Implemented sorting via the ISAMB system. To enable, use sortindex:b
[idzebra-moved-to-github.git] / index / extract.c
1 /* $Id: extract.c,v 1.246 2006-12-18 23:40:07 adam Exp $
2    Copyright (C) 1995-2006
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 logRecord (ZebraHandle zh)
63 {
64     ++zh->records_processed;
65     if (!(zh->records_processed % 1000))
66     {
67         yaz_log(YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
68                 ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT, 
69                 zh->records_processed, zh->records_inserted, 
70                 zh->records_updated, zh->records_deleted);
71     }
72 }
73
74 static void extract_add_index_string (RecWord *p, 
75                                       zinfo_index_category_t cat,
76                                       const char *str, int length);
77
78 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
79
80 static void extract_init(struct recExtractCtrl *p, RecWord *w)
81 {
82     w->seqno = 1;
83     w->index_name = "any";
84     w->index_type = 'w';
85     w->extractCtrl = p;
86     w->record_id = 0;
87     w->section_id = 0;
88     w->segment = 0;
89 }
90
91 static void searchRecordKey(ZebraHandle zh,
92                             zebra_rec_keys_t reckeys,
93                             const char *index_name,
94                             const char **ws, int ws_length)
95 {
96     int i;
97     int ch = -1;
98     zinfo_index_category_t cat = zinfo_index_category_index;
99
100     for (i = 0; i<ws_length; i++)
101         ws[i] = NULL;
102
103     if (ch < 0)
104         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, '0', index_name);
105     if (ch < 0)
106         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'p', index_name);
107     if (ch < 0)
108         ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'w', index_name);
109
110     if (ch < 0)
111         return ;
112
113     if (zebra_rec_keys_rewind(reckeys))
114     {
115         zint startSeq = -1;
116         const char *str;
117         size_t slen;
118         struct it_key key;
119         zint seqno;
120         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
121         {
122             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
123
124             seqno = key.mem[key.len-1];
125             
126             if (key.mem[0] == ch)
127             {
128                 zint woff;
129                 
130                 if (startSeq == -1)
131                     startSeq = seqno;
132                 woff = seqno - startSeq;
133                 if (woff >= 0 && woff < ws_length)
134                     ws[woff] = str;
135             }
136         }
137     }
138 }
139
140 #define FILE_MATCH_BLANK "\t "
141
142 static char *get_match_from_spec(ZebraHandle zh,
143                           zebra_rec_keys_t reckeys,
144                           const char *fname, const char *spec)
145 {
146     static char dstBuf[2048];      /* static here ??? */
147     char *dst = dstBuf;
148     const char *s = spec;
149
150     while (1)
151     {
152         for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
153             ;
154         if (!*s)
155             break;
156         if (*s == '(')
157         {
158             const char *ws[32];
159             char attset_str[64], attname_str[64];
160             int i;
161             int first = 1;
162             
163             for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
164                 ;
165             for (i = 0; *s && *s != ',' && *s != ')' && 
166                      !strchr(FILE_MATCH_BLANK, *s); s++)
167                 if (i+1 < sizeof(attset_str))
168                     attset_str[i++] = *s;
169             attset_str[i] = '\0';
170             
171             for (; strchr(FILE_MATCH_BLANK, *s); s++)
172                 ;
173             if (*s != ',')
174                 strcpy(attname_str, attset_str);
175             else
176             {
177                 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
178                     ;
179                 for (i = 0; *s && *s != ')' && 
180                          !strchr(FILE_MATCH_BLANK, *s); s++)
181                     if (i+1 < sizeof(attname_str))
182                         attname_str[i++] = *s;
183                 attname_str[i] = '\0';
184             }
185
186             searchRecordKey (zh, reckeys, attname_str, ws, 32);
187
188             if (*s != ')')
189             {
190                 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
191                       spec, zh->m_group ? zh->m_group : "none");
192                 return NULL;
193             }
194             s++;
195
196             for (i = 0; i<32; i++)
197                 if (ws[i])
198                 {
199                     if (first)
200                     {
201                         *dst++ = ' ';
202                         first = 0;
203                     }
204                     strcpy (dst, ws[i]);
205                     dst += strlen(ws[i]);
206                 }
207             if (first)
208             {
209                 yaz_log (YLOG_WARN, "Record didn't contain match"
210                       " fields in (%s,%s)", attset_str, attname_str);
211                 return NULL;
212             }
213         }
214         else if (*s == '$')
215         {
216             int spec_len;
217             char special[64];
218             const char *spec_src = NULL;
219             const char *s1 = ++s;
220             while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
221                 s1++;
222
223             spec_len = s1 - s;
224             if (spec_len > sizeof(special)-1)
225                 spec_len = sizeof(special)-1;
226             memcpy (special, s, spec_len);
227             special[spec_len] = '\0';
228             s = s1;
229
230             if (!strcmp (special, "group"))
231                 spec_src = zh->m_group;
232             else if (!strcmp (special, "database"))
233                 spec_src = zh->basenames[0];
