1 /* $Id: extract.c,v 1.198 2005-11-09 11:51:29 adam Exp $
2 Copyright (C) 1995-2005
5 This file is part of the Zebra server.
7 Zebra is free software; you can redistribute it and/or modify it under
8 the terms of the GNU General Public License as published by the Free
9 Software Foundation; either version 2, or (at your option) any later
12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
14 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
17 You should have received a copy of the GNU General Public License
18 along with Zebra; see the file LICENSE.zebra. If not, write to the
19 Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
38 #if _FILE_OFFSET_BITS == 64
39 #define PRINTF_OFF_T "%Ld"
41 #define PRINTF_OFF_T "%ld"
44 #define USE_SHELLSORT 0
47 static void shellsort(void *ar, int r, size_t s,
48 int (*cmp)(const void *a, const void *b))
53 static const int incs[16] = { 1391376, 463792, 198768, 86961, 33936,
54 13776, 4592, 1968, 861, 336,
55 112, 48, 21, 7, 3, 1 };
56 for ( k = 0; k < 16; k++)
57 for (h = incs[k], i = h; i < r; i++)
61 while (j > h && (*cmp)(a + s*(j-h), v) > 0)
63 memcpy (a + s*j, a + s*(j-h), s);
71 static void logRecord (ZebraHandle zh)
73 ++zh->records_processed;
74 if (!(zh->records_processed % 1000))
76 yaz_log (YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
77 ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT,
78 zh->records_processed, zh->records_inserted, zh->records_updated,
83 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
85 static void extract_init (struct recExtractCtrl *p, RecWord *w)
87 w->zebra_maps = p->zebra_maps;
91 w->attrSet = VAL_BIB1;
101 static void searchRecordKey(ZebraHandle zh,
102 zebra_rec_keys_t reckeys,
103 int attrSetS, int attrUseS,
104 const char **ws, int ws_length)
109 for (i = 0; i<ws_length; i++)
112 ch = zebraExplain_lookup_attr_su_any_index(zh->reg->zei,
117 if (zebra_rec_keys_rewind(reckeys))
124 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
126 assert(key.len <= 4 && key.len > 2);
128 seqno = key.mem[key.len-1];
130 if (key.mem[0] == ch)
136 woff = seqno - startSeq;
137 if (woff >= 0 && woff < ws_length)
144 struct file_read_info {
145 off_t file_max; /* maximum offset so far */
146 off_t file_offset; /* current offset */
147 off_t file_moffset; /* offset of rec/rec boundary */
152 static struct file_read_info *file_read_start (int fd)
154 struct file_read_info *fi = (struct file_read_info *)
155 xmalloc (sizeof(*fi));
159 fi->file_moffset = 0;
165 static void file_read_stop (struct file_read_info *fi)
170 static off_t file_seek (void *handle, off_t offset)
172 struct file_read_info *p = (struct file_read_info *) handle;
173 p->file_offset = offset;
174 return lseek (p->fd, offset, SEEK_SET);
177 static off_t file_tell (void *handle)
179 struct file_read_info *p = (struct file_read_info *) handle;
180 return p->file_offset;
183 static int file_read (void *handle, char *buf, size_t count)
185 struct file_read_info *p = (struct file_read_info *) handle;
188 r = read (fd, buf, count);
192 if (p->file_offset > p->file_max)
193 p->file_max = p->file_offset;
198 static void file_end (void *handle, off_t offset)
200 struct file_read_info *p = (struct file_read_info *) handle;
202 if (offset != p->file_moffset)
204 p->file_moffset = offset;
209 static char *fileMatchStr (ZebraHandle zh,
210 zebra_rec_keys_t reckeys,
211 const char *fname, const char *spec)
213 static char dstBuf[2048]; /* static here ??? */
215 const char *s = spec;
219 while (*s == ' ' || *s == '\t')
226 char attset_str[64], attname_str[64];
227 data1_attset *attset;
230 int attSet = 1, attUse = 1;
234 for (i = 0; *s && *s != ',' && *s != ')'; s++)
236 attset_str[i++] = *s;
237 attset_str[i] = '\0';
242 for (i = 0; *s && *s != ')'; s++)
244 attname_str[i++] = *s;
245 attname_str[i] = '\0';
248 if ((attset = data1_get_attset (zh->reg->dh, attset_str)))
251 attSet = attset->reference;
252 att = data1_getattbyname(zh->reg->dh, attset, attname_str);
256 attUse = atoi (attname_str);
258 searchRecordKey (zh, reckeys, attSet, attUse, ws, 32);
262 for (i = 0; i<32; i++)
267 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
268 spec, zh->m_group ? zh->m_group : "none");
273 for (i = 0; i<32; i++)
274 if (matchFlag[i] && ws[i])
282 dst += strlen(ws[i]);
286 yaz_log (YLOG_WARN, "Record didn't contain match"
287 " fields in (%s,%s)", attset_str, attname_str);
295 const char *spec_src = NULL;
296 const char *s1 = ++s;
297 while (*s1 && *s1 != ' ' && *s1 != '\t')
303 memcpy (special, s, spec_len);
304 special[spec_len] = '\0';
307 if (!strcmp (special, "group"))
308 spec_src = zh->m_group;
309 else if (!strcmp (special, "database"))
310 spec_src = zh->basenames[0];
311 else if (!strcmp (special, "filename")) {
314 else if (!strcmp (special, "type"))
315 spec_src = zh->m_record_type;
320 strcpy (dst, spec_src);
321 dst += strlen (spec_src);
324 else if (*s == '\"' || *s == '\'')
326 int stopMarker = *s++;
330 while (*s && *s != stopMarker)
333 tmpString[i++] = *s++;
338 strcpy (dst, tmpString);
339 dst += strlen (tmpString);
343 yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
344 spec, zh->m_group ? zh->m_group : "none");
351 yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
352 fname, zh->m_group ? zh->m_group : "none");
