1 /* $Id: extract.c,v 1.188 2005-08-05 10:40:13 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 const struct recKeys *reckeys,
103 int attrSetS, int attrUseS,
104 const char **ws, int ws_length)
106 void *decode_handle = iscz1_start();
113 for (i = 0; i<ws_length; i++)
116 ch = zebraExplain_lookup_attr_su_any_index(zh->reg->zei,
121 while (off < reckeys->buf_used)
123 const char *src = reckeys->buf + off;
125 char *dst = (char*) &key;
127 iscz1_decode(decode_handle, &dst, &src);
128 assert(key.len <= 4 && key.len > 2);
130 seqno = (int) key.mem[key.len-1];
132 if (key.mem[0] == ch)
138 woff = seqno - startSeq;
139 if (woff >= 0 && woff < ws_length)
145 off = src - reckeys->buf;
147 iscz1_stop(decode_handle);
148 assert (off == reckeys->buf_used);
151 struct file_read_info {
152 off_t file_max; /* maximum offset so far */
153 off_t file_offset; /* current offset */
154 off_t file_moffset; /* offset of rec/rec boundary */
159 static struct file_read_info *file_read_start (int fd)
161 struct file_read_info *fi = (struct file_read_info *)
162 xmalloc (sizeof(*fi));
166 fi->file_moffset = 0;
172 static void file_read_stop (struct file_read_info *fi)
177 static off_t file_seek (void *handle, off_t offset)
179 struct file_read_info *p = (struct file_read_info *) handle;
180 p->file_offset = offset;
181 return lseek (p->fd, offset, SEEK_SET);
184 static off_t file_tell (void *handle)
186 struct file_read_info *p = (struct file_read_info *) handle;
187 return p->file_offset;
190 static int file_read (void *handle, char *buf, size_t count)
192 struct file_read_info *p = (struct file_read_info *) handle;
195 r = read (fd, buf, count);
199 if (p->file_offset > p->file_max)
200 p->file_max = p->file_offset;
205 static void file_end (void *handle, off_t offset)
207 struct file_read_info *p = (struct file_read_info *) handle;
209 if (offset != p->file_moffset)
211 p->file_moffset = offset;
216 static char *fileMatchStr (ZebraHandle zh,
217 struct recKeys *reckeys,
218 const char *fname, const char *spec)
220 static char dstBuf[2048]; /* static here ??? */
222 const char *s = spec;
226 while (*s == ' ' || *s == '\t')
233 char attset_str[64], attname_str[64];
234 data1_attset *attset;
237 int attSet = 1, attUse = 1;
241 for (i = 0; *s && *s != ',' && *s != ')'; s++)
243 attset_str[i++] = *s;
244 attset_str[i] = '\0';
249 for (i = 0; *s && *s != ')'; s++)
251 attname_str[i++] = *s;
252 attname_str[i] = '\0';
255 if ((attset = data1_get_attset (zh->reg->dh, attset_str)))
258 attSet = attset->reference;
259 att = data1_getattbyname(zh->reg->dh, attset, attname_str);
263 attUse = atoi (attname_str);
265 searchRecordKey (zh, reckeys, attSet, attUse, ws, 32);
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] && ws[i])
289 dst += strlen(ws[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 create_rec_keys_codec(&zh->reg->sortKeys);
424 zh->reg->sortKeys.buf_used = 0;
426 recordOffset = fi->file_moffset;
427 extractCtrl.handle = zh;
428 extractCtrl.offset = fi->file_moffset;
429 extractCtrl.readf = file_read;
430 extractCtrl.seekf = file_seek;
431 extractCtrl.tellf = file_tell;
432 extractCtrl.endf = file_end;
434 extractCtrl.init = extract_init;
435 extractCtrl.tokenAdd = extract_token_add;
436 extractCtrl.schemaAdd = extract_schema_add;
437 extractCtrl.dh = zh->reg->dh;
438 extractCtrl.match_criteria[0] = '\0';
439 extractCtrl.first_record = fi->file_offset ? 0 : 1;
441 extract_set_store_data_prepare(&extractCtrl);
443 init_extractCtrl(zh, &extractCtrl);
446 printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset);
450 sprintf (msg, "%s:" PRINTF_OFF_T , fname, recordOffset);
451 yaz_log_init_prefix2 (msg);
454 r = (*recType->extract)(recTypeClientData, &extractCtrl);
456 yaz_log_init_prefix2 (0);
457 if (r == RECCTRL_EXTRACT_EOF)
459 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
461 /* error occured during extraction ... */
463 zh->records_processed < zh->m_file_verbose_limit)
465 yaz_log (YLOG_WARN, "fail %s %s " PRINTF_OFF_T, zh->m_record_type,
466 fname, recordOffset);
470 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
472 /* error occured during extraction ... */
474 zh->records_processed < zh->m_file_verbose_limit)
476 yaz_log (YLOG_WARN, "no filter for %s %s "
477 PRINTF_OFF_T, zh->m_record_type,
478 fname, recordOffset);
482 if (extractCtrl.match_criteria[0])
483 matchStr = extractCtrl.match_criteria;
486 /* perform match if sysno not known and if match criteria is specified */
492 if (matchStr == 0 && zh->m_record_id && *zh->m_record_id)
495 matchStr = fileMatchStr (zh, &zh->reg->keys, fname,
499 yaz_log(YLOG_WARN, "Bad match criteria");
505 char *rinfo = dict_lookup (zh->reg->matchDict, matchStr);
508 assert(*rinfo == sizeof(*sysno));
509 memcpy (sysno, rinfo+1, sizeof(*sysno));
513 if (! *sysno && zh->reg->keys.buf_used == 0)
515 /* the extraction process returned no information - the record
516 is probably empty - unless flagShowRecords is in use */
520 if (zh->records_processed < zh->m_file_verbose_limit)
521 yaz_log (YLOG_WARN, "empty %s %s " PRINTF_OFF_T, zh->m_record_type,
522 fname, recordOffset);
531 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T, zh->m_record_type,
532 fname, recordOffset);
533 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