234             else if (!strcmp (special, "filename")) {
235                 spec_src = fname;
236             }
237             else if (!strcmp (special, "type"))
238                 spec_src = zh->m_record_type;
239             else 
240                 spec_src = NULL;
241             if (spec_src)
242             {
243                 strcpy (dst, spec_src);
244                 dst += strlen (spec_src);
245             }
246         }
247         else if (*s == '\"' || *s == '\'')
248         {
249             int stopMarker = *s++;
250             char tmpString[64];
251             int i = 0;
252
253             while (*s && *s != stopMarker)
254             {
255                 if (i+1 < sizeof(tmpString))
256                     tmpString[i++] = *s++;
257             }
258             if (*s)
259                 s++;
260             tmpString[i] = '\0';
261             strcpy (dst, tmpString);
262             dst += strlen (tmpString);
263         }
264         else
265         {
266             yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
267                   spec, zh->m_group ? zh->m_group : "none");
268             return NULL;
269         }
270         *dst++ = 1;
271     }
272     if (dst == dstBuf)
273     {
274         yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
275               fname, zh->m_group ? zh->m_group : "none");
276         return NULL;
277     }
278     *dst = '\0';
279     return dstBuf;
280 }
281
282 struct recordLogInfo {
283     const char *fname;
284     int recordOffset;
285     struct recordGroup *rGroup;
286 };
287
288 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
289 {
290     int i;
291     for (i = 0; i<256; i++)
292     {
293         if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
294             ctrl->seqno[i] = 1;
295         else
296             ctrl->seqno[i] = 0;
297     }
298     ctrl->flagShowRecords = !zh->m_flag_rw;
299 }
300
301 static void all_matches_add(struct recExtractCtrl *ctrl)
302 {
303     RecWord word;
304     extract_init(ctrl, &word);
305     word.index_name = "_ALLRECORDS";
306     word.index_type = 'w';
307     word.seqno = 1;
308     extract_add_index_string (&word, zinfo_index_category_alwaysmatches,
309                               "", 0);
310 }
311
312 ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname, 
313                              int deleteFlag)
314 {
315     ZEBRA_RES r = ZEBRA_OK;
316     int i, fd;
317     char gprefix[128];
318     char ext[128];
319     char ext_res[128];
320     struct file_read_info *fi = 0;
321     const char *original_record_type = 0;
322     RecType recType;
323     void *recTypeClientData;
324     struct ZebraRecStream stream, *streamp;
325
326     zebra_init_log_level();
327
328     if (!zh->m_group || !*zh->m_group)
329         *gprefix = '\0';
330     else
331         sprintf (gprefix, "%s.", zh->m_group);
332     
333     yaz_log(log_level_extract, "zebra_extract_file %s", fname);
334
335     /* determine file extension */
336     *ext = '\0';
337     for (i = strlen(fname); --i >= 0; )
338         if (fname[i] == '/')
339             break;
340         else if (fname[i] == '.')
341         {
342             strcpy (ext, fname+i+1);
343             break;
344         }
345     /* determine file type - depending on extension */
346     original_record_type = zh->m_record_type;
347     if (!zh->m_record_type)
348     {
349         sprintf (ext_res, "%srecordType.%s", gprefix, ext);
350         zh->m_record_type = res_get (zh->res, ext_res);
351     }
352     if (!zh->m_record_type)
353     {
354         if (zh->records_processed < zh->m_file_verbose_limit)
355             yaz_log (YLOG_LOG, "? %s", fname);
356         return 0;
357     }
358     /* determine match criteria */
359     if (!zh->m_record_id)
360     {
361         sprintf (ext_res, "%srecordId.%s", gprefix, ext);
362         zh->m_record_id = res_get (zh->res, ext_res);
363     }
364
365     if (!(recType =
366           recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
367                           &recTypeClientData)))
368     {
369         yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
370         return ZEBRA_FAIL;
371     }
372
373     switch(recType->version)
374     {
375     case 0:
376         break;
377     default:
378         yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
379     }
380     if (sysno && deleteFlag)
381     {
382         streamp = 0;
383         fi = 0;
384     }
385     else
386     {
387         char full_rep[1024];
388
389         if (zh->path_reg && !yaz_is_abspath (fname))
390         {
391             strcpy (full_rep, zh->path_reg);
392             strcat (full_rep, "/");
393             strcat (full_rep, fname);
394         }
395         else
396             strcpy (full_rep, fname);
397         
398         if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
399         {
400             yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
401             zh->m_record_type = original_record_type;
402             return ZEBRA_FAIL;
403         }
404         streamp = &stream;
405         zebra_create_stream_fd(streamp, fd, 0);
406     }
407     while(1)
408     {
409         int more = 0;
410         r = zebra_extract_record_stream(zh, streamp,
411                                         deleteFlag,
412                                         0, /* tst_mode */
413                                         zh->m_record_type,
414                                         sysno,
415                                         0, /*match_criteria */
416                                         fname,
417                                         1, /* force_update */
418                                         1, /* allow_update */
419                                         recType, recTypeClientData, &more);
420         if (!more)
421             break;
422         if (sysno)
423         {
424             break;
425         }
426     }
427     if (streamp)
428         stream.destroy(streamp);
429     zh->m_record_type = original_record_type;
430     return r;
