1 /* $Id: extract.c,v 1.199 2005-11-10 11:25:47 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);
539 extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
541 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
543 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
544 recordAttr->staticrank);
545 zh->records_inserted++;
549 /* record already exists */
550 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
553 zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
555 struct sortKeys sortKeys;
558 rec = rec_get (zh->reg->records, *sysno);
561 recordAttr = rec_init_attr (zh->reg->zei, rec);
563 zebra_rec_keys_set_buf(delkeys,
564 rec->info[recInfo_delKeys],
565 rec->size[recInfo_delKeys],
569 zebra_rec_keys_set_buf(sortKeys,
570 rec->info[recInfo_sortKeys],
571 rec->size[recInfo_sortKeys],
573 extract_flushSortKeys (zh, *sysno, 0, sortKeys);
575 sortKeys.buf_used = rec->size[recInfo_sortKeys];
576 sortKeys.buf = rec->info[recInfo_sortKeys];
577 extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
580 extract_flushRecordKeys (zh, *sysno, 0, delkeys,
581 recordAttr->staticrank); /* old values */
584 /* record going to be deleted */
585 if (zebra_rec_keys_empty(delkeys))
587 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
588 zh->m_record_type, fname, recordOffset);
589 yaz_log (YLOG_WARN, "cannot delete file above, storeKeys false (1)");
593 if (zh->records_processed < zh->m_file_verbose_limit)
594 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
595 zh->m_record_type, fname, recordOffset);
596 zh->records_deleted++;
598 dict_delete (zh->reg->matchDict, matchStr);
599 rec_del (zh->reg->records, &rec);
607 /* flush new keys for sort&search etc */
608 if (zh->records_processed < zh->m_file_verbose_limit)
609 yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
610 zh->m_record_type, fname, recordOffset);
611 recordAttr->staticrank = extractCtrl.staticrank;
613 extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
615 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
617 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
618 recordAttr->staticrank);
619 zh->records_updated++;
621 zebra_rec_keys_close(delkeys);
623 zebra_rec_keys_close(sortKeys);
626 /* update file type */
627 xfree (rec->info[recInfo_fileType]);
628 rec->info[recInfo_fileType] =
629 rec_strdup (zh->m_record_type, &rec->size[recInfo_fileType]);
631 /* update filename */
632 xfree (rec->info[recInfo_filename]);
633 rec->info[recInfo_filename] =
634 rec_strdup (fname, &rec->size[recInfo_filename]);
636 /* update delete keys */
637 xfree (rec->info[recInfo_delKeys]);
638 if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
640 zebra_rec_keys_get_buf(zh->reg->keys,
641 &rec->info[recInfo_delKeys],
642 &rec->size[recInfo_delKeys]);
646 rec->info[recInfo_delKeys] = NULL;
647 rec->size[recInfo_delKeys] = 0;
650 /* update sort keys */
651 xfree (rec->info[recInfo_sortKeys]);
654 zebra_rec_keys_get_buf(zh->reg->sortKeys,
655 &rec->info[recInfo_sortKeys],
656 &rec->size[recInfo_sortKeys]);
658 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
659 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
660 zh->reg->sortKeys.buf = NULL;
661 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 zebra_rec_keys_reset(zh->reg->keys);
881 zebra_rec_keys_reset(zh->reg->sortKeys);
883 zh->reg->sortKeys.buf_used = 0;
885 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
887 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
888 zh->m_explain_database))
892 if (recordType && *recordType)
894 yaz_log (YLOG_DEBUG, "Record type explicitly specified: %s", recordType);
895 recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