536 if (zh->records_processed < zh->m_file_verbose_limit)
537 yaz_log (YLOG_LOG, "add %s %s " PRINTF_OFF_T, zh->m_record_type,
538 fname, recordOffset);
539 rec = rec_new (zh->reg->records);
543 recordAttr = rec_init_attr (zh->reg->zei, rec);
547 dict_insert (zh->reg->matchDict, matchStr, sizeof(*sysno), sysno);
549 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
550 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
552 zh->records_inserted++;
556 /* record already exists */
557 struct recKeys delkeys;
559 struct recKeys sortKeys;
561 struct sortKeys sortKeys;
564 rec = rec_get (zh->reg->records, *sysno);
567 recordAttr = rec_init_attr (zh->reg->zei, rec);
569 if (!force_update && recordAttr->runNumber ==
570 zebraExplain_runNumberIncrement (zh->reg->zei, 0))
572 yaz_log (YLOG_LOG, "run number = " ZINT_FORMAT,
573 recordAttr->runNumber);
574 yaz_log (YLOG_LOG, "skipped %s %s " PRINTF_OFF_T,
575 zh->m_record_type, fname, recordOffset);
576 extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
581 delkeys.buf_used = rec->size[recInfo_delKeys];
582 delkeys.buf = rec->info[recInfo_delKeys];
584 sortKeys.buf_used = rec->size[recInfo_sortKeys];
585 sortKeys.buf = rec->info[recInfo_sortKeys];
587 extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
588 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
591 /* record going to be deleted */
592 if (!delkeys.buf_used)
594 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
595 zh->m_record_type, fname, recordOffset);
596 yaz_log (YLOG_WARN, "cannot delete file above, storeKeys false");
600 if (zh->records_processed < zh->m_file_verbose_limit)
601 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
602 zh->m_record_type, fname, recordOffset);
603 zh->records_deleted++;
605 dict_delete (zh->reg->matchDict, matchStr);
606 rec_del (zh->reg->records, &rec);
614 /* record going to be updated */
615 if (!delkeys.buf_used)
617 yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
618 zh->m_record_type, fname, recordOffset);
619 yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
623 if (zh->records_processed < zh->m_file_verbose_limit)
624 yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
625 zh->m_record_type, fname, recordOffset);
626 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
627 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
628 zh->records_updated++;
632 /* update file type */
633 xfree (rec->info[recInfo_fileType]);
634 rec->info[recInfo_fileType] =
635 rec_strdup (zh->m_record_type, &rec->size[recInfo_fileType]);
637 /* update filename */
638 xfree (rec->info[recInfo_filename]);
639 rec->info[recInfo_filename] =
640 rec_strdup (fname, &rec->size[recInfo_filename]);
642 /* update delete keys */
643 xfree (rec->info[recInfo_delKeys]);
644 if (zh->reg->keys.buf_used > 0 && zh->m_store_keys == 1)
646 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
647 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
648 zh->reg->keys.buf = NULL;
649 zh->reg->keys.buf_max = 0;
653 rec->info[recInfo_delKeys] = NULL;
654 rec->size[recInfo_delKeys] = 0;
657 /* update sort keys */
658 xfree (rec->info[recInfo_sortKeys]);
660 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
661 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
662 zh->reg->sortKeys.buf = NULL;
663 zh->reg->sortKeys.buf_max = 0;
665 /* save file size of original record */
666 zebraExplain_recordBytesIncrement (zh->reg->zei,
667 - recordAttr->recordSize);
668 recordAttr->recordSize = fi->file_moffset - recordOffset;
669 if (!recordAttr->recordSize)
670 recordAttr->recordSize = fi->file_max - recordOffset;
671 zebraExplain_recordBytesIncrement (zh->reg->zei,
672 recordAttr->recordSize);
674 /* set run-number for this record */
675 recordAttr->runNumber = zebraExplain_runNumberIncrement (zh->reg->zei,
678 /* update store data */
679 xfree (rec->info[recInfo_storeData]);
680 if (zh->store_data_buf)
682 rec->size[recInfo_storeData] = zh->store_data_size;
683 rec->info[recInfo_storeData] = zh->store_data_buf;
684 zh->store_data_buf = 0;
686 else if (zh->m_store_data)
688 rec->size[recInfo_storeData] = recordAttr->recordSize;
689 rec->info[recInfo_storeData] = (char *)
690 xmalloc (recordAttr->recordSize);
691 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
693 yaz_log (YLOG_ERRNO|YLOG_FATAL, "seek to " PRINTF_OFF_T " in %s",
694 recordOffset, fname);
697 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
698 < recordAttr->recordSize)
700 yaz_log (YLOG_ERRNO|YLOG_FATAL, "read %d bytes of %s",
701 recordAttr->recordSize, fname);
707 rec->info[recInfo_storeData] = NULL;
708 rec->size[recInfo_storeData] = 0;
710 /* update database name */
711 xfree (rec->info[recInfo_databaseName]);
712 rec->info[recInfo_databaseName] =
713 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
716 recordAttr->recordOffset = recordOffset;
718 /* commit this record */
719 rec_put (zh->reg->records, &rec);
724 int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
731 struct file_read_info *fi;
732 const char *original_record_type = 0;
734 void *recTypeClientData;
736 if (!zh->m_group || !*zh->m_group)
739 sprintf (gprefix, "%s.", zh->m_group);
741 yaz_log (YLOG_DEBUG, "fileExtract %s", fname);
743 /* determine file extension */
745 for (i = strlen(fname); --i >= 0; )
748 else if (fname[i] == '.')
