1 /* $Id: extract.c,v 1.197 2005-10-28 09:22:50 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 void searchRecordKey(ZebraHandle zh,
102 zebra_rec_keys_t reckeys,
103 int attrSetS, int attrUseS,
104 const char **ws, int ws_length)
109 for (i = 0; i<ws_length; i++)
112 ch = zebraExplain_lookup_attr_su_any_index(zh->reg->zei,
117 if (zebra_rec_keys_rewind(reckeys))
124 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
126 assert(key.len <= 4 && key.len > 2);
128 seqno = key.mem[key.len-1];
130 if (key.mem[0] == ch)
136 woff = seqno - startSeq;
137 if (woff >= 0 && woff < ws_length)
144 struct file_read_info {
145 off_t file_max; /* maximum offset so far */
146 off_t file_offset; /* current offset */
147 off_t file_moffset; /* offset of rec/rec boundary */
152 static struct file_read_info *file_read_start (int fd)
154 struct file_read_info *fi = (struct file_read_info *)
155 xmalloc (sizeof(*fi));
159 fi->file_moffset = 0;
165 static void file_read_stop (struct file_read_info *fi)
170 static off_t file_seek (void *handle, off_t offset)
172 struct file_read_info *p = (struct file_read_info *) handle;
173 p->file_offset = offset;
174 return lseek (p->fd, offset, SEEK_SET);
177 static off_t file_tell (void *handle)
179 struct file_read_info *p = (struct file_read_info *) handle;
180 return p->file_offset;
183 static int file_read (void *handle, char *buf, size_t count)
185 struct file_read_info *p = (struct file_read_info *) handle;
188 r = read (fd, buf, count);
192 if (p->file_offset > p->file_max)
193 p->file_max = p->file_offset;
198 static void file_end (void *handle, off_t offset)
200 struct file_read_info *p = (struct file_read_info *) handle;
202 if (offset != p->file_moffset)
204 p->file_moffset = offset;
209 static char *fileMatchStr (ZebraHandle zh,
210 zebra_rec_keys_t reckeys,
211 const char *fname, const char *spec)
213 static char dstBuf[2048]; /* static here ??? */
215 const char *s = spec;
219 while (*s == ' ' || *s == '\t')
226 char attset_str[64], attname_str[64];
227 data1_attset *attset;
230 int attSet = 1, attUse = 1;
234 for (i = 0; *s && *s != ',' && *s != ')'; s++)
236 attset_str[i++] = *s;
237 attset_str[i] = '\0';
242 for (i = 0; *s && *s != ')'; s++)
244 attname_str[i++] = *s;
245 attname_str[i] = '\0';
248 if ((attset = data1_get_attset (zh->reg->dh, attset_str)))
251 attSet = attset->reference;
252 att = data1_getattbyname(zh->reg->dh, attset, attname_str);
256 attUse = atoi (attname_str);
258 searchRecordKey (zh, reckeys, attSet, attUse, ws, 32);
262 for (i = 0; i<32; i++)
267 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
268 spec, zh->m_group ? zh->m_group : "none");
273 for (i = 0; i<32; i++)
274 if (matchFlag[i] && ws[i])
282 dst += strlen(ws[i]);
286 yaz_log (YLOG_WARN, "Record didn't contain match"
287 " fields in (%s,%s)", attset_str, attname_str);
295 const char *spec_src = NULL;
296 const char *s1 = ++s;
297 while (*s1 && *s1 != ' ' && *s1 != '\t')
303 memcpy (special, s, spec_len);
304 special[spec_len] = '\0';
307 if (!strcmp (special, "group"))
308 spec_src = zh->m_group;
309 else if (!strcmp (special, "database"))
310 spec_src = zh->basenames[0];
311 else if (!strcmp (special, "filename")) {
314 else if (!strcmp (special, "type"))
315 spec_src = zh->m_record_type;
320 strcpy (dst, spec_src);
321 dst += strlen (spec_src);
324 else if (*s == '\"' || *s == '\'')
326 int stopMarker = *s++;
330 while (*s && *s != stopMarker)
333 tmpString[i++] = *s++;
338 strcpy (dst, tmpString);
339 dst += strlen (tmpString);
343 yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
344 spec, zh->m_group ? zh->m_group : "none");
351 yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
352 fname, zh->m_group ? zh->m_group : "none");
359 struct recordLogInfo {
362 struct recordGroup *rGroup;
365 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
368 for (i = 0; i<256; i++)
370 if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
375 ctrl->zebra_maps = zh->reg->zebra_maps;
376 ctrl->flagShowRecords = !zh->m_flag_rw;
379 static int file_extract_record(ZebraHandle zh,
380 SYSNO *sysno, const char *fname,
382 struct file_read_info *fi,
385 void *recTypeClientData)
387 RecordAttr *recordAttr;
389 const char *matchStr = 0;
392 off_t recordOffset = 0;
393 struct recExtractCtrl extractCtrl;
395 /* announce database */
396 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
398 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
399 zh->m_explain_database))
405 /* we are going to read from a file, so prepare the extraction */
406 zebra_rec_keys_reset(zh->reg->keys);
409 zebra_rec_keys_reset(zh->reg->sortKeys);
411 zh->reg->sortKeys.buf_used = 0;
413 recordOffset = fi->file_moffset;