431 }
432
433 /*
434   If sysno is provided, then it's used to identify the reocord.
435   If not, and match_criteria is provided, then sysno is guessed
436   If not, and a record is provided, then sysno is got from there
437   
438  */
439
440 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh, 
441                                       const char *buf, size_t buf_size,
442                                       int delete_flag,
443                                       int test_mode, 
444                                       const char *recordType,
445                                       zint *sysno,
446                                       const char *match_criteria,
447                                       const char *fname,
448                                       int force_update,
449                                       int allow_update)
450 {
451     struct ZebraRecStream stream;
452     ZEBRA_RES res;
453     void *clientData;
454     RecType recType = 0;
455     int more = 0;
456
457     if (recordType && *recordType)
458     {
459         yaz_log(log_level_extract,
460                 "Record type explicitly specified: %s", recordType);
461         recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
462                                   &clientData);
463     } 
464     else
465     {
466         if (!(zh->m_record_type))
467         {
468             yaz_log (YLOG_WARN, "No such record type defined");
469             return ZEBRA_FAIL;
470         }
471         yaz_log(log_level_extract, "Get record type from rgroup: %s",
472                 zh->m_record_type);
473         recType = recType_byName (zh->reg->recTypes, zh->res,
474                                   zh->m_record_type, &clientData);
475         recordType = zh->m_record_type;
476     }
477     
478     if (!recType)
479     {
480         yaz_log (YLOG_WARN, "No such record type: %s", recordType);
481         return ZEBRA_FAIL;
482     }
483
484     zebra_create_stream_mem(&stream, buf, buf_size);
485
486     res = zebra_extract_record_stream(zh, &stream,
487                                       delete_flag,
488                                       test_mode, 
489                                       recordType,
490                                       sysno,
491                                       match_criteria,
492                                       fname,
493                                       force_update,
494                                       allow_update,
495                                       recType, clientData, &more);
496     stream.destroy(&stream);
497     return res;
498 }
499
500
501 ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, 
502                                       struct ZebraRecStream *stream,
503                                       int delete_flag,
504                                       int test_mode, 
505                                       const char *recordType,
506                                       zint *sysno,
507                                       const char *match_criteria,
508                                       const char *fname,
509                                       int force_update,
510                                       int allow_update,
511                                       RecType recType,
512                                       void *recTypeClientData,
513                                       int *more)
514
515 {
516     zint sysno0 = 0;
517     RecordAttr *recordAttr;
518     struct recExtractCtrl extractCtrl;
519     int r;
520     const char *matchStr = 0;
521     Record rec;
522     off_t start_offset = 0, end_offset = 0;
523     const char *pr_fname = fname;  /* filename to print .. */
524     int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
525
526     zebra_init_log_level();
527
528     if (!pr_fname)
529         pr_fname = "<no file>";  /* make it printable if file is omitted */
530
531     zebra_rec_keys_reset(zh->reg->keys);
532     zebra_rec_keys_reset(zh->reg->sortKeys);
533
534     if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
535     {
536         if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0], 
537                                       zh->m_explain_database))
538             return ZEBRA_FAIL;
539     }
540
541     if (stream)
542     {
543         off_t null_offset = 0;
544         extractCtrl.stream = stream;
545
546         start_offset = stream->tellf(stream);
547
548         extractCtrl.first_record = start_offset ? 0 : 1;
549         
550         stream->endf(stream, &null_offset);;
551
552         extractCtrl.init = extract_init;
553         extractCtrl.tokenAdd = extract_token_add;
554         extractCtrl.schemaAdd = extract_schema_add;
555         extractCtrl.dh = zh->reg->dh;
556         extractCtrl.handle = zh;
557         extractCtrl.match_criteria[0] = '\0';
558         extractCtrl.staticrank = 0;
559
560         init_extractCtrl(zh, &extractCtrl);
561         
562         extract_set_store_data_prepare(&extractCtrl);
563         
564         r = (*recType->extract)(recTypeClientData, &extractCtrl);
565         
566         if (r == RECCTRL_EXTRACT_EOF)
567             return ZEBRA_FAIL;
568         else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
569         {
570             /* error occured during extraction ... */
571             yaz_log (YLOG_WARN, "extract error: generic");
572             return ZEBRA_FAIL;
573         }
574         else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
575         {
576             /* error occured during extraction ... */
577             yaz_log (YLOG_WARN, "extract error: no such filter");