900 if (!(zh->m_record_type))
902 yaz_log (YLOG_WARN, "No such record type defined");
905 yaz_log (YLOG_DEBUG, "Get record type from rgroup: %s",zh->m_record_type);
906 recType = recType_byName (zh->reg->recTypes, zh->res,
907 zh->m_record_type, &clientData);
908 recordType = zh->m_record_type;
913 yaz_log (YLOG_WARN, "No such record type: %s", zh->m_record_type);
917 extractCtrl.init = extract_init;
918 extractCtrl.tokenAdd = extract_token_add;
919 extractCtrl.schemaAdd = extract_schema_add;
920 extractCtrl.dh = zh->reg->dh;
921 extractCtrl.handle = zh;
922 extractCtrl.match_criteria[0] = '\0';
923 extractCtrl.staticrank = 0;
925 init_extractCtrl(zh, &extractCtrl);
927 extract_set_store_data_prepare(&extractCtrl);
929 r = (*recType->extract)(clientData, &extractCtrl);
931 if (r == RECCTRL_EXTRACT_EOF)
933 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
935 /* error occured during extraction ... */
936 yaz_log (YLOG_WARN, "extract error: generic");
939 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
941 /* error occured during extraction ... */
942 yaz_log (YLOG_WARN, "extract error: no such filter");
948 if (extractCtrl.match_criteria[0])
949 match_criteria = extractCtrl.match_criteria;
953 if (match_criteria && *match_criteria) {
954 matchStr = match_criteria;
956 if (zh->m_record_id && *zh->m_record_id) {
957 matchStr = fileMatchStr (zh, zh->reg->keys, pr_fname,
961 yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
967 rinfo = dict_lookup (zh->reg->matchDict, matchStr);
970 assert(*rinfo == sizeof(*sysno));
971 memcpy (sysno, rinfo+1, sizeof(*sysno));
975 if (zebra_rec_keys_empty(zh->reg->keys))
977 /* the extraction process returned no information - the record
978 is probably empty - unless flagShowRecords is in use */
988 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
989 pr_fname, (long) recordOffset);
990 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
994 yaz_log (YLOG_LOG, "add %s %s %ld", recordType, pr_fname,
995 (long) recordOffset);
996 rec = rec_new (zh->reg->records);
1000 recordAttr = rec_init_attr (zh->reg->zei, rec);
1001 recordAttr->staticrank = extractCtrl.staticrank;
1005 dict_insert (zh->reg->matchDict, matchStr,
1006 sizeof(*sysno), sysno);
1009 extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
1011 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1015 print_rec_keys(zh, zh->reg->keys);
1017 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
1018 recordAttr->staticrank);
1019 zh->records_inserted++;
1023 /* record already exists */
1024 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1026 zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
1028 struct sortKeys sortKeys;
1033 yaz_log (YLOG_LOG, "skipped %s %s %ld",
1034 recordType, pr_fname, (long) recordOffset);
1039 rec = rec_get (zh->reg->records, *sysno);
1042 recordAttr = rec_init_attr (zh->reg->zei, rec);
1044 zebra_rec_keys_set_buf(delkeys,
1045 rec->info[recInfo_delKeys],
1046 rec->size[recInfo_delKeys],
1049 zebra_rec_keys_set_buf(sortKeys,
1050 rec->info[recInfo_sortKeys],
1051 rec->size[recInfo_sortKeys],
1054 sortKeys.buf_used = rec->size[recInfo_sortKeys];
1055 sortKeys.buf = rec->info[recInfo_sortKeys];
1059 extract_flushSortKeys (zh, *sysno, 0, sortKeys);
1061 extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
1063 extract_flushRecordKeys (zh, *sysno, 0, delkeys,
1064 recordAttr->staticrank);
1067 /* record going to be deleted */
1068 if (zebra_rec_keys_empty(delkeys))
1070 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1071 pr_fname, (long) recordOffset);
1072 yaz_log (YLOG_WARN, "cannot delete file above, "
1073 "storeKeys false (3)");
1078 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1079 pr_fname, (long) recordOffset);