359 struct recordLogInfo {
362 struct recordGroup *rGroup;
365 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
368 for (i = 0; i<256; i++)
370 if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
375 ctrl->zebra_maps = zh->reg->zebra_maps;
376 ctrl->flagShowRecords = !zh->m_flag_rw;
379 static int file_extract_record(ZebraHandle zh,
380 SYSNO *sysno, const char *fname,
382 struct file_read_info *fi,
385 void *recTypeClientData)
387 RecordAttr *recordAttr;
389 const char *matchStr = 0;
392 off_t recordOffset = 0;
393 struct recExtractCtrl extractCtrl;
395 /* announce database */
396 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
398 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
399 zh->m_explain_database))
405 /* we are going to read from a file, so prepare the extraction */
406 zebra_rec_keys_reset(zh->reg->keys);
409 zebra_rec_keys_reset(zh->reg->sortKeys);
411 zh->reg->sortKeys.buf_used = 0;
413 recordOffset = fi->file_moffset;
414 extractCtrl.handle = zh;
415 extractCtrl.offset = fi->file_moffset;
416 extractCtrl.readf = file_read;
417 extractCtrl.seekf = file_seek;
418 extractCtrl.tellf = file_tell;
419 extractCtrl.endf = file_end;
421 extractCtrl.init = extract_init;
422 extractCtrl.tokenAdd = extract_token_add;
423 extractCtrl.schemaAdd = extract_schema_add;
424 extractCtrl.dh = zh->reg->dh;
425 extractCtrl.match_criteria[0] = '\0';
426 extractCtrl.staticrank = 0;
428 extractCtrl.first_record = fi->file_offset ? 0 : 1;
430 extract_set_store_data_prepare(&extractCtrl);
432 init_extractCtrl(zh, &extractCtrl);
435 printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset);
439 sprintf (msg, "%s:" PRINTF_OFF_T , fname, recordOffset);
440 yaz_log_init_prefix2 (msg);
443 r = (*recType->extract)(recTypeClientData, &extractCtrl);
445 yaz_log_init_prefix2 (0);
446 if (r == RECCTRL_EXTRACT_EOF)
448 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
450 /* error occured during extraction ... */
452 zh->records_processed < zh->m_file_verbose_limit)
454 yaz_log (YLOG_WARN, "fail %s %s " PRINTF_OFF_T, zh->m_record_type,
455 fname, recordOffset);
459 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
461 /* error occured during extraction ... */
463 zh->records_processed < zh->m_file_verbose_limit)
465 yaz_log (YLOG_WARN, "no filter for %s %s "
466 PRINTF_OFF_T, zh->m_record_type,
467 fname, recordOffset);
471 if (extractCtrl.match_criteria[0])
472 matchStr = extractCtrl.match_criteria;
475 /* perform match if sysno not known and if match criteria is specified */
481 if (matchStr == 0 && zh->m_record_id && *zh->m_record_id)
483 matchStr = fileMatchStr (zh, zh->reg->keys, fname,
487 yaz_log(YLOG_WARN, "Bad match criteria");
493 char *rinfo = dict_lookup (zh->reg->matchDict, matchStr);
496 assert(*rinfo == sizeof(*sysno));
497 memcpy (sysno, rinfo+1, sizeof(*sysno));
501 if (! *sysno && zebra_rec_keys_empty(zh->reg->keys) )
503 /* the extraction process returned no information - the record
504 is probably empty - unless flagShowRecords is in use */
508 if (zh->records_processed < zh->m_file_verbose_limit)
509 yaz_log (YLOG_WARN, "empty %s %s " PRINTF_OFF_T, zh->m_record_type,
510 fname, recordOffset);
519 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T, zh->m_record_type,
520 fname, recordOffset);
521 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
524 if (zh->records_processed < zh->m_file_verbose_limit)
525 yaz_log (YLOG_LOG, "add %s %s " PRINTF_OFF_T, zh->m_record_type,
526 fname, recordOffset);
527 rec = rec_new (zh->reg->records);
531 recordAttr = rec_init_attr (zh->reg->zei, rec);
532 recordAttr->staticrank = extractCtrl.staticrank;
536 dict_insert (zh->reg->matchDict, matchStr, sizeof(*sysno), sysno);
539 extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
541 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
543 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
544 recordAttr->staticrank);
545 zh->records_inserted++;
549 /* record already exists */
550 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
553 zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
555 struct sortKeys sortKeys;
558 rec = rec_get (zh->reg->records, *sysno);
561 recordAttr = rec_init_attr (zh->reg->zei, rec);
563 zebra_rec_keys_set_buf(delkeys,
564 rec->info[recInfo_delKeys],
565 rec->size[recInfo_delKeys],
569 zebra_rec_keys_set_buf(sortKeys,
570 rec->info[recInfo_sortKeys],
571 rec->size[recInfo_sortKeys],
573 extract_flushSortKeys (zh, *sysno, 0, sortKeys);
575 sortKeys.buf_used = rec->size[recInfo_sortKeys];
576 sortKeys.buf = rec->info[recInfo_sortKeys];
577 extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
580 extract_flushRecordKeys (zh, *sysno, 0, delkeys,
581 recordAttr->staticrank); /* old values */
584 /* record going to be deleted */
585 if (zebra_rec_keys_empty(delkeys))
587 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
588 zh->m_record_type, fname, recordOffset);
589 yaz_log (YLOG_WARN, "cannot delete file above, storeKeys false");
593 if (zh->records_processed < zh->m_file_verbose_limit)
594 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
595 zh->m_record_type, fname, recordOffset);
596 zh->records_deleted++;
598 dict_delete (zh->reg->matchDict, matchStr);
599 rec_del (zh->reg->records, &rec);
