1 /* $Id: extract.c,v 1.247 2007-01-15 15:10:16 adam Exp $
2 Copyright (C) 1995-2007
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 this program; if not, write to the Free Software
19 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
39 static int log_level_extract = 0;
40 static int log_level_details = 0;
41 static int log_level_initialized = 0;
43 static void zebra_init_log_level(void)
45 if (!log_level_initialized)
47 log_level_initialized = 1;
49 log_level_extract = yaz_log_module_level("extract");
50 log_level_details = yaz_log_module_level("indexdetails");
54 static void extract_flush_record_keys(ZebraHandle zh, zint sysno,
55 int cmd, zebra_rec_keys_t reckeys,
57 static void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
58 int cmd, zebra_rec_keys_t skp);
59 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid);
60 static void extract_token_add (RecWord *p);
62 static void logRecord (ZebraHandle zh)
64 ++zh->records_processed;
65 if (!(zh->records_processed % 1000))
67 yaz_log(YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
68 ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT,
69 zh->records_processed, zh->records_inserted,
70 zh->records_updated, zh->records_deleted);
74 static void extract_add_index_string (RecWord *p,
75 zinfo_index_category_t cat,
76 const char *str, int length);
78 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
80 static void extract_init(struct recExtractCtrl *p, RecWord *w)
83 w->index_name = "any";
91 static void searchRecordKey(ZebraHandle zh,
92 zebra_rec_keys_t reckeys,
93 const char *index_name,
94 const char **ws, int ws_length)
98 zinfo_index_category_t cat = zinfo_index_category_index;
100 for (i = 0; i<ws_length; i++)
104 ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, '0', index_name);
106 ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'p', index_name);
108 ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'w', index_name);
113 if (zebra_rec_keys_rewind(reckeys))
120 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
122 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
124 seqno = key.mem[key.len-1];
126 if (key.mem[0] == ch)
132 woff = seqno - startSeq;
133 if (woff >= 0 && woff < ws_length)
140 #define FILE_MATCH_BLANK "\t "
142 static char *get_match_from_spec(ZebraHandle zh,
143 zebra_rec_keys_t reckeys,
144 const char *fname, const char *spec)
146 static char dstBuf[2048]; /* static here ??? */
148 const char *s = spec;
152 for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
159 char attset_str[64], attname_str[64];
163 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
165 for (i = 0; *s && *s != ',' && *s != ')' &&
166 !strchr(FILE_MATCH_BLANK, *s); s++)
167 if (i+1 < sizeof(attset_str))
168 attset_str[i++] = *s;
169 attset_str[i] = '\0';
171 for (; strchr(FILE_MATCH_BLANK, *s); s++)
174 strcpy(attname_str, attset_str);
177 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
179 for (i = 0; *s && *s != ')' &&
180 !strchr(FILE_MATCH_BLANK, *s); s++)
181 if (i+1 < sizeof(attname_str))
182 attname_str[i++] = *s;
183 attname_str[i] = '\0';
186 searchRecordKey (zh, reckeys, attname_str, ws, 32);
190 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
191 spec, zh->m_group ? zh->m_group : "none");
196 for (i = 0; i<32; i++)
205 dst += strlen(ws[i]);
209 yaz_log (YLOG_WARN, "Record didn't contain match"
210 " fields in (%s,%s)", attset_str, attname_str);
218 const char *spec_src = NULL;
219 const char *s1 = ++s;
220 while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
224 if (spec_len > sizeof(special)-1)
225 spec_len = sizeof(special)-1;
226 memcpy (special, s, spec_len);
227 special[spec_len] = '\0';
230 if (!strcmp (special, "group"))
231 spec_src = zh->m_group;
232 else if (!strcmp (special, "database"))
233 spec_src = zh->basenames[0];
234 else if (!strcmp (special, "filename")) {
237 else if (!strcmp (special, "type"))
238 spec_src = zh->m_record_type;
243 strcpy (dst, spec_src);
244 dst += strlen (spec_src);
247 else if (*s == '\"' || *s == '\'')
249 int stopMarker = *s++;
253 while (*s && *s != stopMarker)
255 if (i+1 < sizeof(tmpString))
256 tmpString[i++] = *s++;
261 strcpy (dst, tmpString);
262 dst += strlen (tmpString);
266 yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
267 spec, zh->m_group ? zh->m_group : "none");
274 yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
275 fname, zh->m_group ? zh->m_group : "none");
282 struct recordLogInfo {
285 struct recordGroup *rGroup;
288 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
291 for (i = 0; i<256; i++)
293 if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
298 ctrl->flagShowRecords = !zh->m_flag_rw;
301 static void all_matches_add(struct recExtractCtrl *ctrl)
304 extract_init(ctrl, &word);
305 word.index_name = "_ALLRECORDS";
306 word.index_type = 'w';
308 extract_add_index_string (&word, zinfo_index_category_alwaysmatches,
312 ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
315 ZEBRA_RES r = ZEBRA_OK;
320 struct file_read_info *fi = 0;
321 const char *original_record_type = 0;
323 void *recTypeClientData;
324 struct ZebraRecStream stream, *streamp;
326 zebra_init_log_level();
328 if (!zh->m_group || !*zh->m_group)
331 sprintf (gprefix, "%s.", zh->m_group);
333 yaz_log(log_level_extract, "zebra_extract_file %s", fname);
335 /* determine file extension */
337 for (i = strlen(fname); --i >= 0; )
340 else if (fname[i] == '.')
