1 /* $Id: extract.c,v 1.187 2005-06-23 06:45:46 adam Exp $
2 Copyright (C) 1995-2005
5 This file is part of the Zebra server.
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
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
17 You should have received a copy of the GNU General Public License
18 along with Zebra; see the file LICENSE.zebra. If not, write to the
19 Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
38 #if _FILE_OFFSET_BITS == 64
39 #define PRINTF_OFF_T "%Ld"
41 #define PRINTF_OFF_T "%ld"
44 #define USE_SHELLSORT 0
47 static void shellsort(void *ar, int r, size_t s,
48 int (*cmp)(const void *a, const void *b))
53 static const int incs[16] = { 1391376, 463792, 198768, 86961, 33936,
54 13776, 4592, 1968, 861, 336,
55 112, 48, 21, 7, 3, 1 };
56 for ( k = 0; k < 16; k++)
57 for (h = incs[k], i = h; i < r; i++)
61 while (j > h && (*cmp)(a + s*(j-h), v) > 0)
63 memcpy (a + s*j, a + s*(j-h), s);
71 static void logRecord (ZebraHandle zh)
73 ++zh->records_processed;
74 if (!(zh->records_processed % 1000))
76 yaz_log (YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
77 ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT,
78 zh->records_processed, zh->records_inserted, zh->records_updated,
83 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
85 static void extract_init (struct recExtractCtrl *p, RecWord *w)
87 w->zebra_maps = p->zebra_maps;
91 w->attrSet = VAL_BIB1;
101 static const char **searchRecordKey (ZebraHandle zh,
102 struct recKeys *reckeys,
103 int attrSetS, int attrUseS)
105 static const char *ws[32];
106 void *decode_handle = iscz1_start();
112 for (i = 0; i<32; i++)
115 while (off < reckeys->buf_used)
117 const char *src = reckeys->buf + off;
119 char *dst = (char*) &key;
120 int attrSet, attrUse;
122 iscz1_decode(decode_handle, &dst, &src);
123 assert(key.len <= 4 && key.len > 2);
125 attrSet = (int) key.mem[0] >> 16;
126 attrUse = (int) key.mem[0] & 65535;
127 seqno = (int) key.mem[key.len-1];
129 if (attrUseS == attrUse && attrSetS == attrSet)
135 woff = seqno - startSeq;
136 if (woff >= 0 && woff < 31)
142 off = src - reckeys->buf;
144 iscz1_stop(decode_handle);
145 assert (off == reckeys->buf_used);
149 struct file_read_info {
150 off_t file_max; /* maximum offset so far */
151 off_t file_offset; /* current offset */
152 off_t file_moffset; /* offset of rec/rec boundary */
157 static struct file_read_info *file_read_start (int fd)
159 struct file_read_info *fi = (struct file_read_info *)
160 xmalloc (sizeof(*fi));
164 fi->file_moffset = 0;
170 static void file_read_stop (struct file_read_info *fi)
175 static off_t file_seek (void *handle, off_t offset)
177 struct file_read_info *p = (struct file_read_info *) handle;
178 p->file_offset = offset;
179 return lseek (p->fd, offset, SEEK_SET);
182 static off_t file_tell (void *handle)
184 struct file_read_info *p = (struct file_read_info *) handle;
185 return p->file_offset;
188 static int file_read (void *handle, char *buf, size_t count)
190 struct file_read_info *p = (struct file_read_info *) handle;
193 r = read (fd, buf, count);
197 if (p->file_offset > p->file_max)
198 p->file_max = p->file_offset;
203 static void file_end (void *handle, off_t offset)
205 struct file_read_info *p = (struct file_read_info *) handle;
207 if (offset != p->file_moffset)
209 p->file_moffset = offset;
214 static char *fileMatchStr (ZebraHandle zh,
215 struct recKeys *reckeys,
216 const char *fname, const char *spec)
218 static char dstBuf[2048]; /* static here ??? */
220 const char *s = spec;
221 static const char **w;
225 while (*s == ' ' || *s == '\t')
231 char attset_str[64], attname_str[64];
232 data1_attset *attset;
235 int attSet = 1, attUse = 1;
239 for (i = 0; *s && *s != ',' && *s != ')'; s++)
241 attset_str[i++] = *s;
242 attset_str[i] = '\0';
247 for (i = 0; *s && *s != ')'; s++)
249 attname_str[i++] = *s;
250 attname_str[i] = '\0';
253 if ((attset = data1_get_attset (zh->reg->dh, attset_str)))
256 attSet = attset->reference;
257 att = data1_getattbyname(zh->reg->dh, attset, attname_str);
261 attUse = atoi (attname_str);
263 w = searchRecordKey (zh, reckeys, attSet, attUse);
268 for (i = 0; i<32; i++)
273 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
274 spec, zh->m_group ? zh->m_group : "none");
279 for (i = 0; i<32; i++)
280 if (matchFlag[i] && w[i])
292 yaz_log (YLOG_WARN, "Record didn't contain match"
293 " fields in (%s,%s)", attset_str, attname_str);
301 const char *spec_src = NULL;
302 const char *s1 = ++s;
303 while (*s1 && *s1 != ' ' && *s1 != '\t')
309 memcpy (special, s, spec_len);
310 special[spec_len] = '\0';
313 if (!strcmp (special, "group"))
314 spec_src = zh->m_group;
315 else if (!strcmp (special, "database"))
316 spec_src = zh->basenames[0];
317 else if (!strcmp (special, "filename")) {
320 else if (!strcmp (special, "type"))
321 spec_src = zh->m_record_type;
326 strcpy (dst, spec_src);
327 dst += strlen (spec_src);
330 else if (*s == '\"' || *s == '\'')
332 int stopMarker = *s++;
336 while (*s && *s != stopMarker)
339 tmpString[i++] = *s++;
344 strcpy (dst, tmpString);
345 dst += strlen (tmpString);
349 yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
350 spec, zh->m_group ? zh->m_group : "none");
