1 /* $Id: extract.c,v 1.179 2005-04-28 08:20:39 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
37 #if _FILE_OFFSET_BITS == 64
38 #define PRINTF_OFF_T "%Ld"
40 #define PRINTF_OFF_T "%ld"
43 #define USE_SHELLSORT 0
46 static void shellsort(void *ar, int r, size_t s,
47 int (*cmp)(const void *a, const void *b))
52 static const int incs[16] = { 1391376, 463792, 198768, 86961, 33936,
53 13776, 4592, 1968, 861, 336,
54 112, 48, 21, 7, 3, 1 };
55 for ( k = 0; k < 16; k++)
56 for (h = incs[k], i = h; i < r; i++)
60 while (j > h && (*cmp)(a + s*(j-h), v) > 0)
62 memcpy (a + s*j, a + s*(j-h), s);
70 static void logRecord (ZebraHandle zh)
72 ++zh->records_processed;
73 if (!(zh->records_processed % 1000))
75 yaz_log (YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
76 ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT,
77 zh->records_processed, zh->records_inserted, zh->records_updated,
82 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
84 static void extract_init (struct recExtractCtrl *p, RecWord *w)
86 w->zebra_maps = p->zebra_maps;
88 w->attrSet = VAL_BIB1;
97 static const char **searchRecordKey (ZebraHandle zh,
98 struct recKeys *reckeys,
99 int attrSetS, int attrUseS)
101 static const char *ws[32];
102 void *decode_handle = iscz1_start();
108 for (i = 0; i<32; i++)
111 while (off < reckeys->buf_used)
113 const char *src = reckeys->buf + off;
115 char *dst = (char*) &key;
116 int attrSet, attrUse;
118 iscz1_decode(decode_handle, &dst, &src);
119 assert(key.len < 4 && key.len > 2);
121 attrSet = (int) key.mem[0] >> 16;
122 attrUse = (int) key.mem[0] & 65535;
123 seqno = (int) key.mem[key.len-1];
125 if (attrUseS == attrUse && attrSetS == attrSet)
131 woff = seqno - startSeq;
132 if (woff >= 0 && woff < 31)
138 off = src - reckeys->buf;
140 iscz1_stop(decode_handle);
141 assert (off == reckeys->buf_used);
145 struct file_read_info {
146 off_t file_max; /* maximum offset so far */
147 off_t file_offset; /* current offset */
148 off_t file_moffset; /* offset of rec/rec boundary */
155 static struct file_read_info *file_read_start (int fd)
157 struct file_read_info *fi = (struct file_read_info *)
158 xmalloc (sizeof(*fi));
162 fi->file_moffset = 0;
168 static void file_read_stop (struct file_read_info *fi)
173 static off_t file_seek (void *handle, off_t offset)
175 struct file_read_info *p = (struct file_read_info *) handle;
176 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;
196 if (r > p->sdrmax - p->file_offset)
197 r = p->sdrmax - p->file_offset;
199 memcpy (buf, p->sdrbuf + p->file_offset, r);
202 r = read (fd, buf, count);
206 if (p->file_offset > p->file_max)
207 p->file_max = p->file_offset;
212 static void file_begin (void *handle)
214 struct file_read_info *p = (struct file_read_info *) handle;
216 p->file_offset = p->file_moffset;
217 if (!p->sdrbuf && p->file_moffset)
218 lseek (p->fd, p->file_moffset, SEEK_SET);
222 static void file_end (void *handle, off_t offset)
224 struct file_read_info *p = (struct file_read_info *) handle;
226 assert (p->file_more == 0);
228 p->file_moffset = offset;
231 static char *fileMatchStr (ZebraHandle zh,
232 struct recKeys *reckeys,
233 const char *fname, const char *spec)
235 static char dstBuf[2048]; /* static here ??? */
237 const char *s = spec;
238 static const char **w;
242 while (*s == ' ' || *s == '\t')
248 char attset_str[64], attname_str[64];
249 data1_attset *attset;
252 int attSet = 1, attUse = 1;
256 for (i = 0; *s && *s != ',' && *s != ')'; s++)
258 attset_str[i++] = *s;
259 attset_str[i] = '\0';
264 for (i = 0; *s && *s != ')'; s++)
266 attname_str[i++] = *s;
267 attname_str[i] = '\0';
270 if ((attset = data1_get_attset (zh->reg->dh, attset_str)))
273 attSet = attset->reference;
274 att = data1_getattbyname(zh->reg->dh, attset, attname_str);
278 attUse = atoi (attname_str);
280 w = searchRecordKey (zh, reckeys, attSet, attUse);
285 for (i = 0; i<32; i++)
290 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
291 spec, zh->m_group ? zh->m_group : "none");
296 for (i = 0; i<32; i++)
297 if (matchFlag[i] && w[i])
309 yaz_log (YLOG_WARN, "Record didn't contain match"
310 " fields in (%s,%s)", attset_str, attname_str);
318 const char *spec_src = NULL;
319 const char *s1 = ++s;
320 while (*s1 && *s1 != ' ' && *s1 != '\t')
326 memcpy (special, s, spec_len);
327 special[spec_len] = '\0';
330 if (!strcmp (special, "group"))
331 spec_src = zh->m_group;
332 else if (!strcmp (special, "database"))
333 spec_src = zh->basenames[0];
334 else if (!strcmp (special, "filename")) {
337 else if (!strcmp (special, "type"))
