1 /* $Id: extract.c,v 1.181 2005-04-28 12:13:04 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 */
153 static struct file_read_info *file_read_start (int fd)
155 struct file_read_info *fi = (struct file_read_info *)
156 xmalloc (sizeof(*fi));
160 fi->file_moffset = 0;
166 static void file_read_stop (struct file_read_info *fi)
171 static off_t file_seek (void *handle, off_t offset)
173 struct file_read_info *p = (struct file_read_info *) handle;
174 p->file_offset = offset;
175 return lseek (p->fd, offset, SEEK_SET);
178 static off_t file_tell (void *handle)
180 struct file_read_info *p = (struct file_read_info *) handle;
181 return p->file_offset;
184 static int file_read (void *handle, char *buf, size_t count)
186 struct file_read_info *p = (struct file_read_info *) handle;
189 r = read (fd, buf, count);
193 if (p->file_offset > p->file_max)
194 p->file_max = p->file_offset;
199 static void file_end (void *handle, off_t offset)
201 struct file_read_info *p = (struct file_read_info *) handle;
203 if (offset != p->file_moffset)
205 yaz_log(YLOG_LOG, "file_end OK");
206 p->file_moffset = offset;
211 yaz_log(YLOG_LOG, "file_end REJECTED");
215 static char *fileMatchStr (ZebraHandle zh,
216 struct recKeys *reckeys,
217 const char *fname, const char *spec)
219 static char dstBuf[2048]; /* static here ??? */
221 const char *s = spec;
222 static const char **w;
226 while (*s == ' ' || *s == '\t')
232 char attset_str[64], attname_str[64];
233 data1_attset *attset;
236 int attSet = 1, attUse = 1;
240 for (i = 0; *s && *s != ',' && *s != ')'; s++)
242 attset_str[i++] = *s;
243 attset_str[i] = '\0';
248 for (i = 0; *s && *s != ')'; s++)
250 attname_str[i++] = *s;
251 attname_str[i] = '\0';
254 if ((attset = data1_get_attset (zh->reg->dh, attset_str)))
257 attSet = attset->reference;
258 att = data1_getattbyname(zh->reg->dh, attset, attname_str);
262 attUse = atoi (attname_str);
264 w = searchRecordKey (zh, reckeys, attSet, attUse);
269 for (i = 0; i<32; i++)
274 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
275 spec, zh->m_group ? zh->m_group : "none");
280 for (i = 0; i<32; i++)
281 if (matchFlag[i] && w[i])
293 yaz_log (YLOG_WARN, "Record didn't contain match"
294 " fields in (%s,%s)", attset_str, attname_str);
302 const char *spec_src = NULL;
303 const char *s1 = ++s;
304 while (*s1 && *s1 != ' ' && *s1 != '\t')
310 memcpy (special, s, spec_len);
311 special[spec_len] = '\0';
314 if (!strcmp (special, "group"))
315 spec_src = zh->m_group;
316 else if (!strcmp (special, "database"))
317 spec_src = zh->basenames[0];
318 else if (!strcmp (special, "filename")) {
321 else if (!strcmp (special, "type"))
322 spec_src = zh->m_record_type;
327 strcpy (dst, spec_src);
328 dst += strlen (spec_src);
331 else if (*s == '\"' || *s == '\'')
333 int stopMarker = *s++;
337 while (*s && *s != stopMarker)
340 tmpString[i++] = *s++;
345 strcpy (dst, tmpString);
346 dst += strlen (tmpString);
350 yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
351 spec, zh->m_group ? zh->m_group : "none");
358 yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
359 fname, zh->m_group ? zh->m_group : "none");
366 struct recordLogInfo {
369 struct recordGroup *rGroup;
372 void create_rec_keys_codec(struct recKeys *keys)
375 iscz1_reset(keys->codec_handle);
378 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
381 for (i = 0; i<256; i++)
383 if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
388 ctrl->zebra_maps = zh->reg->zebra_maps;
389 ctrl->flagShowRecords = !zh->m_flag_rw;
392 static int file_extract_record(ZebraHandle zh,
393 SYSNO *sysno, const char *fname,
395 struct file_read_info *fi,
398 void *recTypeClientData)
400 RecordAttr *recordAttr;
402 const char *matchStr = 0;
405 off_t recordOffset = 0;
407 /* announce database */
408 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
410 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
411 zh->m_explain_database))
417 struct recExtractCtrl extractCtrl;
419 /* we are going to read from a file, so prepare the extraction */
420 create_rec_keys_codec(&zh->reg->keys);
422 zh->reg->sortKeys.buf_used = 0;
424 recordOffset = fi->file_moffset;
425 extractCtrl.handle = zh;
426 extractCtrl.offset = fi->file_moffset;
427 extractCtrl.readf = file_read;
428 extractCtrl.seekf = file_seek;
429 extractCtrl.tellf = file_tell;
430 extractCtrl.endf = file_end;
432 extractCtrl.init = extract_init;
433 extractCtrl.tokenAdd = extract_token_add;
434 extractCtrl.schemaAdd = extract_schema_add;