578             return ZEBRA_FAIL;
579         }
580         
581         all_matches_add(&extractCtrl);
582         
583         if (extractCtrl.match_criteria[0])
584             match_criteria = extractCtrl.match_criteria;
585
586
587         end_offset = stream->endf(stream, 0);
588
589         if (!end_offset)
590             end_offset = stream->tellf(stream);
591         else
592             stream->seekf(stream, end_offset);
593
594     }
595
596
597     *more = 1;
598     if (!sysno)
599     {
600         sysno = &sysno0;
601
602         if (match_criteria && *match_criteria) {
603             matchStr = match_criteria;
604         } else {
605             if (zh->m_record_id && *zh->m_record_id) {
606                 matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname, 
607                                                zh->m_record_id);
608                 if (!matchStr)
609                 {
610                     yaz_log (YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
611                              pr_fname, (zint) start_offset);
612                     return ZEBRA_FAIL;
613                 }
614             }
615         }
616         if (matchStr) 
617         {
618             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
619             char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
620                                           matchStr);
621             if (rinfo)
622             {
623                 assert(*rinfo == sizeof(*sysno));
624                 memcpy (sysno, rinfo+1, sizeof(*sysno));
625             }
626         }
627     }
628     if (zebra_rec_keys_empty(zh->reg->keys))
629     {
630         /* the extraction process returned no information - the record
631            is probably empty - unless flagShowRecords is in use */
632         if (test_mode)
633             return ZEBRA_OK;
634     }
635
636     if (! *sysno)
637     {
638         /* new record */
639         if (delete_flag)
640         {
641             yaz_log (YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
642                          pr_fname, (zint) start_offset);
643             yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
644             return ZEBRA_FAIL;
645         }
646         if (show_progress)
647             yaz_log (YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
648                      (zint) start_offset);
649         rec = rec_new (zh->reg->records);
650
651         *sysno = rec->sysno;
652
653         recordAttr = rec_init_attr (zh->reg->zei, rec);
654         if (extractCtrl.staticrank < 0)
655         {
656             yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
657             extractCtrl.staticrank = 0;
658         }
659         recordAttr->staticrank = extractCtrl.staticrank;
660
661         if (matchStr)
662         {
663             int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
664             dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
665                             sizeof(*sysno), sysno);
666         }
667
668         extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
669         extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
670                                   recordAttr->staticrank);
671         zh->records_inserted++;
672     } 
673     else
674     {
675         /* record already exists */
676         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
677         zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
678         if (!allow_update)
679         {
680             yaz_log (YLOG_LOG, "skipped %s %s " ZINT_FORMAT, 
681                          recordType, pr_fname, (zint) start_offset);
682             logRecord(zh);
683             return ZEBRA_FAIL;
684         }
685
686         rec = rec_get (zh->reg->records, *sysno);
687         assert (rec);
688         
689         recordAttr = rec_init_attr (zh->reg->zei, rec);
690
691         /* decrease total size */
692         zebraExplain_recordBytesIncrement (zh->reg->zei,
693                                            - recordAttr->recordSize);
694
695         zebra_rec_keys_set_buf(delkeys,
696                                rec->info[recInfo_delKeys],
697                                rec->size[recInfo_delKeys],
698                                0);
699         zebra_rec_keys_set_buf(sortKeys,
700                                rec->info[recInfo_sortKeys],
701                                rec->size[recInfo_sortKeys],
702                                0);
703
704         extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
705         extract_flush_record_keys(zh, *sysno, 0, delkeys,
706                                   recordAttr->staticrank);
707         if (delete_flag)
708         {
709             /* record going to be deleted */
710             if (zebra_rec_keys_empty(delkeys))
711             {
712                 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
713                         pr_fname, (zint) start_offset);
714                 yaz_log(YLOG_WARN, "cannot delete file above, "
715                         "storeKeys false (3)");
716             }
717             else
718             {
719                 if (show_progress)
720                     yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
721                             pr_fname, (zint) start_offset);
722                 zh->records_deleted++;
723                 if (matchStr)
724                 {
725                     int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
726                     dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
727                 }
728                 rec_del (zh->reg->records, &rec);