338 spec_src = zh->m_record_type;
343 strcpy (dst, spec_src);
344 dst += strlen (spec_src);
347 else if (*s == '\"' || *s == '\'')
349 int stopMarker = *s++;
353 while (*s && *s != stopMarker)
356 tmpString[i++] = *s++;
361 strcpy (dst, tmpString);
362 dst += strlen (tmpString);
366 yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
367 spec, zh->m_group ? zh->m_group : "none");
374 yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
375 fname, zh->m_group ? zh->m_group : "none");
382 struct recordLogInfo {
385 struct recordGroup *rGroup;
388 void create_rec_keys_codec(struct recKeys *keys)
391 iscz1_reset(keys->codec_handle);
394 static int file_extract_record(ZebraHandle zh,
395 SYSNO *sysno, const char *fname,
397 struct file_read_info *fi,
400 void *recTypeClientData)
402 RecordAttr *recordAttr;
404 const char *matchStr = 0;
407 off_t recordOffset = 0;
409 /* announce database */
410 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
412 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
413 zh->m_explain_database))
419 struct recExtractCtrl extractCtrl;
421 /* we are going to read from a file, so prepare the extraction */
424 create_rec_keys_codec(&zh->reg->keys);
426 zh->reg->sortKeys.buf_used = 0;
429 recordOffset = fi->file_moffset;
430 extractCtrl.handle = zh;
431 extractCtrl.offset = fi->file_moffset;
432 extractCtrl.readf = file_read;
433 extractCtrl.seekf = file_seek;
434 extractCtrl.tellf = file_tell;
435 extractCtrl.endf = file_end;
437 extractCtrl.init = extract_init;
438 extractCtrl.tokenAdd = extract_token_add;
439 extractCtrl.schemaAdd = extract_schema_add;
440 extractCtrl.dh = zh->reg->dh;
441 extractCtrl.match_criteria[0] = '\0';
442 extractCtrl.first_record = fi->file_offset ? 0 : 1;
444 extract_set_store_data_prepare(&extractCtrl);
446 for (i = 0; i<256; i++)
448 if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
449 extractCtrl.seqno[i] = 1;
451 extractCtrl.seqno[i] = 0;
453 extractCtrl.zebra_maps = zh->reg->zebra_maps;
454 extractCtrl.flagShowRecords = !zh->m_flag_rw;
457 printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset);
461 sprintf (msg, "%s:" PRINTF_OFF_T , fname, recordOffset);
462 yaz_log_init_prefix2 (msg);
465 r = (*recType->extract)(recTypeClientData, &extractCtrl);
467 yaz_log_init_prefix2 (0);
468 if (r == RECCTRL_EXTRACT_EOF)
470 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
472 /* error occured during extraction ... */
474 zh->records_processed < zh->m_file_verbose_limit)
476 yaz_log (YLOG_WARN, "fail %s %s " PRINTF_OFF_T, zh->m_record_type,
477 fname, recordOffset);
481 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
483 /* error occured during extraction ... */
485 zh->records_processed < zh->m_file_verbose_limit)
487 yaz_log (YLOG_WARN, "no filter for %s %s "
488 PRINTF_OFF_T, zh->m_record_type,
489 fname, recordOffset);
493 if (extractCtrl.match_criteria[0])
494 matchStr = extractCtrl.match_criteria;
497 /* perform match if sysno not known and if match criteria is specified */
503 if (matchStr == 0 && zh->m_record_id && *zh->m_record_id)
506 matchStr = fileMatchStr (zh, &zh->reg->keys, fname,
510 yaz_log(YLOG_WARN, "Bad match criteria");
516 char *rinfo = dict_lookup (zh->reg->matchDict, matchStr);
519 assert(*rinfo == sizeof(*sysno));
520 memcpy (sysno, rinfo+1, sizeof(*sysno));
524 if (! *sysno && zh->reg->keys.buf_used == 0)
526 /* the extraction process returned no information - the record
527 is probably empty - unless flagShowRecords is in use */
531 if (zh->records_processed < zh->m_file_verbose_limit)
532 yaz_log (YLOG_WARN, "empty %s %s " PRINTF_OFF_T, zh->m_record_type,
533 fname, recordOffset);
542 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T, zh->m_record_type,
543 fname, recordOffset);
544 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
547 if (zh->records_processed < zh->m_file_verbose_limit)
548 yaz_log (YLOG_LOG, "add %s %s " PRINTF_OFF_T, zh->m_record_type,
549 fname, recordOffset);
550 rec = rec_new (zh->reg->records);
554 recordAttr = rec_init_attr (zh->reg->zei, rec);
558 dict_insert (zh->reg->matchDict, matchStr, sizeof(*sysno), sysno);
560 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
561 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
563 zh->records_inserted++;
567 /* record already exists */
568 struct recKeys delkeys;
569 struct sortKeys sortKeys;
571 rec = rec_get (zh->reg->records, *sysno);
574 recordAttr = rec_init_attr (zh->reg->zei, rec);
576 if (!force_update && recordAttr->runNumber ==
577 zebraExplain_runNumberIncrement (zh->reg->zei, 0))