357 yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
358 fname, zh->m_group ? zh->m_group : "none");
365 struct recordLogInfo {
368 struct recordGroup *rGroup;
371 void create_rec_keys_codec(struct recKeys *keys)
374 iscz1_reset(keys->codec_handle);
377 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
380 for (i = 0; i<256; i++)
382 if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
387 ctrl->zebra_maps = zh->reg->zebra_maps;
388 ctrl->flagShowRecords = !zh->m_flag_rw;
391 static int file_extract_record(ZebraHandle zh,
392 SYSNO *sysno, const char *fname,
394 struct file_read_info *fi,
397 void *recTypeClientData)
399 RecordAttr *recordAttr;
401 const char *matchStr = 0;
404 off_t recordOffset = 0;
406 /* announce database */
407 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
409 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
410 zh->m_explain_database))
416 struct recExtractCtrl extractCtrl;
418 /* we are going to read from a file, so prepare the extraction */
419 create_rec_keys_codec(&zh->reg->keys);
421 create_rec_keys_codec(&zh->reg->sortKeys);
423 zh->reg->sortKeys.buf_used = 0;
425 recordOffset = fi->file_moffset;
426 extractCtrl.handle = zh;
427 extractCtrl.offset = fi->file_moffset;
428 extractCtrl.readf = file_read;
429 extractCtrl.seekf = file_seek;
430 extractCtrl.tellf = file_tell;
431 extractCtrl.endf = file_end;
433 extractCtrl.init = extract_init;
434 extractCtrl.tokenAdd = extract_token_add;
435 extractCtrl.schemaAdd = extract_schema_add;
436 extractCtrl.dh = zh->reg->dh;
437 extractCtrl.match_criteria[0] = '\0';
438 extractCtrl.first_record = fi->file_offset ? 0 : 1;
440 extract_set_store_data_prepare(&extractCtrl);
442 init_extractCtrl(zh, &extractCtrl);
445 printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset);
449 sprintf (msg, "%s:" PRINTF_OFF_T , fname, recordOffset);
450 yaz_log_init_prefix2 (msg);
453 r = (*recType->extract)(recTypeClientData, &extractCtrl);
455 yaz_log_init_prefix2 (0);
456 if (r == RECCTRL_EXTRACT_EOF)
458 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
460 /* error occured during extraction ... */
462 zh->records_processed < zh->m_file_verbose_limit)
464 yaz_log (YLOG_WARN, "fail %s %s " PRINTF_OFF_T, zh->m_record_type,
465 fname, recordOffset);
469 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
471 /* error occured during extraction ... */
473 zh->records_processed < zh->m_file_verbose_limit)
475 yaz_log (YLOG_WARN, "no filter for %s %s "
476 PRINTF_OFF_T, zh->m_record_type,
477 fname, recordOffset);
481 if (extractCtrl.match_criteria[0])
482 matchStr = extractCtrl.match_criteria;
485 /* perform match if sysno not known and if match criteria is specified */
491 if (matchStr == 0 && zh->m_record_id && *zh->m_record_id)
494 matchStr = fileMatchStr (zh, &zh->reg->keys, fname,
498 yaz_log(YLOG_WARN, "Bad match criteria");
504 char *rinfo = dict_lookup (zh->reg->matchDict, matchStr);
507 assert(*rinfo == sizeof(*sysno));
508 memcpy (sysno, rinfo+1, sizeof(*sysno));
512 if (! *sysno && zh->reg->keys.buf_used == 0)
514 /* the extraction process returned no information - the record
515 is probably empty - unless flagShowRecords is in use */
519 if (zh->records_processed < zh->m_file_verbose_limit)
520 yaz_log (YLOG_WARN, "empty %s %s " PRINTF_OFF_T, zh->m_record_type,
521 fname, recordOffset);
530 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T, zh->m_record_type,
531 fname, recordOffset);
532 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
535 if (zh->records_processed < zh->m_file_verbose_limit)
536 yaz_log (YLOG_LOG, "add %s %s " PRINTF_OFF_T, zh->m_record_type,
537 fname, recordOffset);
538 rec = rec_new (zh->reg->records);
542 recordAttr = rec_init_attr (zh->reg->zei, rec);
546 dict_insert (zh->reg->matchDict, matchStr, sizeof(*sysno), sysno);
548 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
549 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
551 zh->records_inserted++;
555 /* record already exists */
556 struct recKeys delkeys;
558 struct recKeys sortKeys;
560 struct sortKeys sortKeys;
563 rec = rec_get (zh->reg->records, *sysno);
566 recordAttr = rec_init_attr (zh->reg->zei, rec);
568 if (!force_update && recordAttr->runNumber ==
569 zebraExplain_runNumberIncrement (zh->reg->zei, 0))
571 yaz_log (YLOG_LOG, "run number = " ZINT_FORMAT,
572 recordAttr->runNumber);
573 yaz_log (YLOG_LOG, "skipped %s %s " PRINTF_OFF_T,
574 zh->m_record_type, fname, recordOffset);
575 extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
580 delkeys.buf_used = rec->size[recInfo_delKeys];
581 delkeys.buf = rec->info[recInfo_delKeys];
583 sortKeys.buf_used = rec->size[recInfo_sortKeys];
584 sortKeys.buf = rec->info[recInfo_sortKeys];
586 extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
587 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
590 /* record going to be deleted */
591 if (!delkeys.buf_used)
593 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
594 zh->m_record_type, fname, recordOffset);
595 yaz_log (YLOG_WARN, "cannot delete file above, storeKeys false");
599 if (zh->records_processed < zh->m_file_verbose_limit)
600 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