414 extractCtrl.handle = zh;
415 extractCtrl.offset = fi->file_moffset;
416 extractCtrl.readf = file_read;
417 extractCtrl.seekf = file_seek;
418 extractCtrl.tellf = file_tell;
419 extractCtrl.endf = file_end;
421 extractCtrl.init = extract_init;
422 extractCtrl.tokenAdd = extract_token_add;
423 extractCtrl.schemaAdd = extract_schema_add;
424 extractCtrl.dh = zh->reg->dh;
425 extractCtrl.match_criteria[0] = '\0';
426 extractCtrl.staticrank = 0;
428 extractCtrl.first_record = fi->file_offset ? 0 : 1;
430 extract_set_store_data_prepare(&extractCtrl);
432 init_extractCtrl(zh, &extractCtrl);
435 printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset);
439 sprintf (msg, "%s:" PRINTF_OFF_T , fname, recordOffset);
440 yaz_log_init_prefix2 (msg);
443 r = (*recType->extract)(recTypeClientData, &extractCtrl);
445 yaz_log_init_prefix2 (0);
446 if (r == RECCTRL_EXTRACT_EOF)
448 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
450 /* error occured during extraction ... */
452 zh->records_processed < zh->m_file_verbose_limit)
454 yaz_log (YLOG_WARN, "fail %s %s " PRINTF_OFF_T, zh->m_record_type,
455 fname, recordOffset);
459 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
461 /* error occured during extraction ... */
463 zh->records_processed < zh->m_file_verbose_limit)
465 yaz_log (YLOG_WARN, "no filter for %s %s "
466 PRINTF_OFF_T, zh->m_record_type,
467 fname, recordOffset);
471 if (extractCtrl.match_criteria[0])
472 matchStr = extractCtrl.match_criteria;
475 /* perform match if sysno not known and if match criteria is specified */
481 if (matchStr == 0 && zh->m_record_id && *zh->m_record_id)
483 matchStr = fileMatchStr (zh, zh->reg->keys, fname,
487 yaz_log(YLOG_WARN, "Bad match criteria");
493 char *rinfo = dict_lookup (zh->reg->matchDict, matchStr);
496 assert(*rinfo == sizeof(*sysno));
497 memcpy (sysno, rinfo+1, sizeof(*sysno));
501 if (! *sysno && zebra_rec_keys_empty(zh->reg->keys) )
503 /* the extraction process returned no information - the record
504 is probably empty - unless flagShowRecords is in use */
508 if (zh->records_processed < zh->m_file_verbose_limit)
509 yaz_log (YLOG_WARN, "empty %s %s " PRINTF_OFF_T, zh->m_record_type,
510 fname, recordOffset);
519 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T, zh->m_record_type,
520 fname, recordOffset);
521 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
524 if (zh->records_processed < zh->m_file_verbose_limit)
525 yaz_log (YLOG_LOG, "add %s %s " PRINTF_OFF_T, zh->m_record_type,
526 fname, recordOffset);
527 rec = rec_new (zh->reg->records);
531 recordAttr = rec_init_attr (zh->reg->zei, rec);
532 recordAttr->staticrank = extractCtrl.staticrank;
536 dict_insert (zh->reg->matchDict, matchStr, sizeof(*sysno), sysno);
538 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
539 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
540 recordAttr->staticrank);
541 zh->records_inserted++;
545 /* record already exists */
546 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
549 zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
551 struct sortKeys sortKeys;
554 rec = rec_get (zh->reg->records, *sysno);
557 recordAttr = rec_init_attr (zh->reg->zei, rec);
559 zebra_rec_keys_set_buf(delkeys,
560 rec->info[recInfo_delKeys],
561 rec->size[recInfo_delKeys],
565 zebra_rec_keys_set_buf(sortKeys,
566 rec->info[recInfo_sortKeys],
567 rec->size[recInfo_sortKeys],
569 extract_flushSortKeys (zh, *sysno, 0, sortKeys);
571 sortKeys.buf_used = rec->size[recInfo_sortKeys];
572 sortKeys.buf = rec->info[recInfo_sortKeys];
573 extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
576 extract_flushRecordKeys (zh, *sysno, 0, delkeys,
577 recordAttr->staticrank); /* old values */
580 /* record going to be deleted */
581 if (zebra_rec_keys_empty(delkeys))
583 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
584 zh->m_record_type, fname, recordOffset);
585 yaz_log (YLOG_WARN, "cannot delete file above, storeKeys false");
589 if (zh->records_processed < zh->m_file_verbose_limit)
590 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
591 zh->m_record_type, fname, recordOffset);
592 zh->records_deleted++;
594 dict_delete (zh->reg->matchDict, matchStr);
595 rec_del (zh->reg->records, &rec);
603 /* record going to be updated */
604 if (zebra_rec_keys_empty(delkeys))
606 yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
607 zh->m_record_type, fname, recordOffset);
608 yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
612 /* flush new keys for sort&search etc */
613 if (zh->records_processed < zh->m_file_verbose_limit)
614 yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
615 zh->m_record_type, fname, recordOffset);
616 recordAttr->staticrank = extractCtrl.staticrank;