435 extractCtrl.dh = zh->reg->dh;
436 extractCtrl.match_criteria[0] = '\0';
437 extractCtrl.first_record = fi->file_offset ? 0 : 1;
439 extract_set_store_data_prepare(&extractCtrl);
441 init_extractCtrl(zh, &extractCtrl);
444 printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset);
448 sprintf (msg, "%s:" PRINTF_OFF_T , fname, recordOffset);
449 yaz_log_init_prefix2 (msg);
452 r = (*recType->extract)(recTypeClientData, &extractCtrl);
454 yaz_log_init_prefix2 (0);
455 if (r == RECCTRL_EXTRACT_EOF)
457 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
459 /* error occured during extraction ... */
461 zh->records_processed < zh->m_file_verbose_limit)
463 yaz_log (YLOG_WARN, "fail %s %s " PRINTF_OFF_T, zh->m_record_type,
464 fname, recordOffset);
468 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
470 /* error occured during extraction ... */
472 zh->records_processed < zh->m_file_verbose_limit)
474 yaz_log (YLOG_WARN, "no filter for %s %s "
475 PRINTF_OFF_T, zh->m_record_type,
476 fname, recordOffset);
480 if (extractCtrl.match_criteria[0])
481 matchStr = extractCtrl.match_criteria;
484 /* perform match if sysno not known and if match criteria is specified */
490 if (matchStr == 0 && zh->m_record_id && *zh->m_record_id)
493 matchStr = fileMatchStr (zh, &zh->reg->keys, fname,
497 yaz_log(YLOG_WARN, "Bad match criteria");
503 char *rinfo = dict_lookup (zh->reg->matchDict, matchStr);
506 assert(*rinfo == sizeof(*sysno));
507 memcpy (sysno, rinfo+1, sizeof(*sysno));
511 if (! *sysno && zh->reg->keys.buf_used == 0)
513 /* the extraction process returned no information - the record
514 is probably empty - unless flagShowRecords is in use */
518 if (zh->records_processed < zh->m_file_verbose_limit)
519 yaz_log (YLOG_WARN, "empty %s %s " PRINTF_OFF_T, zh->m_record_type,
520 fname, recordOffset);
529 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T, zh->m_record_type,
530 fname, recordOffset);
531 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
534 if (zh->records_processed < zh->m_file_verbose_limit)
535 yaz_log (YLOG_LOG, "add %s %s " PRINTF_OFF_T, zh->m_record_type,
536 fname, recordOffset);
537 rec = rec_new (zh->reg->records);
541 recordAttr = rec_init_attr (zh->reg->zei, rec);
545 dict_insert (zh->reg->matchDict, matchStr, sizeof(*sysno), sysno);
547 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
548 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
550 zh->records_inserted++;
554 /* record already exists */
555 struct recKeys delkeys;
556 struct sortKeys sortKeys;
558 rec = rec_get (zh->reg->records, *sysno);
561 recordAttr = rec_init_attr (zh->reg->zei, rec);
563 if (!force_update && recordAttr->runNumber ==
564 zebraExplain_runNumberIncrement (zh->reg->zei, 0))
566 yaz_log (YLOG_LOG, "run number = " ZINT_FORMAT,
567 recordAttr->runNumber);
568 yaz_log (YLOG_LOG, "skipped %s %s " PRINTF_OFF_T,
569 zh->m_record_type, fname, recordOffset);
570 extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
575 delkeys.buf_used = rec->size[recInfo_delKeys];
576 delkeys.buf = rec->info[recInfo_delKeys];
578 sortKeys.buf_used = rec->size[recInfo_sortKeys];
579 sortKeys.buf = rec->info[recInfo_sortKeys];
581 extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
582 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
585 /* record going to be deleted */
586 if (!delkeys.buf_used)
588 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
589 zh->m_record_type, fname, recordOffset);
590 yaz_log (YLOG_WARN, "cannot delete file above, storeKeys false");
594 if (zh->records_processed < zh->m_file_verbose_limit)
595 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
596 zh->m_record_type, fname, recordOffset);
597 zh->records_deleted++;
599 dict_delete (zh->reg->matchDict, matchStr);
600 rec_del (zh->reg->records, &rec);
608 /* record going to be updated */
609 if (!delkeys.buf_used)
611 yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
612 zh->m_record_type, fname, recordOffset);
613 yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
617 if (zh->records_processed < zh->m_file_verbose_limit)
618 yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
619 zh->m_record_type, fname, recordOffset);
620 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
621 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
622 zh->records_updated++;
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 (zh->reg->keys.buf_used > 0 && zh->m_store_keys == 1)
640 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
