1 /* $Id: extract.c,v 1.252 2007-03-14 11:48:32 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 check_log_limit(ZebraHandle zh)
64 if (zh->records_processed + zh->records_skipped == zh->m_file_verbose_limit)
66 yaz_log(YLOG_LOG, "More than %d file log entries. Omitting rest",
67 zh->m_file_verbose_limit);
71 static void logRecord (ZebraHandle zh)
74 ++zh->records_processed;
75 if (!(zh->records_processed % 1000))
77 yaz_log(YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
78 ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT,
79 zh->records_processed, zh->records_inserted,
80 zh->records_updated, zh->records_deleted);
84 static void extract_add_index_string (RecWord *p,
85 zinfo_index_category_t cat,
86 const char *str, int length);
88 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
90 static void extract_init(struct recExtractCtrl *p, RecWord *w)
93 w->index_name = "any";
101 static void searchRecordKey(ZebraHandle zh,
102 zebra_rec_keys_t reckeys,
103 const char *index_name,
104 const char **ws, int ws_length)
108 zinfo_index_category_t cat = zinfo_index_category_index;
110 for (i = 0; i<ws_length; i++)
114 ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, '0', index_name);
116 ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'p', index_name);
118 ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'w', index_name);
123 if (zebra_rec_keys_rewind(reckeys))
130 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
132 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
134 seqno = key.mem[key.len-1];
136 if (key.mem[0] == ch)
142 woff = seqno - startSeq;
143 if (woff >= 0 && woff < ws_length)
150 #define FILE_MATCH_BLANK "\t "
152 static char *get_match_from_spec(ZebraHandle zh,
153 zebra_rec_keys_t reckeys,
154 const char *fname, const char *spec)
156 static char dstBuf[2048]; /* static here ??? */
158 const char *s = spec;
162 for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
169 char attset_str[64], attname_str[64];
173 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
175 for (i = 0; *s && *s != ',' && *s != ')' &&
176 !strchr(FILE_MATCH_BLANK, *s); s++)
177 if (i+1 < sizeof(attset_str))
178 attset_str[i++] = *s;
179 attset_str[i] = '\0';
181 for (; strchr(FILE_MATCH_BLANK, *s); s++)
184 strcpy(attname_str, attset_str);
187 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
189 for (i = 0; *s && *s != ')' &&
190 !strchr(FILE_MATCH_BLANK, *s); s++)
191 if (i+1 < sizeof(attname_str))
192 attname_str[i++] = *s;
193 attname_str[i] = '\0';
196 searchRecordKey (zh, reckeys, attname_str, ws, 32);
200 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
201 spec, zh->m_group ? zh->m_group : "none");
206 for (i = 0; i<32; i++)
215 dst += strlen(ws[i]);
219 yaz_log (YLOG_WARN, "Record didn't contain match"
220 " fields in (%s,%s)", attset_str, attname_str);
228 const char *spec_src = NULL;
229 const char *s1 = ++s;
230 while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
234 if (spec_len > sizeof(special)-1)
235 spec_len = sizeof(special)-1;
236 memcpy (special, s, spec_len);
237 special[spec_len] = '\0';
240 if (!strcmp (special, "group"))
241 spec_src = zh->m_group;
242 else if (!strcmp (special, "database"))
243 spec_src = zh->basenames[0];
244 else if (!strcmp (special, "filename")) {
247 else if (!strcmp (special, "type"))
248 spec_src = zh->m_record_type;
253 strcpy (dst, spec_src);
254 dst += strlen (spec_src);
257 else if (*s == '\"' || *s == '\'')
259 int stopMarker = *s++;
263 while (*s && *s != stopMarker)
265 if (i+1 < sizeof(tmpString))
266 tmpString[i++] = *s++;
271 strcpy (dst, tmpString);
272 dst += strlen (tmpString);
276 yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
277 spec, zh->m_group ? zh->m_group : "none");
284 yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
285 fname, zh->m_group ? zh->m_group : "none");
292 struct recordLogInfo {
295 struct recordGroup *rGroup;
298 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
301 for (i = 0; i<256; i++)
303 if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
308 ctrl->flagShowRecords = !zh->m_flag_rw;
311 static void all_matches_add(struct recExtractCtrl *ctrl)
314 extract_init(ctrl, &word);
315 word.index_name = "_ALLRECORDS";
316 word.index_type = 'w';
318 extract_add_index_string (&word, zinfo_index_category_alwaysmatches,
322 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
323 struct ZebraRecStream *stream,
324 enum zebra_recctrl_action_t action,
326 const char *recordType,
328 const char *match_criteria,
331 void *recTypeClientData);
334 ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
337 ZEBRA_RES r = ZEBRA_OK;
342 struct file_read_info *fi = 0;
343 const char *original_record_type = 0;
345 void *recTypeClientData;
346 struct ZebraRecStream stream, *streamp;
348 zebra_init_log_level();
350 if (!zh->m_group || !*zh->m_group)
353 sprintf (gprefix, "%s.", zh->m_group);
355 yaz_log(log_level_extract, "zebra_extract_file %s", fname);
357 /* determine file extension */
359 for (i = strlen(fname); --i >= 0; )
362 else if (fname[i] == '.')