342 strcpy (ext, fname+i+1);
345 /* determine file type - depending on extension */
346 original_record_type = zh->m_record_type;
347 if (!zh->m_record_type)
349 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
350 zh->m_record_type = res_get (zh->res, ext_res);
352 if (!zh->m_record_type)
354 if (zh->records_processed < zh->m_file_verbose_limit)
355 yaz_log (YLOG_LOG, "? %s", fname);
358 /* determine match criteria */
359 if (!zh->m_record_id)
361 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
362 zh->m_record_id = res_get (zh->res, ext_res);
366 recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
367 &recTypeClientData)))
369 yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
373 switch(recType->version)
378 yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
380 if (sysno && deleteFlag)
389 if (zh->path_reg && !yaz_is_abspath (fname))
391 strcpy (full_rep, zh->path_reg);
392 strcat (full_rep, "/");
393 strcat (full_rep, fname);
396 strcpy (full_rep, fname);
398 if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
400 yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
401 zh->m_record_type = original_record_type;
405 zebra_create_stream_fd(streamp, fd, 0);
410 r = zebra_extract_record_stream(zh, streamp,
415 0, /*match_criteria */
417 1, /* force_update */
418 1, /* allow_update */
419 recType, recTypeClientData, &more);
428 stream.destroy(streamp);
429 zh->m_record_type = original_record_type;
434 If sysno is provided, then it's used to identify the reocord.
435 If not, and match_criteria is provided, then sysno is guessed
436 If not, and a record is provided, then sysno is got from there
440 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
441 const char *buf, size_t buf_size,
444 const char *recordType,
446 const char *match_criteria,
451 struct ZebraRecStream stream;
457 if (recordType && *recordType)
459 yaz_log(log_level_extract,
460 "Record type explicitly specified: %s", recordType);
461 recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
466 if (!(zh->m_record_type))
468 yaz_log (YLOG_WARN, "No such record type defined");
471 yaz_log(log_level_extract, "Get record type from rgroup: %s",
473 recType = recType_byName (zh->reg->recTypes, zh->res,
474 zh->m_record_type, &clientData);
475 recordType = zh->m_record_type;
480 yaz_log (YLOG_WARN, "No such record type: %s", recordType);
484 zebra_create_stream_mem(&stream, buf, buf_size);
486 res = zebra_extract_record_stream(zh, &stream,
495 recType, clientData, &more);
496 stream.destroy(&stream);
501 ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
502 struct ZebraRecStream *stream,
505 const char *recordType,
507 const char *match_criteria,
512 void *recTypeClientData,
517 RecordAttr *recordAttr;
518 struct recExtractCtrl extractCtrl;
520 const char *matchStr = 0;
522 off_t start_offset = 0, end_offset = 0;
523 const char *pr_fname = fname; /* filename to print .. */
524 int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
526 zebra_init_log_level();
529 pr_fname = "<no file>"; /* make it printable if file is omitted */
531 zebra_rec_keys_reset(zh->reg->keys);
532 zebra_rec_keys_reset(zh->reg->sortKeys);
534 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
536 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
537 zh->m_explain_database))
543 off_t null_offset = 0;
544 extractCtrl.stream = stream;
546 start_offset = stream->tellf(stream);
548 extractCtrl.first_record = start_offset ? 0 : 1;
550 stream->endf(stream, &null_offset);;
552 extractCtrl.init = extract_init;
553 extractCtrl.tokenAdd = extract_token_add;
554 extractCtrl.schemaAdd = extract_schema_add;
555 extractCtrl.dh = zh->reg->dh;
556 extractCtrl.handle = zh;
557 extractCtrl.match_criteria[0] = '\0';
558 extractCtrl.staticrank = 0;
560 init_extractCtrl(zh, &extractCtrl);
562 extract_set_store_data_prepare(&extractCtrl);
564 r = (*recType->extract)(recTypeClientData, &extractCtrl);
566 if (r == RECCTRL_EXTRACT_EOF)
568 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
570 /* error occured during extraction ... */
571 yaz_log (YLOG_WARN, "extract error: generic");
574 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