601 zh->m_record_type, fname, recordOffset);
602 zh->records_deleted++;
604 dict_delete (zh->reg->matchDict, matchStr);
605 rec_del (zh->reg->records, &rec);
613 /* record going to be updated */
614 if (!delkeys.buf_used)
616 yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
617 zh->m_record_type, fname, recordOffset);
618 yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
622 if (zh->records_processed < zh->m_file_verbose_limit)
623 yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
624 zh->m_record_type, fname, recordOffset);
625 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
626 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
627 zh->records_updated++;
631 /* update file type */
632 xfree (rec->info[recInfo_fileType]);
633 rec->info[recInfo_fileType] =
634 rec_strdup (zh->m_record_type, &rec->size[recInfo_fileType]);
636 /* update filename */
637 xfree (rec->info[recInfo_filename]);
638 rec->info[recInfo_filename] =
639 rec_strdup (fname, &rec->size[recInfo_filename]);
641 /* update delete keys */
642 xfree (rec->info[recInfo_delKeys]);
643 if (zh->reg->keys.buf_used > 0 && zh->m_store_keys == 1)
645 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
646 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
647 zh->reg->keys.buf = NULL;
648 zh->reg->keys.buf_max = 0;
652 rec->info[recInfo_delKeys] = NULL;
653 rec->size[recInfo_delKeys] = 0;
656 /* update sort keys */
657 xfree (rec->info[recInfo_sortKeys]);
659 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
660 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
661 zh->reg->sortKeys.buf = NULL;
662 zh->reg->sortKeys.buf_max = 0;
664 /* save file size of original record */
665 zebraExplain_recordBytesIncrement (zh->reg->zei,
666 - recordAttr->recordSize);
667 recordAttr->recordSize = fi->file_moffset - recordOffset;
668 if (!recordAttr->recordSize)
669 recordAttr->recordSize = fi->file_max - recordOffset;
670 zebraExplain_recordBytesIncrement (zh->reg->zei,
671 recordAttr->recordSize);
673 /* set run-number for this record */
674 recordAttr->runNumber = zebraExplain_runNumberIncrement (zh->reg->zei,
677 /* update store data */
678 xfree (rec->info[recInfo_storeData]);
679 if (zh->store_data_buf)
681 rec->size[recInfo_storeData] = zh->store_data_size;
682 rec->info[recInfo_storeData] = zh->store_data_buf;
683 zh->store_data_buf = 0;
685 else if (zh->m_store_data)
687 rec->size[recInfo_storeData] = recordAttr->recordSize;
688 rec->info[recInfo_storeData] = (char *)
689 xmalloc (recordAttr->recordSize);
690 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
692 yaz_log (YLOG_ERRNO|YLOG_FATAL, "seek to " PRINTF_OFF_T " in %s",
693 recordOffset, fname);
696 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
697 < recordAttr->recordSize)
699 yaz_log (YLOG_ERRNO|YLOG_FATAL, "read %d bytes of %s",
700 recordAttr->recordSize, fname);
706 rec->info[recInfo_storeData] = NULL;
707 rec->size[recInfo_storeData] = 0;
709 /* update database name */
710 xfree (rec->info[recInfo_databaseName]);
711 rec->info[recInfo_databaseName] =
712 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
715 recordAttr->recordOffset = recordOffset;
717 /* commit this record */
718 rec_put (zh->reg->records, &rec);
723 int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
730 struct file_read_info *fi;
731 const char *original_record_type = 0;
733 void *recTypeClientData;
735 if (!zh->m_group || !*zh->m_group)
738 sprintf (gprefix, "%s.", zh->m_group);
740 yaz_log (YLOG_DEBUG, "fileExtract %s", fname);
742 /* determine file extension */
744 for (i = strlen(fname); --i >= 0; )
747 else if (fname[i] == '.')
749 strcpy (ext, fname+i+1);
752 /* determine file type - depending on extension */
753 original_record_type = zh->m_record_type;
754 if (!zh->m_record_type)
756 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
757 zh->m_record_type = res_get (zh->res, ext_res);
759 if (!zh->m_record_type)
761 if (zh->records_processed < zh->m_file_verbose_limit)
762 yaz_log (YLOG_LOG, "? %s", fname);
765 /* determine match criteria */
766 if (!zh->m_record_id)
768 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
769 zh->m_record_id = res_get (zh->res, ext_res);
773 recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
774 &recTypeClientData)))
776 yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
780 switch(recType->version)
785 yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
787 if (sysno && deleteFlag)
793 if (zh->path_reg && !yaz_is_abspath (fname))
795 strcpy (full_rep, zh->path_reg);
796 strcat (full_rep, "/");
797 strcat (full_rep, fname);
800 strcpy (full_rep, fname);
803 if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
805 yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
806 zh->m_record_type = original_record_type;
810 fi = file_read_start (fd);
813 fi->file_moffset = fi->file_offset;
814 fi->file_more = 0; /* file_end not called (yet) */
815 r = file_extract_record (zh, sysno, fname, deleteFlag, fi, 1,
816 recType, recTypeClientData);
818 { /* file_end has been called so reset offset .. */
819 fi->file_offset = fi->file_moffset;
820 lseek(fi->fd, fi->file_moffset, SEEK_SET);
827 zh->m_record_type = original_record_type;