750 strcpy (ext, fname+i+1);
753 /* determine file type - depending on extension */
754 original_record_type = zh->m_record_type;
755 if (!zh->m_record_type)
757 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
758 zh->m_record_type = res_get (zh->res, ext_res);
760 if (!zh->m_record_type)
762 if (zh->records_processed < zh->m_file_verbose_limit)
763 yaz_log (YLOG_LOG, "? %s", fname);
766 /* determine match criteria */
767 if (!zh->m_record_id)
769 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
770 zh->m_record_id = res_get (zh->res, ext_res);
774 recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
775 &recTypeClientData)))
777 yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
781 switch(recType->version)
786 yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
788 if (sysno && deleteFlag)
794 if (zh->path_reg && !yaz_is_abspath (fname))
796 strcpy (full_rep, zh->path_reg);
797 strcat (full_rep, "/");
798 strcat (full_rep, fname);
801 strcpy (full_rep, fname);
804 if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
806 yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
807 zh->m_record_type = original_record_type;
811 fi = file_read_start (fd);
814 fi->file_moffset = fi->file_offset;
815 fi->file_more = 0; /* file_end not called (yet) */
816 r = file_extract_record (zh, sysno, fname, deleteFlag, fi, 1,
817 recType, recTypeClientData);
819 { /* file_end has been called so reset offset .. */
820 fi->file_offset = fi->file_moffset;
821 lseek(fi->fd, fi->file_moffset, SEEK_SET);
828 zh->m_record_type = original_record_type;
833 If sysno is provided, then it's used to identify the reocord.
834 If not, and match_criteria is provided, then sysno is guessed
835 If not, and a record is provided, then sysno is got from there
838 ZEBRA_RES buffer_extract_record (ZebraHandle zh,
839 const char *buf, size_t buf_size,
842 const char *recordType,
844 const char *match_criteria,
849 RecordAttr *recordAttr;
850 struct recExtractCtrl extractCtrl;
852 const char *matchStr = 0;
853 RecType recType = NULL;
856 long recordOffset = 0;
857 struct zebra_fetch_control fc;
858 const char *pr_fname = fname; /* filename to print .. */
859 int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
862 pr_fname = "<no file>"; /* make it printable if file is omitted */
865 fc.record_int_buf = buf;
866 fc.record_int_len = buf_size;
867 fc.record_int_pos = 0;
869 fc.record_offset = 0;
871 extractCtrl.offset = 0;
872 extractCtrl.readf = zebra_record_int_read;
873 extractCtrl.seekf = zebra_record_int_seek;
874 extractCtrl.tellf = zebra_record_int_tell;
875 extractCtrl.endf = zebra_record_int_end;
876 extractCtrl.first_record = 1;
877 extractCtrl.fh = &fc;
879 create_rec_keys_codec(&zh->reg->keys);
881 create_rec_keys_codec(&zh->reg->sortKeys);
883 zh->reg->sortKeys.buf_used = 0;
885 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
887 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
888 zh->m_explain_database))
892 if (recordType && *recordType)
894 yaz_log (YLOG_DEBUG, "Record type explicitly specified: %s", recordType);
895 recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
900 if (!(zh->m_record_type))
902 yaz_log (YLOG_WARN, "No such record type defined");
905 yaz_log (YLOG_DEBUG, "Get record type from rgroup: %s",zh->m_record_type);
906 recType = recType_byName (zh->reg->recTypes, zh->res,
907 zh->m_record_type, &clientData);
908 recordType = zh->m_record_type;
913 yaz_log (YLOG_WARN, "No such record type: %s", zh->m_record_type);
917 extractCtrl.init = extract_init;
918 extractCtrl.tokenAdd = extract_token_add;
919 extractCtrl.schemaAdd = extract_schema_add;
920 extractCtrl.dh = zh->reg->dh;
921 extractCtrl.handle = zh;
922 extractCtrl.match_criteria[0] = '\0';
924 init_extractCtrl(zh, &extractCtrl);
926 extract_set_store_data_prepare(&extractCtrl);
928 r = (*recType->extract)(clientData, &extractCtrl);
930 if (r == RECCTRL_EXTRACT_EOF)
932 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
934 /* error occured during extraction ... */
935 yaz_log (YLOG_WARN, "extract error: generic");
938 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
940 /* error occured during extraction ... */
941 yaz_log (YLOG_WARN, "extract error: no such filter");
947 if (extractCtrl.match_criteria[0])
948 match_criteria = extractCtrl.match_criteria;
952 if (match_criteria && *match_criteria) {
953 matchStr = match_criteria;
955 if (zh->m_record_id && *zh->m_record_id) {
956 matchStr = fileMatchStr (zh, &zh->reg->keys, pr_fname,
960 yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
966 rinfo = dict_lookup (zh->reg->matchDict, matchStr);
969 assert(*rinfo == sizeof(*sysno));
970 memcpy (sysno, rinfo+1, sizeof(*sysno));
974 if (zh->reg->keys.buf_used == 0)
976 /* the extraction process returned no information - the record