576 /* error occured during extraction ... */
577 yaz_log (YLOG_WARN, "extract error: no such filter");
581 all_matches_add(&extractCtrl);
583 if (extractCtrl.match_criteria[0])
584 match_criteria = extractCtrl.match_criteria;
587 end_offset = stream->endf(stream, 0);
590 end_offset = stream->tellf(stream);
592 stream->seekf(stream, end_offset);
602 if (match_criteria && *match_criteria) {
603 matchStr = match_criteria;
605 if (zh->m_record_id && *zh->m_record_id) {
606 matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname,
610 yaz_log (YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
611 pr_fname, (zint) start_offset);
618 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
619 char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
623 assert(*rinfo == sizeof(*sysno));
624 memcpy (sysno, rinfo+1, sizeof(*sysno));
628 if (zebra_rec_keys_empty(zh->reg->keys))
630 /* the extraction process returned no information - the record
631 is probably empty - unless flagShowRecords is in use */
641 yaz_log (YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
642 pr_fname, (zint) start_offset);
643 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
647 yaz_log (YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
648 (zint) start_offset);
649 rec = rec_new (zh->reg->records);
653 recordAttr = rec_init_attr (zh->reg->zei, rec);
654 if (extractCtrl.staticrank < 0)
656 yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
657 extractCtrl.staticrank = 0;
659 recordAttr->staticrank = extractCtrl.staticrank;
663 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
664 dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
665 sizeof(*sysno), sysno);
668 extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
669 extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
670 recordAttr->staticrank);
671 zh->records_inserted++;
675 /* record already exists */
676 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
677 zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
680 yaz_log (YLOG_LOG, "skipped %s %s " ZINT_FORMAT,
681 recordType, pr_fname, (zint) start_offset);
686 rec = rec_get (zh->reg->records, *sysno);
689 recordAttr = rec_init_attr (zh->reg->zei, rec);
691 /* decrease total size */
692 zebraExplain_recordBytesIncrement (zh->reg->zei,
693 - recordAttr->recordSize);
695 zebra_rec_keys_set_buf(delkeys,
696 rec->info[recInfo_delKeys],
697 rec->size[recInfo_delKeys],
699 zebra_rec_keys_set_buf(sortKeys,
700 rec->info[recInfo_sortKeys],
701 rec->size[recInfo_sortKeys],
704 extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
705 extract_flush_record_keys(zh, *sysno, 0, delkeys,
706 recordAttr->staticrank);
709 /* record going to be deleted */
710 if (zebra_rec_keys_empty(delkeys))
712 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
713 pr_fname, (zint) start_offset);
714 yaz_log(YLOG_WARN, "cannot delete file above, "
715 "storeKeys false (3)");
720 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
721 pr_fname, (zint) start_offset);
722 zh->records_deleted++;
725 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
726 dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
728 rec_del (zh->reg->records, &rec);
737 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
738 pr_fname, (zint) start_offset);
739 recordAttr->staticrank = extractCtrl.staticrank;
740 extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
741 extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
742 recordAttr->staticrank);
743 zh->records_updated++;
745 zebra_rec_keys_close(delkeys);
746 zebra_rec_keys_close(sortKeys);
748 /* update file type */
749 xfree (rec->info[recInfo_fileType]);
750 rec->info[recInfo_fileType] =
751 rec_strdup (recordType, &rec->size[recInfo_fileType]);
753 /* update filename */
754 xfree (rec->info[recInfo_filename]);
755 rec->info[recInfo_filename] =
756 rec_strdup (fname, &rec->size[recInfo_filename]);
758 /* update delete keys */
759 xfree (rec->info[recInfo_delKeys]);
760 if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
762 zebra_rec_keys_get_buf(zh->reg->keys,
763 &rec->info[recInfo_delKeys],
764 &rec->size[recInfo_delKeys]);
768 rec->info[recInfo_delKeys] = NULL;
769 rec->size[recInfo_delKeys] = 0;