1080 zh->records_deleted++;
1082 dict_delete (zh->reg->matchDict, matchStr);
1083 rec_del (zh->reg->records, &rec);
1092 yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
1093 pr_fname, (long) recordOffset);
1094 recordAttr->staticrank = extractCtrl.staticrank;
1096 extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
1098 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1100 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
1101 recordAttr->staticrank);
1102 zh->records_updated++;
1104 zebra_rec_keys_close(delkeys);
1106 zebra_rec_keys_close(sortKeys);
1109 /* update file type */
1110 xfree (rec->info[recInfo_fileType]);
1111 rec->info[recInfo_fileType] =
1112 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1114 /* update filename */
1115 xfree (rec->info[recInfo_filename]);
1116 rec->info[recInfo_filename] =
1117 rec_strdup (fname, &rec->size[recInfo_filename]);
1119 /* update delete keys */
1120 xfree (rec->info[recInfo_delKeys]);
1121 if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
1123 zebra_rec_keys_get_buf(zh->reg->keys,
1124 &rec->info[recInfo_delKeys],
1125 &rec->size[recInfo_delKeys]);
1129 rec->info[recInfo_delKeys] = NULL;
1130 rec->size[recInfo_delKeys] = 0;
1132 /* update sort keys */
1133 xfree (rec->info[recInfo_sortKeys]);
1136 zebra_rec_keys_get_buf(zh->reg->sortKeys,
1137 &rec->info[recInfo_sortKeys],
1138 &rec->size[recInfo_sortKeys]);
1140 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
1141 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
1142 zh->reg->sortKeys.buf = NULL;
1143 zh->reg->sortKeys.buf_max = 0;
1146 /* save file size of original record */
1147 zebraExplain_recordBytesIncrement (zh->reg->zei,
1148 - recordAttr->recordSize);
1150 recordAttr->recordSize = fi->file_moffset - recordOffset;
1151 if (!recordAttr->recordSize)
1152 recordAttr->recordSize = fi->file_max - recordOffset;
1154 recordAttr->recordSize = buf_size;
1156 zebraExplain_recordBytesIncrement (zh->reg->zei,
1157 recordAttr->recordSize);
1159 /* set run-number for this record */
1160 recordAttr->runNumber =
1161 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
1163 /* update store data */
1164 xfree (rec->info[recInfo_storeData]);
1166 /* update store data */
1167 if (zh->store_data_buf)
1169 rec->size[recInfo_storeData] = zh->store_data_size;
1170 rec->info[recInfo_storeData] = zh->store_data_buf;
1171 zh->store_data_buf = 0;
1173 else if (zh->m_store_data)
1175 rec->size[recInfo_storeData] = recordAttr->recordSize;
1176 rec->info[recInfo_storeData] = (char *)
1177 xmalloc (recordAttr->recordSize);
1178 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1182 rec->info[recInfo_storeData] = NULL;
1183 rec->size[recInfo_storeData] = 0;
1185 /* update database name */
1186 xfree (rec->info[recInfo_databaseName]);
1187 rec->info[recInfo_databaseName] =
1188 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
1191 recordAttr->recordOffset = recordOffset;
1193 /* commit this record */
1194 rec_put (zh->reg->records, &rec);
1199 int explain_extract (void *handle, Record rec, data1_node *n)
1201 ZebraHandle zh = (ZebraHandle) handle;
1202 struct recExtractCtrl extractCtrl;
1204 if (zebraExplain_curDatabase (zh->reg->zei,
1205 rec->info[recInfo_databaseName]))
1208 if (zebraExplain_newDatabase (zh->reg->zei,
1209 rec->info[recInfo_databaseName], 0))
1213 zebra_rec_keys_reset(zh->reg->keys);
1216 zebra_rec_keys_reset(zh->reg->sortKeys);
1218 zh->reg->sortKeys.buf_used = 0;
1220 extractCtrl.init = extract_init;
1221 extractCtrl.tokenAdd = extract_token_add;
1222 extractCtrl.schemaAdd = extract_schema_add;
1223 extractCtrl.dh = zh->reg->dh;