729             }
730             rec_free(&rec);
731             logRecord(zh);
732             return ZEBRA_OK;
733         }
734         else
735         {
736             if (show_progress)
737                 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
738                         pr_fname, (zint) start_offset);
739             recordAttr->staticrank = extractCtrl.staticrank;
740             extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
741             extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys, 
742                                       recordAttr->staticrank);
743             zh->records_updated++;
744         }
745         zebra_rec_keys_close(delkeys);
746         zebra_rec_keys_close(sortKeys);
747     }
748     /* update file type */
749     xfree (rec->info[recInfo_fileType]);
750     rec->info[recInfo_fileType] =
751         rec_strdup (recordType, &rec->size[recInfo_fileType]);
752
753     /* update filename */
754     xfree (rec->info[recInfo_filename]);
755     rec->info[recInfo_filename] =
756         rec_strdup (fname, &rec->size[recInfo_filename]);
757
758     /* update delete keys */
759     xfree (rec->info[recInfo_delKeys]);
760     if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
761     {
762         zebra_rec_keys_get_buf(zh->reg->keys,
763                                &rec->info[recInfo_delKeys],
764                                &rec->size[recInfo_delKeys]);
765     }
766     else
767     {
768         rec->info[recInfo_delKeys] = NULL;
769         rec->size[recInfo_delKeys] = 0;
770     }
771     /* update sort keys */
772     xfree (rec->info[recInfo_sortKeys]);
773
774     zebra_rec_keys_get_buf(zh->reg->sortKeys,
775                            &rec->info[recInfo_sortKeys],
776                            &rec->size[recInfo_sortKeys]);
777
778     if (stream)
779     {
780         recordAttr->recordSize = end_offset - start_offset;
781         zebraExplain_recordBytesIncrement(zh->reg->zei,
782                                           recordAttr->recordSize);
783     }
784
785     /* set run-number for this record */
786     recordAttr->runNumber =
787         zebraExplain_runNumberIncrement (zh->reg->zei, 0);
788
789     /* update store data */
790     xfree (rec->info[recInfo_storeData]);
791
792     /* update store data */
793     if (zh->store_data_buf)
794     {
795         rec->size[recInfo_storeData] = zh->store_data_size;
796         rec->info[recInfo_storeData] = zh->store_data_buf;
797         zh->store_data_buf = 0;
798         recordAttr->recordSize = zh->store_data_size;
799     }
800     else if (zh->m_store_data)
801     {
802         off_t cur_offset = stream->tellf(stream);
803
804         rec->size[recInfo_storeData] = recordAttr->recordSize;
805         rec->info[recInfo_storeData] = (char *)
806             xmalloc (recordAttr->recordSize);
807         stream->seekf(stream, start_offset);
808         stream->readf(stream, rec->info[recInfo_storeData],
809                       recordAttr->recordSize);
810         stream->seekf(stream, cur_offset);
811     }
812     else
813     {
814         rec->info[recInfo_storeData] = NULL;
815         rec->size[recInfo_storeData] = 0;
816     }
817     /* update database name */
818     xfree (rec->info[recInfo_databaseName]);
819     rec->info[recInfo_databaseName] =
820         rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]); 
821
822     /* update offset */
823     recordAttr->recordOffset = start_offset;
824     
825     /* commit this record */
826     rec_put (zh->reg->records, &rec);
827     logRecord(zh);
828     return ZEBRA_OK;
829 }
830
831 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
832 {
833     ZebraHandle zh = (ZebraHandle) handle;
834     struct recExtractCtrl extractCtrl;
835
836     if (zebraExplain_curDatabase (zh->reg->zei,
837                                   rec->info[recInfo_databaseName]))
838     {
839         abort();
840         if (zebraExplain_newDatabase (zh->reg->zei,
841                                       rec->info[recInfo_databaseName], 0))
842             abort ();
843     }
844
845     zebra_rec_keys_reset(zh->reg->keys);
846     zebra_rec_keys_reset(zh->reg->sortKeys);
847
848     extractCtrl.init = extract_init;
849     extractCtrl.tokenAdd = extract_token_add;
850     extractCtrl.schemaAdd = extract_schema_add;
851     extractCtrl.dh = zh->reg->dh;
852
853     init_extractCtrl(zh, &extractCtrl);
854
855     extractCtrl.flagShowRecords = 0;
856     extractCtrl.match_criteria[0] = '\0';
857     extractCtrl.staticrank = 0;
858     extractCtrl.handle = handle;
859     extractCtrl.first_record = 1;
860     
861     extract_set_store_data_prepare(&extractCtrl);
862
863     if (n)
864         grs_extract_tree(&extractCtrl, n);
865
866     if (rec->size[recInfo_delKeys])
867     {
868         zebra_rec_keys_t delkeys = zebra_rec_keys_open();
869         
870         zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
871
872         zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
873                                rec->size[recInfo_delKeys],
874                                0);
875         extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
876         zebra_rec_keys_close(delkeys);
877
878         zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
879                                rec->size[recInfo_sortKeys],
880                                0);
881
882         extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
883         zebra_rec_keys_close(sortkeys);
884     }
885     extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
886     extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