641 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
642 zh->reg->keys.buf = NULL;
643 zh->reg->keys.buf_max = 0;
647 rec->info[recInfo_delKeys] = NULL;
648 rec->size[recInfo_delKeys] = 0;
651 /* update sort keys */
652 xfree (rec->info[recInfo_sortKeys]);
654 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
655 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
656 zh->reg->sortKeys.buf = NULL;
657 zh->reg->sortKeys.buf_max = 0;
659 /* save file size of original record */
660 zebraExplain_recordBytesIncrement (zh->reg->zei,
661 - recordAttr->recordSize);
662 recordAttr->recordSize = fi->file_moffset - recordOffset;
663 if (!recordAttr->recordSize)
664 recordAttr->recordSize = fi->file_max - recordOffset;
665 zebraExplain_recordBytesIncrement (zh->reg->zei,
666 recordAttr->recordSize);
668 /* set run-number for this record */
669 recordAttr->runNumber = zebraExplain_runNumberIncrement (zh->reg->zei,
672 /* update store data */
673 xfree (rec->info[recInfo_storeData]);
674 if (zh->store_data_buf)
676 rec->size[recInfo_storeData] = zh->store_data_size;
677 rec->info[recInfo_storeData] = zh->store_data_buf;
678 zh->store_data_buf = 0;
680 else if (zh->m_store_data)
682 rec->size[recInfo_storeData] = recordAttr->recordSize;
683 rec->info[recInfo_storeData] = (char *)
684 xmalloc (recordAttr->recordSize);
685 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
687 yaz_log (YLOG_ERRNO|YLOG_FATAL, "seek to " PRINTF_OFF_T " in %s",
688 recordOffset, fname);
691 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
692 < recordAttr->recordSize)
694 yaz_log (YLOG_ERRNO|YLOG_FATAL, "read %d bytes of %s",
695 recordAttr->recordSize, fname);
701 rec->info[recInfo_storeData] = NULL;
702 rec->size[recInfo_storeData] = 0;
704 /* update database name */
705 xfree (rec->info[recInfo_databaseName]);
706 rec->info[recInfo_databaseName] =
707 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
710 recordAttr->recordOffset = recordOffset;
712 /* commit this record */
713 rec_put (zh->reg->records, &rec);
718 int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
725 struct file_read_info *fi;
726 const char *original_record_type = 0;
728 void *recTypeClientData;
730 if (!zh->m_group || !*zh->m_group)
733 sprintf (gprefix, "%s.", zh->m_group);
735 yaz_log (YLOG_DEBUG, "fileExtract %s", fname);
737 /* determine file extension */
739 for (i = strlen(fname); --i >= 0; )
742 else if (fname[i] == '.')
744 strcpy (ext, fname+i+1);
747 /* determine file type - depending on extension */
748 original_record_type = zh->m_record_type;
749 if (!zh->m_record_type)
751 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
752 zh->m_record_type = res_get (zh->res, ext_res);
754 if (!zh->m_record_type)
756 if (zh->records_processed < zh->m_file_verbose_limit)
757 yaz_log (YLOG_LOG, "? %s", fname);
760 /* determine match criteria */
761 if (!zh->m_record_id)
763 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
764 zh->m_record_id = res_get (zh->res, ext_res);
768 recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
769 &recTypeClientData)))
771 yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
775 switch(recType->version)
780 yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
782 if (sysno && deleteFlag)
788 if (zh->path_reg && !yaz_is_abspath (fname))
790 strcpy (full_rep, zh->path_reg);
791 strcat (full_rep, "/");
792 strcat (full_rep, fname);
795 strcpy (full_rep, fname);
798 if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
800 yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
801 zh->m_record_type = original_record_type;
805 fi = file_read_start (fd);
808 fi->file_moffset = fi->file_offset;
809 fi->file_more = 0; /* file_end not called (yet) */
810 r = file_extract_record (zh, sysno, fname, deleteFlag, fi, 1,
811 recType, recTypeClientData);
812 yaz_log(YLOG_LOG, "file_extract_record returned %d", r);
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 create_rec_keys_codec(&zh->reg->keys);
876 zh->reg->sortKeys.buf_used = 0;
878 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
880 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
881 zh->m_explain_database))
885 if (recordType && *recordType)
887 yaz_log (YLOG_DEBUG, "Record type explicitly specified: %s", recordType);
888 recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
893 if (!(zh->m_record_type))
895 yaz_log (YLOG_WARN, "No such record type defined");
898 yaz_log (YLOG_DEBUG, "Get record type from rgroup: %s",zh->m_record_type);