579 yaz_log (YLOG_LOG, "run number = " ZINT_FORMAT,
580 recordAttr->runNumber);
581 yaz_log (YLOG_LOG, "skipped %s %s " PRINTF_OFF_T,
582 zh->m_record_type, fname, recordOffset);
583 extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
588 delkeys.buf_used = rec->size[recInfo_delKeys];
589 delkeys.buf = rec->info[recInfo_delKeys];
591 sortKeys.buf_used = rec->size[recInfo_sortKeys];
592 sortKeys.buf = rec->info[recInfo_sortKeys];
594 extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
595 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
598 /* record going to be deleted */
599 if (!delkeys.buf_used)
601 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
602 zh->m_record_type, fname, recordOffset);
603 yaz_log (YLOG_WARN, "cannot delete file above, storeKeys false");
607 if (zh->records_processed < zh->m_file_verbose_limit)
608 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
609 zh->m_record_type, fname, recordOffset);
610 zh->records_deleted++;
612 dict_delete (zh->reg->matchDict, matchStr);
613 rec_del (zh->reg->records, &rec);
621 /* record going to be updated */
622 if (!delkeys.buf_used)
624 yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
625 zh->m_record_type, fname, recordOffset);
626 yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
630 if (zh->records_processed < zh->m_file_verbose_limit)
631 yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
632 zh->m_record_type, fname, recordOffset);
633 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
634 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
635 zh->records_updated++;
639 /* update file type */
640 xfree (rec->info[recInfo_fileType]);
641 rec->info[recInfo_fileType] =
642 rec_strdup (zh->m_record_type, &rec->size[recInfo_fileType]);
644 /* update filename */
645 xfree (rec->info[recInfo_filename]);
646 rec->info[recInfo_filename] =
647 rec_strdup (fname, &rec->size[recInfo_filename]);
649 /* update delete keys */
650 xfree (rec->info[recInfo_delKeys]);
651 if (zh->reg->keys.buf_used > 0 && zh->m_store_keys == 1)
653 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
654 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
655 zh->reg->keys.buf = NULL;
656 zh->reg->keys.buf_max = 0;
660 rec->info[recInfo_delKeys] = NULL;
661 rec->size[recInfo_delKeys] = 0;
664 /* update sort keys */
665 xfree (rec->info[recInfo_sortKeys]);
667 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
668 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
669 zh->reg->sortKeys.buf = NULL;
670 zh->reg->sortKeys.buf_max = 0;
672 /* save file size of original record */
673 zebraExplain_recordBytesIncrement (zh->reg->zei,
674 - recordAttr->recordSize);
675 recordAttr->recordSize = fi->file_moffset - recordOffset;
676 if (!recordAttr->recordSize)
677 recordAttr->recordSize = fi->file_max - recordOffset;
678 zebraExplain_recordBytesIncrement (zh->reg->zei,
679 recordAttr->recordSize);
681 /* set run-number for this record */
682 recordAttr->runNumber = zebraExplain_runNumberIncrement (zh->reg->zei,
685 /* update store data */
686 xfree (rec->info[recInfo_storeData]);
687 if (zh->store_data_buf)
689 rec->size[recInfo_storeData] = zh->store_data_size;
690 rec->info[recInfo_storeData] = zh->store_data_buf;
691 zh->store_data_buf = 0;
692 file_end(fi, fi->file_offset);
694 else if (zh->m_store_data)
696 rec->size[recInfo_storeData] = recordAttr->recordSize;
697 rec->info[recInfo_storeData] = (char *)
698 xmalloc (recordAttr->recordSize);
699 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
701 yaz_log (YLOG_ERRNO|YLOG_FATAL, "seek to " PRINTF_OFF_T " in %s",
702 recordOffset, fname);
705 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
706 < recordAttr->recordSize)
708 yaz_log (YLOG_ERRNO|YLOG_FATAL, "read %d bytes of %s",
709 recordAttr->recordSize, fname);
715 rec->info[recInfo_storeData] = NULL;
716 rec->size[recInfo_storeData] = 0;
718 /* update database name */
719 xfree (rec->info[recInfo_databaseName]);
720 rec->info[recInfo_databaseName] =
721 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
724 recordAttr->recordOffset = recordOffset;
726 /* commit this record */
727 rec_put (zh->reg->records, &rec);
732 int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
739 struct file_read_info *fi;
740 const char *original_record_type = 0;
742 void *recTypeClientData;
744 if (!zh->m_group || !*zh->m_group)
747 sprintf (gprefix, "%s.", zh->m_group);
749 yaz_log (YLOG_DEBUG, "fileExtract %s", fname);
751 /* determine file extension */
753 for (i = strlen(fname); --i >= 0; )
756 else if (fname[i] == '.')