607 /* record going to be updated */
608 if (zebra_rec_keys_empty(delkeys))
610 yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
611 zh->m_record_type, fname, recordOffset);
612 yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
616 /* flush new keys for sort&search etc */
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 recordAttr->staticrank = extractCtrl.staticrank;
622 extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
624 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
626 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
627 recordAttr->staticrank);
628 zh->records_updated++;
631 zebra_rec_keys_close(delkeys);
633 zebra_rec_keys_close(sortKeys);
636 /* update file type */
637 xfree (rec->info[recInfo_fileType]);
638 rec->info[recInfo_fileType] =
639 rec_strdup (zh->m_record_type, &rec->size[recInfo_fileType]);
641 /* update filename */
642 xfree (rec->info[recInfo_filename]);
643 rec->info[recInfo_filename] =
644 rec_strdup (fname, &rec->size[recInfo_filename]);
646 /* update delete keys */
647 xfree (rec->info[recInfo_delKeys]);
648 if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
650 zebra_rec_keys_get_buf(zh->reg->keys,
651 &rec->info[recInfo_delKeys],
652 &rec->size[recInfo_delKeys]);
656 rec->info[recInfo_delKeys] = NULL;
657 rec->size[recInfo_delKeys] = 0;
660 /* update sort keys */
661 xfree (rec->info[recInfo_sortKeys]);
664 zebra_rec_keys_get_buf(zh->reg->sortKeys,
665 &rec->info[recInfo_sortKeys],
666 &rec->size[recInfo_sortKeys]);
668 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
669 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
670 zh->reg->sortKeys.buf = NULL;
671 zh->reg->sortKeys.buf_max = 0;
674 /* save file size of original record */
675 zebraExplain_recordBytesIncrement (zh->reg->zei,
676 - recordAttr->recordSize);
677 recordAttr->recordSize = fi->file_moffset - recordOffset;
678 if (!recordAttr->recordSize)
679 recordAttr->recordSize = fi->file_max - recordOffset;
680 zebraExplain_recordBytesIncrement (zh->reg->zei,
681 recordAttr->recordSize);
683 /* set run-number for this record */
684 recordAttr->runNumber = zebraExplain_runNumberIncrement (zh->reg->zei,
687 /* update store data */
688 xfree (rec->info[recInfo_storeData]);
689 if (zh->store_data_buf)
691 rec->size[recInfo_storeData] = zh->store_data_size;
692 rec->info[recInfo_storeData] = zh->store_data_buf;
693 zh->store_data_buf = 0;
695 else if (zh->m_store_data)
697 rec->size[recInfo_storeData] = recordAttr->recordSize;
698 rec->info[recInfo_storeData] = (char *)
699 xmalloc (recordAttr->recordSize);
700 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
702 yaz_log (YLOG_ERRNO|YLOG_FATAL, "seek to " PRINTF_OFF_T " in %s",
703 recordOffset, fname);
706 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
707 < recordAttr->recordSize)
709 yaz_log (YLOG_ERRNO|YLOG_FATAL, "read %d bytes of %s",
710 recordAttr->recordSize, fname);
716 rec->info[recInfo_storeData] = NULL;
717 rec->size[recInfo_storeData] = 0;
719 /* update database name */
720 xfree (rec->info[recInfo_databaseName]);
721 rec->info[recInfo_databaseName] =
722 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
725 recordAttr->recordOffset = recordOffset;
727 /* commit this record */
728 rec_put (zh->reg->records, &rec);
733 int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
740 struct file_read_info *fi;
741 const char *original_record_type = 0;
743 void *recTypeClientData;
745 if (!zh->m_group || !*zh->m_group)
748 sprintf (gprefix, "%s.", zh->m_group);
750 yaz_log (YLOG_DEBUG, "fileExtract %s", fname);
752 /* determine file extension */
754 for (i = strlen(fname); --i >= 0; )
757 else if (fname[i] == '.')
759 strcpy (ext, fname+i+1);
762 /* determine file type - depending on extension */
763 original_record_type = zh->m_record_type;
764 if (!zh->m_record_type)
766 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
767 zh->m_record_type = res_get (zh->res, ext_res);
769 if (!zh->m_record_type)
771 if (zh->records_processed < zh->m_file_verbose_limit)
772 yaz_log (YLOG_LOG, "? %s", fname);
775 /* determine match criteria */
776 if (!zh->m_record_id)
778 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
779 zh->m_record_id = res_get (zh->res, ext_res);
783 recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
784 &recTypeClientData)))
786 yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
790 switch(recType->version)
795 yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
797 if (sysno && deleteFlag)
803 if (zh->path_reg && !yaz_is_abspath (fname))
805 strcpy (full_rep, zh->path_reg);
806 strcat (full_rep, "/");
807 strcat (full_rep, fname);
810 strcpy (full_rep, fname);
813 if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
815 yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
816 zh->m_record_type = original_record_type;
820 fi = file_read_start (fd);
823 fi->file_moffset = fi->file_offset;
824 fi->file_more = 0; /* file_end not called (yet) */
825 r = file_extract_record (zh, sysno, fname, deleteFlag, fi, 1,
826 recType, recTypeClientData);
828 { /* file_end has been called so reset offset .. */
829 fi->file_offset = fi->file_moffset;
830 lseek(fi->fd, fi->file_moffset, SEEK_SET);
837 zh->m_record_type = original_record_type;