771 /* update sort keys */
772 xfree (rec->info[recInfo_sortKeys]);
774 zebra_rec_keys_get_buf(zh->reg->sortKeys,
775 &rec->info[recInfo_sortKeys],
776 &rec->size[recInfo_sortKeys]);
780 recordAttr->recordSize = end_offset - start_offset;
781 zebraExplain_recordBytesIncrement(zh->reg->zei,
782 recordAttr->recordSize);
785 /* set run-number for this record */
786 recordAttr->runNumber =
787 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
789 /* update store data */
790 xfree (rec->info[recInfo_storeData]);
792 /* update store data */
793 if (zh->store_data_buf)
795 rec->size[recInfo_storeData] = zh->store_data_size;
796 rec->info[recInfo_storeData] = zh->store_data_buf;
797 zh->store_data_buf = 0;
798 recordAttr->recordSize = zh->store_data_size;
800 else if (zh->m_store_data)
802 off_t cur_offset = stream->tellf(stream);
804 rec->size[recInfo_storeData] = recordAttr->recordSize;
805 rec->info[recInfo_storeData] = (char *)
806 xmalloc (recordAttr->recordSize);
807 stream->seekf(stream, start_offset);
808 stream->readf(stream, rec->info[recInfo_storeData],
809 recordAttr->recordSize);
810 stream->seekf(stream, cur_offset);
814 rec->info[recInfo_storeData] = NULL;
815 rec->size[recInfo_storeData] = 0;
817 /* update database name */
818 xfree (rec->info[recInfo_databaseName]);
819 rec->info[recInfo_databaseName] =
820 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
823 recordAttr->recordOffset = start_offset;
825 /* commit this record */
826 rec_put (zh->reg->records, &rec);
831 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
833 ZebraHandle zh = (ZebraHandle) handle;
834 struct recExtractCtrl extractCtrl;
836 if (zebraExplain_curDatabase (zh->reg->zei,
837 rec->info[recInfo_databaseName]))
840 if (zebraExplain_newDatabase (zh->reg->zei,
841 rec->info[recInfo_databaseName], 0))
845 zebra_rec_keys_reset(zh->reg->keys);
846 zebra_rec_keys_reset(zh->reg->sortKeys);
848 extractCtrl.init = extract_init;
849 extractCtrl.tokenAdd = extract_token_add;
850 extractCtrl.schemaAdd = extract_schema_add;
851 extractCtrl.dh = zh->reg->dh;
853 init_extractCtrl(zh, &extractCtrl);
855 extractCtrl.flagShowRecords = 0;
856 extractCtrl.match_criteria[0] = '\0';
857 extractCtrl.staticrank = 0;
858 extractCtrl.handle = handle;
859 extractCtrl.first_record = 1;
861 extract_set_store_data_prepare(&extractCtrl);
864 grs_extract_tree(&extractCtrl, n);
866 if (rec->size[recInfo_delKeys])
868 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
870 zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
872 zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
873 rec->size[recInfo_delKeys],
875 extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
876 zebra_rec_keys_close(delkeys);
878 zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
879 rec->size[recInfo_sortKeys],
882 extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
883 zebra_rec_keys_close(sortkeys);
885 extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
886 extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
888 xfree (rec->info[recInfo_delKeys]);
889 zebra_rec_keys_get_buf(zh->reg->keys,
890 &rec->info[recInfo_delKeys],
891 &rec->size[recInfo_delKeys]);
893 xfree (rec->info[recInfo_sortKeys]);
894 zebra_rec_keys_get_buf(zh->reg->sortKeys,
895 &rec->info[recInfo_sortKeys],
896 &rec->size[recInfo_sortKeys]);
900 void extract_rec_keys_log(ZebraHandle zh, int is_insert,
901 zebra_rec_keys_t reckeys,
904 if (zebra_rec_keys_rewind(reckeys))
909 NMEM nmem = nmem_create();
911 while(zebra_rec_keys_read(reckeys, &str, &slen, &key))
913 char keystr[200]; /* room for zints to print */
915 int ord = CAST_ZINT_TO_INT(key.mem[0]);
917 const char *string_index;
919 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
920 0/* db */, &string_index);
922 zebra_term_untrans_iconv(zh, nmem, index_type,
925 for (i = 0; i<key.len; i++)
927 sprintf(keystr + strlen(keystr), ZINT_FORMAT " ", key.mem[i]);
930 if (*str < CHR_BASE_CHAR)
933 char dst_buf[200]; /* room for special chars */
935 strcpy(dst_buf , "?");
937 if (!strcmp(str, ""))
938 strcpy(dst_buf, "alwaysmatches");
939 if (!strcmp(str, FIRST_IN_FIELD_STR))
940 strcpy(dst_buf, "firstinfield");