758 strcpy (ext, fname+i+1);
761 /* determine file type - depending on extension */
762 original_record_type = zh->m_record_type;
763 if (!zh->m_record_type)
765 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
766 zh->m_record_type = res_get (zh->res, ext_res);
768 if (!zh->m_record_type)
770 if (zh->records_processed < zh->m_file_verbose_limit)
771 yaz_log (YLOG_LOG, "? %s", fname);
774 /* determine match criteria */
775 if (!zh->m_record_id)
777 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
778 zh->m_record_id = res_get (zh->res, ext_res);
782 recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
783 &recTypeClientData)))
785 yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
789 switch(recType->version)
794 yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
796 if (sysno && deleteFlag)
802 if (zh->path_reg && !yaz_is_abspath (fname))
804 strcpy (full_rep, zh->path_reg);
805 strcat (full_rep, "/");
806 strcat (full_rep, fname);
809 strcpy (full_rep, fname);
812 if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
814 yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
815 zh->m_record_type = original_record_type;
819 fi = file_read_start (fd);
823 r = file_extract_record (zh, sysno, fname, deleteFlag, fi, 1,
824 recType, recTypeClientData);
825 } while (r && !sysno && fi->file_more);
829 zh->m_record_type = original_record_type;
834 If sysno is provided, then it's used to identify the reocord.
835 If not, and match_criteria is provided, then sysno is guessed
836 If not, and a record is provided, then sysno is got from there
839 ZEBRA_RES buffer_extract_record (ZebraHandle zh,
840 const char *buf, size_t buf_size,
843 const char *recordType,
845 const char *match_criteria,
850 RecordAttr *recordAttr;
851 struct recExtractCtrl extractCtrl;
853 const char *matchStr = 0;
854 RecType recType = NULL;
857 long recordOffset = 0;
858 struct zebra_fetch_control fc;
859 const char *pr_fname = fname; /* filename to print .. */
860 int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
863 pr_fname = "<no file>"; /* make it printable if file is omitted */
866 fc.record_int_buf = buf;
867 fc.record_int_len = buf_size;
868 fc.record_int_pos = 0;
870 fc.record_offset = 0;
872 extractCtrl.offset = 0;
873 extractCtrl.readf = zebra_record_int_read;
874 extractCtrl.seekf = zebra_record_int_seek;
875 extractCtrl.tellf = zebra_record_int_tell;
876 extractCtrl.endf = zebra_record_int_end;
877 extractCtrl.first_record = 1;
878 extractCtrl.fh = &fc;
880 create_rec_keys_codec(&zh->reg->keys);
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.zebra_maps = zh->reg->zebra_maps;
922 extractCtrl.flagShowRecords = 0;
923 extractCtrl.match_criteria[0] = '\0';
924 for (i = 0; i<256; i++)
926 if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
927 extractCtrl.seqno[i] = 1;
929 extractCtrl.seqno[i] = 0;
931 extract_set_store_data_prepare(&extractCtrl);
933 r = (*recType->extract)(clientData, &extractCtrl);
935 if (r == RECCTRL_EXTRACT_EOF)
937 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
939 /* error occured during extraction ... */
940 yaz_log (YLOG_WARN, "extract error: generic");
943 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
945 /* error occured during extraction ... */
946 yaz_log (YLOG_WARN, "extract error: no such filter");
952 if (extractCtrl.match_criteria[0])
953 match_criteria = extractCtrl.match_criteria;
957 if (match_criteria && *match_criteria) {
958 matchStr = match_criteria;
960 if (zh->m_record_id && *zh->m_record_id) {
961 matchStr = fileMatchStr (zh, &zh->reg->keys, pr_fname,
965 yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
971 rinfo = dict_lookup (zh->reg->matchDict, matchStr);
974 assert(*rinfo == sizeof(*sysno));
975 memcpy (sysno, rinfo+1, sizeof(*sysno));
979 if (zh->reg->keys.buf_used == 0)
981 /* the extraction process returned no information - the record
982 is probably empty - unless flagShowRecords is in use */
993 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
994 pr_fname, (long) recordOffset);
995 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
999 yaz_log (YLOG_LOG, "add %s %s %ld", recordType, pr_fname,
1000 (long) recordOffset);
1001 rec = rec_new (zh->reg->records);
1003 *sysno = rec->sysno;
1005 recordAttr = rec_init_attr (zh->reg->zei, rec);
1009 dict_insert (zh->reg->matchDict, matchStr,
1010 sizeof(*sysno), sysno);
1012 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1013 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
1015 zh->records_inserted++;
1019 /* record already exists */
1020 struct recKeys delkeys;
1021 struct sortKeys sortKeys;
1026 yaz_log (YLOG_LOG, "skipped %s %s %ld",
1027 recordType, pr_fname, (long) recordOffset);
1032 rec = rec_get (zh->reg->records, *sysno);
1035 recordAttr = rec_init_attr (zh->reg->zei, rec);
1037 if (!force_update) {
1038 if (recordAttr->runNumber ==