1225 init_extractCtrl(zh, &extractCtrl);
1227 extractCtrl.flagShowRecords = 0;
1228 extractCtrl.match_criteria[0] = '\0';
1229 extractCtrl.staticrank = 0;
1230 extractCtrl.handle = handle;
1231 extractCtrl.first_record = 1;
1233 extract_set_store_data_prepare(&extractCtrl);
1236 grs_extract_tree(&extractCtrl, n);
1238 if (rec->size[recInfo_delKeys])
1240 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1243 zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
1245 struct sortKeys sortkeys;
1248 zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
1249 rec->size[recInfo_delKeys],
1251 extract_flushRecordKeys (zh, rec->sysno, 0, delkeys, 0);
1252 zebra_rec_keys_close(delkeys);
1254 zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
1255 rec->size[recInfo_sortKeys],
1258 extract_flushSortKeys (zh, rec->sysno, 0, sortkeys);
1259 zebra_rec_keys_close(sortkeys);
1261 sortkeys.buf_used = rec->size[recInfo_sortKeys];
1262 sortkeys.buf = rec->info[recInfo_sortKeys];
1263 extract_flushSortKeys (zh, rec->sysno, 0, &sortkeys);
1266 extract_flushRecordKeys (zh, rec->sysno, 1, zh->reg->keys, 0);
1268 extract_flushSortKeys (zh, rec->sysno, 1, zh->reg->sortKeys);
1270 extract_flushSortKeys (zh, rec->sysno, 1, &zh->reg->sortKeys);
1273 xfree (rec->info[recInfo_delKeys]);
1274 zebra_rec_keys_get_buf(zh->reg->keys,
1275 &rec->info[recInfo_delKeys],
1276 &rec->size[recInfo_delKeys]);
1278 xfree (rec->info[recInfo_sortKeys]);
1280 zebra_rec_keys_get_buf(zh->reg->sortKeys,
1281 &rec->info[recInfo_sortKeys],
1282 &rec->size[recInfo_sortKeys]);
1284 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
1285 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
1286 zh->reg->sortKeys.buf = NULL;
1287 zh->reg->sortKeys.buf_max = 0;
1293 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1295 zebra_rec_keys_t reckeys,
1298 ZebraExplainInfo zei = zh->reg->zei;
1300 if (!zh->reg->key_buf)
1302 int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
1305 yaz_log(YLOG_WARN, "Invalid memory setting, using default 8 MB");
1308 /* FIXME: That "8" should be in a default settings include */
1309 /* not hard-coded here! -H */
1310 zh->reg->key_buf = (char**) xmalloc (mem);
1311 zh->reg->ptr_top = mem/sizeof(char*);
1313 zh->reg->key_buf_used = 0;
1314 zh->reg->key_file_no = 0;
1316 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1318 if (zebra_rec_keys_rewind(reckeys))
1322 struct it_key key_in;
1323 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1326 struct it_key key_out;
1327 zint *keyp = key_out.mem;
1329 assert(key_in.len == 4);
1331 /* check for buffer overflow */
1332 if (zh->reg->key_buf_used + 1024 >
1333 (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
1334 extract_flushWriteKeys (zh, 0);
1337 assert(zh->reg->ptr_i > 0);
1338 (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
1339 (char*)zh->reg->key_buf + zh->reg->key_buf_used;
1341 /* encode the ordinal value (field/use/attribute) .. */
1342 ch = (int) key_in.mem[0];
1343 zh->reg->key_buf_used +=
1344 key_SU_encode(ch, (char*)zh->reg->key_buf +
1345 zh->reg->key_buf_used);
1347 /* copy the 0-terminated stuff from str to output */
1348 memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used, str, slen);
1349 zh->reg->key_buf_used += slen;
1350 ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = '\0';
1352 /* the delete/insert indicator */
1353 ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = cmd;
1355 if (zh->m_staticrank) /* rank config enabled ? */
1357 *keyp++ = staticrank;
1363 if (key_in.mem[1]) /* filter specified record ID */
1364 *keyp++ = key_in.mem[1];
1367 *keyp++ = key_in.mem[2]; /* section_id */
1368 *keyp++ = key_in.mem[3]; /* sequence .. */