832 If sysno is provided, then it's used to identify the reocord.
833 If not, and match_criteria is provided, then sysno is guessed
834 If not, and a record is provided, then sysno is got from there
837 ZEBRA_RES buffer_extract_record (ZebraHandle zh,
838 const char *buf, size_t buf_size,
841 const char *recordType,
843 const char *match_criteria,
848 RecordAttr *recordAttr;
849 struct recExtractCtrl extractCtrl;
851 const char *matchStr = 0;
852 RecType recType = NULL;
855 long recordOffset = 0;
856 struct zebra_fetch_control fc;
857 const char *pr_fname = fname; /* filename to print .. */
858 int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
861 pr_fname = "<no file>"; /* make it printable if file is omitted */
864 fc.record_int_buf = buf;
865 fc.record_int_len = buf_size;
866 fc.record_int_pos = 0;
868 fc.record_offset = 0;
870 extractCtrl.offset = 0;
871 extractCtrl.readf = zebra_record_int_read;
872 extractCtrl.seekf = zebra_record_int_seek;
873 extractCtrl.tellf = zebra_record_int_tell;
874 extractCtrl.endf = zebra_record_int_end;
875 extractCtrl.first_record = 1;
876 extractCtrl.fh = &fc;
878 create_rec_keys_codec(&zh->reg->keys);
880 create_rec_keys_codec(&zh->reg->sortKeys);
882 zh->reg->sortKeys.buf_used = 0;
884 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
886 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
887 zh->m_explain_database))
891 if (recordType && *recordType)
893 yaz_log (YLOG_DEBUG, "Record type explicitly specified: %s", recordType);
894 recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
899 if (!(zh->m_record_type))
901 yaz_log (YLOG_WARN, "No such record type defined");
904 yaz_log (YLOG_DEBUG, "Get record type from rgroup: %s",zh->m_record_type);
905 recType = recType_byName (zh->reg->recTypes, zh->res,
906 zh->m_record_type, &clientData);
907 recordType = zh->m_record_type;
912 yaz_log (YLOG_WARN, "No such record type: %s", zh->m_record_type);
916 extractCtrl.init = extract_init;
917 extractCtrl.tokenAdd = extract_token_add;
918 extractCtrl.schemaAdd = extract_schema_add;
919 extractCtrl.dh = zh->reg->dh;
920 extractCtrl.handle = zh;
921 extractCtrl.match_criteria[0] = '\0';
923 init_extractCtrl(zh, &extractCtrl);
925 extract_set_store_data_prepare(&extractCtrl);
927 r = (*recType->extract)(clientData, &extractCtrl);
929 if (r == RECCTRL_EXTRACT_EOF)
931 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
933 /* error occured during extraction ... */
934 yaz_log (YLOG_WARN, "extract error: generic");
937 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
939 /* error occured during extraction ... */
940 yaz_log (YLOG_WARN, "extract error: no such filter");
946 if (extractCtrl.match_criteria[0])
947 match_criteria = extractCtrl.match_criteria;
951 if (match_criteria && *match_criteria) {
952 matchStr = match_criteria;
954 if (zh->m_record_id && *zh->m_record_id) {
955 matchStr = fileMatchStr (zh, &zh->reg->keys, pr_fname,
959 yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
965 rinfo = dict_lookup (zh->reg->matchDict, matchStr);
968 assert(*rinfo == sizeof(*sysno));
969 memcpy (sysno, rinfo+1, sizeof(*sysno));
973 if (zh->reg->keys.buf_used == 0)
975 /* the extraction process returned no information - the record
976 is probably empty - unless flagShowRecords is in use */
987 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
988 pr_fname, (long) recordOffset);
989 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
993 yaz_log (YLOG_LOG, "add %s %s %ld", recordType, pr_fname,
994 (long) recordOffset);
995 rec = rec_new (zh->reg->records);
999 recordAttr = rec_init_attr (zh->reg->zei, rec);
1003 dict_insert (zh->reg->matchDict, matchStr,
1004 sizeof(*sysno), sysno);
1006 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1007 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
1009 zh->records_inserted++;
1013 /* record already exists */
1014 struct recKeys delkeys;
1016 struct recKeys sortKeys;
1018 struct sortKeys sortKeys;
1024 yaz_log (YLOG_LOG, "skipped %s %s %ld",
1025 recordType, pr_fname, (long) recordOffset);
1030 rec = rec_get (zh->reg->records, *sysno);
1033 recordAttr = rec_init_attr (zh->reg->zei, rec);
1035 if (!force_update) {
1036 if (recordAttr->runNumber ==
1037 zebraExplain_runNumberIncrement (zh->reg->zei, 0))
1040 yaz_log (YLOG_LOG, "skipped %s %s %ld", recordType,
1041 pr_fname, (long) recordOffset);
1042 extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
1049 delkeys.buf_used = rec->size[recInfo_delKeys];
1050 delkeys.buf = rec->info[recInfo_delKeys];
1052 sortKeys.buf_used = rec->size[recInfo_sortKeys];
1053 sortKeys.buf = rec->info[recInfo_sortKeys];
1055 extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
1056 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1059 /* record going to be deleted */
1060 if (!delkeys.buf_used)
1064 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1065 pr_fname, (long) recordOffset);
1066 yaz_log (YLOG_WARN, "cannot delete file above, "
1073 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1074 pr_fname, (long) recordOffset);
1075 zh->records_deleted++;
1077 dict_delete (zh->reg->matchDict, matchStr);
1078 rec_del (zh->reg->records, &rec);
1086 /* record going to be updated */
1087 if (!delkeys.buf_used)
1091 yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
1092 pr_fname, (long) recordOffset);
1093 yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
1099 yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