899 recType = recType_byName (zh->reg->recTypes, zh->res,
900 zh->m_record_type, &clientData);
901 recordType = zh->m_record_type;
906 yaz_log (YLOG_WARN, "No such record type: %s", zh->m_record_type);
910 extractCtrl.init = extract_init;
911 extractCtrl.tokenAdd = extract_token_add;
912 extractCtrl.schemaAdd = extract_schema_add;
913 extractCtrl.dh = zh->reg->dh;
914 extractCtrl.handle = zh;
915 extractCtrl.match_criteria[0] = '\0';
917 init_extractCtrl(zh, &extractCtrl);
919 extract_set_store_data_prepare(&extractCtrl);
921 r = (*recType->extract)(clientData, &extractCtrl);
923 if (r == RECCTRL_EXTRACT_EOF)
925 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
927 /* error occured during extraction ... */
928 yaz_log (YLOG_WARN, "extract error: generic");
931 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
933 /* error occured during extraction ... */
934 yaz_log (YLOG_WARN, "extract error: no such filter");
940 if (extractCtrl.match_criteria[0])
941 match_criteria = extractCtrl.match_criteria;
945 if (match_criteria && *match_criteria) {
946 matchStr = match_criteria;
948 if (zh->m_record_id && *zh->m_record_id) {
949 matchStr = fileMatchStr (zh, &zh->reg->keys, pr_fname,
953 yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
959 rinfo = dict_lookup (zh->reg->matchDict, matchStr);
962 assert(*rinfo == sizeof(*sysno));
963 memcpy (sysno, rinfo+1, sizeof(*sysno));
967 if (zh->reg->keys.buf_used == 0)
969 /* the extraction process returned no information - the record
970 is probably empty - unless flagShowRecords is in use */
981 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
982 pr_fname, (long) recordOffset);
983 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
987 yaz_log (YLOG_LOG, "add %s %s %ld", recordType, pr_fname,
988 (long) recordOffset);
989 rec = rec_new (zh->reg->records);
993 recordAttr = rec_init_attr (zh->reg->zei, rec);
997 dict_insert (zh->reg->matchDict, matchStr,
998 sizeof(*sysno), sysno);
1000 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1001 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
1003 zh->records_inserted++;
1007 /* record already exists */
1008 struct recKeys delkeys;
1009 struct sortKeys sortKeys;
1014 yaz_log (YLOG_LOG, "skipped %s %s %ld",
1015 recordType, pr_fname, (long) recordOffset);
1020 rec = rec_get (zh->reg->records, *sysno);
1023 recordAttr = rec_init_attr (zh->reg->zei, rec);
1025 if (!force_update) {
1026 if (recordAttr->runNumber ==
1027 zebraExplain_runNumberIncrement (zh->reg->zei, 0))
1030 yaz_log (YLOG_LOG, "skipped %s %s %ld", recordType,
1031 pr_fname, (long) recordOffset);
1032 extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
1039 delkeys.buf_used = rec->size[recInfo_delKeys];
1040 delkeys.buf = rec->info[recInfo_delKeys];
1042 sortKeys.buf_used = rec->size[recInfo_sortKeys];
1043 sortKeys.buf = rec->info[recInfo_sortKeys];
1045 extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
1046 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1049 /* record going to be deleted */
1050 if (!delkeys.buf_used)
1054 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1055 pr_fname, (long) recordOffset);
1056 yaz_log (YLOG_WARN, "cannot delete file above, "
1063 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1064 pr_fname, (long) recordOffset);
1065 zh->records_deleted++;
1067 dict_delete (zh->reg->matchDict, matchStr);
1068 rec_del (zh->reg->records, &rec);
1076 /* record going to be updated */
1077 if (!delkeys.buf_used)
1081 yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
1082 pr_fname, (long) recordOffset);
1083 yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
1089 yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
1090 pr_fname, (long) recordOffset);
1091 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1092 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
1093 zh->records_updated++;
1097 /* update file type */
1098 xfree (rec->info[recInfo_fileType]);
1099 rec->info[recInfo_fileType] =
1100 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1102 /* update filename */
1103 xfree (rec->info[recInfo_filename]);
1104 rec->info[recInfo_filename] =
1105 rec_strdup (fname, &rec->size[recInfo_filename]);
1107 /* update delete keys */
1108 xfree (rec->info[recInfo_delKeys]);
1109 if (zh->reg->keys.buf_used > 0 && zh->m_store_keys == 1)
1111 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
1112 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
1113 zh->reg->keys.buf = NULL;
1114 zh->reg->keys.buf_max = 0;