1370 memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used,
1371 &key_out, sizeof(key_out));
1372 (zh->reg->key_buf_used) += sizeof(key_out);
1377 void extract_flushWriteKeys (ZebraHandle zh, int final)
1378 /* optimizing: if final=1, and no files written yet */
1379 /* push the keys directly to merge, sidestepping the */
1380 /* temp file altogether. Speeds small updates */
1383 char out_fname[200];
1385 struct encode_info encode_info;
1386 int ptr_i = zh->reg->ptr_i;
1391 if (!zh->reg->key_buf || ptr_i <= 0)
1393 yaz_log (YLOG_DEBUG, " nothing to flush section=%d buf=%p i=%d",
1394 zh->reg->key_file_no, zh->reg->key_buf, ptr_i);
1395 yaz_log (YLOG_DEBUG, " buf=%p ",
1397 yaz_log (YLOG_DEBUG, " ptr=%d ",zh->reg->ptr_i);
1398 yaz_log (YLOG_DEBUG, " reg=%p ",zh->reg);
1403 (zh->reg->key_file_no)++;
1404 yaz_log (YLOG_LOG, "sorting section %d", (zh->reg->key_file_no));
1405 yaz_log (YLOG_DEBUG, " sort_buff at %p n=%d",
1406 zh->reg->key_buf + zh->reg->ptr_top - ptr_i,ptr_i);
1408 qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
1409 sizeof(char*), key_qsort_compare);
1411 /* zebra.cfg: tempfiles:
1412 Y: always use temp files (old way)
1413 A: use temp files, if more than one (auto)
1414 = if this is both the last and the first
1415 N: never bother with temp files (new) */
1417 temp_policy=toupper(res_get_def(zh->res,"tempfiles","auto")[0]);
1418 if (temp_policy != 'Y' && temp_policy != 'N' && temp_policy != 'A') {
1419 yaz_log (YLOG_WARN, "Illegal tempfiles setting '%c'. using 'Auto' ",
1424 if ( ( temp_policy =='N' ) || /* always from memory */
1425 ( ( temp_policy =='A' ) && /* automatic */
1426 (zh->reg->key_file_no == 1) && /* this is first time */
1427 (final) ) ) /* and last (=only) time */
1428 { /* go directly from memory */
1429 zh->reg->key_file_no =0; /* signal not to read files */
1430 zebra_index_merge(zh);
1432 zh->reg->key_buf_used = 0;
1436 /* Not doing directly from memory, write into a temp file */
1437 extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
1439 if (!(outf = fopen (out_fname, "wb")))
1441 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1444 yaz_log (YLOG_LOG, "writing section %d", zh->reg->key_file_no);
1445 prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1447 encode_key_init (&encode_info);
1448 encode_key_write (cp, &encode_info, outf);
1452 cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1453 if (strcmp (cp, prevcp))
1455 encode_key_flush ( &encode_info, outf);
1456 encode_key_init (&encode_info);
1457 encode_key_write (cp, &encode_info, outf);
1461 encode_key_write (cp + strlen(cp), &encode_info, outf);
1463 encode_key_flush ( &encode_info, outf);
1465 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1466 extract_get_fname_tmp (out_fname, key_file_no);
1468 if (!(outf = fopen (out_fname, "wb")))
1470 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1473 yaz_log (YLOG_LOG, "writing section %d", key_file_no);
1475 prevcp = key_buf[ptr_top-i];
1477 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1479 key_y_len = strlen(prevcp)+1;
1481 yaz_log (YLOG_LOG, "key_y_len: %2d %02x %02x %s",
1482 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1484 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1485 sizeof(char*), key_y_compare);
1486 cp = key_buf[ptr_top-ptr_i];
1488 encode_key_init (&encode_info);
1489 encode_key_write (cp, &encode_info, outf);
1492 cp = key_buf[ptr_top-ptr_i];
1493 encode_key_write (cp+key_y_len, &encode_info, outf);
1495 encode_key_flush ( &encode_info, outf);
1498 prevcp = key_buf[ptr_top-ptr_i];
1503 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fclose %s", out_fname);
1506 yaz_log (YLOG_LOG, "finished section %d", zh->reg->key_file_no);