977 is probably empty - unless flagShowRecords is in use */
988 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
989 pr_fname, (long) recordOffset);
990 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
994 yaz_log (YLOG_LOG, "add %s %s %ld", recordType, pr_fname,
995 (long) recordOffset);
996 rec = rec_new (zh->reg->records);
1000 recordAttr = rec_init_attr (zh->reg->zei, rec);
1004 dict_insert (zh->reg->matchDict, matchStr,
1005 sizeof(*sysno), sysno);
1007 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1008 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
1010 zh->records_inserted++;
1014 /* record already exists */
1015 struct recKeys delkeys;
1017 struct recKeys sortKeys;
1019 struct sortKeys sortKeys;
1025 yaz_log (YLOG_LOG, "skipped %s %s %ld",
1026 recordType, pr_fname, (long) recordOffset);
1031 rec = rec_get (zh->reg->records, *sysno);
1034 recordAttr = rec_init_attr (zh->reg->zei, rec);
1036 if (!force_update) {
1037 if (recordAttr->runNumber ==
1038 zebraExplain_runNumberIncrement (zh->reg->zei, 0))
1041 yaz_log (YLOG_LOG, "skipped %s %s %ld", recordType,
1042 pr_fname, (long) recordOffset);
1043 extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
1050 delkeys.buf_used = rec->size[recInfo_delKeys];
1051 delkeys.buf = rec->info[recInfo_delKeys];
1053 sortKeys.buf_used = rec->size[recInfo_sortKeys];
1054 sortKeys.buf = rec->info[recInfo_sortKeys];
1056 extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
1057 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1060 /* record going to be deleted */
1061 if (!delkeys.buf_used)
1065 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1066 pr_fname, (long) recordOffset);
1067 yaz_log (YLOG_WARN, "cannot delete file above, "
1074 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1075 pr_fname, (long) recordOffset);
1076 zh->records_deleted++;
1078 dict_delete (zh->reg->matchDict, matchStr);
1079 rec_del (zh->reg->records, &rec);
1087 /* record going to be updated */
1088 if (!delkeys.buf_used)
1092 yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
1093 pr_fname, (long) recordOffset);
1094 yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
1100 yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
1101 pr_fname, (long) recordOffset);
1102 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1103 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
1104 zh->records_updated++;
1108 /* update file type */
1109 xfree (rec->info[recInfo_fileType]);
1110 rec->info[recInfo_fileType] =
1111 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1113 /* update filename */
1114 xfree (rec->info[recInfo_filename]);
1115 rec->info[recInfo_filename] =
1116 rec_strdup (fname, &rec->size[recInfo_filename]);
1118 /* update delete keys */
1119 xfree (rec->info[recInfo_delKeys]);
1120 if (zh->reg->keys.buf_used > 0 && zh->m_store_keys == 1)
1122 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
1123 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
1124 zh->reg->keys.buf = NULL;
1125 zh->reg->keys.buf_max = 0;
1129 rec->info[recInfo_delKeys] = NULL;
1130 rec->size[recInfo_delKeys] = 0;
1133 /* update sort keys */
1134 xfree (rec->info[recInfo_sortKeys]);
1136 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
1137 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
1138 zh->reg->sortKeys.buf = NULL;
1139 zh->reg->sortKeys.buf_max = 0;
1141 /* save file size of original record */
1142 zebraExplain_recordBytesIncrement (zh->reg->zei,
1143 - recordAttr->recordSize);
1145 recordAttr->recordSize = fi->file_moffset - recordOffset;
1146 if (!recordAttr->recordSize)
1147 recordAttr->recordSize = fi->file_max - recordOffset;
1149 recordAttr->recordSize = buf_size;
1151 zebraExplain_recordBytesIncrement (zh->reg->zei,
1152 recordAttr->recordSize);
1154 /* set run-number for this record */
1155 recordAttr->runNumber =
1156 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
1158 /* update store data */
1159 xfree (rec->info[recInfo_storeData]);
1161 /* update store data */
1162 if (zh->store_data_buf)
1164 rec->size[recInfo_storeData] = zh->store_data_size;
1165 rec->info[recInfo_storeData] = zh->store_data_buf;
1166 zh->store_data_buf = 0;
1168 else if (zh->m_store_data)
1170 rec->size[recInfo_storeData] = recordAttr->recordSize;
1171 rec->info[recInfo_storeData] = (char *)
1172 xmalloc (recordAttr->recordSize);
1173 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1177 rec->info[recInfo_storeData] = NULL;
1178 rec->size[recInfo_storeData] = 0;
1180 /* update database name */
1181 xfree (rec->info[recInfo_databaseName]);
1182 rec->info[recInfo_databaseName] =
1183 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
1186 recordAttr->recordOffset = recordOffset;
1188 /* commit this record */
1189 rec_put (zh->reg->records, &rec);