1118 rec->info[recInfo_delKeys] = NULL;
1119 rec->size[recInfo_delKeys] = 0;
1122 /* update sort keys */
1123 xfree (rec->info[recInfo_sortKeys]);
1125 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
1126 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
1127 zh->reg->sortKeys.buf = NULL;
1128 zh->reg->sortKeys.buf_max = 0;
1130 /* save file size of original record */
1131 zebraExplain_recordBytesIncrement (zh->reg->zei,
1132 - recordAttr->recordSize);
1134 recordAttr->recordSize = fi->file_moffset - recordOffset;
1135 if (!recordAttr->recordSize)
1136 recordAttr->recordSize = fi->file_max - recordOffset;
1138 recordAttr->recordSize = buf_size;
1140 zebraExplain_recordBytesIncrement (zh->reg->zei,
1141 recordAttr->recordSize);
1143 /* set run-number for this record */
1144 recordAttr->runNumber =
1145 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
1147 /* update store data */
1148 xfree (rec->info[recInfo_storeData]);
1150 /* update store data */
1151 if (zh->store_data_buf)
1153 rec->size[recInfo_storeData] = zh->store_data_size;
1154 rec->info[recInfo_storeData] = zh->store_data_buf;
1155 zh->store_data_buf = 0;
1157 else if (zh->m_store_data)
1159 rec->size[recInfo_storeData] = recordAttr->recordSize;
1160 rec->info[recInfo_storeData] = (char *)
1161 xmalloc (recordAttr->recordSize);
1162 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1166 rec->info[recInfo_storeData] = NULL;
1167 rec->size[recInfo_storeData] = 0;
1169 /* update database name */
1170 xfree (rec->info[recInfo_databaseName]);
1171 rec->info[recInfo_databaseName] =
1172 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
1175 recordAttr->recordOffset = recordOffset;
1177 /* commit this record */
1178 rec_put (zh->reg->records, &rec);
1183 int explain_extract (void *handle, Record rec, data1_node *n)
1185 ZebraHandle zh = (ZebraHandle) handle;
1186 struct recExtractCtrl extractCtrl;
1188 if (zebraExplain_curDatabase (zh->reg->zei,
1189 rec->info[recInfo_databaseName]))
1192 if (zebraExplain_newDatabase (zh->reg->zei,
1193 rec->info[recInfo_databaseName], 0))
1197 create_rec_keys_codec(&zh->reg->keys);
1199 zh->reg->sortKeys.buf_used = 0;
1201 extractCtrl.init = extract_init;
1202 extractCtrl.tokenAdd = extract_token_add;
1203 extractCtrl.schemaAdd = extract_schema_add;
1204 extractCtrl.dh = zh->reg->dh;
1206 init_extractCtrl(zh, &extractCtrl);
1208 extractCtrl.flagShowRecords = 0;
1209 extractCtrl.match_criteria[0] = '\0';
1210 extractCtrl.handle = handle;
1211 extractCtrl.first_record = 1;
1213 extract_set_store_data_prepare(&extractCtrl);
1216 grs_extract_tree(&extractCtrl, n);
1218 if (rec->size[recInfo_delKeys])
1220 struct recKeys delkeys;
1221 struct sortKeys sortkeys;
1223 delkeys.buf_used = rec->size[recInfo_delKeys];
1224 delkeys.buf = rec->info[recInfo_delKeys];
1226 sortkeys.buf_used = rec->size[recInfo_sortKeys];
1227 sortkeys.buf = rec->info[recInfo_sortKeys];
1229 extract_flushSortKeys (zh, rec->sysno, 0, &sortkeys);
1230 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1232 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->reg->keys);
1233 extract_flushSortKeys (zh, rec->sysno, 1, &zh->reg->sortKeys);
1235 xfree (rec->info[recInfo_delKeys]);
1236 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
1237 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
1238 zh->reg->keys.buf = NULL;
1239 zh->reg->keys.buf_max = 0;
1241 xfree (rec->info[recInfo_sortKeys]);
1242 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
1243 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
1244 zh->reg->sortKeys.buf = NULL;
1245 zh->reg->sortKeys.buf_max = 0;
1250 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1251 int cmd, struct recKeys *reckeys)
1253 void *decode_handle = iscz1_start();
1256 ZebraExplainInfo zei = zh->reg->zei;
1258 if (!zh->reg->key_buf)
1260 int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
1263 yaz_log(YLOG_WARN, "Invalid memory setting, using default 8 MB");
1266 /* FIXME: That "8" should be in a default settings include */
1267 /* not hard-coded here! -H */
1268 zh->reg->key_buf = (char**) xmalloc (mem);
1269 zh->reg->ptr_top = mem/sizeof(char*);
1271 zh->reg->key_buf_used = 0;
1272 zh->reg->key_file_no = 0;
1274 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1276 while (off < reckeys->buf_used)
1278 const char *src = reckeys->buf + off;
1280 char *dst = (char*) &key;
1282 iscz1_decode(decode_handle, &dst, &src);
1283 assert(key.len == 4);
1285 if (zh->reg->key_buf_used + 1024 >