1508 zh->reg->key_buf_used = 0;
1511 ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh,
1512 zebra_rec_keys_t reckeys,
1513 zebra_snippets *snippets)
1515 NMEM nmem = nmem_create();
1516 if (zebra_rec_keys_rewind(reckeys))
1521 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1523 char dst_buf[IT_MAX_WORD];
1524 char *dst_term = dst_buf;
1527 assert(key.len <= 4 && key.len > 2);
1528 seqno = (int) key.mem[key.len-1];
1531 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1532 0/* db */, 0/* set */, 0/* use */);
1534 zebra_term_untrans_iconv(zh, nmem, index_type,
1536 zebra_snippets_append(snippets, seqno, ord, dst_term);
1544 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1546 yaz_log(YLOG_LOG, "print_rec_keys");
1547 if (zebra_rec_keys_rewind(reckeys))
1552 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1554 char dst_buf[IT_MAX_WORD];
1558 assert(key.len <= 4 && key.len > 2);
1560 zebraExplain_lookup_ord(zh->reg->zei,
1561 key.mem[0], &index_type, &db, 0, 0);
1563 seqno = (int) key.mem[key.len-1];
1565 zebra_term_untrans(zh, index_type, dst_buf, str);
1567 yaz_log(YLOG_LOG, "ord=" ZINT_FORMAT " seqno=%d term=%s",
1568 key.mem[0], seqno, dst_buf);
1573 void extract_add_index_string (RecWord *p, const char *str, int length)
1577 ZebraHandle zh = p->extractCtrl->handle;
1578 ZebraExplainInfo zei = zh->reg->zei;
1583 ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
1585 ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
1592 ch = zebraExplain_lookup_attr_su(zei, p->index_type,
1593 p->attrSet, p->attrUse);
1595 ch = zebraExplain_add_attr_su(zei, p->index_type,
1596 p->attrSet, p->attrUse);
1601 key.mem[1] = p->record_id;
1602 key.mem[2] = p->section_id;
1603 key.mem[3] = p->seqno;
1612 for (i = 0; i<length && i < 20; i++)
1613 sprintf(strz+strlen(strz), "%02X", str[i] & 0xff);
1614 /* just for debugging .. */
1615 yaz_log(YLOG_LOG, "add: set=%d use=%d "
1616 "record_id=%lld section_id=%lld seqno=%lld %s",
1617 p->attrSet, p->attrUse, p->record_id, p->section_id, p->seqno,
1622 zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1626 static void extract_add_sort_string (RecWord *p, const char *str, int length)
1630 ZebraHandle zh = p->extractCtrl->handle;
1631 ZebraExplainInfo zei = zh->reg->zei;
1636 ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
1638 ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
1646 key.mem[1] = p->record_id;
1647 key.mem[2] = p->section_id;
1648 key.mem[3] = p->seqno;
1650 zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1653 static void extract_add_sort_string (RecWord *p, const char *str, int length)
1655 ZebraHandle zh = p->extractCtrl->handle;
1656 struct sortKeys *sk = &zh->reg->sortKeys;
1659 while (off < sk->buf_used)
1663 off += key_SU_decode(&set, (unsigned char *) sk->buf + off);
1664 off += key_SU_decode(&use, (unsigned char *) sk->buf + off);
1665 off += key_SU_decode(&slen, (unsigned char *) sk->buf + off);
1667 if (p->attrSet == set && p->attrUse == use)
1670 assert (off == sk->buf_used);
1672 if (sk->buf_used + IT_MAX_WORD > sk->buf_max)
1676 b = (char *) xmalloc (sk->buf_max += 128000);
1677 if (sk->buf_used > 0)
1678 memcpy (b, sk->buf, sk->buf_used);
1682 off += key_SU_encode(p->attrSet, sk->buf + off);
1683 off += key_SU_encode(p->attrUse, sk->buf + off);
1684 off += key_SU_encode(length, sk->buf + off);
1685 memcpy (sk->buf + off, str, length);
1686 sk->buf_used = off + length;
1690 void extract_add_string (RecWord *p, const char *string, int length)
1692 assert (length > 0);
1693 if (zebra_maps_is_sort (p->zebra_maps, p->index_type))
1694 extract_add_sort_string (p, string, length);
1696 extract_add_index_string (p, string, length);