1100 pr_fname, (long) recordOffset);
1101 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1102 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
1103 zh->records_updated++;
1107 /* update file type */
1108 xfree (rec->info[recInfo_fileType]);
1109 rec->info[recInfo_fileType] =
1110 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1112 /* update filename */
1113 xfree (rec->info[recInfo_filename]);
1114 rec->info[recInfo_filename] =
1115 rec_strdup (fname, &rec->size[recInfo_filename]);
1117 /* update delete keys */
1118 xfree (rec->info[recInfo_delKeys]);
1119 if (zh->reg->keys.buf_used > 0 && zh->m_store_keys == 1)
1121 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
1122 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
1123 zh->reg->keys.buf = NULL;
1124 zh->reg->keys.buf_max = 0;
1128 rec->info[recInfo_delKeys] = NULL;
1129 rec->size[recInfo_delKeys] = 0;
1132 /* update sort keys */
1133 xfree (rec->info[recInfo_sortKeys]);
1135 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
1136 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
1137 zh->reg->sortKeys.buf = NULL;
1138 zh->reg->sortKeys.buf_max = 0;
1140 /* save file size of original record */
1141 zebraExplain_recordBytesIncrement (zh->reg->zei,
1142 - recordAttr->recordSize);
1144 recordAttr->recordSize = fi->file_moffset - recordOffset;
1145 if (!recordAttr->recordSize)
1146 recordAttr->recordSize = fi->file_max - recordOffset;
1148 recordAttr->recordSize = buf_size;
1150 zebraExplain_recordBytesIncrement (zh->reg->zei,
1151 recordAttr->recordSize);
1153 /* set run-number for this record */
1154 recordAttr->runNumber =
1155 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
1157 /* update store data */
1158 xfree (rec->info[recInfo_storeData]);
1160 /* update store data */
1161 if (zh->store_data_buf)
1163 rec->size[recInfo_storeData] = zh->store_data_size;
1164 rec->info[recInfo_storeData] = zh->store_data_buf;
1165 zh->store_data_buf = 0;
1167 else if (zh->m_store_data)
1169 rec->size[recInfo_storeData] = recordAttr->recordSize;
1170 rec->info[recInfo_storeData] = (char *)
1171 xmalloc (recordAttr->recordSize);
1172 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1176 rec->info[recInfo_storeData] = NULL;
1177 rec->size[recInfo_storeData] = 0;
1179 /* update database name */
1180 xfree (rec->info[recInfo_databaseName]);
1181 rec->info[recInfo_databaseName] =
1182 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
1185 recordAttr->recordOffset = recordOffset;
1187 /* commit this record */
1188 rec_put (zh->reg->records, &rec);
1193 int explain_extract (void *handle, Record rec, data1_node *n)
1195 ZebraHandle zh = (ZebraHandle) handle;
1196 struct recExtractCtrl extractCtrl;
1198 if (zebraExplain_curDatabase (zh->reg->zei,
1199 rec->info[recInfo_databaseName]))
1202 if (zebraExplain_newDatabase (zh->reg->zei,
1203 rec->info[recInfo_databaseName], 0))
1207 create_rec_keys_codec(&zh->reg->keys);
1209 create_rec_keys_codec(&zh->reg->sortKeys);
1211 zh->reg->sortKeys.buf_used = 0;
1213 extractCtrl.init = extract_init;
1214 extractCtrl.tokenAdd = extract_token_add;
1215 extractCtrl.schemaAdd = extract_schema_add;
1216 extractCtrl.dh = zh->reg->dh;
1218 init_extractCtrl(zh, &extractCtrl);
1220 extractCtrl.flagShowRecords = 0;
1221 extractCtrl.match_criteria[0] = '\0';
1222 extractCtrl.handle = handle;
1223 extractCtrl.first_record = 1;
1225 extract_set_store_data_prepare(&extractCtrl);
1228 grs_extract_tree(&extractCtrl, n);
1230 if (rec->size[recInfo_delKeys])
1232 struct recKeys delkeys;
1234 struct recKeys sortkeys;
1236 struct sortKeys sortkeys;
1239 delkeys.buf_used = rec->size[recInfo_delKeys];
1240 delkeys.buf = rec->info[recInfo_delKeys];
1242 sortkeys.buf_used = rec->size[recInfo_sortKeys];
1243 sortkeys.buf = rec->info[recInfo_sortKeys];
1245 extract_flushSortKeys (zh, rec->sysno, 0, &sortkeys);
1246 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1248 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->reg->keys);
1249 extract_flushSortKeys (zh, rec->sysno, 1, &zh->reg->sortKeys);
1251 xfree (rec->info[recInfo_delKeys]);
1252 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
1253 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
1254 zh->reg->keys.buf = NULL;
1255 zh->reg->keys.buf_max = 0;
1257 xfree (rec->info[recInfo_sortKeys]);
1258 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
1259 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
1260 zh->reg->sortKeys.buf = NULL;
1261 zh->reg->sortKeys.buf_max = 0;
1266 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1267 int cmd, struct recKeys *reckeys)
1269 void *decode_handle = iscz1_start();
1272 ZebraExplainInfo zei = zh->reg->zei;
1274 if (!zh->reg->key_buf)
1276 int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
1279 yaz_log(YLOG_WARN, "Invalid memory setting, using default 8 MB");
1282 /* FIXME: That "8" should be in a default settings include */
1283 /* not hard-coded here! -H */
1284 zh->reg->key_buf = (char**) xmalloc (mem);
1285 zh->reg->ptr_top = mem/sizeof(char*);
1287 zh->reg->key_buf_used = 0;
1288 zh->reg->key_file_no = 0;
1290 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1292 while (off < reckeys->buf_used)
1294 const char *src = reckeys->buf + off;
1296 char *dst = (char*) &key;
1298 iscz1_decode(decode_handle, &dst, &src);
1299 assert(key.len == 4);
1301 if (zh->reg->key_buf_used + 1024 >
1302 (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
