1 /* $Id: extract.c,v 1.195 2005-09-16 09:58:38 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
46 #define REC_MULTI_SKIP 0
49 static void shellsort(void *ar, int r, size_t s,
50 int (*cmp)(const void *a, const void *b))
55 static const int incs[16] = { 1391376, 463792, 198768, 86961, 33936,
56 13776, 4592, 1968, 861, 336,
57 112, 48, 21, 7, 3, 1 };
58 for ( k = 0; k < 16; k++)
59 for (h = incs[k], i = h; i < r; i++)
63 while (j > h && (*cmp)(a + s*(j-h), v) > 0)
65 memcpy (a + s*j, a + s*(j-h), s);
73 static void logRecord (ZebraHandle zh)
75 ++zh->records_processed;
76 if (!(zh->records_processed % 1000))
78 yaz_log (YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
79 ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT,
80 zh->records_processed, zh->records_inserted, zh->records_updated,
85 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
87 static void extract_init (struct recExtractCtrl *p, RecWord *w)
89 w->zebra_maps = p->zebra_maps;
93 w->attrSet = VAL_BIB1;
103 static void searchRecordKey(ZebraHandle zh,
104 const struct recKeys *reckeys,
105 int attrSetS, int attrUseS,
106 const char **ws, int ws_length)
108 void *decode_handle = iscz1_start();
115 for (i = 0; i<ws_length; i++)
118 ch = zebraExplain_lookup_attr_su_any_index(zh->reg->zei,
123 while (off < reckeys->buf_used)
125 const char *src = reckeys->buf + off;
127 char *dst = (char*) &key;
129 iscz1_decode(decode_handle, &dst, &src);
130 assert(key.len <= 4 && key.len > 2);
132 seqno = (int) key.mem[key.len-1];
134 if (key.mem[0] == ch)
140 woff = seqno - startSeq;
141 if (woff >= 0 && woff < ws_length)
147 off = src - reckeys->buf;
149 iscz1_stop(decode_handle);
150 assert (off == reckeys->buf_used);
153 struct file_read_info {
154 off_t file_max; /* maximum offset so far */
155 off_t file_offset; /* current offset */
156 off_t file_moffset; /* offset of rec/rec boundary */
161 static struct file_read_info *file_read_start (int fd)
163 struct file_read_info *fi = (struct file_read_info *)
164 xmalloc (sizeof(*fi));
168 fi->file_moffset = 0;
174 static void file_read_stop (struct file_read_info *fi)
179 static off_t file_seek (void *handle, off_t offset)
181 struct file_read_info *p = (struct file_read_info *) handle;
182 p->file_offset = offset;
183 return lseek (p->fd, offset, SEEK_SET);
186 static off_t file_tell (void *handle)
188 struct file_read_info *p = (struct file_read_info *) handle;
189 return p->file_offset;
192 static int file_read (void *handle, char *buf, size_t count)
194 struct file_read_info *p = (struct file_read_info *) handle;
197 r = read (fd, buf, count);
201 if (p->file_offset > p->file_max)
202 p->file_max = p->file_offset;
207 static void file_end (void *handle, off_t offset)
209 struct file_read_info *p = (struct file_read_info *) handle;
211 if (offset != p->file_moffset)
213 p->file_moffset = offset;
218 static char *fileMatchStr (ZebraHandle zh,
219 struct recKeys *reckeys,
220 const char *fname, const char *spec)
222 static char dstBuf[2048]; /* static here ??? */
224 const char *s = spec;
228 while (*s == ' ' || *s == '\t')
235 char attset_str[64], attname_str[64];
236 data1_attset *attset;
239 int attSet = 1, attUse = 1;
243 for (i = 0; *s && *s != ',' && *s != ')'; s++)
245 attset_str[i++] = *s;
246 attset_str[i] = '\0';
251 for (i = 0; *s && *s != ')'; s++)
253 attname_str[i++] = *s;
254 attname_str[i] = '\0';
257 if ((attset = data1_get_attset (zh->reg->dh, attset_str)))
260 attSet = attset->reference;
261 att = data1_getattbyname(zh->reg->dh, attset, attname_str);
265 attUse = atoi (attname_str);
267 searchRecordKey (zh, reckeys, attSet, attUse, ws, 32);
271 for (i = 0; i<32; i++)
276 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
277 spec, zh->m_group ? zh->m_group : "none");
282 for (i = 0; i<32; i++)
283 if (matchFlag[i] && ws[i])
291 dst += strlen(ws[i]);
295 yaz_log (YLOG_WARN, "Record didn't contain match"
296 " fields in (%s,%s)", attset_str, attname_str);
304 const char *spec_src = NULL;
305 const char *s1 = ++s;
306 while (*s1 && *s1 != ' ' && *s1 != '\t')
312 memcpy (special, s, spec_len);
313 special[spec_len] = '\0';
316 if (!strcmp (special, "group"))
317 spec_src = zh->m_group;
318 else if (!strcmp (special, "database"))
319 spec_src = zh->basenames[0];
320 else if (!strcmp (special, "filename")) {
323 else if (!strcmp (special, "type"))
324 spec_src = zh->m_record_type;
329 strcpy (dst, spec_src);
330 dst += strlen (spec_src);
333 else if (*s == '\"' || *s == '\'')
335 int stopMarker = *s++;
339 while (*s && *s != stopMarker)
342 tmpString[i++] = *s++;
347 strcpy (dst, tmpString);
348 dst += strlen (tmpString);
352 yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
353 spec, zh->m_group ? zh->m_group : "none");
360 yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
361 fname, zh->m_group ? zh->m_group : "none");
368 struct recordLogInfo {
371 struct recordGroup *rGroup;
374 void create_rec_keys_codec(struct recKeys *keys)
377 iscz1_reset(keys->codec_handle);
380 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
383 for (i = 0; i<256; i++)