617 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
618 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
619 recordAttr->staticrank);
620 zh->records_updated++;
623 zebra_rec_keys_close(delkeys);
625 zebra_rec_keys_close(sortKeys);
628 /* update file type */
629 xfree (rec->info[recInfo_fileType]);
630 rec->info[recInfo_fileType] =
631 rec_strdup (zh->m_record_type, &rec->size[recInfo_fileType]);
633 /* update filename */
634 xfree (rec->info[recInfo_filename]);
635 rec->info[recInfo_filename] =
636 rec_strdup (fname, &rec->size[recInfo_filename]);
638 /* update delete keys */
639 xfree (rec->info[recInfo_delKeys]);
640 if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
642 zebra_rec_keys_get_buf(zh->reg->keys,
643 &rec->info[recInfo_delKeys],
644 &rec->size[recInfo_delKeys]);
648 rec->info[recInfo_delKeys] = NULL;
649 rec->size[recInfo_delKeys] = 0;
652 /* update sort keys */
653 xfree (rec->info[recInfo_sortKeys]);
655 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
656 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
657 zh->reg->sortKeys.buf = NULL;
658 zh->reg->sortKeys.buf_max = 0;
660 /* save file size of original record */
661 zebraExplain_recordBytesIncrement (zh->reg->zei,
662 - recordAttr->recordSize);
663 recordAttr->recordSize = fi->file_moffset - recordOffset;
664 if (!recordAttr->recordSize)
665 recordAttr->recordSize = fi->file_max - recordOffset;
666 zebraExplain_recordBytesIncrement (zh->reg->zei,
667 recordAttr->recordSize);
669 /* set run-number for this record */
670 recordAttr->runNumber = zebraExplain_runNumberIncrement (zh->reg->zei,
673 /* update store data */
674 xfree (rec->info[recInfo_storeData]);
675 if (zh->store_data_buf)
677 rec->size[recInfo_storeData] = zh->store_data_size;
678 rec->info[recInfo_storeData] = zh->store_data_buf;
679 zh->store_data_buf = 0;
681 else if (zh->m_store_data)
683 rec->size[recInfo_storeData] = recordAttr->recordSize;
684 rec->info[recInfo_storeData] = (char *)
685 xmalloc (recordAttr->recordSize);
686 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
688 yaz_log (YLOG_ERRNO|YLOG_FATAL, "seek to " PRINTF_OFF_T " in %s",
689 recordOffset, fname);
692 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
693 < recordAttr->recordSize)
695 yaz_log (YLOG_ERRNO|YLOG_FATAL, "read %d bytes of %s",
696 recordAttr->recordSize, fname);
702 rec->info[recInfo_storeData] = NULL;
703 rec->size[recInfo_storeData] = 0;
705 /* update database name */
706 xfree (rec->info[recInfo_databaseName]);
707 rec->info[recInfo_databaseName] =
708 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
711 recordAttr->recordOffset = recordOffset;
713 /* commit this record */
714 rec_put (zh->reg->records, &rec);
719 int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
726 struct file_read_info *fi;
727 const char *original_record_type = 0;
729 void *recTypeClientData;
731 if (!zh->m_group || !*zh->m_group)
734 sprintf (gprefix, "%s.", zh->m_group);
736 yaz_log (YLOG_DEBUG, "fileExtract %s", fname);
738 /* determine file extension */
740 for (i = strlen(fname); --i >= 0; )
743 else if (fname[i] == '.')
745 strcpy (ext, fname+i+1);
748 /* determine file type - depending on extension */
749 original_record_type = zh->m_record_type;
750 if (!zh->m_record_type)
752 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
753 zh->m_record_type = res_get (zh->res, ext_res);
755 if (!zh->m_record_type)
757 if (zh->records_processed < zh->m_file_verbose_limit)
758 yaz_log (YLOG_LOG, "? %s", fname);
761 /* determine match criteria */
762 if (!zh->m_record_id)
764 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
765 zh->m_record_id = res_get (zh->res, ext_res);
769 recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
770 &recTypeClientData)))
772 yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
776 switch(recType->version)
781 yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
783 if (sysno && deleteFlag)
789 if (zh->path_reg && !yaz_is_abspath (fname))
791 strcpy (full_rep, zh->path_reg);
792 strcat (full_rep, "/");
793 strcat (full_rep, fname);
796 strcpy (full_rep, fname);
799 if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
801 yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
802 zh->m_record_type = original_record_type;
806 fi = file_read_start (fd);
809 fi->file_moffset = fi->file_offset;
810 fi->file_more = 0; /* file_end not called (yet) */
811 r = file_extract_record (zh, sysno, fname, deleteFlag, fi, 1,
812 recType, recTypeClientData);
814 { /* file_end has been called so reset offset .. */
815 fi->file_offset = fi->file_moffset;
816 lseek(fi->fd, fi->file_moffset, SEEK_SET);
823 zh->m_record_type = original_record_type;
