1 /* $Id: extract.c,v 1.194 2005-08-30 12:23:02 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;
406 struct recExtractCtrl extractCtrl;
408 /* announce database */
409 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
411 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
412 zh->m_explain_database))
418 /* we are going to read from a file, so prepare the extraction */
419 create_rec_keys_codec(&zh->reg->keys);
421 create_rec_keys_codec(&zh->reg->sortKeys);
423 zh->reg->sortKeys.buf_used = 0;
425 recordOffset = fi->file_moffset;
426 extractCtrl.handle = zh;
427 extractCtrl.offset = fi->file_moffset;
428 extractCtrl.readf = file_read;
429 extractCtrl.seekf = file_seek;
430 extractCtrl.tellf = file_tell;
431 extractCtrl.endf = file_end;
433 extractCtrl.init = extract_init;
434 extractCtrl.tokenAdd = extract_token_add;
435 extractCtrl.schemaAdd = extract_schema_add;
436 extractCtrl.dh = zh->reg->dh;
437 extractCtrl.match_criteria[0] = '\0';
438 extractCtrl.staticrank = 0;
440 extractCtrl.first_record = fi->file_offset ? 0 : 1;
442 extract_set_store_data_prepare(&extractCtrl);
444 init_extractCtrl(zh, &extractCtrl);
447 printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset);
451 sprintf (msg, "%s:" PRINTF_OFF_T , fname, recordOffset);
452 yaz_log_init_prefix2 (msg);
455 r = (*recType->extract)(recTypeClientData, &extractCtrl);
457 yaz_log_init_prefix2 (0);
458 if (r == RECCTRL_EXTRACT_EOF)
460 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
462 /* error occured during extraction ... */
464 zh->records_processed < zh->m_file_verbose_limit)
466 yaz_log (YLOG_WARN, "fail %s %s " PRINTF_OFF_T, zh->m_record_type,
467 fname, recordOffset);
471 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
473 /* error occured during extraction ... */
475 zh->records_processed < zh->m_file_verbose_limit)
477 yaz_log (YLOG_WARN, "no filter for %s %s "
478 PRINTF_OFF_T, zh->m_record_type,
479 fname, recordOffset);
483 if (extractCtrl.match_criteria[0])
484 matchStr = extractCtrl.match_criteria;
487 /* perform match if sysno not known and if match criteria is specified */
493 if (matchStr == 0 && zh->m_record_id && *zh->m_record_id)
496 matchStr = fileMatchStr (zh, &zh->reg->keys, fname,
500 yaz_log(YLOG_WARN, "Bad match criteria");
506 char *rinfo = dict_lookup (zh->reg->matchDict, matchStr);
509 assert(*rinfo == sizeof(*sysno));
510 memcpy (sysno, rinfo+1, sizeof(*sysno));
514 if (! *sysno && zh->reg->keys.buf_used == 0)
516 /* the extraction process returned no information - the record
517 is probably empty - unless flagShowRecords is in use */
521 if (zh->records_processed < zh->m_file_verbose_limit)
522 yaz_log (YLOG_WARN, "empty %s %s " PRINTF_OFF_T, zh->m_record_type,
523 fname, recordOffset);
532 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T, zh->m_record_type,
533 fname, recordOffset);
534 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
537 if (zh->records_processed < zh->m_file_verbose_limit)
538 yaz_log (YLOG_LOG, "add %s %s " PRINTF_OFF_T, zh->m_record_type,
539 fname, recordOffset);
540 rec = rec_new (zh->reg->records);
544 recordAttr = rec_init_attr (zh->reg->zei, rec);
545 recordAttr->staticrank = extractCtrl.staticrank;
549 dict_insert (zh->reg->matchDict, matchStr, sizeof(*sysno), sysno);
551 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
552 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys,
553 recordAttr->staticrank);
555 zh->records_inserted++;
559 /* record already exists */
560 struct recKeys delkeys;
562 struct recKeys sortKeys;
564 struct sortKeys sortKeys;
567 rec = rec_get (zh->reg->records, *sysno);
570 recordAttr = rec_init_attr (zh->reg->zei, rec);
572 if (!force_update && recordAttr->runNumber ==
573 zebraExplain_runNumberIncrement (zh->reg->zei, 0))
575 yaz_log (YLOG_LOG, "run number = " ZINT_FORMAT,
576 recordAttr->runNumber);
577 yaz_log (YLOG_LOG, "skipped %s %s " PRINTF_OFF_T,
578 zh->m_record_type, fname, recordOffset);
579 extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
584 /* flush old keys for sort&search etc. */
585 delkeys.buf_used = rec->size[recInfo_delKeys];
586 delkeys.buf = rec->info[recInfo_delKeys];
588 sortKeys.buf_used = rec->size[recInfo_sortKeys];