941 else if (!strcmp(str, CHR_UNKNOWN))
942 strcpy(dst_buf, "unknown");
943 else if (!strcmp(str, CHR_SPACE))
944 strcpy(dst_buf, "space");
946 for (i = 0; i<slen; i++)
948 sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
950 yaz_log(level, "%s%c %s %s", keystr, index_type,
951 string_index, dst_buf);
955 yaz_log(level, "%s%c %s \"%s\"", keystr, index_type,
956 string_index, dst_term);
964 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
965 zebra_rec_keys_t reckeys)
967 ZebraExplainInfo zei = zh->reg->zei;
971 struct ord_stat *next;
974 if (zebra_rec_keys_rewind(reckeys))
976 struct ord_stat *ord_list = 0;
980 struct it_key key_in;
981 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
983 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
985 for (p = ord_list; p ; p = p->next)
993 p = xmalloc(sizeof(*p));
1004 struct ord_stat *p1 = p;
1007 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
1009 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
1016 void extract_flush_record_keys(ZebraHandle zh, zint sysno, int cmd,
1017 zebra_rec_keys_t reckeys,
1020 ZebraExplainInfo zei = zh->reg->zei;
1022 extract_rec_keys_adjust(zh, cmd, reckeys);
1024 if (log_level_details)
1026 yaz_log(log_level_details, "Keys for record " ZINT_FORMAT " %s",
1027 sysno, cmd ? "insert" : "delete");
1028 extract_rec_keys_log(zh, cmd, reckeys, log_level_details);
1031 if (!zh->reg->key_block)
1033 int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1034 const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", ".");
1035 int use_threads = atoi(res_get_def (zh->res, "threads", "1"));
1036 zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1038 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1040 if (zebra_rec_keys_rewind(reckeys))
1044 struct it_key key_in;
1045 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1047 key_block_write(zh->reg->key_block, sysno,
1048 &key_in, cmd, str, slen,
1049 staticrank, zh->m_staticrank);
1055 ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
1056 zebra_rec_keys_t reckeys,
1057 zebra_snippets *snippets)
1059 NMEM nmem = nmem_create();
1060 if (zebra_rec_keys_rewind(reckeys))
1065 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1072 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1073 seqno = key.mem[key.len-1];
1074 ord = CAST_ZINT_TO_INT(key.mem[0]);
1076 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1077 0/* db */, 0 /* string_index */);
1079 zebra_term_untrans_iconv(zh, nmem, index_type,
1081 zebra_snippets_append(snippets, seqno, ord, dst_term);
1089 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1091 yaz_log(YLOG_LOG, "print_rec_keys");
1092 if (zebra_rec_keys_rewind(reckeys))
1097 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1099 char dst_buf[IT_MAX_WORD];
1102 int ord = CAST_ZINT_TO_INT(key.mem[0]);
1104 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1106 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1108 seqno = key.mem[key.len-1];
1110 zebra_term_untrans(zh, index_type, dst_buf, str);
1112 yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
1113 " term=%s", ord, seqno, dst_buf);
1118 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1119 const char *str, int length)
1122 ZebraHandle zh = p->extractCtrl->handle;
1123 ZebraExplainInfo zei = zh->reg->zei;
1126 ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1128 ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1132 key.mem[i++] = p->record_id;
1133 key.mem[i++] = p->section_id;
1135 if (zh->m_segment_indexing)
1136 key.mem[i++] = p->segment;
1137 key.mem[i++] = p->seqno;
1140 zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1143 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1146 ZebraHandle zh = p->extractCtrl->handle;
1147 ZebraExplainInfo zei = zh->reg->zei;
1149 zinfo_index_category_t cat = zinfo_index_category_sort;
1151 ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1153 ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1156 key.mem[1] = p->record_id;
1158 zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1161 static void extract_add_string(RecWord *p, const char *string, int length)
1163 ZebraHandle zh = p->extractCtrl->handle;