828 If sysno is provided, then it's used to identify the reocord.
829 If not, and match_criteria is provided, then sysno is guessed
830 If not, and a record is provided, then sysno is got from there
833 ZEBRA_RES buffer_extract_record(ZebraHandle zh,
834 const char *buf, size_t buf_size,
837 const char *recordType,
839 const char *match_criteria,
844 RecordAttr *recordAttr;
845 struct recExtractCtrl extractCtrl;
847 const char *matchStr = 0;
848 RecType recType = NULL;
851 long recordOffset = 0;
852 struct zebra_fetch_control fc;
853 const char *pr_fname = fname; /* filename to print .. */
854 int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
857 pr_fname = "<no file>"; /* make it printable if file is omitted */
860 fc.record_int_buf = buf;
861 fc.record_int_len = buf_size;
862 fc.record_int_pos = 0;
864 fc.record_offset = 0;
866 extractCtrl.offset = 0;
867 extractCtrl.readf = zebra_record_int_read;
868 extractCtrl.seekf = zebra_record_int_seek;
869 extractCtrl.tellf = zebra_record_int_tell;
870 extractCtrl.endf = zebra_record_int_end;
871 extractCtrl.first_record = 1;
872 extractCtrl.fh = &fc;
874 zebra_rec_keys_reset(zh->reg->keys);
877 zebra_rec_keys_reset(zh->reg->sortKeys);
879 zh->reg->sortKeys.buf_used = 0;
881 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
883 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
884 zh->m_explain_database))
888 if (recordType && *recordType)
890 yaz_log (YLOG_DEBUG, "Record type explicitly specified: %s", recordType);
891 recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
896 if (!(zh->m_record_type))
898 yaz_log (YLOG_WARN, "No such record type defined");
901 yaz_log (YLOG_DEBUG, "Get record type from rgroup: %s",zh->m_record_type);
902 recType = recType_byName (zh->reg->recTypes, zh->res,
903 zh->m_record_type, &clientData);
904 recordType = zh->m_record_type;
909 yaz_log (YLOG_WARN, "No such record type: %s", zh->m_record_type);
913 extractCtrl.init = extract_init;
914 extractCtrl.tokenAdd = extract_token_add;
915 extractCtrl.schemaAdd = extract_schema_add;
916 extractCtrl.dh = zh->reg->dh;
917 extractCtrl.handle = zh;
918 extractCtrl.match_criteria[0] = '\0';
919 extractCtrl.staticrank = 0;
921 init_extractCtrl(zh, &extractCtrl);
923 extract_set_store_data_prepare(&extractCtrl);
925 r = (*recType->extract)(clientData, &extractCtrl);
927 if (r == RECCTRL_EXTRACT_EOF)
929 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
931 /* error occured during extraction ... */
932 yaz_log (YLOG_WARN, "extract error: generic");
935 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
937 /* error occured during extraction ... */
938 yaz_log (YLOG_WARN, "extract error: no such filter");
944 if (extractCtrl.match_criteria[0])
945 match_criteria = extractCtrl.match_criteria;
949 if (match_criteria && *match_criteria) {
950 matchStr = match_criteria;
952 if (zh->m_record_id && *zh->m_record_id) {
953 matchStr = fileMatchStr (zh, zh->reg->keys, pr_fname,
957 yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
963 rinfo = dict_lookup (zh->reg->matchDict, matchStr);
966 assert(*rinfo == sizeof(*sysno));
967 memcpy (sysno, rinfo+1, sizeof(*sysno));
971 if (zebra_rec_keys_empty(zh->reg->keys))
973 /* the extraction process returned no information - the record
974 is probably empty - unless flagShowRecords is in use */
985 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
986 pr_fname, (long) recordOffset);
987 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
991 yaz_log (YLOG_LOG, "add %s %s %ld", recordType, pr_fname,
992 (long) recordOffset);
993 rec = rec_new (zh->reg->records);
997 recordAttr = rec_init_attr (zh->reg->zei, rec);
998 recordAttr->staticrank = extractCtrl.staticrank;
1002 dict_insert (zh->reg->matchDict, matchStr,
1003 sizeof(*sysno), sysno);
1005 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1008 print_rec_keys(zh, zh->reg->keys);
1010 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
1011 recordAttr->staticrank);
1012 zh->records_inserted++;
1016 /* record already exists */
1017 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1019 zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
1021 struct sortKeys sortKeys;
1027 yaz_log (YLOG_LOG, "skipped %s %s %ld",
1028 recordType, pr_fname, (long) recordOffset);
1033 rec = rec_get (zh->reg->records, *sysno);
1036 recordAttr = rec_init_attr (zh->reg->zei, rec);
1038 zebra_rec_keys_set_buf(delkeys,
1039 rec->info[recInfo_delKeys],
1040 rec->size[recInfo_delKeys],
1042 sortKeys.buf_used = rec->size[recInfo_sortKeys];
1043 sortKeys.buf = rec->info[recInfo_sortKeys];
1046 extract_flushSortKeys (zh, *sysno, 0, sortKeys);
1048 extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
1050 extract_flushRecordKeys (zh, *sysno, 0, delkeys,
1051 recordAttr->staticrank);
1054 /* record going to be deleted */
1055 if (zebra_rec_keys_empty(delkeys))
1059 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1060 pr_fname, (long) recordOffset);
1061 yaz_log (YLOG_WARN, "cannot delete file above, "
1068 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1069 pr_fname, (long) recordOffset);
1070 zh->records_deleted++;
1072 dict_delete (zh->reg->matchDict, matchStr);
1073 rec_del (zh->reg->records, &rec);
1081 /* record going to be updated */
1082 if (zebra_rec_keys_empty(delkeys))
1086 yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
1087 pr_fname, (long) recordOffset);