589 sortKeys.buf = rec->info[recInfo_sortKeys];
591 extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
592 extract_flushRecordKeys (zh, *sysno, 0, &delkeys,
593 recordAttr->staticrank); /* old values */
596 /* record going to be deleted */
597 if (!delkeys.buf_used)
599 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
600 zh->m_record_type, fname, recordOffset);
601 yaz_log (YLOG_WARN, "cannot delete file above, storeKeys false");
605 if (zh->records_processed < zh->m_file_verbose_limit)
606 yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
607 zh->m_record_type, fname, recordOffset);
608 zh->records_deleted++;
610 dict_delete (zh->reg->matchDict, matchStr);
611 rec_del (zh->reg->records, &rec);
619 /* record going to be updated */
620 if (!delkeys.buf_used)
622 yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
623 zh->m_record_type, fname, recordOffset);
624 yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
628 /* flush new keys for sort&search etc */
629 if (zh->records_processed < zh->m_file_verbose_limit)
630 yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
631 zh->m_record_type, fname, recordOffset);
632 recordAttr->staticrank = extractCtrl.staticrank;
633 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
634 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys,
635 recordAttr->staticrank);
636 zh->records_updated++;
640 /* update file type */
641 xfree (rec->info[recInfo_fileType]);
642 rec->info[recInfo_fileType] =
643 rec_strdup (zh->m_record_type, &rec->size[recInfo_fileType]);
645 /* update filename */
646 xfree (rec->info[recInfo_filename]);
647 rec->info[recInfo_filename] =
648 rec_strdup (fname, &rec->size[recInfo_filename]);
650 /* update delete keys */
651 xfree (rec->info[recInfo_delKeys]);
652 if (zh->reg->keys.buf_used > 0 && zh->m_store_keys == 1)
654 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
655 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
656 zh->reg->keys.buf = NULL;
657 zh->reg->keys.buf_max = 0;
661 rec->info[recInfo_delKeys] = NULL;
662 rec->size[recInfo_delKeys] = 0;
665 /* update sort keys */
666 xfree (rec->info[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;
673 /* save file size of original record */
674 zebraExplain_recordBytesIncrement (zh->reg->zei,
675 - recordAttr->recordSize);
676 recordAttr->recordSize = fi->file_moffset - recordOffset;
677 if (!recordAttr->recordSize)
678 recordAttr->recordSize = fi->file_max - recordOffset;
679 zebraExplain_recordBytesIncrement (zh->reg->zei,
680 recordAttr->recordSize);
682 /* set run-number for this record */
683 recordAttr->runNumber = zebraExplain_runNumberIncrement (zh->reg->zei,
686 /* update store data */
687 xfree (rec->info[recInfo_storeData]);
688 if (zh->store_data_buf)
690 rec->size[recInfo_storeData] = zh->store_data_size;
691 rec->info[recInfo_storeData] = zh->store_data_buf;
692 zh->store_data_buf = 0;
694 else if (zh->m_store_data)
696 rec->size[recInfo_storeData] = recordAttr->recordSize;
697 rec->info[recInfo_storeData] = (char *)
698 xmalloc (recordAttr->recordSize);
699 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
701 yaz_log (YLOG_ERRNO|YLOG_FATAL, "seek to " PRINTF_OFF_T " in %s",
702 recordOffset, fname);
705 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
706 < recordAttr->recordSize)
708 yaz_log (YLOG_ERRNO|YLOG_FATAL, "read %d bytes of %s",
709 recordAttr->recordSize, fname);
715 rec->info[recInfo_storeData] = NULL;
716 rec->size[recInfo_storeData] = 0;
718 /* update database name */
719 xfree (rec->info[recInfo_databaseName]);
720 rec->info[recInfo_databaseName] =
721 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
724 recordAttr->recordOffset = recordOffset;
726 /* commit this record */
727 rec_put (zh->reg->records, &rec);
732 int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
739 struct file_read_info *fi;
740 const char *original_record_type = 0;
742 void *recTypeClientData;
744 if (!zh->m_group || !*zh->m_group)
747 sprintf (gprefix, "%s.", zh->m_group);
749 yaz_log (YLOG_DEBUG, "fileExtract %s", fname);
751 /* determine file extension */
753 for (i = strlen(fname); --i >= 0; )
756 else if (fname[i] == '.')