842 If sysno is provided, then it's used to identify the reocord.
843 If not, and match_criteria is provided, then sysno is guessed
844 If not, and a record is provided, then sysno is got from there
847 ZEBRA_RES buffer_extract_record(ZebraHandle zh,
848 const char *buf, size_t buf_size,
851 const char *recordType,
853 const char *match_criteria,
858 RecordAttr *recordAttr;
859 struct recExtractCtrl extractCtrl;
861 const char *matchStr = 0;
862 RecType recType = NULL;
865 long recordOffset = 0;
866 struct zebra_fetch_control fc;
867 const char *pr_fname = fname; /* filename to print .. */
868 int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
871 pr_fname = "<no file>"; /* make it printable if file is omitted */
874 fc.record_int_buf = buf;
875 fc.record_int_len = buf_size;
876 fc.record_int_pos = 0;
878 fc.record_offset = 0;
880 extractCtrl.offset = 0;
881 extractCtrl.readf = zebra_record_int_read;
882 extractCtrl.seekf = zebra_record_int_seek;
883 extractCtrl.tellf = zebra_record_int_tell;
884 extractCtrl.endf = zebra_record_int_end;
885 extractCtrl.first_record = 1;
886 extractCtrl.fh = &fc;
888 zebra_rec_keys_reset(zh->reg->keys);
891 zebra_rec_keys_reset(zh->reg->sortKeys);
893 zh->reg->sortKeys.buf_used = 0;
895 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
897 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
898 zh->m_explain_database))
902 if (recordType && *recordType)
904 yaz_log (YLOG_DEBUG, "Record type explicitly specified: %s", recordType);
905 recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
910 if (!(zh->m_record_type))
912 yaz_log (YLOG_WARN, "No such record type defined");
915 yaz_log (YLOG_DEBUG, "Get record type from rgroup: %s",zh->m_record_type);
916 recType = recType_byName (zh->reg->recTypes, zh->res,
917 zh->m_record_type, &clientData);
918 recordType = zh->m_record_type;
923 yaz_log (YLOG_WARN, "No such record type: %s", zh->m_record_type);
927 extractCtrl.init = extract_init;
928 extractCtrl.tokenAdd = extract_token_add;
929 extractCtrl.schemaAdd = extract_schema_add;
930 extractCtrl.dh = zh->reg->dh;
931 extractCtrl.handle = zh;
932 extractCtrl.match_criteria[0] = '\0';
933 extractCtrl.staticrank = 0;
935 init_extractCtrl(zh, &extractCtrl);
937 extract_set_store_data_prepare(&extractCtrl);
939 r = (*recType->extract)(clientData, &extractCtrl);
941 if (r == RECCTRL_EXTRACT_EOF)
943 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
945 /* error occured during extraction ... */
946 yaz_log (YLOG_WARN, "extract error: generic");
949 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
951 /* error occured during extraction ... */
952 yaz_log (YLOG_WARN, "extract error: no such filter");
958 if (extractCtrl.match_criteria[0])
959 match_criteria = extractCtrl.match_criteria;
963 if (match_criteria && *match_criteria) {
964 matchStr = match_criteria;
966 if (zh->m_record_id && *zh->m_record_id) {
967 matchStr = fileMatchStr (zh, zh->reg->keys, pr_fname,
971 yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
977 rinfo = dict_lookup (zh->reg->matchDict, matchStr);
980 assert(*rinfo == sizeof(*sysno));
981 memcpy (sysno, rinfo+1, sizeof(*sysno));
985 if (zebra_rec_keys_empty(zh->reg->keys))
987 /* the extraction process returned no information - the record
988 is probably empty - unless flagShowRecords is in use */
999 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1000 pr_fname, (long) recordOffset);
1001 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
1005 yaz_log (YLOG_LOG, "add %s %s %ld", recordType, pr_fname,
1006 (long) recordOffset);
1007 rec = rec_new (zh->reg->records);
1009 *sysno = rec->sysno;
1011 recordAttr = rec_init_attr (zh->reg->zei, rec);
1012 recordAttr->staticrank = extractCtrl.staticrank;
1016 dict_insert (zh->reg->matchDict, matchStr,
1017 sizeof(*sysno), sysno);
1020 extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
1022 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1026 print_rec_keys(zh, zh->reg->keys);
1028 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
1029 recordAttr->staticrank);
1030 zh->records_inserted++;
1034 /* record already exists */
1035 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1037 zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
1039 struct sortKeys sortKeys;
1045 yaz_log (YLOG_LOG, "skipped %s %s %ld",
1046 recordType, pr_fname, (long) recordOffset);
1051 rec = rec_get (zh->reg->records, *sysno);
1054 recordAttr = rec_init_attr (zh->reg->zei, rec);
1056 zebra_rec_keys_set_buf(delkeys,
1057 rec->info[recInfo_delKeys],
1058 rec->size[recInfo_delKeys],
1061 zebra_rec_keys_set_buf(sortKeys,
1062 rec->info[recInfo_sortKeys],
1063 rec->size[recInfo_sortKeys],
1066 sortKeys.buf_used = rec->size[recInfo_sortKeys];
1067 sortKeys.buf = rec->info[recInfo_sortKeys];
1071 extract_flushSortKeys (zh, *sysno, 0, sortKeys);
1073 extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
1075 extract_flushRecordKeys (zh, *sysno, 0, delkeys,
1076 recordAttr->staticrank);
1079 /* record going to be deleted */
1080 if (zebra_rec_keys_empty(delkeys))
1084 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1085 pr_fname, (long) recordOffset);
1086 yaz_log (YLOG_WARN, "cannot delete file above, "
1093 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1094 pr_fname, (long) recordOffset);
1095 zh->records_deleted++;
1097 dict_delete (zh->reg->matchDict, matchStr);
1098 rec_del (zh->reg->records, &rec);
1106 /* record going to be updated */
1107 if (zebra_rec_keys_empty(delkeys))
1111 yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