1039 zebraExplain_runNumberIncrement (zh->reg->zei, 0))
1042 yaz_log (YLOG_LOG, "skipped %s %s %ld", recordType,
1043 pr_fname, (long) recordOffset);
1044 extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
1051 delkeys.buf_used = rec->size[recInfo_delKeys];
1052 delkeys.buf = rec->info[recInfo_delKeys];
1054 sortKeys.buf_used = rec->size[recInfo_sortKeys];
1055 sortKeys.buf = rec->info[recInfo_sortKeys];
1057 extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
1058 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1061 /* record going to be deleted */
1062 if (!delkeys.buf_used)
1066 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1067 pr_fname, (long) recordOffset);
1068 yaz_log (YLOG_WARN, "cannot delete file above, "
1075 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1076 pr_fname, (long) recordOffset);
1077 zh->records_deleted++;
1079 dict_delete (zh->reg->matchDict, matchStr);
1080 rec_del (zh->reg->records, &rec);
1088 /* record going to be updated */
1089 if (!delkeys.buf_used)
1093 yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
1094 pr_fname, (long) recordOffset);
1095 yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
1101 yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
1102 pr_fname, (long) recordOffset);
1103 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1104 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
1105 zh->records_updated++;
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 (zh->reg->keys.buf_used > 0 && zh->m_store_keys == 1)
1123 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
1124 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
1125 zh->reg->keys.buf = NULL;
1126 zh->reg->keys.buf_max = 0;
1130 rec->info[recInfo_delKeys] = NULL;
1131 rec->size[recInfo_delKeys] = 0;
1134 /* update sort keys */
1135 xfree (rec->info[recInfo_sortKeys]);
1137 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
1138 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
1139 zh->reg->sortKeys.buf = NULL;
1140 zh->reg->sortKeys.buf_max = 0;
1142 /* save file size of original record */
1143 zebraExplain_recordBytesIncrement (zh->reg->zei,
1144 - recordAttr->recordSize);
1146 recordAttr->recordSize = fi->file_moffset - recordOffset;
1147 if (!recordAttr->recordSize)
1148 recordAttr->recordSize = fi->file_max - recordOffset;
1150 recordAttr->recordSize = buf_size;
1152 zebraExplain_recordBytesIncrement (zh->reg->zei,
1153 recordAttr->recordSize);
1155 /* set run-number for this record */
1156 recordAttr->runNumber =
1157 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
1159 /* update store data */
1160 xfree (rec->info[recInfo_storeData]);
1161 if (zh->m_store_data)
1163 rec->size[recInfo_storeData] = recordAttr->recordSize;
1164 rec->info[recInfo_storeData] = (char *)
1165 xmalloc (recordAttr->recordSize);
1166 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1170 rec->info[recInfo_storeData] = NULL;
1171 rec->size[recInfo_storeData] = 0;
1173 /* update database name */
1174 xfree (rec->info[recInfo_databaseName]);
1175 rec->info[recInfo_databaseName] =
1176 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
1179 recordAttr->recordOffset = recordOffset;
1181 /* commit this record */
1182 rec_put (zh->reg->records, &rec);
1187 int explain_extract (void *handle, Record rec, data1_node *n)
1189 ZebraHandle zh = (ZebraHandle) handle;
1190 struct recExtractCtrl extractCtrl;
1193 if (zebraExplain_curDatabase (zh->reg->zei,
1194 rec->info[recInfo_databaseName]))
1197 if (zebraExplain_newDatabase (zh->reg->zei,
1198 rec->info[recInfo_databaseName], 0))
1202 create_rec_keys_codec(&zh->reg->keys);
1204 zh->reg->sortKeys.buf_used = 0;
1206 extractCtrl.init = extract_init;
1207 extractCtrl.tokenAdd = extract_token_add;
1208 extractCtrl.schemaAdd = extract_schema_add;
1209 extractCtrl.dh = zh->reg->dh;
1210 for (i = 0; i<256; i++)
1211 extractCtrl.seqno[i] = 0;
1212 extractCtrl.zebra_maps = zh->reg->zebra_maps;
1213 extractCtrl.flagShowRecords = 0;
1214 extractCtrl.match_criteria[0] = '\0';
1215 extractCtrl.handle = handle;
1216 extractCtrl.first_record = 1;
1218 extract_set_store_data_prepare(&extractCtrl);
1221 grs_extract_tree(&extractCtrl, n);
1223 if (rec->size[recInfo_delKeys])
1225 struct recKeys delkeys;
1226 struct sortKeys sortkeys;
1228 delkeys.buf_used = rec->size[recInfo_delKeys];
1229 delkeys.buf = rec->info[recInfo_delKeys];
1231 sortkeys.buf_used = rec->size[recInfo_sortKeys];
1232 sortkeys.buf = rec->info[recInfo_sortKeys];
1234 extract_flushSortKeys (zh, rec->sysno, 0, &sortkeys);
1235 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1237 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->reg->keys);
1238 extract_flushSortKeys (zh, rec->sysno, 1, &zh->reg->sortKeys);
1240 xfree (rec->info[recInfo_delKeys]);
1241 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
1242 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
1243 zh->reg->keys.buf = NULL;