1286 (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
1287 extract_flushWriteKeys (zh,0);
1289 assert(zh->reg->ptr_i > 0);
1290 (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
1291 (char*)zh->reg->key_buf + zh->reg->key_buf_used;
1293 ch = (int) key.mem[0]; /* ordinal for field/use/attribute */
1295 zh->reg->key_buf_used +=
1296 key_SU_encode (ch,((char*)zh->reg->key_buf) +
1297 zh->reg->key_buf_used);
1299 ((char*)zh->reg->key_buf) [(zh->reg->key_buf_used)++] = *src++;
1301 ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = '\0';
1302 ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = cmd;
1305 if (key.mem[1]) /* filter specified record ID */
1306 key.mem[0] = key.mem[1];
1309 key.mem[1] = key.mem[2]; /* section_id */
1310 key.mem[2] = key.mem[3]; /* sequence .. */
1312 memcpy ((char*)zh->reg->key_buf + zh->reg->key_buf_used,
1314 (zh->reg->key_buf_used) += sizeof(key);
1315 off = src - reckeys->buf;
1317 assert (off == reckeys->buf_used);
1318 iscz1_stop(decode_handle);
1321 void extract_flushWriteKeys (ZebraHandle zh, int final)
1322 /* optimizing: if final=1, and no files written yet */
1323 /* push the keys directly to merge, sidestepping the */
1324 /* temp file altogether. Speeds small updates */
1327 char out_fname[200];
1329 struct encode_info encode_info;
1330 int ptr_i = zh->reg->ptr_i;
1335 if (!zh->reg->key_buf || ptr_i <= 0)
1337 yaz_log (YLOG_DEBUG, " nothing to flush section=%d buf=%p i=%d",
1338 zh->reg->key_file_no, zh->reg->key_buf, ptr_i);
1339 yaz_log (YLOG_DEBUG, " buf=%p ",
1341 yaz_log (YLOG_DEBUG, " ptr=%d ",zh->reg->ptr_i);
1342 yaz_log (YLOG_DEBUG, " reg=%p ",zh->reg);
1347 (zh->reg->key_file_no)++;
1348 yaz_log (YLOG_LOG, "sorting section %d", (zh->reg->key_file_no));
1349 yaz_log (YLOG_DEBUG, " sort_buff at %p n=%d",
1350 zh->reg->key_buf + zh->reg->ptr_top - ptr_i,ptr_i);
1352 qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
1353 sizeof(char*), key_qsort_compare);
1355 /* zebra.cfg: tempfiles:
1356 Y: always use temp files (old way)
1357 A: use temp files, if more than one (auto)
1358 = if this is both the last and the first
1359 N: never bother with temp files (new) */
1361 temp_policy=toupper(res_get_def(zh->res,"tempfiles","auto")[0]);
1362 if (temp_policy != 'Y' && temp_policy != 'N' && temp_policy != 'A') {
1363 yaz_log (YLOG_WARN, "Illegal tempfiles setting '%c'. using 'Auto' ",
1368 if ( ( temp_policy =='N' ) || /* always from memory */
1369 ( ( temp_policy =='A' ) && /* automatic */
1370 (zh->reg->key_file_no == 1) && /* this is first time */
1371 (final) ) ) /* and last (=only) time */
1372 { /* go directly from memory */
1373 zh->reg->key_file_no =0; /* signal not to read files */
1374 zebra_index_merge(zh);
1376 zh->reg->key_buf_used = 0;
1380 /* Not doing directly from memory, write into a temp file */
1381 extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
1383 if (!(outf = fopen (out_fname, "wb")))
1385 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1388 yaz_log (YLOG_LOG, "writing section %d", zh->reg->key_file_no);
1389 prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1391 encode_key_init (&encode_info);
1392 encode_key_write (cp, &encode_info, outf);
1396 cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1397 if (strcmp (cp, prevcp))
1399 encode_key_flush ( &encode_info, outf);
1400 encode_key_init (&encode_info);
1401 encode_key_write (cp, &encode_info, outf);
1405 encode_key_write (cp + strlen(cp), &encode_info, outf);
1407 encode_key_flush ( &encode_info, outf);
1409 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1410 extract_get_fname_tmp (out_fname, key_file_no);
1412 if (!(outf = fopen (out_fname, "wb")))
1414 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1417 yaz_log (YLOG_LOG, "writing section %d", key_file_no);
1419 prevcp = key_buf[ptr_top-i];
1421 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1423 key_y_len = strlen(prevcp)+1;
1425 yaz_log (YLOG_LOG, "key_y_len: %2d %02x %02x %s",
1426 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1428 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1429 sizeof(char*), key_y_compare);
1430 cp = key_buf[ptr_top-ptr_i];
1432 encode_key_init (&encode_info);
1433 encode_key_write (cp, &encode_info, outf);
1436 cp = key_buf[ptr_top-ptr_i];
1437 encode_key_write (cp+key_y_len, &encode_info, outf);
1439 encode_key_flush ( &encode_info, outf);
1442 prevcp = key_buf[ptr_top-ptr_i];
1447 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fclose %s", out_fname);