1194 int explain_extract (void *handle, Record rec, data1_node *n)
1196 ZebraHandle zh = (ZebraHandle) handle;
1197 struct recExtractCtrl extractCtrl;
1199 if (zebraExplain_curDatabase (zh->reg->zei,
1200 rec->info[recInfo_databaseName]))
1203 if (zebraExplain_newDatabase (zh->reg->zei,
1204 rec->info[recInfo_databaseName], 0))
1208 create_rec_keys_codec(&zh->reg->keys);
1210 create_rec_keys_codec(&zh->reg->sortKeys);
1212 zh->reg->sortKeys.buf_used = 0;
1214 extractCtrl.init = extract_init;
1215 extractCtrl.tokenAdd = extract_token_add;
1216 extractCtrl.schemaAdd = extract_schema_add;
1217 extractCtrl.dh = zh->reg->dh;
1219 init_extractCtrl(zh, &extractCtrl);
1221 extractCtrl.flagShowRecords = 0;
1222 extractCtrl.match_criteria[0] = '\0';
1223 extractCtrl.handle = handle;
1224 extractCtrl.first_record = 1;
1226 extract_set_store_data_prepare(&extractCtrl);
1229 grs_extract_tree(&extractCtrl, n);
1231 if (rec->size[recInfo_delKeys])
1233 struct recKeys delkeys;
1235 struct recKeys sortkeys;
1237 struct sortKeys sortkeys;
1240 delkeys.buf_used = rec->size[recInfo_delKeys];
1241 delkeys.buf = rec->info[recInfo_delKeys];
1243 sortkeys.buf_used = rec->size[recInfo_sortKeys];
1244 sortkeys.buf = rec->info[recInfo_sortKeys];
1246 extract_flushSortKeys (zh, rec->sysno, 0, &sortkeys);
1247 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1249 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->reg->keys);
1250 extract_flushSortKeys (zh, rec->sysno, 1, &zh->reg->sortKeys);
1252 xfree (rec->info[recInfo_delKeys]);
1253 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
1254 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
1255 zh->reg->keys.buf = NULL;
1256 zh->reg->keys.buf_max = 0;
1258 xfree (rec->info[recInfo_sortKeys]);
1259 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
1260 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
1261 zh->reg->sortKeys.buf = NULL;
1262 zh->reg->sortKeys.buf_max = 0;
1267 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1268 int cmd, struct recKeys *reckeys)
1270 void *decode_handle = iscz1_start();
1273 ZebraExplainInfo zei = zh->reg->zei;
1275 if (!zh->reg->key_buf)
1277 int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
1280 yaz_log(YLOG_WARN, "Invalid memory setting, using default 8 MB");
1283 /* FIXME: That "8" should be in a default settings include */
1284 /* not hard-coded here! -H */
1285 zh->reg->key_buf = (char**) xmalloc (mem);
1286 zh->reg->ptr_top = mem/sizeof(char*);
1288 zh->reg->key_buf_used = 0;
1289 zh->reg->key_file_no = 0;
1291 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1293 while (off < reckeys->buf_used)
1295 const char *src = reckeys->buf + off;
1297 char *dst = (char*) &key;
1299 iscz1_decode(decode_handle, &dst, &src);
1300 assert(key.len == 4);
1302 if (zh->reg->key_buf_used + 1024 >
1303 (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
1304 extract_flushWriteKeys (zh,0);
1306 assert(zh->reg->ptr_i > 0);
1307 (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
1308 (char*)zh->reg->key_buf + zh->reg->key_buf_used;
1310 ch = (int) key.mem[0]; /* ordinal for field/use/attribute */
1312 zh->reg->key_buf_used +=
1313 key_SU_encode (ch,((char*)zh->reg->key_buf) +
1314 zh->reg->key_buf_used);
1316 ((char*)zh->reg->key_buf) [(zh->reg->key_buf_used)++] = *src++;
1318 ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = '\0';
1319 ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = cmd;
1322 if (key.mem[1]) /* filter specified record ID */
1323 key.mem[0] = key.mem[1];
1326 key.mem[1] = key.mem[2]; /* section_id */
1327 key.mem[2] = key.mem[3]; /* sequence .. */
1329 memcpy ((char*)zh->reg->key_buf + zh->reg->key_buf_used,
1331 (zh->reg->key_buf_used) += sizeof(key);
1332 off = src - reckeys->buf;
1334 assert (off == reckeys->buf_used);
1335 iscz1_stop(decode_handle);
1338 void extract_flushWriteKeys (ZebraHandle zh, int final)
1339 /* optimizing: if final=1, and no files written yet */
1340 /* push the keys directly to merge, sidestepping the */
1341 /* temp file altogether. Speeds small updates */
1344 char out_fname[200];
1346 struct encode_info encode_info;
1347 int ptr_i = zh->reg->ptr_i;
1352 if (!zh->reg->key_buf || ptr_i <= 0)
1354 yaz_log (YLOG_DEBUG, " nothing to flush section=%d buf=%p i=%d",
1355 zh->reg->key_file_no, zh->reg->key_buf, ptr_i);
1356 yaz_log (YLOG_DEBUG, " buf=%p ",
1358 yaz_log (YLOG_DEBUG, " ptr=%d ",zh->reg->ptr_i);
1359 yaz_log (YLOG_DEBUG, " reg=%p ",zh->reg);
1364 (zh->reg->key_file_no)++;
1365 yaz_log (YLOG_LOG, "sorting section %d", (zh->reg->key_file_no));
1366 yaz_log (YLOG_DEBUG, " sort_buff at %p n=%d",
1367 zh->reg->key_buf + zh->reg->ptr_top - ptr_i,ptr_i);
1369 qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
1370 sizeof(char*), key_qsort_compare);
