPACE;
1310         while (map && *map && **map != *CHR_SPACE)
1311         {
1312             const char *cp = *map;
1313
1314             if (**map == *CHR_CUT)
1315             {
1316                 i = 0;
1317             }
1318             else
1319             {
1320                 if (i >= IT_MAX_WORD)
1321                     break;
1322                 while (i < IT_MAX_WORD && *cp)
1323                     buf[i++] = *(cp++);
1324             }
1325             remain = p->term_len  - (b - p->term_buf);
1326             if (remain > 0)
1327             {
1328                 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
1329                                         remain, 0);
1330             }
1331             else
1332                 map = 0;
1333         }
1334     }
1335     if (!i)
1336         return;
1337     extract_add_string (p, buf, i);
1338 }
1339
1340 static void extract_token_add(RecWord *p)
1341 {
1342     ZebraHandle zh = p->extractCtrl->handle;
1343     WRBUF wrbuf;
1344
1345     if (log_level_extract)
1346     {
1347         yaz_log(log_level_extract, "extract_token_add "
1348                 "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
1349                 p->index_type, p->index_name, 
1350                 p->seqno, p->term_len, p->term_buf);
1351     }
1352     if ((wrbuf = zebra_replace(zh->reg->zebra_maps, p->index_type, 0,
1353                                p->term_buf, p->term_len)))
1354     {
1355         p->term_buf = wrbuf_buf(wrbuf);
1356         p->term_len = wrbuf_len(wrbuf);
1357     }
1358     if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type))
1359         extract_add_complete_field (p);
1360     else
1361         extract_add_incomplete_field(p);
1362 }
1363
1364 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1365                                       void *buf, size_t sz)
1366 {
1367     ZebraHandle zh = (ZebraHandle) p->handle;
1368
1369     xfree(zh->store_data_buf);
1370     zh->store_data_buf = 0;
1371     zh->store_data_size = 0;
1372     if (buf && sz)
1373     {
1374         zh->store_data_buf = xmalloc(sz);
1375         zh->store_data_size = sz;
1376         memcpy(zh->store_data_buf, buf, sz);
1377     }
1378 }
1379
1380 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1381 {
1382     ZebraHandle zh = (ZebraHandle) p->handle;
1383     xfree(zh->store_data_buf);
1384     zh->store_data_buf = 0;
1385     zh->store_data_size = 0;
1386     p->setStoreData = extract_set_store_data_cb;
1387 }
1388
1389 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
1390 {
1391     ZebraHandle zh = (ZebraHandle) p->handle;
1392     zebraExplain_addSchema (zh->reg->zei, oid);
1393 }
1394
1395 void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
1396                              int cmd, zebra_rec_keys_t reckeys)
1397 {
1398 #if 0
1399     yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT,
1400             cmd, sysno);
1401     extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG);
1402 #endif
1403
1404     if (zebra_rec_keys_rewind(reckeys))
1405     {
1406         zebra_sort_index_t si = zh->reg->sort_index;
1407         size_t slen;
1408         const char *str;
1409         struct it_key key_in;
1410
1411         zebra_sort_sysno(si, sysno);
1412
1413         while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1414         {
1415             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1416             
1417             zebra_sort_type(si, ord);
1418             if (cmd == 1)
1419                 zebra_sort_add(si, str, slen);
1420             else
1421                 zebra_sort_delete(si);
1422         }
1423     }
1424 }
1425
1426 /*
1427  * Local variables:
1428  * c-basic-offset: 4
1429  * indent-tabs-mode: nil
1430  * End:
1431  * vim: shiftwidth=4 tabstop=8 expandtab
1432  */
1433