1112 pr_fname, (long) recordOffset);
1113 yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
1119 yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
1120 pr_fname, (long) recordOffset);
1121 recordAttr->staticrank = extractCtrl.staticrank;
1123 extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
1125 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1127 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
1128 recordAttr->staticrank);
1129 zh->records_updated++;
1132 zebra_rec_keys_close(delkeys);
1134 zebra_rec_keys_close(sortKeys);
1137 /* update file type */
1138 xfree (rec->info[recInfo_fileType]);
1139 rec->info[recInfo_fileType] =
1140 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1142 /* update filename */
1143 xfree (rec->info[recInfo_filename]);
1144 rec->info[recInfo_filename] =
1145 rec_strdup (fname, &rec->size[recInfo_filename]);
1147 /* update delete keys */
1148 xfree (rec->info[recInfo_delKeys]);
1149 if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
1151 zebra_rec_keys_get_buf(zh->reg->keys,
1152 &rec->info[recInfo_delKeys],
1153 &rec->size[recInfo_delKeys]);
1157 rec->info[recInfo_delKeys] = NULL;
1158 rec->size[recInfo_delKeys] = 0;
1160 /* update sort keys */
1161 xfree (rec->info[recInfo_sortKeys]);
1164 zebra_rec_keys_get_buf(zh->reg->sortKeys,
1165 &rec->info[recInfo_sortKeys],
1166 &rec->size[recInfo_sortKeys]);
1168 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
1169 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
1170 zh->reg->sortKeys.buf = NULL;
1171 zh->reg->sortKeys.buf_max = 0;
1174 /* save file size of original record */
1175 zebraExplain_recordBytesIncrement (zh->reg->zei,
1176 - recordAttr->recordSize);
1178 recordAttr->recordSize = fi->file_moffset - recordOffset;
1179 if (!recordAttr->recordSize)
1180 recordAttr->recordSize = fi->file_max - recordOffset;
1182 recordAttr->recordSize = buf_size;
1184 zebraExplain_recordBytesIncrement (zh->reg->zei,
1185 recordAttr->recordSize);
1187 /* set run-number for this record */
1188 recordAttr->runNumber =
1189 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
1191 /* update store data */
1192 xfree (rec->info[recInfo_storeData]);
1194 /* update store data */
1195 if (zh->store_data_buf)
1197 rec->size[recInfo_storeData] = zh->store_data_size;
1198 rec->info[recInfo_storeData] = zh->store_data_buf;
1199 zh->store_data_buf = 0;
1201 else if (zh->m_store_data)
1203 rec->size[recInfo_storeData] = recordAttr->recordSize;
1204 rec->info[recInfo_storeData] = (char *)
1205 xmalloc (recordAttr->recordSize);
1206 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1210 rec->info[recInfo_storeData] = NULL;
1211 rec->size[recInfo_storeData] = 0;
1213 /* update database name */
1214 xfree (rec->info[recInfo_databaseName]);
1215 rec->info[recInfo_databaseName] =
1216 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
1219 recordAttr->recordOffset = recordOffset;
1221 /* commit this record */
1222 rec_put (zh->reg->records, &rec);
1227 int explain_extract (void *handle, Record rec, data1_node *n)
1229 ZebraHandle zh = (ZebraHandle) handle;
1230 struct recExtractCtrl extractCtrl;
1232 if (zebraExplain_curDatabase (zh->reg->zei,
1233 rec->info[recInfo_databaseName]))
1236 if (zebraExplain_newDatabase (zh->reg->zei,
1237 rec->info[recInfo_databaseName], 0))
1241 zebra_rec_keys_reset(zh->reg->keys);
1244 zebra_rec_keys_reset(zh->reg->sortKeys);
1246 zh->reg->sortKeys.buf_used = 0;
1248 extractCtrl.init = extract_init;
1249 extractCtrl.tokenAdd = extract_token_add;
1250 extractCtrl.schemaAdd = extract_schema_add;
1251 extractCtrl.dh = zh->reg->dh;
1253 init_extractCtrl(zh, &extractCtrl);
1255 extractCtrl.flagShowRecords = 0;
1256 extractCtrl.match_criteria[0] = '\0';
1257 extractCtrl.staticrank = 0;
1258 extractCtrl.handle = handle;
1259 extractCtrl.first_record = 1;
1261 extract_set_store_data_prepare(&extractCtrl);
1264 grs_extract_tree(&extractCtrl, n);
1266 if (rec->size[recInfo_delKeys])
1268 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1271 zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
1273 struct sortKeys sortkeys;
1276 zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
1277 rec->size[recInfo_delKeys],
1279 extract_flushRecordKeys (zh, rec->sysno, 0, delkeys, 0);
1280 zebra_rec_keys_close(delkeys);
1282 zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
1283 rec->size[recInfo_sortKeys],
1286 extract_flushSortKeys (zh, rec->sysno, 0, sortkeys);
1287 zebra_rec_keys_close(sortkeys);
1289 sortkeys.buf_used = rec->size[recInfo_sortKeys];
1290 sortkeys.buf = rec->info[recInfo_sortKeys];
1291 extract_flushSortKeys (zh, rec->sysno, 0, &sortkeys);
1294 extract_flushRecordKeys (zh, rec->sysno, 1, zh->reg->keys, 0);
1296 extract_flushSortKeys (zh, rec->sysno, 1, zh->reg->sortKeys);
1298 extract_flushSortKeys (zh, rec->sysno, 1, &zh->reg->sortKeys);
1301 xfree (rec->info[recInfo_delKeys]);
1302 zebra_rec_keys_get_buf(zh->reg->keys,
1303 &rec->info[recInfo_delKeys],
1304 &rec->size[recInfo_delKeys]);
1306 xfree (rec->info[recInfo_sortKeys]);
1308 zebra_rec_keys_get_buf(zh->reg->sortKeys,
1309 &rec->info[recInfo_sortKeys],
1310 &rec->size[recInfo_sortKeys]);
1312 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
1313 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
1314 zh->reg->sortKeys.buf = NULL;
1315 zh->reg->sortKeys.buf_max = 0;
1321 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1323 zebra_rec_keys_t reckeys,
1326 ZebraExplainInfo zei = zh->reg->zei;