1303 extract_flushWriteKeys (zh,0);
1305 assert(zh->reg->ptr_i > 0);
1306 (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
1307 (char*)zh->reg->key_buf + zh->reg->key_buf_used;
1309 ch = (int) key.mem[0]; /* ordinal for field/use/attribute */
1311 zh->reg->key_buf_used +=
1312 key_SU_encode (ch,((char*)zh->reg->key_buf) +
1313 zh->reg->key_buf_used);
1315 ((char*)zh->reg->key_buf) [(zh->reg->key_buf_used)++] = *src++;
1317 ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = '\0';
1318 ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = cmd;
1321 if (key.mem[1]) /* filter specified record ID */
1322 key.mem[0] = key.mem[1];
1325 key.mem[1] = key.mem[2]; /* section_id */
1326 key.mem[2] = key.mem[3]; /* sequence .. */
1328 memcpy ((char*)zh->reg->key_buf + zh->reg->key_buf_used,
1330 (zh->reg->key_buf_used) += sizeof(key);
1331 off = src - reckeys->buf;
1333 assert (off == reckeys->buf_used);
1334 iscz1_stop(decode_handle);
1337 void extract_flushWriteKeys (ZebraHandle zh, int final)
1338 /* optimizing: if final=1, and no files written yet */
1339 /* push the keys directly to merge, sidestepping the */
1340 /* temp file altogether. Speeds small updates */
1343 char out_fname[200];
1345 struct encode_info encode_info;
1346 int ptr_i = zh->reg->ptr_i;
1351 if (!zh->reg->key_buf || ptr_i <= 0)
1353 yaz_log (YLOG_DEBUG, " nothing to flush section=%d buf=%p i=%d",
1354 zh->reg->key_file_no, zh->reg->key_buf, ptr_i);
1355 yaz_log (YLOG_DEBUG, " buf=%p ",
1357 yaz_log (YLOG_DEBUG, " ptr=%d ",zh->reg->ptr_i);
1358 yaz_log (YLOG_DEBUG, " reg=%p ",zh->reg);
1363 (zh->reg->key_file_no)++;
1364 yaz_log (YLOG_LOG, "sorting section %d", (zh->reg->key_file_no));
1365 yaz_log (YLOG_DEBUG, " sort_buff at %p n=%d",
1366 zh->reg->key_buf + zh->reg->ptr_top - ptr_i,ptr_i);
1368 qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
1369 sizeof(char*), key_qsort_compare);
1371 /* zebra.cfg: tempfiles:
1372 Y: always use temp files (old way)
1373 A: use temp files, if more than one (auto)
1374 = if this is both the last and the first
1375 N: never bother with temp files (new) */
1377 temp_policy=toupper(res_get_def(zh->res,"tempfiles","auto")[0]);
1378 if (temp_policy != 'Y' && temp_policy != 'N' && temp_policy != 'A') {
1379 yaz_log (YLOG_WARN, "Illegal tempfiles setting '%c'. using 'Auto' ",
1384 if ( ( temp_policy =='N' ) || /* always from memory */
1385 ( ( temp_policy =='A' ) && /* automatic */
1386 (zh->reg->key_file_no == 1) && /* this is first time */
1387 (final) ) ) /* and last (=only) time */
1388 { /* go directly from memory */
1389 zh->reg->key_file_no =0; /* signal not to read files */
1390 zebra_index_merge(zh);
1392 zh->reg->key_buf_used = 0;
1396 /* Not doing directly from memory, write into a temp file */
1397 extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
1399 if (!(outf = fopen (out_fname, "wb")))
1401 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1404 yaz_log (YLOG_LOG, "writing section %d", zh->reg->key_file_no);
1405 prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1407 encode_key_init (&encode_info);
1408 encode_key_write (cp, &encode_info, outf);
1412 cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1413 if (strcmp (cp, prevcp))
1415 encode_key_flush ( &encode_info, outf);
1416 encode_key_init (&encode_info);
1417 encode_key_write (cp, &encode_info, outf);
1421 encode_key_write (cp + strlen(cp), &encode_info, outf);
1423 encode_key_flush ( &encode_info, outf);
1425 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1426 extract_get_fname_tmp (out_fname, key_file_no);
1428 if (!(outf = fopen (out_fname, "wb")))
1430 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1433 yaz_log (YLOG_LOG, "writing section %d", key_file_no);
1435 prevcp = key_buf[ptr_top-i];
1437 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1439 key_y_len = strlen(prevcp)+1;
1441 yaz_log (YLOG_LOG, "key_y_len: %2d %02x %02x %s",
1442 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1444 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1445 sizeof(char*), key_y_compare);
1446 cp = key_buf[ptr_top-ptr_i];
1448 encode_key_init (&encode_info);
1449 encode_key_write (cp, &encode_info, outf);
1452 cp = key_buf[ptr_top-ptr_i];
1453 encode_key_write (cp+key_y_len, &encode_info, outf);
1455 encode_key_flush ( &encode_info, outf);
1458 prevcp = key_buf[ptr_top-ptr_i];
1463 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fclose %s", out_fname);
1466 yaz_log (YLOG_LOG, "finished section %d", zh->reg->key_file_no);
1468 zh->reg->key_buf_used = 0;
1471 void extract_add_it_key (ZebraHandle zh,
1472 struct recKeys *keys,
1474 const char *str, int slen, struct it_key *key)
1477 const char *src = (char*) key;
1479 if (keys->buf_used+1024 > keys->buf_max)
1481 char *b = (char *) xmalloc (keys->buf_max += 128000);
1482 if (keys->buf_used > 0)
1483 memcpy (b, keys->buf, keys->buf_used);
1487 dst = keys->buf + keys->buf_used;
1489 iscz1_encode(keys->codec_handle, &dst, &src);
1494 memcpy (dst, str, slen);
1497 keys->buf_used = dst - keys->buf;
1500 ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh, struct recKeys *reckeys,
1501 zebra_snippets *snippets)
1503 void *decode_handle = iscz1_start();
1506 NMEM nmem = nmem_create();
1508 yaz_log(YLOG_LOG, "zebra_rec_keys_snippets buf=%p sz=%d", reckeys->buf,
1510 assert(reckeys->buf);
1511 while (off < reckeys->buf_used)
1513 const char *src = reckeys->buf + off;