1244 zh->reg->keys.buf_max = 0;
1246 xfree (rec->info[recInfo_sortKeys]);
1247 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
1248 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
1249 zh->reg->sortKeys.buf = NULL;
1250 zh->reg->sortKeys.buf_max = 0;
1255 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1256 int cmd, struct recKeys *reckeys)
1258 void *decode_handle = iscz1_start();
1261 ZebraExplainInfo zei = zh->reg->zei;
1263 if (!zh->reg->key_buf)
1265 int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
1268 yaz_log(YLOG_WARN, "Invalid memory setting, using default 8 MB");
1271 /* FIXME: That "8" should be in a default settings include */
1272 /* not hard-coded here! -H */
1273 zh->reg->key_buf = (char**) xmalloc (mem);
1274 zh->reg->ptr_top = mem/sizeof(char*);
1276 zh->reg->key_buf_used = 0;
1277 zh->reg->key_file_no = 0;
1279 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1281 while (off < reckeys->buf_used)
1283 const char *src = reckeys->buf + off;
1285 char *dst = (char*) &key;
1287 iscz1_decode(decode_handle, &dst, &src);
1288 assert(key.len == 4);
1290 if (zh->reg->key_buf_used + 1024 >
1291 (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
1292 extract_flushWriteKeys (zh,0);
1294 assert(zh->reg->ptr_i > 0);
1295 (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
1296 (char*)zh->reg->key_buf + zh->reg->key_buf_used;
1298 ch = (int) key.mem[0]; /* ordinal for field/use/attribute */
1300 zh->reg->key_buf_used +=
1301 key_SU_encode (ch,((char*)zh->reg->key_buf) +
1302 zh->reg->key_buf_used);
1304 ((char*)zh->reg->key_buf) [(zh->reg->key_buf_used)++] = *src++;
1306 ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = '\0';
1307 ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = cmd;
1310 if (key.mem[1]) /* filter specified record ID */
1311 key.mem[0] = key.mem[1];
1314 key.mem[1] = key.mem[2]; /* section_id */
1315 key.mem[2] = key.mem[3]; /* sequence .. */
1317 memcpy ((char*)zh->reg->key_buf + zh->reg->key_buf_used,
1319 (zh->reg->key_buf_used) += sizeof(key);
1320 off = src - reckeys->buf;
1322 assert (off == reckeys->buf_used);
1323 iscz1_stop(decode_handle);
1326 void extract_flushWriteKeys (ZebraHandle zh, int final)
1327 /* optimizing: if final=1, and no files written yet */
1328 /* push the keys directly to merge, sidestepping the */
1329 /* temp file altogether. Speeds small updates */
1332 char out_fname[200];
1334 struct encode_info encode_info;
1335 int ptr_i = zh->reg->ptr_i;
1340 if (!zh->reg->key_buf || ptr_i <= 0)
1342 yaz_log (YLOG_DEBUG, " nothing to flush section=%d buf=%p i=%d",
1343 zh->reg->key_file_no, zh->reg->key_buf, ptr_i);
1344 yaz_log (YLOG_DEBUG, " buf=%p ",
1346 yaz_log (YLOG_DEBUG, " ptr=%d ",zh->reg->ptr_i);
1347 yaz_log (YLOG_DEBUG, " reg=%p ",zh->reg);
1352 (zh->reg->key_file_no)++;
1353 yaz_log (YLOG_LOG, "sorting section %d", (zh->reg->key_file_no));
1354 yaz_log (YLOG_DEBUG, " sort_buff at %p n=%d",
1355 zh->reg->key_buf + zh->reg->ptr_top - ptr_i,ptr_i);
1357 qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
1358 sizeof(char*), key_qsort_compare);
1360 /* zebra.cfg: tempfiles:
1361 Y: always use temp files (old way)
1362 A: use temp files, if more than one (auto)
1363 = if this is both the last and the first
1364 N: never bother with temp files (new) */
1366 temp_policy=toupper(res_get_def(zh->res,"tempfiles","auto")[0]);
1367 if (temp_policy != 'Y' && temp_policy != 'N' && temp_policy != 'A') {
1368 yaz_log (YLOG_WARN, "Illegal tempfiles setting '%c'. using 'Auto' ",
1373 if ( ( temp_policy =='N' ) || /* always from memory */
1374 ( ( temp_policy =='A' ) && /* automatic */
1375 (zh->reg->key_file_no == 1) && /* this is first time */
1376 (final) ) ) /* and last (=only) time */
1377 { /* go directly from memory */
1378 zh->reg->key_file_no =0; /* signal not to read files */
1379 zebra_index_merge(zh);
1381 zh->reg->key_buf_used = 0;
1385 /* Not doing directly from memory, write into a temp file */
1386 extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
1388 if (!(outf = fopen (out_fname, "wb")))
1390 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1393 yaz_log (YLOG_LOG, "writing section %d", zh->reg->key_file_no);
1394 prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1396 encode_key_init (&encode_info);
1397 encode_key_write (cp, &encode_info, outf);
1401 cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1402 if (strcmp (cp, prevcp))
1404 encode_key_flush ( &encode_info, outf);
1405 encode_key_init (&encode_info);
1406 encode_key_write (cp, &encode_info, outf);
1410 encode_key_write (cp + strlen(cp), &encode_info, outf);
1412 encode_key_flush ( &encode_info, outf);
1414 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1415 extract_get_fname_tmp (out_fname, key_file_no);
1417 if (!(outf = fopen (out_fname, "wb")))
1419 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1422 yaz_log (YLOG_LOG, "writing section %d", key_file_no);