1164 assert (length > 0);
1169 if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
1170 extract_add_sort_string(p, string, length);
1173 extract_add_index_string(p, zinfo_index_category_index,
1175 if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type))
1178 memcpy(&word, p, sizeof(word));
1181 extract_add_index_string(
1182 &word, zinfo_index_category_alwaysmatches, "", 0);
1187 static void extract_add_incomplete_field(RecWord *p)
1189 ZebraHandle zh = p->extractCtrl->handle;
1190 const char *b = p->term_buf;
1191 int remain = p->term_len;
1193 const char **map = 0;
1196 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1200 char buf[IT_MAX_WORD+1];
1204 while (map && *map && **map == *CHR_SPACE)
1206 remain = p->term_len - (b - p->term_buf);
1208 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
1216 while (map && *map && **map != *CHR_SPACE)
1218 const char *cp = *map;
1220 while (i < IT_MAX_WORD && *cp)
1222 remain = p->term_len - (b - p->term_buf);
1224 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1234 if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
1236 /* first in field marker */
1237 extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1241 extract_add_string (p, buf, i);
1246 static void extract_add_complete_field (RecWord *p)
1248 ZebraHandle zh = p->extractCtrl->handle;
1249 const char *b = p->term_buf;
1250 char buf[IT_MAX_WORD+1];
1251 const char **map = 0;
1252 int i = 0, remain = p->term_len;
1255 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
1257 while (remain > 0 && i < IT_MAX_WORD)
1259 while (map && *map && **map == *CHR_SPACE)
1261 remain = p->term_len - (b - p->term_buf);
1265 int first = i ? 0 : 1; /* first position */
1266 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, first);
1274 if (i && i < IT_MAX_WORD)
1275 buf[i++] = *CHR_SPACE;
1276 while (map && *map && **map != *CHR_SPACE)
1278 const char *cp = *map;
1280 if (**map == *CHR_CUT)
1286 if (i >= IT_MAX_WORD)
1288 while (i < IT_MAX_WORD && *cp)
1291 remain = p->term_len - (b - p->term_buf);
1294 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
1303 extract_add_string (p, buf, i);
1306 static void extract_token_add(RecWord *p)
1308 ZebraHandle zh = p->extractCtrl->handle;
1311 if (log_level_extract)
1313 yaz_log(log_level_extract, "extract_token_add "
1314 "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
1315 p->index_type, p->index_name,
1316 p->seqno, p->term_len, p->term_buf);
1318 if ((wrbuf = zebra_replace(zh->reg->zebra_maps, p->index_type, 0,
1319 p->term_buf, p->term_len)))
1321 p->term_buf = wrbuf_buf(wrbuf);
1322 p->term_len = wrbuf_len(wrbuf);
1324 if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type))
1325 extract_add_complete_field (p);
1327 extract_add_incomplete_field(p);
1330 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1331 void *buf, size_t sz)
1333 ZebraHandle zh = (ZebraHandle) p->handle;
1335 xfree(zh->store_data_buf);
1336 zh->store_data_buf = 0;
1337 zh->store_data_size = 0;
1340 zh->store_data_buf = xmalloc(sz);
1341 zh->store_data_size = sz;
1342 memcpy(zh->store_data_buf, buf, sz);
1346 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1348 ZebraHandle zh = (ZebraHandle) p->handle;
1349 xfree(zh->store_data_buf);
1350 zh->store_data_buf = 0;
1351 zh->store_data_size = 0;
1352 p->setStoreData = extract_set_store_data_cb;
1355 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
1357 ZebraHandle zh = (ZebraHandle) p->handle;
1358 zebraExplain_addSchema (zh->reg->zei, oid);
1361 void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
1362 int cmd, zebra_rec_keys_t reckeys)
1365 yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT,
1367 extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG);
1370 if (zebra_rec_keys_rewind(reckeys))
1372 zebra_sort_index_t si = zh->reg->sort_index;
1375 struct it_key key_in;
1377 zebra_sort_sysno(si, sysno);
1379 while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1381 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1383 zebra_sort_type(si, ord);
1385 zebra_sort_add(si, str, slen);
1387 zebra_sort_delete(si);
1395 * indent-tabs-mode: nil
1397 * vim: shiftwidth=4 tabstop=8 expandtab