364 strcpy (ext, fname+i+1);
367 /* determine file type - depending on extension */
368 original_record_type = zh->m_record_type;
369 if (!zh->m_record_type)
371 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
372 zh->m_record_type = res_get (zh->res, ext_res);
374 if (!zh->m_record_type)
377 if (zh->records_processed + zh->records_skipped
378 < zh->m_file_verbose_limit)
379 yaz_log (YLOG_LOG, "? %s", fname);
380 zh->records_skipped++;
383 /* determine match criteria */
384 if (!zh->m_record_id)
386 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
387 zh->m_record_id = res_get (zh->res, ext_res);
391 recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
392 &recTypeClientData)))
394 yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
398 switch(recType->version)
403 yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
405 if (sysno && deleteFlag)
414 if (zh->path_reg && !yaz_is_abspath (fname))
416 strcpy (full_rep, zh->path_reg);
417 strcat (full_rep, "/");
418 strcat (full_rep, fname);
421 strcpy (full_rep, fname);
423 if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
425 yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
426 zh->m_record_type = original_record_type;
430 zebra_create_stream_fd(streamp, fd, 0);
432 r = zebra_extract_records_stream(zh, streamp,
434 action_delete : action_update,
438 0, /*match_criteria */
440 recType, recTypeClientData);
442 stream.destroy(streamp);
443 zh->m_record_type = original_record_type;
448 If sysno is provided, then it's used to identify the reocord.
449 If not, and match_criteria is provided, then sysno is guessed
450 If not, and a record is provided, then sysno is got from there
454 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
455 const char *buf, size_t buf_size,
456 enum zebra_recctrl_action_t action,
458 const char *recordType,
460 const char *match_criteria,
463 struct ZebraRecStream stream;
468 if (recordType && *recordType)
470 yaz_log(log_level_extract,
471 "Record type explicitly specified: %s", recordType);
472 recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
477 if (!(zh->m_record_type))
479 yaz_log (YLOG_WARN, "No such record type defined");
482 yaz_log(log_level_extract, "Get record type from rgroup: %s",
484 recType = recType_byName (zh->reg->recTypes, zh->res,
485 zh->m_record_type, &clientData);
486 recordType = zh->m_record_type;
491 yaz_log (YLOG_WARN, "No such record type: %s", recordType);
495 zebra_create_stream_mem(&stream, buf, buf_size);
497 res = zebra_extract_records_stream(zh, &stream,
504 recType, clientData);
505 stream.destroy(&stream);
509 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
510 struct ZebraRecStream *stream,
511 enum zebra_recctrl_action_t action,
513 const char *recordType,
515 const char *match_criteria,
518 void *recTypeClientData)
520 ZEBRA_RES res = ZEBRA_OK;
524 res = zebra_extract_record_stream(zh, stream,
531 recType, recTypeClientData, &more);
546 ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
547 struct ZebraRecStream *stream,
548 enum zebra_recctrl_action_t action,
550 const char *recordType,
552 const char *match_criteria,
555 void *recTypeClientData,
560 RecordAttr *recordAttr;
561 struct recExtractCtrl extractCtrl;
563 const char *matchStr = 0;
565 off_t start_offset = 0, end_offset = 0;
566 const char *pr_fname = fname; /* filename to print .. */
567 int show_progress = zh->records_processed + zh->records_skipped
568 < zh->m_file_verbose_limit ? 1:0;
570 zebra_init_log_level();
573 pr_fname = "<no file>"; /* make it printable if file is omitted */
575 zebra_rec_keys_reset(zh->reg->keys);
576 zebra_rec_keys_reset(zh->reg->sortKeys);
578 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