385 if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
390 ctrl->zebra_maps = zh->reg->zebra_maps;
391 ctrl->flagShowRecords = !zh->m_flag_rw;
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;
408 struct recExtractCtrl extractCtrl;
410 /* announce database */
411 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
413 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
414 zh->m_explain_database))
420 /* we are going to read from a file, so prepare the extraction */
421 create_rec_keys_codec(&zh->reg->keys);
423 create_rec_keys_codec(&zh->reg->sortKeys);
425 zh->reg->sortKeys.buf_used = 0;
427 recordOffset = fi->file_moffset;
428 extractCtrl.handle = zh;
429 extractCtrl.offset = fi->file_moffset;
430 extractCtrl.readf = file_read;
431 extractCtrl.seekf = file_seek;
432 extractCtrl.tellf = file_tell;
433 extractCtrl.endf = file_end;
435 extractCtrl.init = extract_init;
436 extractCtrl.tokenAdd = extract_token_add;
437 extractCtrl.schemaAdd = extract_schema_add;
438 extractCtrl.dh = zh->reg->dh;
439 extractCtrl.match_criteria[0] = '\0';
440 extractCtrl.staticrank = 0;
442 extractCtrl.first_record = fi->file_offset ? 0 : 1;
444 extract_set_store_data_prepare(&extractCtrl);
446 init_extractCtrl(zh, &extractCtrl);
449 printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset);
453 sprintf (msg, "%s:" PRINTF_OFF_T , fname, recordOffset);
454 yaz_log_init_prefix2 (msg);
457 r = (*recType->extract)(recTypeClientData, &extractCtrl);
459 yaz_log_init_prefix2 (0);
460 if (r == RECCTRL_EXTRACT_EOF)
462 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
464 /* error occured during extraction ... */
466 zh->records_processed < zh->m_file_verbose_limit)
468 yaz_log (YLOG_WARN, "fail %s %s " PRINTF_OFF_T, zh->m_record_type,
469 fname, recordOffset);
473 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
475 /* error occured during extraction ... */
477 zh->records_processed < zh->m_file_verbose_limit)
479 yaz_log (YLOG_WARN, "no filter for %s %s "
480 PRINTF_OFF_T, zh->m_record_type,
481 fname, recordOffset);
485 if (extractCtrl.match_criteria[0])
486 matchStr = extractCtrl.match_criteria;
489 /* perform match if sysno not known and if match criteria is specified */
495 if (matchStr == 0 && zh->m_record_id && *zh->m_record_id)
498 matchStr = fileMatchStr (zh, &zh->reg->keys, fname,
502 yaz_log(YLOG_WARN, "Bad match criteria");
508 char *rinfo = dict_lookup (zh->reg->matchDict, matchStr);
511 assert(*rinfo == sizeof(*sysno));
512 memcpy (sysno, rinfo+1, sizeof(*sysno));
516 if (! *sysno && zh->reg->keys.buf_used == 0)
518 /* the extraction process returned no information - the record
519 is probably empty - unless flagShowRecords is in use */
523 if (zh->records_processed < zh->m_file_verbose_limit)
524 yaz_log (YLOG_WARN, "empty %s %s " PRINTF_OFF_T, zh->m_record_type,
525 fname, recordOffset);
534 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T, zh->m_record_type,
535 fname, recordOffset);
536 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
539 if (zh->records_processed < zh->m_file_verbose_limit)
540 yaz_log (YLOG_LOG, "add %s %s " PRINTF_OFF_T, zh->m_record_type,
541 fname, recordOffset);
542 rec = rec_new (zh->reg->records);
546 recordAttr = rec_init_attr (zh->reg->zei, rec);
547 recordAttr->staticrank = extractCtrl.staticrank;
551 dict_insert (zh->reg->matchDict, matchStr, sizeof(*sysno), sysno);
553 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
554 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys,
555 recordAttr->staticrank);
557 zh->records_inserted++;
561 /* record already exists */
562 struct recKeys delkeys;
564 struct recKeys sortKeys;
566 struct sortKeys sortKeys;
569 rec = rec_get (zh->reg->records, *sysno);
572 recordAttr = rec_init_attr (zh->reg->zei, rec);
575 if (!force_update && recordAttr->runNumber ==
576 zebraExplain_runNumberIncrement (zh->reg->zei, 0))
578 yaz_log (YLOG_LOG, "run number = " ZINT_FORMAT,
579 recordAttr->runNumber);
580 yaz_log (YLOG_LOG, "skipped %s %s " PRINTF_OFF_T,
581 zh->m_record_type, fname, recordOffset);
582 extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
588 /* flush old keys for sort&search etc. */
589 delkeys.buf_used = rec->size[recInfo_delKeys];
590 delkeys.buf = rec->info[recInfo_delKeys];
592 sortKeys.buf_used = rec->size[recInfo_sortKeys];
593 sortKeys.buf = rec->info[recInfo_sortKeys];
595 extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
596 extract_flushRecordKeys (zh, *sysno, 0, &delkeys,
597 recordAttr->staticrank); /* old values */
600 /* record going to be deleted */
601 if (!delkeys.buf_used)
603 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
604 zh->m_record_type, fname, recordOffset);
605 yaz_log (YLOG_WARN, "cannot delete file above, storeKeys false");
609 if (zh->records_processed < zh->m_file_verbose_limit)
610 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
611 zh->m_record_type, fname, recordOffset);
612 zh->records_deleted++;
614 dict_delete (zh->reg->matchDict, matchStr);
615 rec_del (zh->reg->records, &rec);
623 /* record going to be updated */
624 if (!delkeys.buf_used)