1328 if (!zh->reg->key_buf)
1330 int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
1333 yaz_log(YLOG_WARN, "Invalid memory setting, using default 8 MB");
1336 /* FIXME: That "8" should be in a default settings include */
1337 /* not hard-coded here! -H */
1338 zh->reg->key_buf = (char**) xmalloc (mem);
1339 zh->reg->ptr_top = mem/sizeof(char*);
1341 zh->reg->key_buf_used = 0;
1342 zh->reg->key_file_no = 0;
1344 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1346 if (zebra_rec_keys_rewind(reckeys))
1350 struct it_key key_in;
1351 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1354 struct it_key key_out;
1355 zint *keyp = key_out.mem;
1357 assert(key_in.len == 4);
1359 /* check for buffer overflow */
1360 if (zh->reg->key_buf_used + 1024 >
1361 (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
1362 extract_flushWriteKeys (zh, 0);
1365 assert(zh->reg->ptr_i > 0);
1366 (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
1367 (char*)zh->reg->key_buf + zh->reg->key_buf_used;
1369 /* encode the ordinal value (field/use/attribute) .. */
1370 ch = (int) key_in.mem[0];
1371 zh->reg->key_buf_used +=
1372 key_SU_encode(ch, (char*)zh->reg->key_buf +
1373 zh->reg->key_buf_used);
1375 /* copy the 0-terminated stuff from str to output */
1376 memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used, str, slen);
1377 zh->reg->key_buf_used += slen;
1378 ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = '\0';
1380 /* the delete/insert indicator */
1381 ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = cmd;
1383 if (zh->m_staticrank) /* rank config enabled ? */
1385 *keyp++ = staticrank;
1391 if (key_in.mem[1]) /* filter specified record ID */
1392 *keyp++ = key_in.mem[1];
1395 *keyp++ = key_in.mem[2]; /* section_id */
1396 *keyp++ = key_in.mem[3]; /* sequence .. */
1398 memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used,
1399 &key_out, sizeof(key_out));
1400 (zh->reg->key_buf_used) += sizeof(key_out);
1405 void extract_flushWriteKeys (ZebraHandle zh, int final)
1406 /* optimizing: if final=1, and no files written yet */
1407 /* push the keys directly to merge, sidestepping the */
1408 /* temp file altogether. Speeds small updates */
1411 char out_fname[200];
1413 struct encode_info encode_info;
1414 int ptr_i = zh->reg->ptr_i;
1419 if (!zh->reg->key_buf || ptr_i <= 0)
1421 yaz_log (YLOG_DEBUG, " nothing to flush section=%d buf=%p i=%d",
1422 zh->reg->key_file_no, zh->reg->key_buf, ptr_i);
1423 yaz_log (YLOG_DEBUG, " buf=%p ",
1425 yaz_log (YLOG_DEBUG, " ptr=%d ",zh->reg->ptr_i);
1426 yaz_log (YLOG_DEBUG, " reg=%p ",zh->reg);
1431 (zh->reg->key_file_no)++;
1432 yaz_log (YLOG_LOG, "sorting section %d", (zh->reg->key_file_no));
1433 yaz_log (YLOG_DEBUG, " sort_buff at %p n=%d",
1434 zh->reg->key_buf + zh->reg->ptr_top - ptr_i,ptr_i);
1436 qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
1437 sizeof(char*), key_qsort_compare);
1439 /* zebra.cfg: tempfiles:
1440 Y: always use temp files (old way)
1441 A: use temp files, if more than one (auto)
1442 = if this is both the last and the first
1443 N: never bother with temp files (new) */
1445 temp_policy=toupper(res_get_def(zh->res,"tempfiles","auto")[0]);
1446 if (temp_policy != 'Y' && temp_policy != 'N' && temp_policy != 'A') {
1447 yaz_log (YLOG_WARN, "Illegal tempfiles setting '%c'. using 'Auto' ",
1452 if ( ( temp_policy =='N' ) || /* always from memory */
1453 ( ( temp_policy =='A' ) && /* automatic */
1454 (zh->reg->key_file_no == 1) && /* this is first time */
1455 (final) ) ) /* and last (=only) time */
1456 { /* go directly from memory */
1457 zh->reg->key_file_no =0; /* signal not to read files */
1458 zebra_index_merge(zh);
1460 zh->reg->key_buf_used = 0;
1464 /* Not doing directly from memory, write into a temp file */
1465 extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
1467 if (!(outf = fopen (out_fname, "wb")))
1469 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1472 yaz_log (YLOG_LOG, "writing section %d", zh->reg->key_file_no);
1473 prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1475 encode_key_init (&encode_info);
1476 encode_key_write (cp, &encode_info, outf);
1480 cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1481 if (strcmp (cp, prevcp))
1483 encode_key_flush ( &encode_info, outf);
1484 encode_key_init (&encode_info);
1485 encode_key_write (cp, &encode_info, outf);
1489 encode_key_write (cp + strlen(cp), &encode_info, outf);
1491 encode_key_flush ( &encode_info, outf);
1493 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1494 extract_get_fname_tmp (out_fname, key_file_no);
1496 if (!(outf = fopen (out_fname, "wb")))
1498 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1501 yaz_log (YLOG_LOG, "writing section %d", key_file_no);
1503 prevcp = key_buf[ptr_top-i];
1505 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1507 key_y_len = strlen(prevcp)+1;
1509 yaz_log (YLOG_LOG, "key_y_len: %2d %02x %02x %s",
1510 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1512 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1513 sizeof(char*), key_y_compare);
1514 cp = key_buf[ptr_top-ptr_i];
1516 encode_key_init (&encode_info);
1517 encode_key_write (cp, &encode_info, outf);
1520 cp = key_buf[ptr_top-ptr_i];
1521 encode_key_write (cp+key_y_len, &encode_info, outf);
1523 encode_key_flush ( &encode_info, outf);
1526 prevcp = key_buf[ptr_top-ptr_i];
1531 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fclose %s", out_fname);
1534 yaz_log (YLOG_LOG, "finished section %d", zh->reg->key_file_no);