1515 char *dst = (char*) &key;
1516 char dst_buf[IT_MAX_WORD];
1517 char *dst_term = dst_buf;
1519 iscz1_decode(decode_handle, &dst, &src);
1520 assert(key.len <= 4 && key.len > 2);
1522 seqno = (int) key.mem[key.len-1];
1524 zebra_term_untrans_iconv(zh, nmem, src[0], &dst_term, src+1);
1525 zebra_snippets_append(snippets, seqno, src[0], key.mem[0], dst_term);
1528 off = src - reckeys->buf;
1532 iscz1_stop(decode_handle);
1536 void print_rec_keys(ZebraHandle zh, struct recKeys *reckeys)
1538 void *decode_handle = iscz1_start();
1541 NMEM nmem = nmem_create();
1543 yaz_log(YLOG_LOG, "print_rec_keys buf=%p sz=%d", reckeys->buf,
1545 assert(reckeys->buf);
1546 while (off < reckeys->buf_used)
1548 const char *src = reckeys->buf + off;
1550 char *dst = (char*) &key;
1551 int attrSet, attrUse;
1552 char dst_buf[IT_MAX_WORD];
1553 char *dst_term = dst_buf;
1555 iscz1_decode(decode_handle, &dst, &src);
1556 assert(key.len <= 4 && key.len > 2);
1558 attrSet = (int) key.mem[0] >> 16;
1559 attrUse = (int) key.mem[0] & 65535;
1560 seqno = (int) key.mem[key.len-1];
1562 zebra_term_untrans_iconv(zh, nmem, src[0], &dst_term, src+1);
1564 yaz_log(YLOG_LOG, "ord=" ZINT_FORMAT " seqno=%d term=%s",
1565 key.mem[0], seqno, dst_term);
1568 off = src - reckeys->buf;
1572 iscz1_stop(decode_handle);
1575 void extract_add_index_string (RecWord *p, const char *str, int length)
1579 ZebraHandle zh = p->extractCtrl->handle;
1580 ZebraExplainInfo zei = zh->reg->zei;
1585 ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
1587 ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
1594 ch = zebraExplain_lookup_attr_su(zei, p->index_type,
1595 p->attrSet, p->attrUse);
1597 ch = zebraExplain_add_attr_su(zei, p->index_type,
1598 p->attrSet, p->attrUse);
1603 key.mem[1] = p->record_id;
1604 key.mem[2] = p->section_id;
1605 key.mem[3] = p->seqno;
1608 /* just for debugging .. */
1609 yaz_log(YLOG_LOG, "add: set=%d use=%d "
1610 "record_id=%lld section_id=%lld seqno=%lld",
1611 p->attrSet, p->attrUse, p->record_id, p->section_id, p->seqno);
1614 extract_add_it_key(p->extractCtrl->handle,
1621 static void extract_add_sort_string (RecWord *p, const char *str, int length)
1625 ZebraHandle zh = p->extractCtrl->handle;
1626 ZebraExplainInfo zei = zh->reg->zei;
1631 ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
1633 ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
1641 key.mem[1] = p->record_id;
1642 key.mem[2] = p->section_id;
1643 key.mem[3] = p->seqno;
1645 extract_add_it_key(p->extractCtrl->handle,
1651 static void extract_add_sort_string (RecWord *p, const char *str, int length)
1653 ZebraHandle zh = p->extractCtrl->handle;
1654 struct sortKeys *sk = &zh->reg->sortKeys;
1657 while (off < sk->buf_used)
1661 off += key_SU_decode(&set, sk->buf + off);
1662 off += key_SU_decode(&use, sk->buf + off);
1663 off += key_SU_decode(&slen, sk->buf + off);
1665 if (p->attrSet == set && p->attrUse == use)
1668 assert (off == sk->buf_used);
1670 if (sk->buf_used + IT_MAX_WORD > sk->buf_max)
1674 b = (char *) xmalloc (sk->buf_max += 128000);
1675 if (sk->buf_used > 0)
1676 memcpy (b, sk->buf, sk->buf_used);
1680 off += key_SU_encode(p->attrSet, sk->buf + off);
1681 off += key_SU_encode(p->attrUse, sk->buf + off);
1682 off += key_SU_encode(length, sk->buf + off);
1683 memcpy (sk->buf + off, str, length);
1684 sk->buf_used = off + length;
1688 void extract_add_string (RecWord *p, const char *string, int length)
1690 assert (length > 0);
1691 if (zebra_maps_is_sort (p->zebra_maps, p->index_type))
1692 extract_add_sort_string (p, string, length);
1694 extract_add_index_string (p, string, length);
1697 static void extract_add_incomplete_field (RecWord *p)
1699 const char *b = p->term_buf;
1700 int remain = p->term_len;
1701 const char **map = 0;
1703 yaz_log(YLOG_DEBUG, "Incomplete field, w='%.*s'", p->term_len, p->term_buf);
1706 map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, 0);
1710 char buf[IT_MAX_WORD+1];
1714 while (map && *map && **map == *CHR_SPACE)
1716 remain = p->term_len - (b - p->term_buf);
1718 map = zebra_maps_input(p->zebra_maps, p->index_type, &b,
1726 while (map && *map && **map != *CHR_SPACE)
1728 const char *cp = *map;
1730 while (i < IT_MAX_WORD && *cp)
1732 remain = p->term_len - (b - p->term_buf);
1734 map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, 0);
1740 extract_add_string (p, buf, i);
1745 static void extract_add_complete_field (RecWord *p)
1747 const char *b = p->term_buf;
1748 char buf[IT_MAX_WORD+1];
1749 const char **map = 0;
1750 int i = 0, remain = p->term_len;
1752 yaz_log(YLOG_DEBUG, "Complete field, w='%.*s'",
1753 p->term_len, p->term_buf);
1756 map = zebra_maps_input (p->zebra_maps, p->index_type, &b, remain, 1);
1758 while (remain > 0 && i < IT_MAX_WORD)
1760 while (map && *map && **map == *CHR_SPACE)
1762 remain = p->term_len - (b - p->term_buf);
1766 int first = i ? 0 : 1; /* first position */
1767 map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, first);
1775 if (i && i < IT_MAX_WORD)
1776 buf[i++] = *CHR_SPACE;
1777 while (map && *map && **map != *CHR_SPACE)
1779 const char *cp = *map;
1781 if (**map == *CHR_CUT)
1787 if (i >= IT_MAX_WORD)
1789 yaz_log(YLOG_DEBUG, "Adding string to index '%d'", **map);
1790 while (i < IT_MAX_WORD && *cp)
1793 remain = p->term_len - (b - p->term_buf);
1796 map = zebra_maps_input (p->zebra_maps, p->index_type, &b,
1805 extract_add_string (p, buf, i);
1808 void extract_token_add (RecWord *p)