626 yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
627 zh->m_record_type, fname, recordOffset);
628 yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
632 /* flush new keys for sort&search etc */
633 if (zh->records_processed < zh->m_file_verbose_limit)
634 yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
635 zh->m_record_type, fname, recordOffset);
636 recordAttr->staticrank = extractCtrl.staticrank;
637 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
638 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys,
639 recordAttr->staticrank);
640 zh->records_updated++;
644 /* update file type */
645 xfree (rec->info[recInfo_fileType]);
646 rec->info[recInfo_fileType] =
647 rec_strdup (zh->m_record_type, &rec->size[recInfo_fileType]);
649 /* update filename */
650 xfree (rec->info[recInfo_filename]);
651 rec->info[recInfo_filename] =
652 rec_strdup (fname, &rec->size[recInfo_filename]);
654 /* update delete keys */
655 xfree (rec->info[recInfo_delKeys]);
656 if (zh->reg->keys.buf_used > 0 && zh->m_store_keys == 1)
658 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
659 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
660 zh->reg->keys.buf = NULL;
661 zh->reg->keys.buf_max = 0;
665 rec->info[recInfo_delKeys] = NULL;
666 rec->size[recInfo_delKeys] = 0;
669 /* update sort keys */
670 xfree (rec->info[recInfo_sortKeys]);
672 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
673 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
674 zh->reg->sortKeys.buf = NULL;
675 zh->reg->sortKeys.buf_max = 0;
677 /* save file size of original record */
678 zebraExplain_recordBytesIncrement (zh->reg->zei,
679 - recordAttr->recordSize);
680 recordAttr->recordSize = fi->file_moffset - recordOffset;
681 if (!recordAttr->recordSize)
682 recordAttr->recordSize = fi->file_max - recordOffset;
683 zebraExplain_recordBytesIncrement (zh->reg->zei,
684 recordAttr->recordSize);
686 /* set run-number for this record */
687 recordAttr->runNumber = zebraExplain_runNumberIncrement (zh->reg->zei,
690 /* update store data */
691 xfree (rec->info[recInfo_storeData]);
692 if (zh->store_data_buf)
694 rec->size[recInfo_storeData] = zh->store_data_size;
695 rec->info[recInfo_storeData] = zh->store_data_buf;
696 zh->store_data_buf = 0;
698 else if (zh->m_store_data)
700 rec->size[recInfo_storeData] = recordAttr->recordSize;
701 rec->info[recInfo_storeData] = (char *)
702 xmalloc (recordAttr->recordSize);
703 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
705 yaz_log (YLOG_ERRNO|YLOG_FATAL, "seek to " PRINTF_OFF_T " in %s",
706 recordOffset, fname);
709 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
710 < recordAttr->recordSize)
712 yaz_log (YLOG_ERRNO|YLOG_FATAL, "read %d bytes of %s",
713 recordAttr->recordSize, fname);
719 rec->info[recInfo_storeData] = NULL;
720 rec->size[recInfo_storeData] = 0;
722 /* update database name */
723 xfree (rec->info[recInfo_databaseName]);
724 rec->info[recInfo_databaseName] =
725 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
728 recordAttr->recordOffset = recordOffset;
730 /* commit this record */
731 rec_put (zh->reg->records, &rec);
736 int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
743 struct file_read_info *fi;
744 const char *original_record_type = 0;
746 void *recTypeClientData;
748 if (!zh->m_group || !*zh->m_group)
751 sprintf (gprefix, "%s.", zh->m_group);
753 yaz_log (YLOG_DEBUG, "fileExtract %s", fname);
755 /* determine file extension */
757 for (i = strlen(fname); --i >= 0; )
760 else if (fname[i] == '.')
762 strcpy (ext, fname+i+1);
765 /* determine file type - depending on extension */
766 original_record_type = zh->m_record_type;
767 if (!zh->m_record_type)
769 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
770 zh->m_record_type = res_get (zh->res, ext_res);
772 if (!zh->m_record_type)
774 if (zh->records_processed < zh->m_file_verbose_limit)
775 yaz_log (YLOG_LOG, "? %s", fname);
778 /* determine match criteria */
779 if (!zh->m_record_id)
781 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
782 zh->m_record_id = res_get (zh->res, ext_res);
786 recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
787 &recTypeClientData)))
789 yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
793 switch(recType->version)
798 yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
800 if (sysno && deleteFlag)
806 if (zh->path_reg && !yaz_is_abspath (fname))
808 strcpy (full_rep, zh->path_reg);
809 strcat (full_rep, "/");
810 strcat (full_rep, fname);
813 strcpy (full_rep, fname);
816 if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
818 yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
819 zh->m_record_type = original_record_type;
823 fi = file_read_start (fd);
826 fi->file_moffset = fi->file_offset;
827 fi->file_more = 0; /* file_end not called (yet) */
828 r = file_extract_record (zh, sysno, fname, deleteFlag, fi, 1,
829 recType, recTypeClientData);
831 { /* file_end has been called so reset offset .. */
832 fi->file_offset = fi->file_moffset;
833 lseek(fi->fd, fi->file_moffset, SEEK_SET);
840 zh->m_record_type = original_record_type;