1536 zh->reg->key_buf_used = 0;
1539 ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh,
1540 zebra_rec_keys_t reckeys,
1541 zebra_snippets *snippets)
1543 NMEM nmem = nmem_create();
1544 if (zebra_rec_keys_rewind(reckeys))
1549 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1551 char dst_buf[IT_MAX_WORD];
1552 char *dst_term = dst_buf;
1555 assert(key.len <= 4 && key.len > 2);
1556 seqno = (int) key.mem[key.len-1];
1559 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1560 0/* db */, 0/* set */, 0/* use */);
1562 zebra_term_untrans_iconv(zh, nmem, index_type,
1564 zebra_snippets_append(snippets, seqno, ord, dst_term);
1572 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1574 yaz_log(YLOG_LOG, "print_rec_keys");
1575 if (zebra_rec_keys_rewind(reckeys))
1580 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1582 char dst_buf[IT_MAX_WORD];
1586 assert(key.len <= 4 && key.len > 2);
1588 zebraExplain_lookup_ord(zh->reg->zei,
1589 key.mem[0], &index_type, &db, 0, 0);
1591 seqno = (int) key.mem[key.len-1];
1593 zebra_term_untrans(zh, index_type, dst_buf, str);
1595 yaz_log(YLOG_LOG, "ord=" ZINT_FORMAT " seqno=%d term=%s",
1596 key.mem[0], seqno, dst_buf);
1601 void extract_add_index_string (RecWord *p, const char *str, int length)
1605 ZebraHandle zh = p->extractCtrl->handle;
1606 ZebraExplainInfo zei = zh->reg->zei;
1611 ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
1613 ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
1620 ch = zebraExplain_lookup_attr_su(zei, p->index_type,
1621 p->attrSet, p->attrUse);
1623 ch = zebraExplain_add_attr_su(zei, p->index_type,
1624 p->attrSet, p->attrUse);
1629 key.mem[1] = p->record_id;
1630 key.mem[2] = p->section_id;
1631 key.mem[3] = p->seqno;
1640 for (i = 0; i<length && i < 20; i++)
1641 sprintf(strz+strlen(strz), "%02X", str[i] & 0xff);
1642 /* just for debugging .. */
1643 yaz_log(YLOG_LOG, "add: set=%d use=%d "
1644 "record_id=%lld section_id=%lld seqno=%lld %s",
1645 p->attrSet, p->attrUse, p->record_id, p->section_id, p->seqno,
1650 zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1654 static void extract_add_sort_string (RecWord *p, const char *str, int length)
1658 ZebraHandle zh = p->extractCtrl->handle;
1659 ZebraExplainInfo zei = zh->reg->zei;
1664 ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
1666 ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
1674 key.mem[1] = p->record_id;
1675 key.mem[2] = p->section_id;
1676 key.mem[3] = p->seqno;
1678 zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1681 static void extract_add_sort_string (RecWord *p, const char *str, int length)
1683 ZebraHandle zh = p->extractCtrl->handle;
1684 struct sortKeys *sk = &zh->reg->sortKeys;
1687 while (off < sk->buf_used)
1691 off += key_SU_decode(&set, (unsigned char *) sk->buf + off);
1692 off += key_SU_decode(&use, (unsigned char *) sk->buf + off);
1693 off += key_SU_decode(&slen, (unsigned char *) sk->buf + off);
1695 if (p->attrSet == set && p->attrUse == use)
1698 assert (off == sk->buf_used);
1700 if (sk->buf_used + IT_MAX_WORD > sk->buf_max)
1704 b = (char *) xmalloc (sk->buf_max += 128000);
1705 if (sk->buf_used > 0)
1706 memcpy (b, sk->buf, sk->buf_used);
1710 off += key_SU_encode(p->attrSet, sk->buf + off);
1711 off += key_SU_encode(p->attrUse, sk->buf + off);
1712 off += key_SU_encode(length, sk->buf + off);
1713 memcpy (sk->buf + off, str, length);
1714 sk->buf_used = off + length;
1718 void extract_add_string (RecWord *p, const char *string, int length)
1720 assert (length > 0);
1721 if (zebra_maps_is_sort (p->zebra_maps, p->index_type))
1722 extract_add_sort_string (p, string, length);
1724 extract_add_index_string (p, string, length);
1727 static void extract_add_incomplete_field (RecWord *p)
1729 const char *b = p->term_buf;
1730 int remain = p->term_len;
1731 const char **map = 0;
1733 yaz_log(YLOG_DEBUG, "Incomplete field, w='%.*s'", p->term_len, p->term_buf);
1736 map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, 0);
1740 char buf[IT_MAX_WORD+1];
1744 while (map && *map && **map == *CHR_SPACE)
1746 remain = p->term_len - (b - p->term_buf);
1748 map = zebra_maps_input(p->zebra_maps, p->index_type, &b,
1756 while (map && *map && **map != *CHR_SPACE)
1758 const char *cp = *map;
1760 while (i < IT_MAX_WORD && *cp)
1762 remain = p->term_len - (b - p->term_buf);
1764 map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, 0);
1770 extract_add_string (p, buf, i);
1775 static void extract_add_complete_field (RecWord *p)
1777 const char *b = p->term_buf;
1778 char buf[IT_MAX_WORD+1];
1779 const char **map = 0;
1780 int i = 0, remain = p->term_len;
1782 yaz_log(YLOG_DEBUG, "Complete field, w='%.*s'",
1783 p->term_len, p->term_buf);
1786 map = zebra_maps_input (p->zebra_maps, p->index_type, &b, remain, 1);
1788 while (remain > 0 && i < IT_MAX_WORD)
1790 while (map && *map && **map == *CHR_SPACE)
1792 remain = p->term_len - (b - p->term_buf);
1796 int first = i ? 0 : 1; /* first position */
1797 map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, first);
1805 if (i && i < IT_MAX_WORD)
1806 buf[i++] = *CHR_SPACE;
1807 while (map && *map && **map != *CHR_SPACE)
1809 const char *cp = *map;
1811 if (**map == *CHR_CUT)
1817 if (i >= IT_MAX_WORD)
1819 yaz_log(YLOG_DEBUG, "Adding string to index '%d'", **map);
1820 while (i < IT_MAX_WORD && *cp)
1823 remain = p->term_len - (b - p->term_buf);
1826 map = zebra_maps_input (p->zebra_maps, p->index_type, &b,
1835 extract_add_string (p, buf, i);
1838 void extract_token_add (RecWord *p)