580 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
581 zh->m_explain_database))
587 off_t null_offset = 0;
588 extractCtrl.stream = stream;
590 start_offset = stream->tellf(stream);
592 extractCtrl.first_record = start_offset ? 0 : 1;
594 stream->endf(stream, &null_offset);;
596 extractCtrl.init = extract_init;
597 extractCtrl.tokenAdd = extract_token_add;
598 extractCtrl.schemaAdd = extract_schema_add;
599 extractCtrl.dh = zh->reg->dh;
600 extractCtrl.handle = zh;
601 extractCtrl.match_criteria[0] = '\0';
602 extractCtrl.staticrank = 0;
604 init_extractCtrl(zh, &extractCtrl);
606 extract_set_store_data_prepare(&extractCtrl);
608 r = (*recType->extract)(recTypeClientData, &extractCtrl);
612 case RECCTRL_EXTRACT_EOF:
614 case RECCTRL_EXTRACT_ERROR_GENERIC:
615 /* error occured during extraction ... */
616 yaz_log (YLOG_WARN, "extract error: generic");
618 case RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER:
619 /* error occured during extraction ... */
620 yaz_log (YLOG_WARN, "extract error: no such filter");
622 case RECCTRL_EXTRACT_SKIP:
624 yaz_log (YLOG_LOG, "skip %s %s " ZINT_FORMAT,
625 recordType, pr_fname, (zint) start_offset);
628 end_offset = stream->endf(stream, 0);
630 stream->seekf(stream, end_offset);
633 case RECCTRL_EXTRACT_OK:
636 yaz_log (YLOG_WARN, "extract error: unknown error: %d", r);
639 end_offset = stream->endf(stream, 0);
641 stream->seekf(stream, end_offset);
643 end_offset = stream->tellf(stream);
645 all_matches_add(&extractCtrl);
647 if (extractCtrl.match_criteria[0])
648 match_criteria = extractCtrl.match_criteria;
656 if (match_criteria && *match_criteria) {
657 matchStr = match_criteria;
659 if (zh->m_record_id && *zh->m_record_id) {
660 matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname,
664 yaz_log (YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
665 pr_fname, (zint) start_offset);
672 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
673 char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
677 assert(*rinfo == sizeof(*sysno));
678 memcpy (sysno, rinfo+1, sizeof(*sysno));
682 if (zebra_rec_keys_empty(zh->reg->keys))
684 /* the extraction process returned no information - the record
685 is probably empty - unless flagShowRecords is in use */
693 if (action == action_delete)
695 yaz_log (YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
696 pr_fname, (zint) start_offset);
697 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
700 else if (action == action_replace)
702 yaz_log (YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
703 pr_fname, (zint) start_offset);
704 yaz_log (YLOG_WARN, "cannot update record above (seems new)");
708 yaz_log (YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
709 (zint) start_offset);
710 rec = rec_new (zh->reg->records);
714 recordAttr = rec_init_attr (zh->reg->zei, rec);
715 if (extractCtrl.staticrank < 0)
717 yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
718 extractCtrl.staticrank = 0;
720 recordAttr->staticrank = extractCtrl.staticrank;
724 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
725 dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
726 sizeof(*sysno), sysno);
729 extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
730 extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
731 recordAttr->staticrank);
732 zh->records_inserted++;
736 /* record already exists */
737 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
738 zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
739 if (action == action_insert)
741 yaz_log (YLOG_LOG, "skipped %s %s " ZINT_FORMAT,
742 recordType, pr_fname, (zint) start_offset);
747 rec = rec_get (zh->reg->records, *sysno);
750 recordAttr = rec_init_attr (zh->reg->zei, rec);