887     
888     xfree (rec->info[recInfo_delKeys]);
889     zebra_rec_keys_get_buf(zh->reg->keys,
890                            &rec->info[recInfo_delKeys], 
891                            &rec->size[recInfo_delKeys]);
892
893     xfree (rec->info[recInfo_sortKeys]);
894     zebra_rec_keys_get_buf(zh->reg->sortKeys,
895                            &rec->info[recInfo_sortKeys],
896                            &rec->size[recInfo_sortKeys]);
897     return ZEBRA_OK;
898 }
899
900 void extract_rec_keys_log(ZebraHandle zh, int is_insert,
901                           zebra_rec_keys_t reckeys,
902                           int level)
903 {
904     if (zebra_rec_keys_rewind(reckeys))
905     {
906         size_t slen;
907         const char *str;
908         struct it_key key;
909         NMEM nmem = nmem_create();
910
911         while(zebra_rec_keys_read(reckeys, &str, &slen, &key))
912         {
913             char keystr[200]; /* room for zints to print */
914             char *dst_term = 0;
915             int ord = CAST_ZINT_TO_INT(key.mem[0]);
916             int index_type, i;
917             const char *string_index;
918             
919             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
920                                     0/* db */, &string_index);
921             assert(index_type);
922             zebra_term_untrans_iconv(zh, nmem, index_type,
923                                      &dst_term, str);
924             *keystr = '\0';
925             for (i = 0; i<key.len; i++)
926             {
927                 sprintf(keystr + strlen(keystr), ZINT_FORMAT " ", key.mem[i]);
928             }
929
930             if (*str < CHR_BASE_CHAR)
931             {
932                 int i;
933                 char dst_buf[200]; /* room for special chars */
934
935                 strcpy(dst_buf , "?");
936
937                 if (!strcmp(str, ""))
938                     strcpy(dst_buf, "alwaysmatches");
939                 if (!strcmp(str, FIRST_IN_FIELD_STR))
940                     strcpy(dst_buf, "firstinfield");
941                 else if (!strcmp(str, CHR_UNKNOWN))
942                     strcpy(dst_buf, "unknown");
943                 else if (!strcmp(str, CHR_SPACE))
944                     strcpy(dst_buf, "space");
945                 
946                 for (i = 0; i<slen; i++)
947                 {
948                     sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
949                 }
950                 yaz_log(level, "%s%c %s %s", keystr, index_type,
951                         string_index, dst_buf);
952                 
953             }
954             else
955                 yaz_log(level, "%s%c %s \"%s\"", keystr, index_type,
956                         string_index, dst_term);
957
958             nmem_reset(nmem);
959         }
960         nmem_destroy(nmem);
961     }
962 }
963
964 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
965                              zebra_rec_keys_t reckeys)
966 {
967     ZebraExplainInfo zei = zh->reg->zei;
968     struct ord_stat {
969         int no;
970         int ord;
971         struct ord_stat *next;
972     };
973
974     if (zebra_rec_keys_rewind(reckeys))
975     {
976         struct ord_stat *ord_list = 0;
977         struct ord_stat *p;
978         size_t slen;
979         const char *str;
980         struct it_key key_in;
981         while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
982         {
983             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
984
985             for (p = ord_list; p ; p = p->next)
986                 if (p->ord == ord)
987                 {
988                     p->no++;
989                     break;
990                 }
991             if (!p)
992             {
993                 p = xmalloc(sizeof(*p));
994                 p->no = 1;
995                 p->ord = ord;
996                 p->next = ord_list;
997                 ord_list = p;
998             }
999         }
1000
1001         p = ord_list;
1002         while (p)
1003         {
1004             struct ord_stat *p1 = p;
1005
1006             if (is_insert)
1007                 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
1008             else
1009                 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
1010             p = p->next;
1011             xfree(p1);
1012         }
1013     }
1014 }
1015
1016 void extract_flush_record_keys(ZebraHandle zh, zint sysno, int cmd,
1017                                zebra_rec_keys_t reckeys,
1018                                zint staticrank)
1019 {
1020     ZebraExplainInfo zei = zh->reg->zei;
1021
1022     extract_rec_keys_adjust(zh, cmd, reckeys);
1023
1024     if (log_level_details)
1025     {
1026         yaz_log(log_level_details, "Keys for record " ZINT_FORMAT " %s",
1027                 sysno, cmd ? "insert" : "delete");
1028         extract_rec_keys_log(zh, cmd, reckeys, log_level_details);
1029     }
1030
1031     if (!zh->reg->key_block)
1032     {
1033         int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1034         const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", ".");
1035         int use_threads = atoi(res_get_def (zh->res, "threads", "1"));
1036         zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1037     }
1038     zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1039
1040     if (zebra_rec_keys_rewind(reckeys))
1041     {
1042         size_t slen;
1043         const char *str;
1044         struct it_key key_in;
1045         while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1046         {
1047             key_block_write(zh->reg->key_block, sysno, 