845 If sysno is provided, then it's used to identify the reocord.
846 If not, and match_criteria is provided, then sysno is guessed
847 If not, and a record is provided, then sysno is got from there
850 ZEBRA_RES buffer_extract_record (ZebraHandle zh,
851 const char *buf, size_t buf_size,
854 const char *recordType,
856 const char *match_criteria,
861 RecordAttr *recordAttr;
862 struct recExtractCtrl extractCtrl;
864 const char *matchStr = 0;
865 RecType recType = NULL;
868 long recordOffset = 0;
869 struct zebra_fetch_control fc;
870 const char *pr_fname = fname; /* filename to print .. */
871 int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
874 pr_fname = "<no file>"; /* make it printable if file is omitted */
877 fc.record_int_buf = buf;
878 fc.record_int_len = buf_size;
879 fc.record_int_pos = 0;
881 fc.record_offset = 0;
883 extractCtrl.offset = 0;
884 extractCtrl.readf = zebra_record_int_read;
885 extractCtrl.seekf = zebra_record_int_seek;
886 extractCtrl.tellf = zebra_record_int_tell;
887 extractCtrl.endf = zebra_record_int_end;
888 extractCtrl.first_record = 1;
889 extractCtrl.fh = &fc;
891 create_rec_keys_codec(&zh->reg->keys);
893 create_rec_keys_codec(&zh->reg->sortKeys);
895 zh->reg->sortKeys.buf_used = 0;
897 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
899 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
900 zh->m_explain_database))
904 if (recordType && *recordType)
906 yaz_log (YLOG_DEBUG, "Record type explicitly specified: %s", recordType);
907 recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
912 if (!(zh->m_record_type))
914 yaz_log (YLOG_WARN, "No such record type defined");
917 yaz_log (YLOG_DEBUG, "Get record type from rgroup: %s",zh->m_record_type);
918 recType = recType_byName (zh->reg->recTypes, zh->res,
919 zh->m_record_type, &clientData);
920 recordType = zh->m_record_type;
925 yaz_log (YLOG_WARN, "No such record type: %s", zh->m_record_type);
929 extractCtrl.init = extract_init;
930 extractCtrl.tokenAdd = extract_token_add;
931 extractCtrl.schemaAdd = extract_schema_add;
932 extractCtrl.dh = zh->reg->dh;
933 extractCtrl.handle = zh;
934 extractCtrl.match_criteria[0] = '\0';
935 extractCtrl.staticrank = 0;
937 init_extractCtrl(zh, &extractCtrl);
939 extract_set_store_data_prepare(&extractCtrl);
941 r = (*recType->extract)(clientData, &extractCtrl);
943 if (r == RECCTRL_EXTRACT_EOF)
945 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
947 /* error occured during extraction ... */
948 yaz_log (YLOG_WARN, "extract error: generic");
951 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
953 /* error occured during extraction ... */
954 yaz_log (YLOG_WARN, "extract error: no such filter");
960 if (extractCtrl.match_criteria[0])
961 match_criteria = extractCtrl.match_criteria;
965 if (match_criteria && *match_criteria) {
966 matchStr = match_criteria;
968 if (zh->m_record_id && *zh->m_record_id) {
969 matchStr = fileMatchStr (zh, &zh->reg->keys, pr_fname,
973 yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
979 rinfo = dict_lookup (zh->reg->matchDict, matchStr);
982 assert(*rinfo == sizeof(*sysno));
983 memcpy (sysno, rinfo+1, sizeof(*sysno));
987 if (zh->reg->keys.buf_used == 0)
989 /* the extraction process returned no information - the record
990 is probably empty - unless flagShowRecords is in use */
1001 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1002 pr_fname, (long) recordOffset);
1003 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
1007 yaz_log (YLOG_LOG, "add %s %s %ld", recordType, pr_fname,
1008 (long) recordOffset);
1009 rec = rec_new (zh->reg->records);
1011 *sysno = rec->sysno;
1013 recordAttr = rec_init_attr (zh->reg->zei, rec);
1014 recordAttr->staticrank = extractCtrl.staticrank;
1018 dict_insert (zh->reg->matchDict, matchStr,
1019 sizeof(*sysno), sysno);
1021 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1022 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys,
1023 recordAttr->staticrank);
1025 zh->records_inserted++;
1029 /* record already exists */
1030 struct recKeys delkeys;
1032 struct recKeys sortKeys;
1034 struct sortKeys sortKeys;
1040 yaz_log (YLOG_LOG, "skipped %s %s %ld",
1041 recordType, pr_fname, (long) recordOffset);
1046 rec = rec_get (zh->reg->records, *sysno);
1049 recordAttr = rec_init_attr (zh->reg->zei, rec);
1052 if (!force_update) {
1053 if (recordAttr->runNumber ==
1054 zebraExplain_runNumberIncrement (zh->reg->zei, 0))
1057 yaz_log (YLOG_LOG, "skipped %s %s %ld", recordType,
1058 pr_fname, (long) recordOffset);
1059 extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
1067 delkeys.buf_used = rec->size[recInfo_delKeys];
1068 delkeys.buf = rec->info[recInfo_delKeys];
1070 sortKeys.buf_used = rec->size[recInfo_sortKeys];
1071 sortKeys.buf = rec->info[recInfo_sortKeys];
1073 extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
1074 extract_flushRecordKeys (zh, *sysno, 0, &delkeys,
1075 recordAttr->staticrank);
1078 /* record going to be deleted */
1079 if (!delkeys.buf_used)
1083 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1084 pr_fname, (long) recordOffset);
1085 yaz_log (YLOG_WARN, "cannot delete file above, "
1092 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1093 pr_fname, (long) recordOffset);
1094 zh->records_deleted++;
1096 dict_delete (zh->reg->matchDict, matchStr);
1097 rec_del (zh->reg->records, &rec);
1105 /* record going to be updated */
1106 if (!delkeys.buf_used)
1110 yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
1111 pr_fname, (long) recordOffset);
1112 yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