752 /* decrease total size */
753 zebraExplain_recordBytesIncrement (zh->reg->zei,
754 - recordAttr->recordSize);
756 zebra_rec_keys_set_buf(delkeys,
757 rec->info[recInfo_delKeys],
758 rec->size[recInfo_delKeys],
760 zebra_rec_keys_set_buf(sortKeys,
761 rec->info[recInfo_sortKeys],
762 rec->size[recInfo_sortKeys],
765 extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
766 extract_flush_record_keys(zh, *sysno, 0, delkeys,
767 recordAttr->staticrank);
768 if (action == action_delete)
770 /* record going to be deleted */
771 if (zebra_rec_keys_empty(delkeys))
773 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
774 pr_fname, (zint) start_offset);
775 yaz_log(YLOG_WARN, "cannot delete file above, "
776 "storeKeys false (3)");
781 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
782 pr_fname, (zint) start_offset);
783 zh->records_deleted++;
786 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
787 dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
789 rec_del (zh->reg->records, &rec);
796 { /* update or special_update */
798 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
799 pr_fname, (zint) start_offset);
800 recordAttr->staticrank = extractCtrl.staticrank;
801 extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
802 extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
803 recordAttr->staticrank);
804 zh->records_updated++;
806 zebra_rec_keys_close(delkeys);
807 zebra_rec_keys_close(sortKeys);
809 /* update file type */
810 xfree (rec->info[recInfo_fileType]);
811 rec->info[recInfo_fileType] =
812 rec_strdup (recordType, &rec->size[recInfo_fileType]);
814 /* update filename */
815 xfree (rec->info[recInfo_filename]);
816 rec->info[recInfo_filename] =
817 rec_strdup (fname, &rec->size[recInfo_filename]);
819 /* update delete keys */
820 xfree (rec->info[recInfo_delKeys]);
821 if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
823 zebra_rec_keys_get_buf(zh->reg->keys,
824 &rec->info[recInfo_delKeys],
825 &rec->size[recInfo_delKeys]);
829 rec->info[recInfo_delKeys] = NULL;
830 rec->size[recInfo_delKeys] = 0;
832 /* update sort keys */
833 xfree (rec->info[recInfo_sortKeys]);
835 zebra_rec_keys_get_buf(zh->reg->sortKeys,
836 &rec->info[recInfo_sortKeys],
837 &rec->size[recInfo_sortKeys]);
841 recordAttr->recordSize = end_offset - start_offset;
842 zebraExplain_recordBytesIncrement(zh->reg->zei,
843 recordAttr->recordSize);
846 /* set run-number for this record */
847 recordAttr->runNumber =
848 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
850 /* update store data */
851 xfree (rec->info[recInfo_storeData]);
853 /* update store data */
854 if (zh->store_data_buf)
856 rec->size[recInfo_storeData] = zh->store_data_size;
857 rec->info[recInfo_storeData] = zh->store_data_buf;
858 zh->store_data_buf = 0;
859 recordAttr->recordSize = zh->store_data_size;
861 else if (zh->m_store_data)
863 off_t cur_offset = stream->tellf(stream);
865 rec->size[recInfo_storeData] = recordAttr->recordSize;
866 rec->info[recInfo_storeData] = (char *)
867 xmalloc (recordAttr->recordSize);
868 stream->seekf(stream, start_offset);
869 stream->readf(stream, rec->info[recInfo_storeData],
870 recordAttr->recordSize);
871 stream->seekf(stream, cur_offset);
875 rec->info[recInfo_storeData] = NULL;
876 rec->size[recInfo_storeData] = 0;
878 /* update database name */
879 xfree (rec->info[recInfo_databaseName]);
880 rec->info[recInfo_databaseName] =
881 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
884 recordAttr->recordOffset = start_offset;
886 /* commit this record */
887 rec_put (zh->reg->records, &rec);
892 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
894 ZebraHandle zh = (ZebraHandle) handle;