1372 /* zebra.cfg: tempfiles:
1373 Y: always use temp files (old way)
1374 A: use temp files, if more than one (auto)
1375 = if this is both the last and the first
1376 N: never bother with temp files (new) */
1378 temp_policy=toupper(res_get_def(zh->res,"tempfiles","auto")[0]);
1379 if (temp_policy != 'Y' && temp_policy != 'N' && temp_policy != 'A') {
1380 yaz_log (YLOG_WARN, "Illegal tempfiles setting '%c'. using 'Auto' ",
1385 if ( ( temp_policy =='N' ) || /* always from memory */
1386 ( ( temp_policy =='A' ) && /* automatic */
1387 (zh->reg->key_file_no == 1) && /* this is first time */
1388 (final) ) ) /* and last (=only) time */
1389 { /* go directly from memory */
1390 zh->reg->key_file_no =0; /* signal not to read files */
1391 zebra_index_merge(zh);
1393 zh->reg->key_buf_used = 0;
1397 /* Not doing directly from memory, write into a temp file */
1398 extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
1400 if (!(outf = fopen (out_fname, "wb")))
1402 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1405 yaz_log (YLOG_LOG, "writing section %d", zh->reg->key_file_no);
1406 prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1408 encode_key_init (&encode_info);
1409 encode_key_write (cp, &encode_info, outf);
1413 cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1414 if (strcmp (cp, prevcp))
1416 encode_key_flush ( &encode_info, outf);
1417 encode_key_init (&encode_info);
1418 encode_key_write (cp, &encode_info, outf);
1422 encode_key_write (cp + strlen(cp), &encode_info, outf);
1424 encode_key_flush ( &encode_info, outf);
1426 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1427 extract_get_fname_tmp (out_fname, key_file_no);
1429 if (!(outf = fopen (out_fname, "wb")))
1431 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1434 yaz_log (YLOG_LOG, "writing section %d", key_file_no);
1436 prevcp = key_buf[ptr_top-i];
1438 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1440 key_y_len = strlen(prevcp)+1;
1442 yaz_log (YLOG_LOG, "key_y_len: %2d %02x %02x %s",
1443 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1445 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1446 sizeof(char*), key_y_compare);
1447 cp = key_buf[ptr_top-ptr_i];
1449 encode_key_init (&encode_info);
1450 encode_key_write (cp, &encode_info, outf);
1453 cp = key_buf[ptr_top-ptr_i];
1454 encode_key_write (cp+key_y_len, &encode_info, outf);
1456 encode_key_flush ( &encode_info, outf);
1459 prevcp = key_buf[ptr_top-ptr_i];
1464 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fclose %s", out_fname);
1467 yaz_log (YLOG_LOG, "finished section %d", zh->reg->key_file_no);
1469 zh->reg->key_buf_used = 0;
1472 void extract_add_it_key (ZebraHandle zh,
1473 struct recKeys *keys,
1475 const char *str, int slen, struct it_key *key)
1478 const char *src = (char*) key;
1480 if (keys->buf_used+1024 > keys->buf_max)
1482 char *b = (char *) xmalloc (keys->buf_max += 128000);
1483 if (keys->buf_used > 0)
1484 memcpy (b, keys->buf, keys->buf_used);
1488 dst = keys->buf + keys->buf_used;
1490 iscz1_encode(keys->codec_handle, &dst, &src);
1495 memcpy (dst, str, slen);
1498 keys->buf_used = dst - keys->buf;
1501 ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh, struct recKeys *reckeys,
1502 zebra_snippets *snippets)
1504 void *decode_handle = iscz1_start();
1507 NMEM nmem = nmem_create();
1509 yaz_log(YLOG_LOG, "zebra_rec_keys_snippets buf=%p sz=%d", reckeys->buf,
1511 assert(reckeys->buf);
1512 while (off < reckeys->buf_used)
1514 const char *src = reckeys->buf + off;
1516 char *dst = (char*) &key;
1517 char dst_buf[IT_MAX_WORD];
1518 char *dst_term = dst_buf;
1520 iscz1_decode(decode_handle, &dst, &src);
1521 assert(key.len <= 4 && key.len > 2);
1523 seqno = (int) key.mem[key.len-1];
1525 zebra_term_untrans_iconv(zh, nmem, src[0], &dst_term, src+1);
1526 zebra_snippets_append(snippets, seqno, src[0], key.mem[0], dst_term);
1529 off = src - reckeys->buf;
1533 iscz1_stop(decode_handle);
1537 void print_rec_keys(ZebraHandle zh, struct recKeys *reckeys)
1539 void *decode_handle = iscz1_start();
1542 NMEM nmem = nmem_create();
1544 yaz_log(YLOG_LOG, "print_rec_keys buf=%p sz=%d", reckeys->buf,
1546 assert(reckeys->buf);
1547 while (off < reckeys->buf_used)
1549 const char *src = reckeys->buf + off;
1551 char *dst = (char*) &key;
1552 int attrSet, attrUse;
1553 char dst_buf[IT_MAX_WORD];
1554 char *dst_term = dst_buf;
1556 iscz1_decode(decode_handle, &dst, &src);
1557 assert(key.len <= 4 && key.len > 2);
1559 attrSet = (int) key.mem[0] >> 16;
1560 attrUse = (int) key.mem[0] & 65535;
1561 seqno = (int) key.mem[key.len-1];
1563 zebra_term_untrans_iconv(zh, nmem, src[0], &dst_term, src+1);
1565 yaz_log(YLOG_LOG, "ord=" ZINT_FORMAT " seqno=%d term=%s",
1566 key.mem[0], seqno, dst_term);
1569 off = src - reckeys->buf;
1573 iscz1_stop(decode_handle);