1118 yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
1119 pr_fname, (long) recordOffset);
1120 recordAttr->staticrank = extractCtrl.staticrank;
1121 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1122 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys,
1123 recordAttr->staticrank);
1124 zh->records_updated++;
1128 /* update file type */
1129 xfree (rec->info[recInfo_fileType]);
1130 rec->info[recInfo_fileType] =
1131 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1133 /* update filename */
1134 xfree (rec->info[recInfo_filename]);
1135 rec->info[recInfo_filename] =
1136 rec_strdup (fname, &rec->size[recInfo_filename]);
1138 /* update delete keys */
1139 xfree (rec->info[recInfo_delKeys]);
1140 if (zh->reg->keys.buf_used > 0 && zh->m_store_keys == 1)
1142 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
1143 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
1144 zh->reg->keys.buf = NULL;
1145 zh->reg->keys.buf_max = 0;
1149 rec->info[recInfo_delKeys] = NULL;
1150 rec->size[recInfo_delKeys] = 0;
1153 /* update sort keys */
1154 xfree (rec->info[recInfo_sortKeys]);
1156 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
1157 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
1158 zh->reg->sortKeys.buf = NULL;
1159 zh->reg->sortKeys.buf_max = 0;
1161 /* save file size of original record */
1162 zebraExplain_recordBytesIncrement (zh->reg->zei,
1163 - recordAttr->recordSize);
1165 recordAttr->recordSize = fi->file_moffset - recordOffset;
1166 if (!recordAttr->recordSize)
1167 recordAttr->recordSize = fi->file_max - recordOffset;
1169 recordAttr->recordSize = buf_size;
1171 zebraExplain_recordBytesIncrement (zh->reg->zei,
1172 recordAttr->recordSize);
1174 /* set run-number for this record */
1175 recordAttr->runNumber =
1176 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
1178 /* update store data */
1179 xfree (rec->info[recInfo_storeData]);
1181 /* update store data */
1182 if (zh->store_data_buf)
1184 rec->size[recInfo_storeData] = zh->store_data_size;
1185 rec->info[recInfo_storeData] = zh->store_data_buf;
1186 zh->store_data_buf = 0;
1188 else if (zh->m_store_data)
1190 rec->size[recInfo_storeData] = recordAttr->recordSize;
1191 rec->info[recInfo_storeData] = (char *)
1192 xmalloc (recordAttr->recordSize);
1193 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1197 rec->info[recInfo_storeData] = NULL;
1198 rec->size[recInfo_storeData] = 0;
1200 /* update database name */
1201 xfree (rec->info[recInfo_databaseName]);
1202 rec->info[recInfo_databaseName] =
1203 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
1206 recordAttr->recordOffset = recordOffset;
1208 /* commit this record */
1209 rec_put (zh->reg->records, &rec);
1214 int explain_extract (void *handle, Record rec, data1_node *n)
1216 ZebraHandle zh = (ZebraHandle) handle;
1217 struct recExtractCtrl extractCtrl;
1219 if (zebraExplain_curDatabase (zh->reg->zei,
1220 rec->info[recInfo_databaseName]))
1223 if (zebraExplain_newDatabase (zh->reg->zei,
1224 rec->info[recInfo_databaseName], 0))
1228 create_rec_keys_codec(&zh->reg->keys);
1230 create_rec_keys_codec(&zh->reg->sortKeys);
1232 zh->reg->sortKeys.buf_used = 0;
1234 extractCtrl.init = extract_init;
1235 extractCtrl.tokenAdd = extract_token_add;
1236 extractCtrl.schemaAdd = extract_schema_add;
1237 extractCtrl.dh = zh->reg->dh;
1239 init_extractCtrl(zh, &extractCtrl);
1241 extractCtrl.flagShowRecords = 0;
1242 extractCtrl.match_criteria[0] = '\0';
1243 extractCtrl.staticrank = 0;
1244 extractCtrl.handle = handle;
1245 extractCtrl.first_record = 1;
1247 extract_set_store_data_prepare(&extractCtrl);
1250 grs_extract_tree(&extractCtrl, n);
1252 if (rec->size[recInfo_delKeys])
1254 struct recKeys delkeys;
1256 struct recKeys sortkeys;
1258 struct sortKeys sortkeys;
1261 delkeys.buf_used = rec->size[recInfo_delKeys];
1262 delkeys.buf = rec->info[recInfo_delKeys];
1264 sortkeys.buf_used = rec->size[recInfo_sortKeys];
1265 sortkeys.buf = rec->info[recInfo_sortKeys];
1267 extract_flushSortKeys (zh, rec->sysno, 0, &sortkeys);
1268 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys, 0);
1270 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->reg->keys, 0);
1271 extract_flushSortKeys (zh, rec->sysno, 1, &zh->reg->sortKeys);
1273 xfree (rec->info[recInfo_delKeys]);
1274 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
1275 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
1276 zh->reg->keys.buf = NULL;
1277 zh->reg->keys.buf_max = 0;
1279 xfree (rec->info[recInfo_sortKeys]);
1280 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
1281 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
1282 zh->reg->sortKeys.buf = NULL;
1283 zh->reg->sortKeys.buf_max = 0;
1288 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1289 int cmd, struct recKeys *reckeys,
1292 void *decode_handle = iscz1_start();
1295 ZebraExplainInfo zei = zh->reg->zei;
1297 if (!zh->reg->key_buf)
1299 int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
1302 yaz_log(YLOG_WARN, "Invalid memory setting, using default 8 MB");
1305 /* FIXME: That "8" should be in a default settings include */
1306 /* not hard-coded here! -H */
1307 zh->reg->key_buf = (char**) xmalloc (mem);
1308 zh->reg->ptr_top = mem/sizeof(char*);
1310 zh->reg->key_buf_used = 0;
1311 zh->reg->key_file_no = 0;
1313 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1315 while (off < reckeys->buf_used)
1317 const char *src = reckeys->buf + off;
1318 struct it_key key_in;
1319 struct it_key key_out;
1320 char *dst = (char*) &key_in;
1321 zint *keyp = key_out.mem;