758 strcpy (ext, fname+i+1);
761 /* determine file type - depending on extension */
762 original_record_type = zh->m_record_type;
763 if (!zh->m_record_type)
765 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
766 zh->m_record_type = res_get (zh->res, ext_res);
768 if (!zh->m_record_type)
770 if (zh->records_processed < zh->m_file_verbose_limit)
771 yaz_log (YLOG_LOG, "? %s", fname);
774 /* determine match criteria */
775 if (!zh->m_record_id)
777 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
778 zh->m_record_id = res_get (zh->res, ext_res);
782 recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
783 &recTypeClientData)))
785 yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
789 switch(recType->version)
794 yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
796 if (sysno && deleteFlag)
802 if (zh->path_reg && !yaz_is_abspath (fname))
804 strcpy (full_rep, zh->path_reg);
805 strcat (full_rep, "/");
806 strcat (full_rep, fname);
809 strcpy (full_rep, fname);
812 if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
814 yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
815 zh->m_record_type = original_record_type;
819 fi = file_read_start (fd);
822 fi->file_moffset = fi->file_offset;
823 fi->file_more = 0; /* file_end not called (yet) */
824 r = file_extract_record (zh, sysno, fname, deleteFlag, fi, 1,
825 recType, recTypeClientData);
827 { /* file_end has been called so reset offset .. */
828 fi->file_offset = fi->file_moffset;
829 lseek(fi->fd, fi->file_moffset, SEEK_SET);
836 zh->m_record_type = original_record_type;
841 If sysno is provided, then it's used to identify the reocord.
842 If not, and match_criteria is provided, then sysno is guessed
843 If not, and a record is provided, then sysno is got from there
846 ZEBRA_RES buffer_extract_record (ZebraHandle zh,
847 const char *buf, size_t buf_size,
850 const char *recordType,
852 const char *match_criteria,
857 RecordAttr *recordAttr;
858 struct recExtractCtrl extractCtrl;
860 const char *matchStr = 0;
861 RecType recType = NULL;
864 long recordOffset = 0;
865 struct zebra_fetch_control fc;
866 const char *pr_fname = fname; /* filename to print .. */
867 int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
870 pr_fname = "<no file>"; /* make it printable if file is omitted */
873 fc.record_int_buf = buf;
874 fc.record_int_len = buf_size;
875 fc.record_int_pos = 0;
877 fc.record_offset = 0;
879 extractCtrl.offset = 0;
880 extractCtrl.readf = zebra_record_int_read;
881 extractCtrl.seekf = zebra_record_int_seek;
882 extractCtrl.tellf = zebra_record_int_tell;
883 extractCtrl.endf = zebra_record_int_end;
884 extractCtrl.first_record = 1;
885 extractCtrl.fh = &fc;
887 create_rec_keys_codec(&zh->reg->keys);
889 create_rec_keys_codec(&zh->reg->sortKeys);
891 zh->reg->sortKeys.buf_used = 0;
893 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
895 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
896 zh->m_explain_database))
900 if (recordType && *recordType)
902 yaz_log (YLOG_DEBUG, "Record type explicitly specified: %s", recordType);
903 recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
908 if (!(zh->m_record_type))
910 yaz_log (YLOG_WARN, "No such record type defined");
913 yaz_log (YLOG_DEBUG, "Get record type from rgroup: %s",zh->m_record_type);
914 recType = recType_byName (zh->reg->recTypes, zh->res,
915 zh->m_record_type, &clientData);
916 recordType = zh->m_record_type;
921 yaz_log (YLOG_WARN, "No such record type: %s", zh->m_record_type);
925 extractCtrl.init = extract_init;
926 extractCtrl.tokenAdd = extract_token_add;
927 extractCtrl.schemaAdd = extract_schema_add;
928 extractCtrl.dh = zh->reg->dh;
929 extractCtrl.handle = zh;
930 extractCtrl.match_criteria[0] = '\0';
931 extractCtrl.staticrank = 0;
933 init_extractCtrl(zh, &extractCtrl);
935 extract_set_store_data_prepare(&extractCtrl);
937 r = (*recType->extract)(clientData, &extractCtrl);
939 if (r == RECCTRL_EXTRACT_EOF)
941 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
943 /* error occured during extraction ... */
944 yaz_log (YLOG_WARN, "extract error: generic");
947 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
949 /* error occured during extraction ... */