1088 yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
1094 yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
1095 pr_fname, (long) recordOffset);
1096 recordAttr->staticrank = extractCtrl.staticrank;
1097 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1098 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
1099 recordAttr->staticrank);
1100 zh->records_updated++;
1103 zebra_rec_keys_close(delkeys);
1105 zebra_rec_keys_close(sortKeys);
1108 /* update file type */
1109 xfree (rec->info[recInfo_fileType]);
1110 rec->info[recInfo_fileType] =
1111 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1113 /* update filename */
1114 xfree (rec->info[recInfo_filename]);
1115 rec->info[recInfo_filename] =
1116 rec_strdup (fname, &rec->size[recInfo_filename]);
1118 /* update delete keys */
1119 xfree (rec->info[recInfo_delKeys]);
1120 if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
1122 zebra_rec_keys_get_buf(zh->reg->keys,
1123 &rec->info[recInfo_delKeys],
1124 &rec->size[recInfo_delKeys]);
1128 rec->info[recInfo_delKeys] = NULL;
1129 rec->size[recInfo_delKeys] = 0;
1131 /* update sort keys */
1132 xfree (rec->info[recInfo_sortKeys]);
1134 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
1135 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
1136 zh->reg->sortKeys.buf = NULL;
1137 zh->reg->sortKeys.buf_max = 0;
1139 /* save file size of original record */
1140 zebraExplain_recordBytesIncrement (zh->reg->zei,
1141 - recordAttr->recordSize);
1143 recordAttr->recordSize = fi->file_moffset - recordOffset;
1144 if (!recordAttr->recordSize)
1145 recordAttr->recordSize = fi->file_max - recordOffset;
1147 recordAttr->recordSize = buf_size;
1149 zebraExplain_recordBytesIncrement (zh->reg->zei,
1150 recordAttr->recordSize);
1152 /* set run-number for this record */
1153 recordAttr->runNumber =
1154 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
1156 /* update store data */
1157 xfree (rec->info[recInfo_storeData]);
1159 /* update store data */
1160 if (zh->store_data_buf)
1162 rec->size[recInfo_storeData] = zh->store_data_size;
1163 rec->info[recInfo_storeData] = zh->store_data_buf;
1164 zh->store_data_buf = 0;
1166 else if (zh->m_store_data)
1168 rec->size[recInfo_storeData] = recordAttr->recordSize;
1169 rec->info[recInfo_storeData] = (char *)
1170 xmalloc (recordAttr->recordSize);
1171 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1175 rec->info[recInfo_storeData] = NULL;
1176 rec->size[recInfo_storeData] = 0;
1178 /* update database name */
1179 xfree (rec->info[recInfo_databaseName]);
1180 rec->info[recInfo_databaseName] =
1181 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
1184 recordAttr->recordOffset = recordOffset;
1186 /* commit this record */
1187 rec_put (zh->reg->records, &rec);
1192 int explain_extract (void *handle, Record rec, data1_node *n)
1194 ZebraHandle zh = (ZebraHandle) handle;
1195 struct recExtractCtrl extractCtrl;
1197 if (zebraExplain_curDatabase (zh->reg->zei,
1198 rec->info[recInfo_databaseName]))
1201 if (zebraExplain_newDatabase (zh->reg->zei,
1202 rec->info[recInfo_databaseName], 0))
1206 zebra_rec_keys_reset(zh->reg->keys);
1209 zebra_rec_keys_reset(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.staticrank = 0;
1223 extractCtrl.handle = handle;
1224 extractCtrl.first_record = 1;
1226 extract_set_store_data_prepare(&extractCtrl);
1229 grs_extract_tree(&extractCtrl, n);
1231 if (rec->size[recInfo_delKeys])
1233 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1236 zebra_rec_keys_t sortkeys = zzebra_rec_keys_open();
1238 struct sortKeys sortkeys;
1241 zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
1242 rec->size[recInfo_delKeys],
1244 extract_flushRecordKeys (zh, rec->sysno, 0, delkeys, 0);
1245 zebra_rec_keys_close(delkeys);
1247 zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
1248 rec->size[recInfo_sortKeys],
1251 extract_flushSortKeys (zh, rec->sysno, 0, sortkeys);
1252 zebra_rec_keys_close(sortkeys);
1254 sortkeys.buf_used = rec->size[recInfo_sortKeys];
1255 sortkeys.buf = rec->info[recInfo_sortKeys];
1256 extract_flushSortKeys (zh, rec->sysno, 0, &sortkeys);
1259 extract_flushRecordKeys (zh, rec->sysno, 1, zh->reg->keys, 0);
1261 extract_flushSortKeys (zh, rec->sysno, 1, zh->reg->sortKeys);
1263 extract_flushSortKeys (zh, rec->sysno, 1, &zh->reg->sortKeys);
1266 xfree (rec->info[recInfo_delKeys]);
1267 zebra_rec_keys_get_buf(zh->reg->keys,
1268 &rec->info[recInfo_delKeys],
1269 &rec->size[recInfo_delKeys]);
1271 xfree (rec->info[recInfo_sortKeys]);
1272 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
1273 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
1274 zh->reg->sortKeys.buf = NULL;
1275 zh->reg->sortKeys.buf_max = 0;
1280 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1282 zebra_rec_keys_t reckeys,
1285 ZebraExplainInfo zei = zh->reg->zei;
1287 if (!zh->reg->key_buf)
1289 int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
1292 yaz_log(YLOG_WARN, "Invalid memory setting, using default 8 MB");
1295 /* FIXME: That "8" should be in a default settings include */
1296 /* not hard-coded here! -H */
1297 zh->reg->key_buf = (char**) xmalloc (mem);
1298 zh->reg->ptr_top = mem/sizeof(char*);
1300 zh->reg->key_buf_used = 0;
1301 zh->reg->key_file_no = 0;
1303 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1305 if (zebra_rec_keys_rewind(reckeys))