1323 iscz1_decode(decode_handle, &dst, &src);
1324 assert(key_in.len == 4);
1326 if (zh->reg->key_buf_used + 1024 >
1327 (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
1328 extract_flushWriteKeys (zh,0);
1330 assert(zh->reg->ptr_i > 0);
1331 (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
1332 (char*)zh->reg->key_buf + zh->reg->key_buf_used;
1334 ch = (int) key_in.mem[0]; /* ordinal for field/use/attribute */
1336 zh->reg->key_buf_used +=
1337 key_SU_encode (ch,((char*)zh->reg->key_buf) +
1338 zh->reg->key_buf_used);
1340 ((char*)zh->reg->key_buf) [(zh->reg->key_buf_used)++] = *src++;
1342 ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = '\0';
1343 ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = cmd;
1345 if (zh->m_staticrank) /* rank config enabled ? */
1347 *keyp++ = staticrank;
1353 if (key_in.mem[1]) /* filter specified record ID */
1354 *keyp++ = key_in.mem[1];
1357 *keyp++ = key_in.mem[2]; /* section_id */
1358 *keyp++ = key_in.mem[3]; /* sequence .. */
1360 memcpy ((char*)zh->reg->key_buf + zh->reg->key_buf_used,
1361 &key_out, sizeof(key_out));
1362 (zh->reg->key_buf_used) += sizeof(key_out);
1363 off = src - reckeys->buf;
1365 assert (off == reckeys->buf_used);
1366 iscz1_stop(decode_handle);
1369 void extract_flushWriteKeys (ZebraHandle zh, int final)
1370 /* optimizing: if final=1, and no files written yet */
1371 /* push the keys directly to merge, sidestepping the */
1372 /* temp file altogether. Speeds small updates */
1375 char out_fname[200];
1377 struct encode_info encode_info;
1378 int ptr_i = zh->reg->ptr_i;
1383 if (!zh->reg->key_buf || ptr_i <= 0)
1385 yaz_log (YLOG_DEBUG, " nothing to flush section=%d buf=%p i=%d",
1386 zh->reg->key_file_no, zh->reg->key_buf, ptr_i);
1387 yaz_log (YLOG_DEBUG, " buf=%p ",
1389 yaz_log (YLOG_DEBUG, " ptr=%d ",zh->reg->ptr_i);
1390 yaz_log (YLOG_DEBUG, " reg=%p ",zh->reg);
1395 (zh->reg->key_file_no)++;
1396 yaz_log (YLOG_LOG, "sorting section %d", (zh->reg->key_file_no));
1397 yaz_log (YLOG_DEBUG, " sort_buff at %p n=%d",
1398 zh->reg->key_buf + zh->reg->ptr_top - ptr_i,ptr_i);
1400 qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
1401 sizeof(char*), key_qsort_compare);
1403 /* zebra.cfg: tempfiles:
1404 Y: always use temp files (old way)
1405 A: use temp files, if more than one (auto)
1406 = if this is both the last and the first
1407 N: never bother with temp files (new) */
1409 temp_policy=toupper(res_get_def(zh->res,"tempfiles","auto")[0]);
1410 if (temp_policy != 'Y' && temp_policy != 'N' && temp_policy != 'A') {
1411 yaz_log (YLOG_WARN, "Illegal tempfiles setting '%c'. using 'Auto' ",
1416 if ( ( temp_policy =='N' ) || /* always from memory */
1417 ( ( temp_policy =='A' ) && /* automatic */
1418 (zh->reg->key_file_no == 1) && /* this is first time */
1419 (final) ) ) /* and last (=only) time */
1420 { /* go directly from memory */
1421 zh->reg->key_file_no =0; /* signal not to read files */
1422 zebra_index_merge(zh);
1424 zh->reg->key_buf_used = 0;
1428 /* Not doing directly from memory, write into a temp file */
1429 extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
1431 if (!(outf = fopen (out_fname, "wb")))
1433 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1436 yaz_log (YLOG_LOG, "writing section %d", zh->reg->key_file_no);
1437 prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1439 encode_key_init (&encode_info);
1440 encode_key_write (cp, &encode_info, outf);
1444 cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1445 if (strcmp (cp, prevcp))
1447 encode_key_flush ( &encode_info, outf);
1448 encode_key_init (&encode_info);
1449 encode_key_write (cp, &encode_info, outf);
1453 encode_key_write (cp + strlen(cp), &encode_info, outf);
1455 encode_key_flush ( &encode_info, outf);
1457 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1458 extract_get_fname_tmp (out_fname, key_file_no);
1460 if (!(outf = fopen (out_fname, "wb")))
1462 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1465 yaz_log (YLOG_LOG, "writing section %d", key_file_no);
1467 prevcp = key_buf[ptr_top-i];
1469 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1471 key_y_len = strlen(prevcp)+1;
1473 yaz_log (YLOG_LOG, "key_y_len: %2d %02x %02x %s",
1474 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1476 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1477 sizeof(char*), key_y_compare);
1478 cp = key_buf[ptr_top-ptr_i];
1480 encode_key_init (&encode_info);
1481 encode_key_write (cp, &encode_info, outf);
1484 cp = key_buf[ptr_top-ptr_i];
1485 encode_key_write (cp+key_y_len, &encode_info, outf);
1487 encode_key_flush ( &encode_info, outf);
1490 prevcp = key_buf[ptr_top-ptr_i];
1495 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fclose %s", out_fname);
1498 yaz_log (YLOG_LOG, "finished section %d", zh->reg->key_file_no);
1500 zh->reg->key_buf_used = 0;
1503 void extract_add_it_key (ZebraHandle zh,
1504 struct recKeys *keys,
1506 const char *str, int slen, struct it_key *key)
1509 const char *src = (char*) key;
1511 if (keys->buf_used+1024 > keys->buf_max)
1513 char *b = (char *) xmalloc (keys->buf_max += 128000);
1514 if (keys->buf_used > 0)
1515 memcpy (b, keys->buf, keys->buf_used);
1519 dst = keys->buf + keys->buf_used;
1521 iscz1_encode(keys->codec_handle, &dst, &src);
1526 memcpy (dst, str, slen);
1529 keys->buf_used = dst - keys->buf;
1532 ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh, struct recKeys *reckeys,
1533 zebra_snippets *snippets)
1535 void *decode_handle = iscz1_start();
1538 NMEM nmem = nmem_create();
1540 assert(reckeys->buf);
1541 while (off < reckeys->buf_used)