1699 static void extract_add_incomplete_field (RecWord *p)
1701 const char *b = p->term_buf;
1702 int remain = p->term_len;
1703 const char **map = 0;
1705 yaz_log(YLOG_DEBUG, "Incomplete field, w='%.*s'", p->term_len, p->term_buf);
1708 map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, 0);
1712 char buf[IT_MAX_WORD+1];
1716 while (map && *map && **map == *CHR_SPACE)
1718 remain = p->term_len - (b - p->term_buf);
1720 map = zebra_maps_input(p->zebra_maps, p->index_type, &b,
1728 while (map && *map && **map != *CHR_SPACE)
1730 const char *cp = *map;
1732 while (i < IT_MAX_WORD && *cp)
1734 remain = p->term_len - (b - p->term_buf);
1736 map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, 0);
1742 extract_add_string (p, buf, i);
1747 static void extract_add_complete_field (RecWord *p)
1749 const char *b = p->term_buf;
1750 char buf[IT_MAX_WORD+1];
1751 const char **map = 0;
1752 int i = 0, remain = p->term_len;
1754 yaz_log(YLOG_DEBUG, "Complete field, w='%.*s'",
1755 p->term_len, p->term_buf);
1758 map = zebra_maps_input (p->zebra_maps, p->index_type, &b, remain, 1);
1760 while (remain > 0 && i < IT_MAX_WORD)
1762 while (map && *map && **map == *CHR_SPACE)
1764 remain = p->term_len - (b - p->term_buf);
1768 int first = i ? 0 : 1; /* first position */
1769 map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, first);
1777 if (i && i < IT_MAX_WORD)
1778 buf[i++] = *CHR_SPACE;
1779 while (map && *map && **map != *CHR_SPACE)
1781 const char *cp = *map;
1783 if (**map == *CHR_CUT)
1789 if (i >= IT_MAX_WORD)
1791 yaz_log(YLOG_DEBUG, "Adding string to index '%d'", **map);
1792 while (i < IT_MAX_WORD && *cp)
1795 remain = p->term_len - (b - p->term_buf);
1798 map = zebra_maps_input (p->zebra_maps, p->index_type, &b,
1807 extract_add_string (p, buf, i);
1810 void extract_token_add (RecWord *p)
1814 yaz_log (YLOG_LOG, "token_add "
1815 "reg_type=%c attrSet=%d attrUse=%d seqno=%d s=%.*s",
1816 p->reg_type, p->attrSet, p->attrUse, p->seqno, p->length,
1819 if ((wrbuf = zebra_replace(p->zebra_maps, p->index_type, 0,
1820 p->term_buf, p->term_len)))
1822 p->term_buf = wrbuf_buf(wrbuf);
1823 p->term_len = wrbuf_len(wrbuf);
1825 if (zebra_maps_is_complete (p->zebra_maps, p->index_type))
1826 extract_add_complete_field (p);
1828 extract_add_incomplete_field(p);
1831 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1832 void *buf, size_t sz)
1834 ZebraHandle zh = (ZebraHandle) p->handle;
1836 xfree(zh->store_data_buf);
1837 zh->store_data_buf = 0;
1838 zh->store_data_size = 0;
1841 zh->store_data_buf = xmalloc(sz);
1842 zh->store_data_size = sz;
1843 memcpy(zh->store_data_buf, buf, sz);
1847 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1849 ZebraHandle zh = (ZebraHandle) p->handle;
1850 xfree(zh->store_data_buf);
1851 zh->store_data_buf = 0;
1852 zh->store_data_size = 0;
1853 p->setStoreData = extract_set_store_data_cb;
1856 void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1858 ZebraHandle zh = (ZebraHandle) p->handle;
1859 zebraExplain_addSchema (zh->reg->zei, oid);
1863 #error not done yet with zebra_rec_keys_t
1864 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1865 int cmd, struct recKeys *reckeys)
1867 SortIdx sortIdx = zh->reg->sortIdx;
1868 void *decode_handle = iscz1_start();
1872 while (off < reckeys->buf_used)
1874 const char *src = reckeys->buf + off;
1876 char *dst = (char*) &key;
1878 iscz1_decode(decode_handle, &dst, &src);
1879 assert(key.len == 4);
1881 ch = (int) key.mem[0]; /* ordinal for field/use/attribute */
1883 sortIdx_type(sortIdx, ch);
1885 sortIdx_add(sortIdx, src, strlen(src));
1887 sortIdx_add(sortIdx, "", 1);
1892 off = src - reckeys->buf;
1894 assert (off == reckeys->buf_used);