1309 struct it_key key_in;
1310 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1313 struct it_key key_out;
1314 zint *keyp = key_out.mem;
1316 assert(key_in.len == 4);
1318 /* check for buffer overflow */
1319 if (zh->reg->key_buf_used + 1024 >
1320 (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
1321 extract_flushWriteKeys (zh, 0);
1324 assert(zh->reg->ptr_i > 0);
1325 (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
1326 (char*)zh->reg->key_buf + zh->reg->key_buf_used;
1328 /* encode the ordinal value (field/use/attribute) .. */
1329 ch = (int) key_in.mem[0];
1330 zh->reg->key_buf_used +=
1331 key_SU_encode(ch, (char*)zh->reg->key_buf +
1332 zh->reg->key_buf_used);
1334 /* copy the 0-terminated stuff from str to output */
1335 memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used, str, slen);
1336 zh->reg->key_buf_used += slen;
1337 ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = '\0';
1339 /* the delete/insert indicator */
1340 ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = cmd;
1342 if (zh->m_staticrank) /* rank config enabled ? */
1344 *keyp++ = staticrank;
1350 if (key_in.mem[1]) /* filter specified record ID */
1351 *keyp++ = key_in.mem[1];
1354 *keyp++ = key_in.mem[2]; /* section_id */
1355 *keyp++ = key_in.mem[3]; /* sequence .. */
1357 memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used,
1358 &key_out, sizeof(key_out));
1359 (zh->reg->key_buf_used) += sizeof(key_out);
1364 void extract_flushWriteKeys (ZebraHandle zh, int final)
1365 /* optimizing: if final=1, and no files written yet */
1366 /* push the keys directly to merge, sidestepping the */
1367 /* temp file altogether. Speeds small updates */
1370 char out_fname[200];
1372 struct encode_info encode_info;
1373 int ptr_i = zh->reg->ptr_i;
1378 if (!zh->reg->key_buf || ptr_i <= 0)
1380 yaz_log (YLOG_DEBUG, " nothing to flush section=%d buf=%p i=%d",
1381 zh->reg->key_file_no, zh->reg->key_buf, ptr_i);
1382 yaz_log (YLOG_DEBUG, " buf=%p ",
1384 yaz_log (YLOG_DEBUG, " ptr=%d ",zh->reg->ptr_i);
1385 yaz_log (YLOG_DEBUG, " reg=%p ",zh->reg);
1390 (zh->reg->key_file_no)++;
1391 yaz_log (YLOG_LOG, "sorting section %d", (zh->reg->key_file_no));
1392 yaz_log (YLOG_DEBUG, " sort_buff at %p n=%d",
1393 zh->reg->key_buf + zh->reg->ptr_top - ptr_i,ptr_i);
1395 qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
1396 sizeof(char*), key_qsort_compare);
1398 /* zebra.cfg: tempfiles:
1399 Y: always use temp files (old way)
1400 A: use temp files, if more than one (auto)
1401 = if this is both the last and the first
1402 N: never bother with temp files (new) */
1404 temp_policy=toupper(res_get_def(zh->res,"tempfiles","auto")[0]);
1405 if (temp_policy != 'Y' && temp_policy != 'N' && temp_policy != 'A') {
1406 yaz_log (YLOG_WARN, "Illegal tempfiles setting '%c'. using 'Auto' ",
1411 if ( ( temp_policy =='N' ) || /* always from memory */
1412 ( ( temp_policy =='A' ) && /* automatic */
1413 (zh->reg->key_file_no == 1) && /* this is first time */
1414 (final) ) ) /* and last (=only) time */
1415 { /* go directly from memory */
1416 zh->reg->key_file_no =0; /* signal not to read files */
1417 zebra_index_merge(zh);
1419 zh->reg->key_buf_used = 0;
1423 /* Not doing directly from memory, write into a temp file */
1424 extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
1426 if (!(outf = fopen (out_fname, "wb")))
1428 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1431 yaz_log (YLOG_LOG, "writing section %d", zh->reg->key_file_no);
1432 prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1434 encode_key_init (&encode_info);
1435 encode_key_write (cp, &encode_info, outf);
1439 cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1440 if (strcmp (cp, prevcp))
1442 encode_key_flush ( &encode_info, outf);
1443 encode_key_init (&encode_info);
1444 encode_key_write (cp, &encode_info, outf);
1448 encode_key_write (cp + strlen(cp), &encode_info, outf);
1450 encode_key_flush ( &encode_info, outf);
1452 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1453 extract_get_fname_tmp (out_fname, key_file_no);
1455 if (!(outf = fopen (out_fname, "wb")))
1457 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1460 yaz_log (YLOG_LOG, "writing section %d", key_file_no);
1462 prevcp = key_buf[ptr_top-i];
1464 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1466 key_y_len = strlen(prevcp)+1;
1468 yaz_log (YLOG_LOG, "key_y_len: %2d %02x %02x %s",
1469 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1471 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1472 sizeof(char*), key_y_compare);
1473 cp = key_buf[ptr_top-ptr_i];
1475 encode_key_init (&encode_info);
1476 encode_key_write (cp, &encode_info, outf);
1479 cp = key_buf[ptr_top-ptr_i];
1480 encode_key_write (cp+key_y_len, &encode_info, outf);
1482 encode_key_flush ( &encode_info, outf);
1485 prevcp = key_buf[ptr_top-ptr_i];
1490 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fclose %s", out_fname);
1493 yaz_log (YLOG_LOG, "finished section %d", zh->reg->key_file_no);
1495 zh->reg->key_buf_used = 0;
1498 void extract_add_it_key (ZebraHandle zh,
1499 zebra_rec_keys_t *keys,
1501 const char *str, int slen, struct it_key *key)
1503 zebra_rec_keys_write(*keys, reg_type, str, slen, key);
1506 ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh,
1507 zebra_rec_keys_t reckeys,
1508 zebra_snippets *snippets)
1510 NMEM nmem = nmem_create();