1424 prevcp = key_buf[ptr_top-i];
1426 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1428 key_y_len = strlen(prevcp)+1;
1430 yaz_log (YLOG_LOG, "key_y_len: %2d %02x %02x %s",
1431 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1433 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1434 sizeof(char*), key_y_compare);
1435 cp = key_buf[ptr_top-ptr_i];
1437 encode_key_init (&encode_info);
1438 encode_key_write (cp, &encode_info, outf);
1441 cp = key_buf[ptr_top-ptr_i];
1442 encode_key_write (cp+key_y_len, &encode_info, outf);
1444 encode_key_flush ( &encode_info, outf);
1447 prevcp = key_buf[ptr_top-ptr_i];
1452 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fclose %s", out_fname);
1455 yaz_log (YLOG_LOG, "finished section %d", zh->reg->key_file_no);
1457 zh->reg->key_buf_used = 0;
1460 void extract_add_it_key (ZebraHandle zh,
1462 const char *str, int slen, struct it_key *key)
1465 struct recKeys *keys = &zh->reg->keys;
1466 const char *src = (char*) key;
1468 if (keys->buf_used+1024 > keys->buf_max)
1470 char *b = (char *) xmalloc (keys->buf_max += 128000);
1471 if (keys->buf_used > 0)
1472 memcpy (b, keys->buf, keys->buf_used);
1476 dst = keys->buf + keys->buf_used;
1478 iscz1_encode(keys->codec_handle, &dst, &src);
1481 memcpy (dst, str, slen);
1484 keys->buf_used = dst - keys->buf;
1487 void extract_add_index_string (RecWord *p, const char *str, int length)
1491 ZebraHandle zh = p->extractCtrl->handle;
1492 ZebraExplainInfo zei = zh->reg->zei;
1497 ch = zebraExplain_lookup_attr_str(zei, p->attrStr);
1499 ch = zebraExplain_add_attr_str(zei, p->attrStr);
1503 ch = zebraExplain_lookup_attr_su(zei, p->attrSet, p->attrUse);
1505 ch = zebraExplain_add_attr_su(zei, p->attrSet, p->attrUse);
1509 key.mem[1] = p->record_id;
1510 key.mem[2] = p->section_id;
1511 key.mem[3] = p->seqno;
1514 /* just for debugging .. */
1515 yaz_log(YLOG_LOG, "add: set=%d use=%d "
1516 "record_id=%lld section_id=%lld seqno=%lld",
1517 p->attrSet, p->attrUse, p->record_id, p->section_id, p->seqno);
1520 extract_add_it_key(p->extractCtrl->handle, p->reg_type, str,
1524 static void extract_add_sort_string (RecWord *p, const char *str,
1527 ZebraHandle zh = p->extractCtrl->handle;
1528 struct sortKeys *sk = &zh->reg->sortKeys;
1531 while (off < sk->buf_used)
1535 off += key_SU_decode(&set, sk->buf + off);
1536 off += key_SU_decode(&use, sk->buf + off);
1537 off += key_SU_decode(&slen, sk->buf + off);
1539 if (p->attrSet == set && p->attrUse == use)
1542 assert (off == sk->buf_used);
1544 if (sk->buf_used + IT_MAX_WORD > sk->buf_max)
1548 b = (char *) xmalloc (sk->buf_max += 128000);
1549 if (sk->buf_used > 0)
1550 memcpy (b, sk->buf, sk->buf_used);
1554 off += key_SU_encode(p->attrSet, sk->buf + off);
1555 off += key_SU_encode(p->attrUse, sk->buf + off);
1556 off += key_SU_encode(length, sk->buf + off);
1557 memcpy (sk->buf + off, str, length);
1558 sk->buf_used = off + length;
1561 void extract_add_string (RecWord *p, const char *string, int length)
1563 assert (length > 0);
1564 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
1565 extract_add_sort_string (p, string, length);
1567 extract_add_index_string (p, string, length);
1570 static void extract_add_incomplete_field (RecWord *p)
1572 const char *b = p->term_buf;
1573 int remain = p->term_len;
1574 const char **map = 0;
1576 yaz_log(YLOG_DEBUG, "Incomplete field, w='%.*s'", p->term_len, p->term_buf);
1579 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
1583 char buf[IT_MAX_WORD+1];
1587 while (map && *map && **map == *CHR_SPACE)
1589 remain = p->term_len - (b - p->term_buf);
1591 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
1598 while (map && *map && **map != *CHR_SPACE)
1600 const char *cp = *map;
1602 while (i < IT_MAX_WORD && *cp)
1604 remain = p->term_len - (b - p->term_buf);
1606 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
1612 extract_add_string (p, buf, i);
1617 static void extract_add_complete_field (RecWord *p)
1619 const char *b = p->term_buf;
1620 char buf[IT_MAX_WORD+1];
1621 const char **map = 0;
1622 int i = 0, remain = p->term_len;
1624 yaz_log(YLOG_DEBUG, "Complete field, w='%.*s'",
1625 p->term_len, p->term_buf);
1628 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain, 1);
1630 while (remain > 0 && i < IT_MAX_WORD)
1632 while (map && *map && **map == *CHR_SPACE)
1634 remain = p->term_len - (b - p->term_buf);
1638 int first = i ? 0 : 1; /* first position */
1639 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, first);
1647 if (i && i < IT_MAX_WORD)
1648 buf[i++] = *CHR_SPACE;
1649 while (map && *map && **map != *CHR_SPACE)
1651 const char *cp = *map;
1653 if (**map == *CHR_CUT)
1659 if (i >= IT_MAX_WORD)
1661 yaz_log(YLOG_DEBUG, "Adding string to index '%d'", **map);
1662 while (i < IT_MAX_WORD && *cp)
1665 remain = p->term_len - (b - p->term_buf);