950 yaz_log (YLOG_WARN, "extract error: no such filter");
956 if (extractCtrl.match_criteria[0])
957 match_criteria = extractCtrl.match_criteria;
961 if (match_criteria && *match_criteria) {
962 matchStr = match_criteria;
964 if (zh->m_record_id && *zh->m_record_id) {
965 matchStr = fileMatchStr (zh, &zh->reg->keys, pr_fname,
969 yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
975 rinfo = dict_lookup (zh->reg->matchDict, matchStr);
978 assert(*rinfo == sizeof(*sysno));
979 memcpy (sysno, rinfo+1, sizeof(*sysno));
983 if (zh->reg->keys.buf_used == 0)
985 /* the extraction process returned no information - the record
986 is probably empty - unless flagShowRecords is in use */
997 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
998 pr_fname, (long) recordOffset);
999 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
1003 yaz_log (YLOG_LOG, "add %s %s %ld", recordType, pr_fname,
1004 (long) recordOffset);
1005 rec = rec_new (zh->reg->records);
1007 *sysno = rec->sysno;
1009 recordAttr = rec_init_attr (zh->reg->zei, rec);
1010 recordAttr->staticrank = extractCtrl.staticrank;
1014 dict_insert (zh->reg->matchDict, matchStr,
1015 sizeof(*sysno), sysno);
1017 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1018 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys,
1019 recordAttr->staticrank);
1021 zh->records_inserted++;
1025 /* record already exists */
1026 struct recKeys delkeys;
1028 struct recKeys sortKeys;
1030 struct sortKeys sortKeys;
1036 yaz_log (YLOG_LOG, "skipped %s %s %ld",
1037 recordType, pr_fname, (long) recordOffset);
1042 rec = rec_get (zh->reg->records, *sysno);
1045 recordAttr = rec_init_attr (zh->reg->zei, rec);
1047 if (!force_update) {
1048 if (recordAttr->runNumber ==
1049 zebraExplain_runNumberIncrement (zh->reg->zei, 0))
1052 yaz_log (YLOG_LOG, "skipped %s %s %ld", recordType,
1053 pr_fname, (long) recordOffset);
1054 extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
1061 delkeys.buf_used = rec->size[recInfo_delKeys];
1062 delkeys.buf = rec->info[recInfo_delKeys];
1064 sortKeys.buf_used = rec->size[recInfo_sortKeys];
1065 sortKeys.buf = rec->info[recInfo_sortKeys];
1067 extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
1068 extract_flushRecordKeys (zh, *sysno, 0, &delkeys,
1069 recordAttr->staticrank);
1072 /* record going to be deleted */
1073 if (!delkeys.buf_used)
1077 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1078 pr_fname, (long) recordOffset);
1079 yaz_log (YLOG_WARN, "cannot delete file above, "
1086 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1087 pr_fname, (long) recordOffset);
1088 zh->records_deleted++;
1090 dict_delete (zh->reg->matchDict, matchStr);
1091 rec_del (zh->reg->records, &rec);
1099 /* record going to be updated */
1100 if (!delkeys.buf_used)
1104 yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
1105 pr_fname, (long) recordOffset);
1106 yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
1112 yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
1113 pr_fname, (long) recordOffset);
1114 recordAttr->staticrank = extractCtrl.staticrank;
1115 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1116 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys,
1117 recordAttr->staticrank);
1118 zh->records_updated++;
1122 /* update file type */
1123 xfree (rec->info[recInfo_fileType]);
1124 rec->info[recInfo_fileType] =
1125 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1127 /* update filename */
1128 xfree (rec->info[recInfo_filename]);
1129 rec->info[recInfo_filename] =
1130 rec_strdup (fname, &rec->size[recInfo_filename]);
1132 /* update delete keys */
1133 xfree (rec->info[recInfo_delKeys]);
1134 if (zh->reg->keys.buf_used > 0 && zh->m_store_keys == 1)
1136 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
1137 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
1138 zh->reg->keys.buf = NULL;
1139 zh->reg->keys.buf_max = 0;
1143 rec->info[recInfo_delKeys] = NULL;
1144 rec->size[recInfo_delKeys] = 0;
1147 /* update sort keys */
1148 xfree (rec->info[recInfo_sortKeys]);
1150 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
1151 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
1152 zh->reg->sortKeys.buf = NULL;
1153 zh->reg->sortKeys.buf_max = 0;