1511 if (zebra_rec_keys_rewind(reckeys))
1516 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1518 char dst_buf[IT_MAX_WORD];
1519 char *dst_term = dst_buf;
1522 assert(key.len <= 4 && key.len > 2);
1523 seqno = (int) key.mem[key.len-1];
1526 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1527 0/* db */, 0/* set */, 0/* use */);
1529 zebra_term_untrans_iconv(zh, nmem, index_type,
1531 zebra_snippets_append(snippets, seqno, ord, dst_term);
1539 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1541 yaz_log(YLOG_LOG, "print_rec_keys");
1542 if (zebra_rec_keys_rewind(reckeys))
1547 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1549 char dst_buf[IT_MAX_WORD];
1553 assert(key.len <= 4 && key.len > 2);
1555 zebraExplain_lookup_ord(zh->reg->zei,
1556 key.mem[0], &index_type, &db, 0, 0);
1558 seqno = (int) key.mem[key.len-1];
1560 zebra_term_untrans(zh, index_type, dst_buf, str);
1562 yaz_log(YLOG_LOG, "ord=" ZINT_FORMAT " seqno=%d term=%s",
1563 key.mem[0], seqno, dst_buf);
1568 void extract_add_index_string (RecWord *p, const char *str, int length)
1572 ZebraHandle zh = p->extractCtrl->handle;
1573 ZebraExplainInfo zei = zh->reg->zei;
1578 ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
1580 ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
1587 ch = zebraExplain_lookup_attr_su(zei, p->index_type,
1588 p->attrSet, p->attrUse);
1590 ch = zebraExplain_add_attr_su(zei, p->index_type,
1591 p->attrSet, p->attrUse);
1596 key.mem[1] = p->record_id;
1597 key.mem[2] = p->section_id;
1598 key.mem[3] = p->seqno;
1601 /* just for debugging .. */
1602 yaz_log(YLOG_LOG, "add: set=%d use=%d "
1603 "record_id=%lld section_id=%lld seqno=%lld",
1604 p->attrSet, p->attrUse, p->record_id, p->section_id, p->seqno);
1607 extract_add_it_key(p->extractCtrl->handle,
1614 static void extract_add_sort_string (RecWord *p, const char *str, int length)
1618 ZebraHandle zh = p->extractCtrl->handle;
1619 ZebraExplainInfo zei = zh->reg->zei;
1624 ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
1626 ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
1634 key.mem[1] = p->record_id;
1635 key.mem[2] = p->section_id;
1636 key.mem[3] = p->seqno;
1638 extract_add_it_key(p->extractCtrl->handle,
1644 static void extract_add_sort_string (RecWord *p, const char *str, int length)
1646 ZebraHandle zh = p->extractCtrl->handle;
1647 struct sortKeys *sk = &zh->reg->sortKeys;
1650 while (off < sk->buf_used)
1654 off += key_SU_decode(&set, (unsigned char *) sk->buf + off);
1655 off += key_SU_decode(&use, (unsigned char *) sk->buf + off);
1656 off += key_SU_decode(&slen, (unsigned char *) sk->buf + off);
1658 if (p->attrSet == set && p->attrUse == use)
1661 assert (off == sk->buf_used);
1663 if (sk->buf_used + IT_MAX_WORD > sk->buf_max)
1667 b = (char *) xmalloc (sk->buf_max += 128000);
1668 if (sk->buf_used > 0)
1669 memcpy (b, sk->buf, sk->buf_used);
1673 off += key_SU_encode(p->attrSet, sk->buf + off);
1674 off += key_SU_encode(p->attrUse, sk->buf + off);
1675 off += key_SU_encode(length, sk->buf + off);
1676 memcpy (sk->buf + off, str, length);
1677 sk->buf_used = off + length;
1681 void extract_add_string (RecWord *p, const char *string, int length)
1683 assert (length > 0);
1684 if (zebra_maps_is_sort (p->zebra_maps, p->index_type))
1685 extract_add_sort_string (p, string, length);
1687 extract_add_index_string (p, string, length);
1690 static void extract_add_incomplete_field (RecWord *p)
1692 const char *b = p->term_buf;
1693 int remain = p->term_len;
1694 const char **map = 0;
1696 yaz_log(YLOG_DEBUG, "Incomplete field, w='%.*s'", p->term_len, p->term_buf);
1699 map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, 0);
1703 char buf[IT_MAX_WORD+1];
1707 while (map && *map && **map == *CHR_SPACE)
1709 remain = p->term_len - (b - p->term_buf);
1711 map = zebra_maps_input(p->zebra_maps, p->index_type, &b,
1719 while (map && *map && **map != *CHR_SPACE)
1721 const char *cp = *map;
1723 while (i < IT_MAX_WORD && *cp)
1725 remain = p->term_len - (b - p->term_buf);
1727 map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, 0);
1733 extract_add_string (p, buf, i);
1738 static void extract_add_complete_field (RecWord *p)
1740 const char *b = p->term_buf;
1741 char buf[IT_MAX_WORD+1];
1742 const char **map = 0;
1743 int i = 0, remain = p->term_len;
1745 yaz_log(YLOG_DEBUG, "Complete field, w='%.*s'",
1746 p->term_len, p->term_buf);
1749 map = zebra_maps_input (p->zebra_maps, p->index_type, &b, remain, 1);
1751 while (remain > 0 && i < IT_MAX_WORD)
1753 while (map && *map && **map == *CHR_SPACE)
1755 remain = p->term_len - (b - p->term_buf);
1759 int first = i ? 0 : 1; /* first position */
1760 map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, first);
1768 if (i && i < IT_MAX_WORD)
1769 buf[i++] = *CHR_SPACE;
1770 while (map && *map && **map != *CHR_SPACE)
1772 const char *cp = *map;
1774 if (**map == *CHR_CUT)
1780 if (i >= IT_MAX_WORD)
1782 yaz_log(YLOG_DEBUG, "Adding string to index '%d'", **map);
1783 while (i < IT_MAX_WORD && *cp)
1786 remain = p->term_len - (b - p->term_buf);
1789 map = zebra_maps_input (p->zebra_maps, p->index_type, &b,
1798 extract_add_string (p, buf, i);
1801 void extract_token_add (RecWord *p)
1805 yaz_log (YLOG_LOG, "token_add "
1806 "reg_type=%c attrSet=%d attrUse=%d seqno=%d s=%.*s",
1807 p->reg_type, p->attrSet, p->attrUse, p->seqno, p->length,