1450 yaz_log (YLOG_LOG, "finished section %d", zh->reg->key_file_no);
1452 zh->reg->key_buf_used = 0;
1455 void extract_add_it_key (ZebraHandle zh,
1457 const char *str, int slen, struct it_key *key)
1460 struct recKeys *keys = &zh->reg->keys;
1461 const char *src = (char*) key;
1463 if (keys->buf_used+1024 > keys->buf_max)
1465 char *b = (char *) xmalloc (keys->buf_max += 128000);
1466 if (keys->buf_used > 0)
1467 memcpy (b, keys->buf, keys->buf_used);
1471 dst = keys->buf + keys->buf_used;
1473 iscz1_encode(keys->codec_handle, &dst, &src);
1476 memcpy (dst, str, slen);
1479 keys->buf_used = dst - keys->buf;
1482 void extract_add_index_string (RecWord *p, const char *str, int length)
1486 ZebraHandle zh = p->extractCtrl->handle;
1487 ZebraExplainInfo zei = zh->reg->zei;
1492 ch = zebraExplain_lookup_attr_str(zei, p->attrStr);
1494 ch = zebraExplain_add_attr_str(zei, p->attrStr);
1498 ch = zebraExplain_lookup_attr_su(zei, p->attrSet, p->attrUse);
1500 ch = zebraExplain_add_attr_su(zei, p->attrSet, p->attrUse);
1504 key.mem[1] = p->record_id;
1505 key.mem[2] = p->section_id;
1506 key.mem[3] = p->seqno;
1509 /* just for debugging .. */
1510 yaz_log(YLOG_LOG, "add: set=%d use=%d "
1511 "record_id=%lld section_id=%lld seqno=%lld",
1512 p->attrSet, p->attrUse, p->record_id, p->section_id, p->seqno);
1515 extract_add_it_key(p->extractCtrl->handle, p->reg_type, str,
1519 static void extract_add_sort_string (RecWord *p, const char *str,
1522 ZebraHandle zh = p->extractCtrl->handle;
1523 struct sortKeys *sk = &zh->reg->sortKeys;
1526 while (off < sk->buf_used)
1530 off += key_SU_decode(&set, sk->buf + off);
1531 off += key_SU_decode(&use, sk->buf + off);
1532 off += key_SU_decode(&slen, sk->buf + off);
1534 if (p->attrSet == set && p->attrUse == use)
1537 assert (off == sk->buf_used);
1539 if (sk->buf_used + IT_MAX_WORD > sk->buf_max)
1543 b = (char *) xmalloc (sk->buf_max += 128000);
1544 if (sk->buf_used > 0)
1545 memcpy (b, sk->buf, sk->buf_used);
1549 off += key_SU_encode(p->attrSet, sk->buf + off);
1550 off += key_SU_encode(p->attrUse, sk->buf + off);
1551 off += key_SU_encode(length, sk->buf + off);
1552 memcpy (sk->buf + off, str, length);
1553 sk->buf_used = off + length;
1556 void extract_add_string (RecWord *p, const char *string, int length)
1558 assert (length > 0);
1559 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
1560 extract_add_sort_string (p, string, length);
1562 extract_add_index_string (p, string, length);
1565 static void extract_add_incomplete_field (RecWord *p)
1567 const char *b = p->term_buf;
1568 int remain = p->term_len;
1569 const char **map = 0;
1571 yaz_log(YLOG_DEBUG, "Incomplete field, w='%.*s'", p->term_len, p->term_buf);
1574 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
1578 char buf[IT_MAX_WORD+1];
1582 while (map && *map && **map == *CHR_SPACE)
1584 remain = p->term_len - (b - p->term_buf);
1586 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
1593 while (map && *map && **map != *CHR_SPACE)
1595 const char *cp = *map;
1597 while (i < IT_MAX_WORD && *cp)
1599 remain = p->term_len - (b - p->term_buf);
1601 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
1607 extract_add_string (p, buf, i);
1612 static void extract_add_complete_field (RecWord *p)
1614 const char *b = p->term_buf;
1615 char buf[IT_MAX_WORD+1];
1616 const char **map = 0;
1617 int i = 0, remain = p->term_len;
1619 yaz_log(YLOG_DEBUG, "Complete field, w='%.*s'",
1620 p->term_len, p->term_buf);
1623 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain, 1);
1625 while (remain > 0 && i < IT_MAX_WORD)
1627 while (map && *map && **map == *CHR_SPACE)
1629 remain = p->term_len - (b - p->term_buf);
1633 int first = i ? 0 : 1; /* first position */
1634 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, first);
1642 if (i && i < IT_MAX_WORD)
1643 buf[i++] = *CHR_SPACE;
1644 while (map && *map && **map != *CHR_SPACE)
1646 const char *cp = *map;
1648 if (**map == *CHR_CUT)
1654 if (i >= IT_MAX_WORD)
1656 yaz_log(YLOG_DEBUG, "Adding string to index '%d'", **map);
1657 while (i < IT_MAX_WORD && *cp)
1660 remain = p->term_len - (b - p->term_buf);
1663 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
1672 extract_add_string (p, buf, i);
1675 void extract_token_add (RecWord *p)
1679 yaz_log (YLOG_LOG, "token_add "
1680 "reg_type=%c attrSet=%d attrUse=%d seqno=%d s=%.*s",
1681 p->reg_type, p->attrSet, p->attrUse, p->seqno, p->length,
1684 if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
1685 p->term_buf, p->term_len)))
1687 p->term_buf = wrbuf_buf(wrbuf);