1895 iscz1_stop(decode_handle);
1898 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1899 int cmd, struct sortKeys *sk)
1901 SortIdx sortIdx = zh->reg->sortIdx;
1904 sortIdx_sysno (sortIdx, sysno);
1906 while (off < sk->buf_used)
1910 off += key_SU_decode(&set, (unsigned char *) sk->buf + off);
1911 off += key_SU_decode(&use, (unsigned char *) sk->buf + off);
1912 off += key_SU_decode(&slen, (unsigned char *) sk->buf + off);
1914 sortIdx_type(sortIdx, use);
1916 sortIdx_add(sortIdx, sk->buf + off, slen);
1918 sortIdx_add(sortIdx, "", 1);
1924 void encode_key_init (struct encode_info *i)
1933 i->encode_handle = iscz1_start();
1939 /* this is the old encode_key_write
1940 * may be deleted once we are confident that the new works
1943 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1946 char *bp = i->buf, *bp0;
1947 const char *src = (char *) &key;
1949 /* copy term to output buf */
1950 while ((*bp++ = *k++))
1952 /* and copy & align key so we can mangle */
1953 memcpy (&key, k+1, sizeof(struct it_key)); /* *k is insert/delete */
1956 iscz1_encode(i->encode_handle, &bp, &src);
1957 *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */
1958 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1960 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
1965 void encode_key_flush (struct encode_info *i, FILE *outf)
1966 { /* dummy routine */
1967 iscz1_stop(i->encode_handle);
1972 /* new encode_key_write
1973 * The idea is to buffer one more key, and compare them
1974 * If we are going to delete and insert the same key,
1975 * we may as well not bother. Should make a difference in
1976 * updates with small modifications (appending to a mbox)
1978 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1983 if (*k) /* first time for new key */
1986 while ((*bp++ = *k++))
1988 i->keylen= bp - i->buf -1;
1989 assert(i->keylen+1+sizeof(struct it_key) < ENCODE_BUFLEN);
1993 bp=i->buf + i->keylen;
1998 memcpy (&key, k+1, sizeof(struct it_key));
1999 if (0==i->prevsys) /* no previous filter, fill up */
2001 i->prevsys=key.sysno;
2002 i->prevseq=key.seqno;
2005 else if ( (i->prevsys==key.sysno) &&
2006 (i->prevseq==key.seqno) &&
2008 { /* same numbers, diff cmd, they cancel out */
2012 { /* different stuff, write previous, move buf */
2013 bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
2014 if (i->sysno != i->prevsys)
2016 i->sysno = i->prevsys;
2019 else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
2021 return; /* ??? Filters some sort of duplicates away */
2022 /* ??? Can this ever happen -H 15oct02 */
2024 bp = encode_key_int (i->prevseq - i->seqno, bp);
2025 i->seqno = i->prevseq;
2026 i->cmd = i->prevcmd;
2027 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
2029 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
2032 i->keylen=0; /* ok, it's written, forget it */
2033 i->prevsys=key.sysno;
2034 i->prevseq=key.seqno;
2039 void encode_key_flush (struct encode_info *i, FILE *outf)
2040 { /* flush the last key from i */
2041 char *bp =i->buf + i->keylen;
2044 return; /* nothing to flush */
2047 bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
2048 if (i->sysno != i->prevsys)
2050 i->sysno = i->prevsys;
2053 else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
2055 return; /* ??? Filters some sort of duplicates away */
2056 /* ??? Can this ever happen -H 15oct02 */
2058 bp = encode_key_int (i->prevseq - i->seqno, bp);
2059 i->seqno = i->prevseq;
2060 i->cmd = i->prevcmd;
2061 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
2063 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
2066 i->keylen=0; /* ok, it's written, forget it */
2067 i->prevsys=0; /* forget the values too */