1155 /* save file size of original record */
1156 zebraExplain_recordBytesIncrement (zh->reg->zei,
1157 - recordAttr->recordSize);
1159 recordAttr->recordSize = fi->file_moffset - recordOffset;
1160 if (!recordAttr->recordSize)
1161 recordAttr->recordSize = fi->file_max - recordOffset;
1163 recordAttr->recordSize = buf_size;
1165 zebraExplain_recordBytesIncrement (zh->reg->zei,
1166 recordAttr->recordSize);
1168 /* set run-number for this record */
1169 recordAttr->runNumber =
1170 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
1172 /* update store data */
1173 xfree (rec->info[recInfo_storeData]);
1175 /* update store data */
1176 if (zh->store_data_buf)
1178 rec->size[recInfo_storeData] = zh->store_data_size;
1179 rec->info[recInfo_storeData] = zh->store_data_buf;
1180 zh->store_data_buf = 0;
1182 else if (zh->m_store_data)
1184 rec->size[recInfo_storeData] = recordAttr->recordSize;
1185 rec->info[recInfo_storeData] = (char *)
1186 xmalloc (recordAttr->recordSize);
1187 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1191 rec->info[recInfo_storeData] = NULL;
1192 rec->size[recInfo_storeData] = 0;
1194 /* update database name */
1195 xfree (rec->info[recInfo_databaseName]);
1196 rec->info[recInfo_databaseName] =
1197 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
1200 recordAttr->recordOffset = recordOffset;
1202 /* commit this record */
1203 rec_put (zh->reg->records, &rec);
1208 int explain_extract (void *handle, Record rec, data1_node *n)
1210 ZebraHandle zh = (ZebraHandle) handle;
1211 struct recExtractCtrl extractCtrl;
1213 if (zebraExplain_curDatabase (zh->reg->zei,
1214 rec->info[recInfo_databaseName]))
1217 if (zebraExplain_newDatabase (zh->reg->zei,
1218 rec->info[recInfo_databaseName], 0))
1222 create_rec_keys_codec(&zh->reg->keys);
1224 create_rec_keys_codec(&zh->reg->sortKeys);
1226 zh->reg->sortKeys.buf_used = 0;
1228 extractCtrl.init = extract_init;
1229 extractCtrl.tokenAdd = extract_token_add;
1230 extractCtrl.schemaAdd = extract_schema_add;
1231 extractCtrl.dh = zh->reg->dh;
1233 init_extractCtrl(zh, &extractCtrl);
1235 extractCtrl.flagShowRecords = 0;
1236 extractCtrl.match_criteria[0] = '\0';
1237 extractCtrl.staticrank = 0;
1238 extractCtrl.handle = handle;
1239 extractCtrl.first_record = 1;
1241 extract_set_store_data_prepare(&extractCtrl);
1244 grs_extract_tree(&extractCtrl, n);
1246 if (rec->size[recInfo_delKeys])
1248 struct recKeys delkeys;
1250 struct recKeys sortkeys;
1252 struct sortKeys sortkeys;
1255 delkeys.buf_used = rec->size[recInfo_delKeys];
1256 delkeys.buf = rec->info[recInfo_delKeys];
1258 sortkeys.buf_used = rec->size[recInfo_sortKeys];
1259 sortkeys.buf = rec->info[recInfo_sortKeys];
1261 extract_flushSortKeys (zh, rec->sysno, 0, &sortkeys);
1262 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys, 0);
1264 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->reg->keys, 0);
1265 extract_flushSortKeys (zh, rec->sysno, 1, &zh->reg->sortKeys);
1267 xfree (rec->info[recInfo_delKeys]);
1268 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
1269 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
1270 zh->reg->keys.buf = NULL;
1271 zh->reg->keys.buf_max = 0;
1273 xfree (rec->info[recInfo_sortKeys]);
1274 rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
1275 rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
1276 zh->reg->sortKeys.buf = NULL;
1277 zh->reg->sortKeys.buf_max = 0;
1282 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1283 int cmd, struct recKeys *reckeys,
1286 void *decode_handle = iscz1_start();
1289 ZebraExplainInfo zei = zh->reg->zei;
1291 if (!zh->reg->key_buf)
1293 int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
1296 yaz_log(YLOG_WARN, "Invalid memory setting, using default 8 MB");
1299 /* FIXME: That "8" should be in a default settings include */
1300 /* not hard-coded here! -H */
1301 zh->reg->key_buf = (char**) xmalloc (mem);
1302 zh->reg->ptr_top = mem/sizeof(char*);
1304 zh->reg->key_buf_used = 0;
1305 zh->reg->key_file_no = 0;
1307 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1309 while (off < reckeys->buf_used)
1311 const char *src = reckeys->buf + off;
1312 struct it_key key_in;
1313 struct it_key key_out;