1810 if ((wrbuf = zebra_replace(p->zebra_maps, p->index_type, 0,
1811 p->term_buf, p->term_len)))
1813 p->term_buf = wrbuf_buf(wrbuf);
1814 p->term_len = wrbuf_len(wrbuf);
1816 if (zebra_maps_is_complete (p->zebra_maps, p->index_type))
1817 extract_add_complete_field (p);
1819 extract_add_incomplete_field(p);
1822 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1823 void *buf, size_t sz)
1825 ZebraHandle zh = (ZebraHandle) p->handle;
1827 xfree(zh->store_data_buf);
1828 zh->store_data_buf = 0;
1829 zh->store_data_size = 0;
1832 zh->store_data_buf = xmalloc(sz);
1833 zh->store_data_size = sz;
1834 memcpy(zh->store_data_buf, buf, sz);
1838 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1840 ZebraHandle zh = (ZebraHandle) p->handle;
1841 xfree(zh->store_data_buf);
1842 zh->store_data_buf = 0;
1843 zh->store_data_size = 0;
1844 p->setStoreData = extract_set_store_data_cb;
1847 void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1849 ZebraHandle zh = (ZebraHandle) p->handle;
1850 zebraExplain_addSchema (zh->reg->zei, oid);
1854 #error not done yet with zebra_rec_keys_t
1855 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1856 int cmd, struct recKeys *reckeys)
1858 SortIdx sortIdx = zh->reg->sortIdx;
1859 void *decode_handle = iscz1_start();
1863 while (off < reckeys->buf_used)
1865 const char *src = reckeys->buf + off;
1867 char *dst = (char*) &key;
1869 iscz1_decode(decode_handle, &dst, &src);
1870 assert(key.len == 4);
1872 ch = (int) key.mem[0]; /* ordinal for field/use/attribute */
1874 sortIdx_type(sortIdx, ch);
1876 sortIdx_add(sortIdx, src, strlen(src));
1878 sortIdx_add(sortIdx, "", 1);
1883 off = src - reckeys->buf;
1885 assert (off == reckeys->buf_used);
1886 iscz1_stop(decode_handle);
1889 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1890 int cmd, struct sortKeys *sk)
1892 SortIdx sortIdx = zh->reg->sortIdx;
1895 sortIdx_sysno (sortIdx, sysno);
1897 while (off < sk->buf_used)
1901 off += key_SU_decode(&set, (unsigned char *) sk->buf + off);
1902 off += key_SU_decode(&use, (unsigned char *) sk->buf + off);
1903 off += key_SU_decode(&slen, (unsigned char *) sk->buf + off);
1905 sortIdx_type(sortIdx, use);
1907 sortIdx_add(sortIdx, sk->buf + off, slen);
1909 sortIdx_add(sortIdx, "", 1);
1915 void encode_key_init (struct encode_info *i)
1924 i->encode_handle = iscz1_start();
1930 /* this is the old encode_key_write
1931 * may be deleted once we are confident that the new works
1934 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1937 char *bp = i->buf, *bp0;
1938 const char *src = (char *) &key;
1940 /* copy term to output buf */
1941 while ((*bp++ = *k++))
1943 /* and copy & align key so we can mangle */
1944 memcpy (&key, k+1, sizeof(struct it_key)); /* *k is insert/delete */
1947 iscz1_encode(i->encode_handle, &bp, &src);
1948 *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */
1949 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1951 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
1956 void encode_key_flush (struct encode_info *i, FILE *outf)
1957 { /* dummy routine */
1958 iscz1_stop(i->encode_handle);
1963 /* new encode_key_write
1964 * The idea is to buffer one more key, and compare them
1965 * If we are going to delete and insert the same key,
1966 * we may as well not bother. Should make a difference in
1967 * updates with small modifications (appending to a mbox)
1969 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1974 if (*k) /* first time for new key */
1977 while ((*bp++ = *k++))
1979 i->keylen= bp - i->buf -1;
1980 assert(i->keylen+1+sizeof(struct it_key) < ENCODE_BUFLEN);
1984 bp=i->buf + i->keylen;
1989 memcpy (&key, k+1, sizeof(struct it_key));
1990 if (0==i->prevsys) /* no previous filter, fill up */
1992 i->prevsys=key.sysno;
1993 i->prevseq=key.seqno;
1996 else if ( (i->prevsys==key.sysno) &&
1997 (i->prevseq==key.seqno) &&
1999 { /* same numbers, diff cmd, they cancel out */
2003 { /* different stuff, write previous, move buf */
2004 bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
2005 if (i->sysno != i->prevsys)
2007 i->sysno = i->prevsys;
2010 else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
2012 return; /* ??? Filters some sort of duplicates away */
2013 /* ??? Can this ever happen -H 15oct02 */
2015 bp = encode_key_int (i->prevseq - i->seqno, bp);
2016 i->seqno = i->prevseq;
2017 i->cmd = i->prevcmd;
2018 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
2020 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
2023 i->keylen=0; /* ok, it's written, forget it */
2024 i->prevsys=key.sysno;
2025 i->prevseq=key.seqno;
2030 void encode_key_flush (struct encode_info *i, FILE *outf)
2031 { /* flush the last key from i */
2032 char *bp =i->buf + i->keylen;
2035 return; /* nothing to flush */
2038 bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
2039 if (i->sysno != i->prevsys)
2041 i->sysno = i->prevsys;
2044 else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
2046 return; /* ??? Filters some sort of duplicates away */
2047 /* ??? Can this ever happen -H 15oct02 */
2049 bp = encode_key_int (i->prevseq - i->seqno, bp);
2050 i->seqno = i->prevseq;
2051 i->cmd = i->prevcmd;
2052 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
2054 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
2057 i->keylen=0; /* ok, it's written, forget it */
2058 i->prevsys=0; /* forget the values too */