895 struct recExtractCtrl extractCtrl;
897 if (zebraExplain_curDatabase (zh->reg->zei,
898 rec->info[recInfo_databaseName]))
901 if (zebraExplain_newDatabase (zh->reg->zei,
902 rec->info[recInfo_databaseName], 0))
906 zebra_rec_keys_reset(zh->reg->keys);
907 zebra_rec_keys_reset(zh->reg->sortKeys);
909 extractCtrl.init = extract_init;
910 extractCtrl.tokenAdd = extract_token_add;
911 extractCtrl.schemaAdd = extract_schema_add;
912 extractCtrl.dh = zh->reg->dh;
914 init_extractCtrl(zh, &extractCtrl);
916 extractCtrl.flagShowRecords = 0;
917 extractCtrl.match_criteria[0] = '\0';
918 extractCtrl.staticrank = 0;
919 extractCtrl.handle = handle;
920 extractCtrl.first_record = 1;
922 extract_set_store_data_prepare(&extractCtrl);
925 grs_extract_tree(&extractCtrl, n);
927 if (rec->size[recInfo_delKeys])
929 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
931 zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
933 zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
934 rec->size[recInfo_delKeys],
936 extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
937 zebra_rec_keys_close(delkeys);
939 zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
940 rec->size[recInfo_sortKeys],
943 extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
944 zebra_rec_keys_close(sortkeys);
946 extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
947 extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
949 xfree (rec->info[recInfo_delKeys]);
950 zebra_rec_keys_get_buf(zh->reg->keys,
951 &rec->info[recInfo_delKeys],
952 &rec->size[recInfo_delKeys]);
954 xfree (rec->info[recInfo_sortKeys]);
955 zebra_rec_keys_get_buf(zh->reg->sortKeys,
956 &rec->info[recInfo_sortKeys],
957 &rec->size[recInfo_sortKeys]);
961 void extract_rec_keys_log(ZebraHandle zh, int is_insert,
962 zebra_rec_keys_t reckeys,
965 if (zebra_rec_keys_rewind(reckeys))
970 NMEM nmem = nmem_create();
972 while(zebra_rec_keys_read(reckeys, &str, &slen, &key))
974 char keystr[200]; /* room for zints to print */
976 int ord = CAST_ZINT_TO_INT(key.mem[0]);
978 const char *string_index;
980 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
981 0/* db */, &string_index);
983 zebra_term_untrans_iconv(zh, nmem, index_type,
986 for (i = 0; i<key.len; i++)
988 sprintf(keystr + strlen(keystr), ZINT_FORMAT " ", key.mem[i]);
991 if (*str < CHR_BASE_CHAR)
994 char dst_buf[200]; /* room for special chars */
996 strcpy(dst_buf , "?");
998 if (!strcmp(str, ""))
999 strcpy(dst_buf, "alwaysmatches");
1000 if (!strcmp(str, FIRST_IN_FIELD_STR))
1001 strcpy(dst_buf, "firstinfield");
1002 else if (!strcmp(str, CHR_UNKNOWN))
1003 strcpy(dst_buf, "unknown");
1004 else if (!strcmp(str, CHR_SPACE))
1005 strcpy(dst_buf, "space");
1007 for (i = 0; i<slen; i++)
1009 sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
1011 yaz_log(level, "%s%c %s %s", keystr, index_type,
1012 string_index, dst_buf);
1016 yaz_log(level, "%s%c %s \"%s\"", keystr, index_type,
1017 string_index, dst_term);
1025 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
1026 zebra_rec_keys_t reckeys)
1028 ZebraExplainInfo zei = zh->reg->zei;
1032 struct ord_stat *next;
1035 if (zebra_rec_keys_rewind(reckeys))
1037 struct ord_stat *ord_list = 0;
1041 struct it_key key_in;
1042 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1044 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1046 for (p = ord_list; p ; p = p->next)
1054 p = xmalloc(sizeof(*p));
1065 struct ord_stat *p1 = p;
1068 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
1070 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
1077 void extract_flush_record_keys(ZebraHandle zh, zint sysno, int cmd,
1078 zebra_rec_keys_t reckeys,
1081 ZebraExplainInfo zei = zh->reg->zei;