1314 char *dst = (char*) &key_in;
1315 zint *keyp = key_out.mem;
1317 iscz1_decode(decode_handle, &dst, &src);
1318 assert(key_in.len == 4);
1320 if (zh->reg->key_buf_used + 1024 >
1321 (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
1322 extract_flushWriteKeys (zh,0);
1324 assert(zh->reg->ptr_i > 0);
1325 (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
1326 (char*)zh->reg->key_buf + zh->reg->key_buf_used;
1328 ch = (int) key_in.mem[0]; /* ordinal for field/use/attribute */
1330 zh->reg->key_buf_used +=
1331 key_SU_encode (ch,((char*)zh->reg->key_buf) +
1332 zh->reg->key_buf_used);
1334 ((char*)zh->reg->key_buf) [(zh->reg->key_buf_used)++] = *src++;
1336 ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = '\0';
1337 ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = cmd;
1339 if (zh->m_staticrank) /* rank config enabled ? */
1341 *keyp++ = staticrank;
1347 if (key_in.mem[1]) /* filter specified record ID */
1348 *keyp++ = key_in.mem[1];
1351 *keyp++ = key_in.mem[2]; /* section_id */
1352 *keyp++ = key_in.mem[3]; /* sequence .. */
1354 memcpy ((char*)zh->reg->key_buf + zh->reg->key_buf_used,
1355 &key_out, sizeof(key_out));
1356 (zh->reg->key_buf_used) += sizeof(key_out);
1357 off = src - reckeys->buf;
1359 assert (off == reckeys->buf_used);
1360 iscz1_stop(decode_handle);
1363 void extract_flushWriteKeys (ZebraHandle zh, int final)
1364 /* optimizing: if final=1, and no files written yet */
1365 /* push the keys directly to merge, sidestepping the */
1366 /* temp file altogether. Speeds small updates */
1369 char out_fname[200];
1371 struct encode_info encode_info;
1372 int ptr_i = zh->reg->ptr_i;
1377 if (!zh->reg->key_buf || ptr_i <= 0)
1379 yaz_log (YLOG_DEBUG, " nothing to flush section=%d buf=%p i=%d",
1380 zh->reg->key_file_no, zh->reg->key_buf, ptr_i);
1381 yaz_log (YLOG_DEBUG, " buf=%p ",
1383 yaz_log (YLOG_DEBUG, " ptr=%d ",zh->reg->ptr_i);
1384 yaz_log (YLOG_DEBUG, " reg=%p ",zh->reg);
1389 (zh->reg->key_file_no)++;
1390 yaz_log (YLOG_LOG, "sorting section %d", (zh->reg->key_file_no));
1391 yaz_log (YLOG_DEBUG, " sort_buff at %p n=%d",
1392 zh->reg->key_buf + zh->reg->ptr_top - ptr_i,ptr_i);
1394 qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
1395 sizeof(char*), key_qsort_compare);
1397 /* zebra.cfg: tempfiles:
1398 Y: always use temp files (old way)
1399 A: use temp files, if more than one (auto)
1400 = if this is both the last and the first
1401 N: never bother with temp files (new) */
1403 temp_policy=toupper(res_get_def(zh->res,"tempfiles","auto")[0]);
1404 if (temp_policy != 'Y' && temp_policy != 'N' && temp_policy != 'A') {
1405 yaz_log (YLOG_WARN, "Illegal tempfiles setting '%c'. using 'Auto' ",
1410 if ( ( temp_policy =='N' ) || /* always from memory */
1411 ( ( temp_policy =='A' ) && /* automatic */
1412 (zh->reg->key_file_no == 1) && /* this is first time */
1413 (final) ) ) /* and last (=only) time */
1414 { /* go directly from memory */
1415 zh->reg->key_file_no =0; /* signal not to read files */
1416 zebra_index_merge(zh);
1418 zh->reg->key_buf_used = 0;
1422 /* Not doing directly from memory, write into a temp file */
1423 extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
1425 if (!(outf = fopen (out_fname, "wb")))
1427 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1430 yaz_log (YLOG_LOG, "writing section %d", zh->reg->key_file_no);
1431 prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1433 encode_key_init (&encode_info);
1434 encode_key_write (cp, &encode_info, outf);
1438 cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1439 if (strcmp (cp, prevcp))
1441 encode_key_flush ( &encode_info, outf);
1442 encode_key_init (&encode_info);
1443 encode_key_write (cp, &encode_info, outf);
1447 encode_key_write (cp + strlen(cp), &encode_info, outf);
1449 encode_key_flush ( &encode_info, outf);
1451 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1452 extract_get_fname_tmp (out_fname, key_file_no);
1454 if (!(outf = fopen (out_fname, "wb")))
1456 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1459 yaz_log (YLOG_LOG, "writing section %d", key_file_no);
1461 prevcp = key_buf[ptr_top-i];
1463 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1465 key_y_len = strlen(prevcp)+1;