1576 void extract_add_index_string (RecWord *p, const char *str, int length)
1580 ZebraHandle zh = p->extractCtrl->handle;
1581 ZebraExplainInfo zei = zh->reg->zei;
1586 ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
1588 ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
1595 ch = zebraExplain_lookup_attr_su(zei, p->index_type,
1596 p->attrSet, p->attrUse);
1598 ch = zebraExplain_add_attr_su(zei, p->index_type,
1599 p->attrSet, p->attrUse);
1604 key.mem[1] = p->record_id;
1605 key.mem[2] = p->section_id;
1606 key.mem[3] = p->seqno;
1609 /* just for debugging .. */
1610 yaz_log(YLOG_LOG, "add: set=%d use=%d "
1611 "record_id=%lld section_id=%lld seqno=%lld",
1612 p->attrSet, p->attrUse, p->record_id, p->section_id, p->seqno);
1615 extract_add_it_key(p->extractCtrl->handle,
1622 static void extract_add_sort_string (RecWord *p, const char *str, int length)
1626 ZebraHandle zh = p->extractCtrl->handle;
1627 ZebraExplainInfo zei = zh->reg->zei;
1632 ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
1634 ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
1642 key.mem[1] = p->record_id;
1643 key.mem[2] = p->section_id;
1644 key.mem[3] = p->seqno;
1646 extract_add_it_key(p->extractCtrl->handle,
1652 static void extract_add_sort_string (RecWord *p, const char *str, int length)
1654 ZebraHandle zh = p->extractCtrl->handle;
1655 struct sortKeys *sk = &zh->reg->sortKeys;
1658 while (off < sk->buf_used)
1662 off += key_SU_decode(&set, sk->buf + off);
1663 off += key_SU_decode(&use, sk->buf + off);
1664 off += key_SU_decode(&slen, sk->buf + off);
1666 if (p->attrSet == set && p->attrUse == use)
1669 assert (off == sk->buf_used);
1671 if (sk->buf_used + IT_MAX_WORD > sk->buf_max)
1675 b = (char *) xmalloc (sk->buf_max += 128000);
1676 if (sk->buf_used > 0)
1677 memcpy (b, sk->buf, sk->buf_used);
1681 off += key_SU_encode(p->attrSet, sk->buf + off);
1682 off += key_SU_encode(p->attrUse, sk->buf + off);
1683 off += key_SU_encode(length, sk->buf + off);
1684 memcpy (sk->buf + off, str, length);
1685 sk->buf_used = off + length;
1689 void extract_add_string (RecWord *p, const char *string, int length)
1691 assert (length > 0);
1692 if (zebra_maps_is_sort (p->zebra_maps, p->index_type))
1693 extract_add_sort_string (p, string, length);
1695 extract_add_index_string (p, string, length);
1698 static void extract_add_incomplete_field (RecWord *p)
1700 const char *b = p->term_buf;
1701 int remain = p->term_len;
1702 const char **map = 0;
1704 yaz_log(YLOG_DEBUG, "Incomplete field, w='%.*s'", p->term_len, p->term_buf);
1707 map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, 0);
1711 char buf[IT_MAX_WORD+1];
1715 while (map && *map && **map == *CHR_SPACE)
1717 remain = p->term_len - (b - p->term_buf);
1719 map = zebra_maps_input(p->zebra_maps, p->index_type, &b,
1727 while (map && *map && **map != *CHR_SPACE)
1729 const char *cp = *map;
1731 while (i < IT_MAX_WORD && *cp)
1733 remain = p->term_len - (b - p->term_buf);
1735 map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, 0);
1741 extract_add_string (p, buf, i);
1746 static void extract_add_complete_field (RecWord *p)
1748 const char *b = p->term_buf;
1749 char buf[IT_MAX_WORD+1];
1750 const char **map = 0;
1751 int i = 0, remain = p->term_len;
1753 yaz_log(YLOG_DEBUG, "Complete field, w='%.*s'",
1754 p->term_len, p->term_buf);
1757 map = zebra_maps_input (p->zebra_maps, p->index_type, &b, remain, 1);
1759 while (remain > 0 && i < IT_MAX_WORD)
1761 while (map && *map && **map == *CHR_SPACE)
1763 remain = p->term_len - (b - p->term_buf);
1767 int first = i ? 0 : 1; /* first position */
1768 map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, first);
1776 if (i && i < IT_MAX_WORD)
1777 buf[i++] = *CHR_SPACE;
1778 while (map && *map && **map != *CHR_SPACE)
1780 const char *cp = *map;
1782 if (**map == *CHR_CUT)
1788 if (i >= IT_MAX_WORD)
1790 yaz_log(YLOG_DEBUG, "Adding string to index '%d'", **map);
1791 while (i < IT_MAX_WORD && *cp)
1794 remain = p->term_len - (b - p->term_buf);
1797 map = zebra_maps_input (p->zebra_maps, p->index_type, &b,
1806 extract_add_string (p, buf, i);
1809 void extract_token_add (RecWord *p)
1813 yaz_log (YLOG_LOG, "token_add "
1814 "reg_type=%c attrSet=%d attrUse=%d seqno=%d s=%.*s",
1815 p->reg_type, p->attrSet, p->attrUse, p->seqno, p->length,
1818 if ((wrbuf = zebra_replace(p->zebra_maps, p->index_type, 0,
1819 p->term_buf, p->term_len)))
1821 p->term_buf = wrbuf_buf(wrbuf);
1822 p->term_len = wrbuf_len(wrbuf);
1824 if (zebra_maps_is_complete (p->zebra_maps, p->index_type))
1825 extract_add_complete_field (p);
1827 extract_add_incomplete_field(p);
1830 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1831 void *buf, size_t sz)
1833 ZebraHandle zh = (ZebraHandle) p->handle;
1835 xfree(zh->store_data_buf);