1048                             &key_in, cmd, str, slen,
1049                             staticrank, zh->m_staticrank);
1050         }
1051     }
1052 }
1053
1054
1055 ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
1056                                      zebra_rec_keys_t reckeys,
1057                                      zebra_snippets *snippets)
1058 {
1059     NMEM nmem = nmem_create();
1060     if (zebra_rec_keys_rewind(reckeys)) 
1061     {
1062         const char *str;
1063         size_t slen;
1064         struct it_key key;
1065         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1066         {
1067             char *dst_term = 0;
1068             int ord;
1069             zint seqno;
1070             int index_type;
1071
1072             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1073             seqno = key.mem[key.len-1];
1074             ord = CAST_ZINT_TO_INT(key.mem[0]);
1075             
1076             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1077                                     0/* db */, 0 /* string_index */);
1078             assert(index_type);
1079             zebra_term_untrans_iconv(zh, nmem, index_type,
1080                                      &dst_term, str);
1081             zebra_snippets_append(snippets, seqno, ord, dst_term);
1082             nmem_reset(nmem);
1083         }
1084     }
1085     nmem_destroy(nmem);
1086     return ZEBRA_OK;
1087 }
1088
1089 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1090 {
1091     yaz_log(YLOG_LOG, "print_rec_keys");
1092     if (zebra_rec_keys_rewind(reckeys))
1093     {
1094         const char *str;
1095         size_t slen;
1096         struct it_key key;
1097         while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1098         {
1099             char dst_buf[IT_MAX_WORD];
1100             zint seqno;
1101             int index_type;
1102             int ord = CAST_ZINT_TO_INT(key.mem[0]);
1103             const char *db = 0;
1104             assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1105
1106             zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1107             
1108             seqno = key.mem[key.len-1];
1109             
1110             zebra_term_untrans(zh, index_type, dst_buf, str);
1111             
1112             yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT 
1113                     " term=%s", ord, seqno, dst_buf); 
1114         }
1115     }
1116 }
1117
1118 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1119                                      const char *str, int length)
1120 {
1121     struct it_key key;
1122     ZebraHandle zh = p->extractCtrl->handle;
1123     ZebraExplainInfo zei = zh->reg->zei;
1124     int ch, i;
1125
1126     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1127     if (ch < 0)
1128         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1129
1130     i = 0;
1131     key.mem[i++] = ch;
1132     key.mem[i++] = p->record_id;
1133     key.mem[i++] = p->section_id;
1134
1135     if (zh->m_segment_indexing)
1136         key.mem[i++] = p->segment;
1137     key.mem[i++] = p->seqno;
1138     key.len = i;
1139
1140     zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1141 }
1142
1143 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1144 {
1145     struct it_key key;
1146     ZebraHandle zh = p->extractCtrl->handle;
1147     ZebraExplainInfo zei = zh->reg->zei;
1148     int ch;
1149     zinfo_index_category_t cat = zinfo_index_category_sort;
1150
1151     ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1152     if (ch < 0)
1153         ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1154     key.len = 2;
1155     key.mem[0] = ch;
1156     key.mem[1] = p->record_id;
1157
1158     zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1159 }
1160
1161 static void extract_add_string(RecWord *p, const char *string, int length)
1162 {
1163     ZebraHandle zh = p->extractCtrl->handle;
1164     assert (length > 0);
1165
1166     if (!p->index_name)
1167         return;
1168
1169     if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
1170         extract_add_sort_string(p, string, length);
1171     else
1172     {
1173         extract_add_index_string(p, zinfo_index_category_index,
1174                                  string, length);
1175         if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type))
1176         {
1177             RecWord word;
1178             memcpy(&word, p, sizeof(word));
1179
1180             word.seqno = 1;
1181             extract_add_index_string(
1182                 &word, zinfo_index_category_alwaysmatches, "", 0);
1183         }
1184     }
1185 }
1186
1187 static void extract_add_incomplete_field(RecWord *p)
1188 {
1189     ZebraHandle zh = p->extractCtrl->handle;
1190     const char *b = p->term_buf;
1191     int remain = p->term_len;
1192     int first = 1;
1193     const char **map = 0;
1194     
1195     if (remain > 0)
1196         map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1197
1198     while (map)
1199     {
1200         char buf[IT_MAX_WORD+1];
1201         int i, remain;
1202
1203         /* Skip spaces */
1204         while (map && *map && **map == *CHR_SPACE)
1205         {
1206             remain = p->term_len - (b - p->term_buf);
1207             if (remain > 0)
1208                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
1209                                        remain, 0);
1210             else
1211                 map = 0;
1212         }
1213         if (!map)
1214             break;
1215         i = 0;
1216         while (map && *map && **map != *CHR_SPACE)
1217         {