1467 yaz_log (YLOG_LOG, "key_y_len: %2d %02x %02x %s",
1468 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1470 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1471 sizeof(char*), key_y_compare);
1472 cp = key_buf[ptr_top-ptr_i];
1474 encode_key_init (&encode_info);
1475 encode_key_write (cp, &encode_info, outf);
1478 cp = key_buf[ptr_top-ptr_i];
1479 encode_key_write (cp+key_y_len, &encode_info, outf);
1481 encode_key_flush ( &encode_info, outf);
1484 prevcp = key_buf[ptr_top-ptr_i];
1489 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fclose %s", out_fname);
1492 yaz_log (YLOG_LOG, "finished section %d", zh->reg->key_file_no);
1494 zh->reg->key_buf_used = 0;
1497 void extract_add_it_key (ZebraHandle zh,
1498 struct recKeys *keys,
1500 const char *str, int slen, struct it_key *key)
1503 const char *src = (char*) key;
1505 if (keys->buf_used+1024 > keys->buf_max)
1507 char *b = (char *) xmalloc (keys->buf_max += 128000);
1508 if (keys->buf_used > 0)
1509 memcpy (b, keys->buf, keys->buf_used);
1513 dst = keys->buf + keys->buf_used;
1515 iscz1_encode(keys->codec_handle, &dst, &src);
1520 memcpy (dst, str, slen);
1523 keys->buf_used = dst - keys->buf;
1526 ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh, struct recKeys *reckeys,
1527 zebra_snippets *snippets)
1529 void *decode_handle = iscz1_start();
1532 NMEM nmem = nmem_create();
1534 assert(reckeys->buf);
1535 while (off < reckeys->buf_used)
1537 const char *src = reckeys->buf + off;
1539 char *dst = (char*) &key;
1540 char dst_buf[IT_MAX_WORD];
1541 char *dst_term = dst_buf;
1542 int index_type = 0, ord;
1544 iscz1_decode(decode_handle, &dst, &src);
1545 assert(key.len <= 4 && key.len > 2);
1547 seqno = (int) key.mem[key.len-1];
1550 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1551 0/* db */, 0/* set */, 0/* use */);
1553 zebra_term_untrans_iconv(zh, nmem, index_type,
1555 zebra_snippets_append(snippets, seqno, ord, dst_term);
1558 off = src - reckeys->buf;
1562 iscz1_stop(decode_handle);
1566 void print_rec_keys(ZebraHandle zh, struct recKeys *reckeys)
1568 void *decode_handle = iscz1_start();
1571 NMEM nmem = nmem_create();
1573 yaz_log(YLOG_LOG, "print_rec_keys buf=%p sz=%d", reckeys->buf,
1575 assert(reckeys->buf);
1576 while (off < reckeys->buf_used)
1578 const char *src = reckeys->buf + off;
1580 char *dst = (char*) &key;
1581 int attrSet, attrUse;
1582 char dst_buf[IT_MAX_WORD];
1583 char *dst_term = dst_buf;
1585 iscz1_decode(decode_handle, &dst, &src);
1586 assert(key.len <= 4 && key.len > 2);
1588 attrSet = (int) key.mem[0] >> 16;
1589 attrUse = (int) key.mem[0] & 65535;
1590 seqno = (int) key.mem[key.len-1];
1592 zebra_term_untrans_iconv(zh, nmem, src[0], &dst_term, src+1);
1594 yaz_log(YLOG_LOG, "ord=" ZINT_FORMAT " seqno=%d term=%s",
1595 key.mem[0], seqno, dst_term);
1598 off = src - reckeys->buf;
1602 iscz1_stop(decode_handle);
1605 void extract_add_index_string (RecWord *p, const char *str, int length)
1609 ZebraHandle zh = p->extractCtrl->handle;
1610 ZebraExplainInfo zei = zh->reg->zei;
1615 ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
1617 ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
1624 ch = zebraExplain_lookup_attr_su(zei, p->index_type,
1625 p->attrSet, p->attrUse);
1627 ch = zebraExplain_add_attr_su(zei, p->index_type,
1628 p->attrSet, p->attrUse);
1633 key.mem[1] = p->record_id;
1634 key.mem[2] = p->section_id;
1635 key.mem[3] = p->seqno;
1638 /* just for debugging .. */
1639 yaz_log(YLOG_LOG, "add: set=%d use=%d "
1640 "record_id=%lld section_id=%lld seqno=%lld",
1641 p->attrSet, p->attrUse, p->record_id, p->section_id, p->seqno);
1644 extract_add_it_key(p->extractCtrl->handle,
1651 static void extract_add_sort_string (RecWord *p, const char *str, int length)
1655 ZebraHandle zh = p->extractCtrl->handle;
1656 ZebraExplainInfo zei = zh->reg->zei;
1661 ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
1663 ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
1671 key.mem[1] = p->record_id;
1672 key.mem[2] = p->section_id;
1673 key.mem[3] = p->seqno;
1675 extract_add_it_key(p->extractCtrl->handle,
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 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1892 int cmd, struct recKeys *reckeys)