1688 p->term_len = wrbuf_len(wrbuf);
1690 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
1691 extract_add_complete_field (p);
1693 extract_add_incomplete_field(p);
1696 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1697 void *buf, size_t sz)
1699 ZebraHandle zh = (ZebraHandle) p->handle;
1701 xfree(zh->store_data_buf);
1702 zh->store_data_buf = 0;
1703 zh->store_data_size = 0;
1706 zh->store_data_buf = xmalloc(sz);
1707 zh->store_data_size = sz;
1708 memcpy(zh->store_data_buf, buf, sz);
1712 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1714 ZebraHandle zh = (ZebraHandle) p->handle;
1715 xfree(zh->store_data_buf);
1716 zh->store_data_buf = 0;
1717 zh->store_data_size = 0;
1718 p->setStoreData = extract_set_store_data_cb;
1721 void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1723 ZebraHandle zh = (ZebraHandle) p->handle;
1724 zebraExplain_addSchema (zh->reg->zei, oid);
1727 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1728 int cmd, struct sortKeys *sk)
1730 SortIdx sortIdx = zh->reg->sortIdx;
1733 sortIdx_sysno (sortIdx, sysno);
1735 while (off < sk->buf_used)
1739 off += key_SU_decode(&set, sk->buf + off);
1740 off += key_SU_decode(&use, sk->buf + off);
1741 off += key_SU_decode(&slen, sk->buf + off);
1743 sortIdx_type(sortIdx, use);
1745 sortIdx_add(sortIdx, sk->buf + off, slen);
1747 sortIdx_add(sortIdx, "", 1);
1752 void encode_key_init (struct encode_info *i)
1761 i->encode_handle = iscz1_start();
1767 /* this is the old encode_key_write
1768 * may be deleted once we are confident that the new works
1771 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1774 char *bp = i->buf, *bp0;
1775 const char *src = (char *) &key;
1777 /* copy term to output buf */
1778 while ((*bp++ = *k++))
1780 /* and copy & align key so we can mangle */
1781 memcpy (&key, k+1, sizeof(struct it_key)); /* *k is insert/delete */
1784 iscz1_encode(i->encode_handle, &bp, &src);
1785 *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */
1786 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1788 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
1793 void encode_key_flush (struct encode_info *i, FILE *outf)
1794 { /* dummy routine */
1795 iscz1_stop(i->encode_handle);
1800 /* new encode_key_write
1801 * The idea is to buffer one more key, and compare them
1802 * If we are going to delete and insert the same key,
1803 * we may as well not bother. Should make a difference in
1804 * updates with small modifications (appending to a mbox)
1806 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1811 if (*k) /* first time for new key */
1814 while ((*bp++ = *k++))
1816 i->keylen= bp - i->buf -1;
1817 assert(i->keylen+1+sizeof(struct it_key) < ENCODE_BUFLEN);
1821 bp=i->buf + i->keylen;
1826 memcpy (&key, k+1, sizeof(struct it_key));
1827 if (0==i->prevsys) /* no previous filter, fill up */
1829 i->prevsys=key.sysno;
1830 i->prevseq=key.seqno;
1833 else if ( (i->prevsys==key.sysno) &&
1834 (i->prevseq==key.seqno) &&
1836 { /* same numbers, diff cmd, they cancel out */
1840 { /* different stuff, write previous, move buf */
1841 bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
1842 if (i->sysno != i->prevsys)
1844 i->sysno = i->prevsys;
1847 else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
1849 return; /* ??? Filters some sort of duplicates away */
1850 /* ??? Can this ever happen -H 15oct02 */
1852 bp = encode_key_int (i->prevseq - i->seqno, bp);
1853 i->seqno = i->prevseq;
1854 i->cmd = i->prevcmd;
1855 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1857 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
1860 i->keylen=0; /* ok, it's written, forget it */
1861 i->prevsys=key.sysno;
1862 i->prevseq=key.seqno;
1867 void encode_key_flush (struct encode_info *i, FILE *outf)
1868 { /* flush the last key from i */
1869 char *bp =i->buf + i->keylen;
1872 return; /* nothing to flush */
1875 bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
1876 if (i->sysno != i->prevsys)
1878 i->sysno = i->prevsys;
1881 else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
1883 return; /* ??? Filters some sort of duplicates away */
1884 /* ??? Can this ever happen -H 15oct02 */
1886 bp = encode_key_int (i->prevseq - i->seqno, bp);
1887 i->seqno = i->prevseq;
1888 i->cmd = i->prevcmd;
1889 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1891 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
1894 i->keylen=0; /* ok, it's written, forget it */
1895 i->prevsys=0; /* forget the values too */