1836 zh->store_data_buf = 0;
1837 zh->store_data_size = 0;
1840 zh->store_data_buf = xmalloc(sz);
1841 zh->store_data_size = sz;
1842 memcpy(zh->store_data_buf, buf, sz);
1846 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1848 ZebraHandle zh = (ZebraHandle) p->handle;
1849 xfree(zh->store_data_buf);
1850 zh->store_data_buf = 0;
1851 zh->store_data_size = 0;
1852 p->setStoreData = extract_set_store_data_cb;
1855 void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1857 ZebraHandle zh = (ZebraHandle) p->handle;
1858 zebraExplain_addSchema (zh->reg->zei, oid);
1862 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1863 int cmd, struct recKeys *reckeys)
1865 SortIdx sortIdx = zh->reg->sortIdx;
1866 void *decode_handle = iscz1_start();
1870 while (off < reckeys->buf_used)
1872 const char *src = reckeys->buf + off;
1874 char *dst = (char*) &key;
1876 iscz1_decode(decode_handle, &dst, &src);
1877 assert(key.len == 4);
1879 ch = (int) key.mem[0]; /* ordinal for field/use/attribute */
1881 sortIdx_type(sortIdx, ch);
1883 sortIdx_add(sortIdx, src, strlen(src));
1885 sortIdx_add(sortIdx, "", 1);
1890 off = src - reckeys->buf;
1892 assert (off == reckeys->buf_used);
1893 iscz1_stop(decode_handle);
1896 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1897 int cmd, struct sortKeys *sk)
1899 SortIdx sortIdx = zh->reg->sortIdx;
1902 sortIdx_sysno (sortIdx, sysno);
1904 while (off < sk->buf_used)
1908 off += key_SU_decode(&set, sk->buf + off);
1909 off += key_SU_decode(&use, sk->buf + off);
1910 off += key_SU_decode(&slen, sk->buf + off);
1912 sortIdx_type(sortIdx, use);
1914 sortIdx_add(sortIdx, sk->buf + off, slen);
1916 sortIdx_add(sortIdx, "", 1);
1922 void encode_key_init (struct encode_info *i)
1931 i->encode_handle = iscz1_start();
1937 /* this is the old encode_key_write
1938 * may be deleted once we are confident that the new works
1941 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1944 char *bp = i->buf, *bp0;
1945 const char *src = (char *) &key;
1947 /* copy term to output buf */
1948 while ((*bp++ = *k++))
1950 /* and copy & align key so we can mangle */
1951 memcpy (&key, k+1, sizeof(struct it_key)); /* *k is insert/delete */
1954 iscz1_encode(i->encode_handle, &bp, &src);
1955 *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */
1956 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1958 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
1963 void encode_key_flush (struct encode_info *i, FILE *outf)
1964 { /* dummy routine */
1965 iscz1_stop(i->encode_handle);
1970 /* new encode_key_write
1971 * The idea is to buffer one more key, and compare them
1972 * If we are going to delete and insert the same key,
1973 * we may as well not bother. Should make a difference in
1974 * updates with small modifications (appending to a mbox)
1976 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1981 if (*k) /* first time for new key */
1984 while ((*bp++ = *k++))
1986 i->keylen= bp - i->buf -1;
1987 assert(i->keylen+1+sizeof(struct it_key) < ENCODE_BUFLEN);
1991 bp=i->buf + i->keylen;
1996 memcpy (&key, k+1, sizeof(struct it_key));
1997 if (0==i->prevsys) /* no previous filter, fill up */
1999 i->prevsys=key.sysno;
2000 i->prevseq=key.seqno;
2003 else if ( (i->prevsys==key.sysno) &&
2004 (i->prevseq==key.seqno) &&
2006 { /* same numbers, diff cmd, they cancel out */
2010 { /* different stuff, write previous, move buf */
2011 bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
2012 if (i->sysno != i->prevsys)
2014 i->sysno = i->prevsys;
2017 else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
2019 return; /* ??? Filters some sort of duplicates away */
2020 /* ??? Can this ever happen -H 15oct02 */
2022 bp = encode_key_int (i->prevseq - i->seqno, bp);
2023 i->seqno = i->prevseq;
2024 i->cmd = i->prevcmd;
2025 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
2027 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
2030 i->keylen=0; /* ok, it's written, forget it */
2031 i->prevsys=key.sysno;
2032 i->prevseq=key.seqno;
2037 void encode_key_flush (struct encode_info *i, FILE *outf)
2038 { /* flush the last key from i */
2039 char *bp =i->buf + i->keylen;
2042 return; /* nothing to flush */
2045 bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
2046 if (i->sysno != i->prevsys)
2048 i->sysno = i->prevsys;
2051 else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
2053 return; /* ??? Filters some sort of duplicates away */
2054 /* ??? Can this ever happen -H 15oct02 */
2056 bp = encode_key_int (i->prevseq - i->seqno, bp);
2057 i->seqno = i->prevseq;
2058 i->cmd = i->prevcmd;
2059 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
2061 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
2064 i->keylen=0; /* ok, it's written, forget it */
2065 i->prevsys=0; /* forget the values too */