1894 SortIdx sortIdx = zh->reg->sortIdx;
1895 void *decode_handle = iscz1_start();
1899 while (off < reckeys->buf_used)
1901 const char *src = reckeys->buf + off;
1903 char *dst = (char*) &key;
1905 iscz1_decode(decode_handle, &dst, &src);
1906 assert(key.len == 4);
1908 ch = (int) key.mem[0]; /* ordinal for field/use/attribute */
1910 sortIdx_type(sortIdx, ch);
1912 sortIdx_add(sortIdx, src, strlen(src));
1914 sortIdx_add(sortIdx, "", 1);
1919 off = src - reckeys->buf;
1921 assert (off == reckeys->buf_used);
1922 iscz1_stop(decode_handle);
1925 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1926 int cmd, struct sortKeys *sk)
1928 SortIdx sortIdx = zh->reg->sortIdx;
1931 sortIdx_sysno (sortIdx, sysno);
1933 while (off < sk->buf_used)
1937 off += key_SU_decode(&set, (unsigned char *) sk->buf + off);
1938 off += key_SU_decode(&use, (unsigned char *) sk->buf + off);
1939 off += key_SU_decode(&slen, (unsigned char *) sk->buf + off);
1941 sortIdx_type(sortIdx, use);
1943 sortIdx_add(sortIdx, sk->buf + off, slen);
1945 sortIdx_add(sortIdx, "", 1);
1951 void encode_key_init (struct encode_info *i)
1960 i->encode_handle = iscz1_start();
1966 /* this is the old encode_key_write
1967 * may be deleted once we are confident that the new works
1970 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1973 char *bp = i->buf, *bp0;
1974 const char *src = (char *) &key;
1976 /* copy term to output buf */
1977 while ((*bp++ = *k++))
1979 /* and copy & align key so we can mangle */
1980 memcpy (&key, k+1, sizeof(struct it_key)); /* *k is insert/delete */
1983 iscz1_encode(i->encode_handle, &bp, &src);
1984 *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */
1985 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1987 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
1992 void encode_key_flush (struct encode_info *i, FILE *outf)
1993 { /* dummy routine */
1994 iscz1_stop(i->encode_handle);
1999 /* new encode_key_write
2000 * The idea is to buffer one more key, and compare them
2001 * If we are going to delete and insert the same key,
2002 * we may as well not bother. Should make a difference in
2003 * updates with small modifications (appending to a mbox)
2005 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
2010 if (*k) /* first time for new key */
2013 while ((*bp++ = *k++))
2015 i->keylen= bp - i->buf -1;
2016 assert(i->keylen+1+sizeof(struct it_key) < ENCODE_BUFLEN);
2020 bp=i->buf + i->keylen;
2025 memcpy (&key, k+1, sizeof(struct it_key));
2026 if (0==i->prevsys) /* no previous filter, fill up */
2028 i->prevsys=key.sysno;
2029 i->prevseq=key.seqno;
2032 else if ( (i->prevsys==key.sysno) &&
2033 (i->prevseq==key.seqno) &&
2035 { /* same numbers, diff cmd, they cancel out */
2039 { /* different stuff, write previous, move buf */
2040 bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
2041 if (i->sysno != i->prevsys)
2043 i->sysno = i->prevsys;
2046 else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
2048 return; /* ??? Filters some sort of duplicates away */
2049 /* ??? Can this ever happen -H 15oct02 */
2051 bp = encode_key_int (i->prevseq - i->seqno, bp);
2052 i->seqno = i->prevseq;
2053 i->cmd = i->prevcmd;
2054 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
2056 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
2059 i->keylen=0; /* ok, it's written, forget it */
2060 i->prevsys=key.sysno;
2061 i->prevseq=key.seqno;
2066 void encode_key_flush (struct encode_info *i, FILE *outf)
2067 { /* flush the last key from i */
2068 char *bp =i->buf + i->keylen;
2071 return; /* nothing to flush */
2074 bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
2075 if (i->sysno != i->prevsys)
2077 i->sysno = i->prevsys;
2080 else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
2082 return; /* ??? Filters some sort of duplicates away */
2083 /* ??? Can this ever happen -H 15oct02 */
2085 bp = encode_key_int (i->prevseq - i->seqno, bp);
2086 i->seqno = i->prevseq;
2087 i->cmd = i->prevcmd;
2088 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
2090 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
2093 i->keylen=0; /* ok, it's written, forget it */
2094 i->prevsys=0; /* forget the values too */