1543 const char *src = reckeys->buf + off;
1545 char *dst = (char*) &key;
1546 char dst_buf[IT_MAX_WORD];
1547 char *dst_term = dst_buf;
1548 int index_type = 0, ord;
1550 iscz1_decode(decode_handle, &dst, &src);
1551 assert(key.len <= 4 && key.len > 2);
1553 seqno = (int) key.mem[key.len-1];
1556 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1557 0/* db */, 0/* set */, 0/* use */);
1559 zebra_term_untrans_iconv(zh, nmem, index_type,
1561 zebra_snippets_append(snippets, seqno, ord, dst_term);
1564 off = src - reckeys->buf;
1568 iscz1_stop(decode_handle);
1572 void print_rec_keys(ZebraHandle zh, struct recKeys *reckeys)
1574 void *decode_handle = iscz1_start();
1577 NMEM nmem = nmem_create();
1579 yaz_log(YLOG_LOG, "print_rec_keys buf=%p sz=%d", reckeys->buf,
1581 assert(reckeys->buf);
1582 while (off < reckeys->buf_used)
1584 const char *src = reckeys->buf + off;
1586 char *dst = (char*) &key;
1587 int attrSet, attrUse;
1588 char dst_buf[IT_MAX_WORD];
1589 char *dst_term = dst_buf;
1591 iscz1_decode(decode_handle, &dst, &src);
1592 assert(key.len <= 4 && key.len > 2);
1594 attrSet = (int) key.mem[0] >> 16;
1595 attrUse = (int) key.mem[0] & 65535;
1596 seqno = (int) key.mem[key.len-1];
1598 zebra_term_untrans_iconv(zh, nmem, src[0], &dst_term, src+1);
1600 yaz_log(YLOG_LOG, "ord=" ZINT_FORMAT " seqno=%d term=%s",
1601 key.mem[0], seqno, dst_term);
1604 off = src - reckeys->buf;
1608 iscz1_stop(decode_handle);
1611 void extract_add_index_string (RecWord *p, const char *str, int length)
1615 ZebraHandle zh = p->extractCtrl->handle;
1616 ZebraExplainInfo zei = zh->reg->zei;
1621 ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
1623 ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
1630 ch = zebraExplain_lookup_attr_su(zei, p->index_type,
1631 p->attrSet, p->attrUse);
1633 ch = zebraExplain_add_attr_su(zei, p->index_type,
1634 p->attrSet, p->attrUse);
1639 key.mem[1] = p->record_id;
1640 key.mem[2] = p->section_id;
1641 key.mem[3] = p->seqno;
1644 /* just for debugging .. */
1645 yaz_log(YLOG_LOG, "add: set=%d use=%d "
1646 "record_id=%lld section_id=%lld seqno=%lld",
1647 p->attrSet, p->attrUse, p->record_id, p->section_id, p->seqno);
1650 extract_add_it_key(p->extractCtrl->handle,
1657 static void extract_add_sort_string (RecWord *p, const char *str, int length)
1661 ZebraHandle zh = p->extractCtrl->handle;
1662 ZebraExplainInfo zei = zh->reg->zei;
1667 ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
1669 ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
1677 key.mem[1] = p->record_id;
1678 key.mem[2] = p->section_id;
1679 key.mem[3] = p->seqno;
1681 extract_add_it_key(p->extractCtrl->handle,
1687 static void extract_add_sort_string (RecWord *p, const char *str, int length)
1689 ZebraHandle zh = p->extractCtrl->handle;
1690 struct sortKeys *sk = &zh->reg->sortKeys;
1693 while (off < sk->buf_used)
1697 off += key_SU_decode(&set, (unsigned char *) sk->buf + off);
1698 off += key_SU_decode(&use, (unsigned char *) sk->buf + off);
1699 off += key_SU_decode(&slen, (unsigned char *) sk->buf + off);
1701 if (p->attrSet == set && p->attrUse == use)
1704 assert (off == sk->buf_used);
1706 if (sk->buf_used + IT_MAX_WORD > sk->buf_max)
1710 b = (char *) xmalloc (sk->buf_max += 128000);
1711 if (sk->buf_used > 0)
1712 memcpy (b, sk->buf, sk->buf_used);
1716 off += key_SU_encode(p->attrSet, sk->buf + off);
1717 off += key_SU_encode(p->attrUse, sk->buf + off);
1718 off += key_SU_encode(length, sk->buf + off);
1719 memcpy (sk->buf + off, str, length);
1720 sk->buf_used = off + length;
1724 void extract_add_string (RecWord *p, const char *string, int length)
1726 assert (length > 0);
1727 if (zebra_maps_is_sort (p->zebra_maps, p->index_type))
1728 extract_add_sort_string (p, string, length);
1730 extract_add_index_string (p, string, length);
1733 static void extract_add_incomplete_field (RecWord *p)
1735 const char *b = p->term_buf;
1736 int remain = p->term_len;
1737 const char **map = 0;
1739 yaz_log(YLOG_DEBUG, "Incomplete field, w='%.*s'", p->term_len, p->term_buf);
1742 map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, 0);
1746 char buf[IT_MAX_WORD+1];
1750 while (map && *map && **map == *CHR_SPACE)
1752 remain = p->term_len - (b - p->term_buf);
1754 map = zebra_maps_input(p->zebra_maps, p->index_type, &b,
1762 while (map && *map && **map != *CHR_SPACE)
1764 const char *cp = *map;
1766 while (i < IT_MAX_WORD && *cp)
1768 remain = p->term_len - (b - p->term_buf);
1770 map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, 0);
1776 extract_add_string (p, buf, i);
1781 static void extract_add_complete_field (RecWord *p)
1783 const char *b = p->term_buf;
1784 char buf[IT_MAX_WORD+1];
1785 const char **map = 0;
1786 int i = 0, remain = p->term_len;
1788 yaz_log(YLOG_DEBUG, "Complete field, w='%.*s'",
1789 p->term_len, p->term_buf);
1792 map = zebra_maps_input (p->zebra_maps, p->index_type, &b, remain, 1);
1794 while (remain > 0 && i < IT_MAX_WORD)
1796 while (map && *map && **map == *CHR_SPACE)
1798 remain = p->term_len - (b - p->term_buf);
1802 int first = i ? 0 : 1; /* first position */
1803 map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, first);
1811 if (i && i < IT_MAX_WORD)
1812 buf[i++] = *CHR_SPACE;
1813 while (map && *map && **map != *CHR_SPACE)
1815 const char *cp = *map;
1817 if (**map == *CHR_CUT)
1823 if (i >= IT_MAX_WORD)
1825 yaz_log(YLOG_DEBUG, "Adding string to index '%d'", **map);
1826 while (i < IT_MAX_WORD && *cp)
1829 remain = p->term_len - (b - p->term_buf);
1832 map = zebra_maps_input (p->zebra_maps, p->index_type, &b,