1668 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
1677 extract_add_string (p, buf, i);
1680 void extract_token_add (RecWord *p)
1684 yaz_log (YLOG_LOG, "token_add "
1685 "reg_type=%c attrSet=%d attrUse=%d seqno=%d s=%.*s",
1686 p->reg_type, p->attrSet, p->attrUse, p->seqno, p->length,
1689 if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
1690 p->term_buf, p->term_len)))
1692 p->term_buf = wrbuf_buf(wrbuf);
1693 p->term_len = wrbuf_len(wrbuf);
1695 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
1696 extract_add_complete_field (p);
1698 extract_add_incomplete_field(p);
1701 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1702 void *buf, size_t sz)
1704 ZebraHandle zh = (ZebraHandle) p->handle;
1706 xfree(zh->store_data_buf);
1707 zh->store_data_buf = 0;
1708 zh->store_data_size = 0;
1711 zh->store_data_buf = xmalloc(sz);
1712 zh->store_data_size = sz;
1713 memcpy(zh->store_data_buf, buf, sz);
1717 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1719 ZebraHandle zh = (ZebraHandle) p->handle;
1720 xfree(zh->store_data_buf);
1721 zh->store_data_buf = 0;
1722 zh->store_data_size = 0;
1723 p->setStoreData = extract_set_store_data_cb;
1726 void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1728 ZebraHandle zh = (ZebraHandle) p->handle;
1729 zebraExplain_addSchema (zh->reg->zei, oid);
1732 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1733 int cmd, struct sortKeys *sk)
1735 SortIdx sortIdx = zh->reg->sortIdx;
1738 sortIdx_sysno (sortIdx, sysno);
1740 while (off < sk->buf_used)
1744 off += key_SU_decode(&set, sk->buf + off);
1745 off += key_SU_decode(&use, sk->buf + off);
1746 off += key_SU_decode(&slen, sk->buf + off);
1748 sortIdx_type(sortIdx, use);
1750 sortIdx_add(sortIdx, sk->buf + off, slen);
1752 sortIdx_add(sortIdx, "", 1);
1757 void encode_key_init (struct encode_info *i)
1766 i->encode_handle = iscz1_start();
1772 /* this is the old encode_key_write
1773 * may be deleted once we are confident that the new works
1776 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1779 char *bp = i->buf, *bp0;
1780 const char *src = (char *) &key;
1782 /* copy term to output buf */
1783 while ((*bp++ = *k++))
1785 /* and copy & align key so we can mangle */
1786 memcpy (&key, k+1, sizeof(struct it_key)); /* *k is insert/delete */
1789 iscz1_encode(i->encode_handle, &bp, &src);
1790 *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */
1791 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1793 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
1798 void encode_key_flush (struct encode_info *i, FILE *outf)
1799 { /* dummy routine */
1800 iscz1_stop(i->encode_handle);
1805 /* new encode_key_write
1806 * The idea is to buffer one more key, and compare them
1807 * If we are going to delete and insert the same key,
1808 * we may as well not bother. Should make a difference in
1809 * updates with small modifications (appending to a mbox)
1811 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1816 if (*k) /* first time for new key */
1819 while ((*bp++ = *k++))
1821 i->keylen= bp - i->buf -1;
1822 assert(i->keylen+1+sizeof(struct it_key) < ENCODE_BUFLEN);
1826 bp=i->buf + i->keylen;
1831 memcpy (&key, k+1, sizeof(struct it_key));
1832 if (0==i->prevsys) /* no previous filter, fill up */
1834 i->prevsys=key.sysno;
1835 i->prevseq=key.seqno;
1838 else if ( (i->prevsys==key.sysno) &&
1839 (i->prevseq==key.seqno) &&
1841 { /* same numbers, diff cmd, they cancel out */
1845 { /* different stuff, write previous, move buf */
1846 bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
1847 if (i->sysno != i->prevsys)
1849 i->sysno = i->prevsys;
1852 else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
1854 return; /* ??? Filters some sort of duplicates away */
1855 /* ??? Can this ever happen -H 15oct02 */
1857 bp = encode_key_int (i->prevseq - i->seqno, bp);
1858 i->seqno = i->prevseq;
1859 i->cmd = i->prevcmd;
1860 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1862 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
1865 i->keylen=0; /* ok, it's written, forget it */
1866 i->prevsys=key.sysno;
1867 i->prevseq=key.seqno;
1872 void encode_key_flush (struct encode_info *i, FILE *outf)
1873 { /* flush the last key from i */
1874 char *bp =i->buf + i->keylen;
1877 return; /* nothing to flush */
1880 bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
1881 if (i->sysno != i->prevsys)
1883 i->sysno = i->prevsys;
1886 else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
1888 return; /* ??? Filters some sort of duplicates away */
1889 /* ??? Can this ever happen -H 15oct02 */
1891 bp = encode_key_int (i->prevseq - i->seqno, bp);
1892 i->seqno = i->prevseq;
1893 i->cmd = i->prevcmd;
1894 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1896 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
1899 i->keylen=0; /* ok, it's written, forget it */
1900 i->prevsys=0; /* forget the values too */