1218             const char *cp = *map;
1219
1220             while (i < IT_MAX_WORD && *cp)
1221                 buf[i++] = *(cp++);
1222             remain = p->term_len - (b - p->term_buf);
1223             if (remain > 0)
1224                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1225             else
1226                 map = 0;
1227         }
1228         if (!i)
1229             return;
1230
1231         if (first)
1232         {   
1233             first = 0;
1234             if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
1235             {
1236                 /* first in field marker */
1237                 extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1238                 p->seqno++;
1239             }
1240         }
1241         extract_add_string (p, buf, i);
1242         p->seqno++;
1243     }
1244 }
1245
1246 static void extract_add_complete_field (RecWord *p)
1247 {
1248     ZebraHandle zh = p->extractCtrl->handle;
1249     const char *b = p->term_buf;
1250     char buf[IT_MAX_WORD+1];
1251     const char **map = 0;
1252     int i = 0, remain = p->term_len;
1253
1254     if (remain > 0)
1255         map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
1256
1257     while (remain > 0 && i < IT_MAX_WORD)
1258     {
1259         while (map && *map && **map == *CHR_SPACE)
1260         {
1261             remain = p->term_len - (b - p->term_buf);
1262
1263             if (remain > 0)
1264             {
1265                 int first = i ? 0 : 1;  /* first position */
1266                 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, first);
1267             }
1268             else
1269                 map = 0;
1270         }
1271         if (!map)
1272             break;
1273
1274         if (i && i < IT_MAX_WORD)
1275             buf[i++] = *CHR_SPACE;
1276         while (map && *map && **map != *CHR_SPACE)
1277         {
1278             const char *cp = *map;
1279
1280             if (**map == *CHR_CUT)
1281             {
1282                 i = 0;
1283             }
1284             else
1285             {
1286                 if (i >= IT_MAX_WORD)
1287                     break;
1288                 while (i < IT_MAX_WORD && *cp)
1289                     buf[i++] = *(cp++);
1290             }
1291             remain = p->term_len  - (b - p->term_buf);
1292             if (remain > 0)
1293             {
1294                 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
1295                                         remain, 0);
1296             }
1297             else
1298                 map = 0;
1299         }
1300     }
1301     if (!i)
1302         return;
1303     extract_add_string (p, buf, i);
1304 }
1305
1306 static void extract_token_add(RecWord *p)
1307 {
1308     ZebraHandle zh = p->extractCtrl->handle;
1309     WRBUF wrbuf;
1310
1311     if (log_level_extract)
1312     {
1313         yaz_log(log_level_extract, "extract_token_add "
1314                 "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
1315                 p->index_type, p->index_name, 
1316                 p->seqno, p->term_len, p->term_buf);
1317     }
1318     if ((wrbuf = zebra_replace(zh->reg->zebra_maps, p->index_type, 0,
1319                                p->term_buf, p->term_len)))
1320     {
1321         p->term_buf = wrbuf_buf(wrbuf);
1322         p->term_len = wrbuf_len(wrbuf);
1323     }
1324     if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type))
1325         extract_add_complete_field (p);
1326     else
1327         extract_add_incomplete_field(p);
1328 }
1329
1330 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1331                                       void *buf, size_t sz)
1332 {
1333     ZebraHandle zh = (ZebraHandle) p->handle;
1334
1335     xfree(zh->store_data_buf);
1336     zh->store_data_buf = 0;
1337     zh->store_data_size = 0;
1338     if (buf && sz)
1339     {
1340         zh->store_data_buf = xmalloc(sz);
1341         zh->store_data_size = sz;
1342         memcpy(zh->store_data_buf, buf, sz);
1343     }
1344 }
1345
1346 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1347 {
1348     ZebraHandle zh = (ZebraHandle) p->handle;
1349     xfree(zh->store_data_buf);
1350     zh->store_data_buf = 0;
1351     zh->store_data_size = 0;
1352     p->setStoreData = extract_set_store_data_cb;
1353 }
1354
1355 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
1356 {
1357     ZebraHandle zh = (ZebraHandle) p->handle;
1358     zebraExplain_addSchema (zh->reg->zei, oid);
1359 }
1360
1361 void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
1362                              int cmd, zebra_rec_keys_t reckeys)
1363 {
1364 #if 0
1365     yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT,
1366             cmd, sysno);
1367     extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG);
1368 #endif
1369
1370     if (zebra_rec_keys_rewind(reckeys))
1371     {
1372         zebra_sort_index_t si = zh->reg->sort_index;
1373         size_t slen;
1374         const char *str;
1375         struct it_key key_in;
1376
1377         zebra_sort_sysno(si, sysno);
1378
1379         while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1380         {
1381             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1382             
1383             zebra_sort_type(si, ord);
1384             if (cmd == 1)
1385                 zebra_sort_add(si, str, slen);
1386             else
1387                 zebra_sort_delete(si);
1388         }
1389     }
1390 }
1391
1392 /*
1393  * Local variables:
1394  * c-basic-offset: 4
1395  * indent-tabs-mode: nil
1396  * End:
1397  * vim: shiftwidth=4 tabstop=8 expandtab
1398  */
1399