1841 extract_add_string (p, buf, i);
1844 void extract_token_add (RecWord *p)
1848 yaz_log (YLOG_LOG, "token_add "
1849 "reg_type=%c attrSet=%d attrUse=%d seqno=%d s=%.*s",
1850 p->reg_type, p->attrSet, p->attrUse, p->seqno, p->length,
1853 if ((wrbuf = zebra_replace(p->zebra_maps, p->index_type, 0,
1854 p->term_buf, p->term_len)))
1856 p->term_buf = wrbuf_buf(wrbuf);
1857 p->term_len = wrbuf_len(wrbuf);
1859 if (zebra_maps_is_complete (p->zebra_maps, p->index_type))
1860 extract_add_complete_field (p);
1862 extract_add_incomplete_field(p);
1865 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1866 void *buf, size_t sz)
1868 ZebraHandle zh = (ZebraHandle) p->handle;
1870 xfree(zh->store_data_buf);
1871 zh->store_data_buf = 0;
1872 zh->store_data_size = 0;
1875 zh->store_data_buf = xmalloc(sz);
1876 zh->store_data_size = sz;
1877 memcpy(zh->store_data_buf, buf, sz);
1881 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1883 ZebraHandle zh = (ZebraHandle) p->handle;
1884 xfree(zh->store_data_buf);
1885 zh->store_data_buf = 0;
1886 zh->store_data_size = 0;
1887 p->setStoreData = extract_set_store_data_cb;
1890 void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1892 ZebraHandle zh = (ZebraHandle) p->handle;
1893 zebraExplain_addSchema (zh->reg->zei, oid);
1897 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1898 int cmd, struct recKeys *reckeys)
1900 SortIdx sortIdx = zh->reg->sortIdx;
1901 void *decode_handle = iscz1_start();
1905 while (off < reckeys->buf_used)
1907 const char *src = reckeys->buf + off;
1909 char *dst = (char*) &key;
1911 iscz1_decode(decode_handle, &dst, &src);
1912 assert(key.len == 4);
1914 ch = (int) key.mem[0]; /* ordinal for field/use/attribute */
1916 sortIdx_type(sortIdx, ch);
1918 sortIdx_add(sortIdx, src, strlen(src));
1920 sortIdx_add(sortIdx, "", 1);
1925 off = src - reckeys->buf;
1927 assert (off == reckeys->buf_used);
1928 iscz1_stop(decode_handle);
1931 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1932 int cmd, struct sortKeys *sk)
1934 SortIdx sortIdx = zh->reg->sortIdx;
1937 sortIdx_sysno (sortIdx, sysno);
1939 while (off < sk->buf_used)
1943 off += key_SU_decode(&set, (unsigned char *) sk->buf + off);
1944 off += key_SU_decode(&use, (unsigned char *) sk->buf + off);
1945 off += key_SU_decode(&slen, (unsigned char *) sk->buf + off);
1947 sortIdx_type(sortIdx, use);
1949 sortIdx_add(sortIdx, sk->buf + off, slen);
1951 sortIdx_add(sortIdx, "", 1);
1957 void encode_key_init (struct encode_info *i)
1966 i->encode_handle = iscz1_start();
1972 /* this is the old encode_key_write
1973 * may be deleted once we are confident that the new works
1976 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1979 char *bp = i->buf, *bp0;
1980 const char *src = (char *) &key;
1982 /* copy term to output buf */
1983 while ((*bp++ = *k++))
1985 /* and copy & align key so we can mangle */
1986 memcpy (&key, k+1, sizeof(struct it_key)); /* *k is insert/delete */
1989 iscz1_encode(i->encode_handle, &bp, &src);
1990 *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */
1991 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1993 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
1998 void encode_key_flush (struct encode_info *i, FILE *outf)
1999 { /* dummy routine */
2000 iscz1_stop(i->encode_handle);
2005 /* new encode_key_write
2006 * The idea is to buffer one more key, and compare them
2007 * If we are going to delete and insert the same key,
2008 * we may as well not bother. Should make a difference in
2009 * updates with small modifications (appending to a mbox)
2011 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
2016 if (*k) /* first time for new key */
2019 while ((*bp++ = *k++))
2021 i->keylen= bp - i->buf -1;
2022 assert(i->keylen+1+sizeof(struct it_key) < ENCODE_BUFLEN);
2026 bp=i->buf + i->keylen;
2031 memcpy (&key, k+1, sizeof(struct it_key));
2032 if (0==i->prevsys) /* no previous filter, fill up */
2034 i->prevsys=key.sysno;
2035 i->prevseq=key.seqno;
2038 else if ( (i->prevsys==key.sysno) &&
2039 (i->prevseq==key.seqno) &&
2041 { /* same numbers, diff cmd, they cancel out */
2045 { /* different stuff, write previous, move buf */
2046 bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
2047 if (i->sysno != i->prevsys)
2049 i->sysno = i->prevsys;
2052 else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
2054 return; /* ??? Filters some sort of duplicates away */
2055 /* ??? Can this ever happen -H 15oct02 */
2057 bp = encode_key_int (i->prevseq - i->seqno, bp);
2058 i->seqno = i->prevseq;
2059 i->cmd = i->prevcmd;
2060 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
2062 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
2065 i->keylen=0; /* ok, it's written, forget it */
2066 i->prevsys=key.sysno;
2067 i->prevseq=key.seqno;
2072 void encode_key_flush (struct encode_info *i, FILE *outf)
2073 { /* flush the last key from i */
2074 char *bp =i->buf + i->keylen;
2077 return; /* nothing to flush */
2080 bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
2081 if (i->sysno != i->prevsys)
2083 i->sysno = i->prevsys;
2086 else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
2088 return; /* ??? Filters some sort of duplicates away */
2089 /* ??? Can this ever happen -H 15oct02 */
2091 bp = encode_key_int (i->prevseq - i->seqno, bp);
2092 i->seqno = i->prevseq;
2093 i->cmd = i->prevcmd;
2094 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
2096 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
2099 i->keylen=0; /* ok, it's written, forget it */
2100 i->prevsys=0; /* forget the values too */