1812 yaz_log (YLOG_LOG, "token_add "
1813 "reg_type=%c attrSet=%d attrUse=%d seqno=%d s=%.*s",
1814 p->reg_type, p->attrSet, p->attrUse, p->seqno, p->length,
1817 if ((wrbuf = zebra_replace(p->zebra_maps, p->index_type, 0,
1818 p->term_buf, p->term_len)))
1820 p->term_buf = wrbuf_buf(wrbuf);
1821 p->term_len = wrbuf_len(wrbuf);
1823 if (zebra_maps_is_complete (p->zebra_maps, p->index_type))
1824 extract_add_complete_field (p);
1826 extract_add_incomplete_field(p);
1829 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1830 void *buf, size_t sz)
1832 ZebraHandle zh = (ZebraHandle) p->handle;
1834 xfree(zh->store_data_buf);
1835 zh->store_data_buf = 0;
1836 zh->store_data_size = 0;
1839 zh->store_data_buf = xmalloc(sz);
1840 zh->store_data_size = sz;
1841 memcpy(zh->store_data_buf, buf, sz);
1845 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1847 ZebraHandle zh = (ZebraHandle) p->handle;
1848 xfree(zh->store_data_buf);
1849 zh->store_data_buf = 0;
1850 zh->store_data_size = 0;
1851 p->setStoreData = extract_set_store_data_cb;
1854 void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1856 ZebraHandle zh = (ZebraHandle) p->handle;
1857 zebraExplain_addSchema (zh->reg->zei, oid);
1861 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1862 int cmd, struct recKeys *reckeys)
1864 SortIdx sortIdx = zh->reg->sortIdx;
1865 void *decode_handle = iscz1_start();
1869 while (off < reckeys->buf_used)
1871 const char *src = reckeys->buf + off;
1873 char *dst = (char*) &key;
1875 iscz1_decode(decode_handle, &dst, &src);
1876 assert(key.len == 4);
1878 ch = (int) key.mem[0]; /* ordinal for field/use/attribute */
1880 sortIdx_type(sortIdx, ch);
1882 sortIdx_add(sortIdx, src, strlen(src));
1884 sortIdx_add(sortIdx, "", 1);
1889 off = src - reckeys->buf;
1891 assert (off == reckeys->buf_used);
1892 iscz1_stop(decode_handle);
1895 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1896 int cmd, struct sortKeys *sk)
1898 SortIdx sortIdx = zh->reg->sortIdx;
1901 sortIdx_sysno (sortIdx, sysno);
1903 while (off < sk->buf_used)
1907 off += key_SU_decode(&set, sk->buf + off);
1908 off += key_SU_decode(&use, sk->buf + off);
1909 off += key_SU_decode(&slen, sk->buf + off);
1911 sortIdx_type(sortIdx, use);
1913 sortIdx_add(sortIdx, sk->buf + off, slen);
1915 sortIdx_add(sortIdx, "", 1);
1921 void encode_key_init (struct encode_info *i)
1930 i->encode_handle = iscz1_start();
1936 /* this is the old encode_key_write
1937 * may be deleted once we are confident that the new works
1940 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1943 char *bp = i->buf, *bp0;
1944 const char *src = (char *) &key;
1946 /* copy term to output buf */
1947 while ((*bp++ = *k++))
1949 /* and copy & align key so we can mangle */
1950 memcpy (&key, k+1, sizeof(struct it_key)); /* *k is insert/delete */
1953 iscz1_encode(i->encode_handle, &bp, &src);
1954 *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */
1955 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1957 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
1962 void encode_key_flush (struct encode_info *i, FILE *outf)
1963 { /* dummy routine */
1964 iscz1_stop(i->encode_handle);
1969 /* new encode_key_write
1970 * The idea is to buffer one more key, and compare them
1971 * If we are going to delete and insert the same key,
1972 * we may as well not bother. Should make a difference in
1973 * updates with small modifications (appending to a mbox)
1975 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1980 if (*k) /* first time for new key */
1983 while ((*bp++ = *k++))
1985 i->keylen= bp - i->buf -1;
1986 assert(i->keylen+1+sizeof(struct it_key) < ENCODE_BUFLEN);
1990 bp=i->buf + i->keylen;
1995 memcpy (&key, k+1, sizeof(struct it_key));
1996 if (0==i->prevsys) /* no previous filter, fill up */
1998 i->prevsys=key.sysno;
1999 i->prevseq=key.seqno;
2002 else if ( (i->prevsys==key.sysno) &&
2003 (i->prevseq==key.seqno) &&
2005 { /* same numbers, diff cmd, they cancel out */
2009 { /* different stuff, write previous, move buf */
2010 bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
2011 if (i->sysno != i->prevsys)
2013 i->sysno = i->prevsys;
2016 else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
2018 return; /* ??? Filters some sort of duplicates away */
2019 /* ??? Can this ever happen -H 15oct02 */
2021 bp = encode_key_int (i->prevseq - i->seqno, bp);
2022 i->seqno = i->prevseq;
2023 i->cmd = i->prevcmd;
2024 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
2026 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
2029 i->keylen=0; /* ok, it's written, forget it */
2030 i->prevsys=key.sysno;
2031 i->prevseq=key.seqno;
2036 void encode_key_flush (struct encode_info *i, FILE *outf)
2037 { /* flush the last key from i */
2038 char *bp =i->buf + i->keylen;
2041 return; /* nothing to flush */
2044 bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
2045 if (i->sysno != i->prevsys)
2047 i->sysno = i->prevsys;
2050 else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
2052 return; /* ??? Filters some sort of duplicates away */
2053 /* ??? Can this ever happen -H 15oct02 */
2055 bp = encode_key_int (i->prevseq - i->seqno, bp);
2056 i->seqno = i->prevseq;
2057 i->cmd = i->prevcmd;
2058 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
2060 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
2063 i->keylen=0; /* ok, it's written, forget it */
2064 i->prevsys=0; /* forget the values too */