1842 yaz_log (YLOG_LOG, "token_add "
1843 "reg_type=%c attrSet=%d attrUse=%d seqno=%d s=%.*s",
1844 p->reg_type, p->attrSet, p->attrUse, p->seqno, p->length,
1847 if ((wrbuf = zebra_replace(p->zebra_maps, p->index_type, 0,
1848 p->term_buf, p->term_len)))
1850 p->term_buf = wrbuf_buf(wrbuf);
1851 p->term_len = wrbuf_len(wrbuf);
1853 if (zebra_maps_is_complete (p->zebra_maps, p->index_type))
1854 extract_add_complete_field (p);
1856 extract_add_incomplete_field(p);
1859 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1860 void *buf, size_t sz)
1862 ZebraHandle zh = (ZebraHandle) p->handle;
1864 xfree(zh->store_data_buf);
1865 zh->store_data_buf = 0;
1866 zh->store_data_size = 0;
1869 zh->store_data_buf = xmalloc(sz);
1870 zh->store_data_size = sz;
1871 memcpy(zh->store_data_buf, buf, sz);
1875 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1877 ZebraHandle zh = (ZebraHandle) p->handle;
1878 xfree(zh->store_data_buf);
1879 zh->store_data_buf = 0;
1880 zh->store_data_size = 0;
1881 p->setStoreData = extract_set_store_data_cb;
1884 void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1886 ZebraHandle zh = (ZebraHandle) p->handle;
1887 zebraExplain_addSchema (zh->reg->zei, oid);
1891 #error not done yet with zebra_rec_keys_t
1892 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1893 int cmd, struct recKeys *reckeys)
1895 SortIdx sortIdx = zh->reg->sortIdx;
1896 void *decode_handle = iscz1_start();
1900 while (off < reckeys->buf_used)
1902 const char *src = reckeys->buf + off;
1904 char *dst = (char*) &key;
1906 iscz1_decode(decode_handle, &dst, &src);
1907 assert(key.len == 4);
1909 ch = (int) key.mem[0]; /* ordinal for field/use/attribute */
1911 sortIdx_type(sortIdx, ch);
1913 sortIdx_add(sortIdx, src, strlen(src));
1915 sortIdx_add(sortIdx, "", 1);
1920 off = src - reckeys->buf;
1922 assert (off == reckeys->buf_used);
1923 iscz1_stop(decode_handle);
1926 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1927 int cmd, struct sortKeys *sk)
1929 SortIdx sortIdx = zh->reg->sortIdx;
1932 sortIdx_sysno (sortIdx, sysno);
1934 while (off < sk->buf_used)
1938 off += key_SU_decode(&set, (unsigned char *) sk->buf + off);
1939 off += key_SU_decode(&use, (unsigned char *) sk->buf + off);
1940 off += key_SU_decode(&slen, (unsigned char *) sk->buf + off);
1942 sortIdx_type(sortIdx, use);
1944 sortIdx_add(sortIdx, sk->buf + off, slen);
1946 sortIdx_add(sortIdx, "", 1);
1952 void encode_key_init (struct encode_info *i)
1961 i->encode_handle = iscz1_start();
1967 /* this is the old encode_key_write
1968 * may be deleted once we are confident that the new works
1971 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1974 char *bp = i->buf, *bp0;
1975 const char *src = (char *) &key;
1977 /* copy term to output buf */
1978 while ((*bp++ = *k++))
1980 /* and copy & align key so we can mangle */
1981 memcpy (&key, k+1, sizeof(struct it_key)); /* *k is insert/delete */
1984 iscz1_encode(i->encode_handle, &bp, &src);
1985 *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */
1986 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1988 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
1993 void encode_key_flush (struct encode_info *i, FILE *outf)
1994 { /* dummy routine */
1995 iscz1_stop(i->encode_handle);
2000 /* new encode_key_write
2001 * The idea is to buffer one more key, and compare them
2002 * If we are going to delete and insert the same key,
2003 * we may as well not bother. Should make a difference in
2004 * updates with small modifications (appending to a mbox)
2006 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
2011 if (*k) /* first time for new key */
2014 while ((*bp++ = *k++))
2016 i->keylen= bp - i->buf -1;
2017 assert(i->keylen+1+sizeof(struct it_key) < ENCODE_BUFLEN);
2021 bp=i->buf + i->keylen;
2026 memcpy (&key, k+1, sizeof(struct it_key));
2027 if (0==i->prevsys) /* no previous filter, fill up */
2029 i->prevsys=key.sysno;
2030 i->prevseq=key.seqno;
2033 else if ( (i->prevsys==key.sysno) &&
2034 (i->prevseq==key.seqno) &&
2036 { /* same numbers, diff cmd, they cancel out */
2040 { /* different stuff, write previous, move buf */
2041 bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
2042 if (i->sysno != i->prevsys)
2044 i->sysno = i->prevsys;
2047 else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
2049 return; /* ??? Filters some sort of duplicates away */
2050 /* ??? Can this ever happen -H 15oct02 */
2052 bp = encode_key_int (i->prevseq - i->seqno, bp);
2053 i->seqno = i->prevseq;
2054 i->cmd = i->prevcmd;
2055 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
2057 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
2060 i->keylen=0; /* ok, it's written, forget it */
2061 i->prevsys=key.sysno;
2062 i->prevseq=key.seqno;
2067 void encode_key_flush (struct encode_info *i, FILE *outf)
2068 { /* flush the last key from i */
2069 char *bp =i->buf + i->keylen;
2072 return; /* nothing to flush */
2075 bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
2076 if (i->sysno != i->prevsys)
2078 i->sysno = i->prevsys;
2081 else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
2083 return; /* ??? Filters some sort of duplicates away */
2084 /* ??? Can this ever happen -H 15oct02 */
2086 bp = encode_key_int (i->prevseq - i->seqno, bp);
2087 i->seqno = i->prevseq;
2088 i->cmd = i->prevcmd;
2089 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
2091 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
2094 i->keylen=0; /* ok, it's written, forget it */
2095 i->prevsys=0; /* forget the values too */