1083 extract_rec_keys_adjust(zh, cmd, reckeys);
1085 if (log_level_details)
1087 yaz_log(log_level_details, "Keys for record " ZINT_FORMAT " %s",
1088 sysno, cmd ? "insert" : "delete");
1089 extract_rec_keys_log(zh, cmd, reckeys, log_level_details);
1092 if (!zh->reg->key_block)
1094 int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1095 const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", ".");
1096 int use_threads = atoi(res_get_def (zh->res, "threads", "1"));
1097 zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1099 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1101 if (zebra_rec_keys_rewind(reckeys))
1105 struct it_key key_in;
1106 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1108 key_block_write(zh->reg->key_block, sysno,
1109 &key_in, cmd, str, slen,
1110 staticrank, zh->m_staticrank);
1116 ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
1117 zebra_rec_keys_t reckeys,
1118 zebra_snippets *snippets)
1120 NMEM nmem = nmem_create();
1121 if (zebra_rec_keys_rewind(reckeys))
1126 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1133 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1134 seqno = key.mem[key.len-1];
1135 ord = CAST_ZINT_TO_INT(key.mem[0]);
1137 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1138 0/* db */, 0 /* string_index */);
1140 zebra_term_untrans_iconv(zh, nmem, index_type,
1142 zebra_snippets_append(snippets, seqno, ord, dst_term);
1150 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1152 yaz_log(YLOG_LOG, "print_rec_keys");
1153 if (zebra_rec_keys_rewind(reckeys))
1158 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1160 char dst_buf[IT_MAX_WORD];
1163 int ord = CAST_ZINT_TO_INT(key.mem[0]);
1165 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1167 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1169 seqno = key.mem[key.len-1];
1171 zebra_term_untrans(zh, index_type, dst_buf, str);
1173 yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
1174 " term=%s", ord, seqno, dst_buf);
1179 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1180 const char *str, int length)
1183 ZebraHandle zh = p->extractCtrl->handle;
1184 ZebraExplainInfo zei = zh->reg->zei;
1187 ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1189 ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1193 key.mem[i++] = p->record_id;
1194 key.mem[i++] = p->section_id;
1196 if (zh->m_segment_indexing)
1197 key.mem[i++] = p->segment;
1198 key.mem[i++] = p->seqno;
1201 zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1204 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1207 ZebraHandle zh = p->extractCtrl->handle;
1208 ZebraExplainInfo zei = zh->reg->zei;
1210 zinfo_index_category_t cat = zinfo_index_category_sort;
1212 ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1214 ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1217 key.mem[1] = p->record_id;
1219 zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1222 static void extract_add_staticrank_string(RecWord *p,
1223 const char *str, int length)
1226 struct recExtractCtrl *ctrl = p->extractCtrl;
1228 if (length > sizeof(valz)-1)
1229 length = sizeof(valz)-1;
1231 memcpy(valz, str, length);
1232 valz[length] = '\0';
1233 ctrl->staticrank = atozint(valz);
1236 static void extract_add_string(RecWord *p, const char *string, int length)
1238 ZebraHandle zh = p->extractCtrl->handle;
1239 assert (length > 0);
1244 if (zebra_maps_is_index(zh->reg->zebra_maps, p->index_type))
1246 extract_add_index_string(p, zinfo_index_category_index,
1248 if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type))
1251 memcpy(&word, p, sizeof(word));
1254 extract_add_index_string(
1255 &word, zinfo_index_category_alwaysmatches, "", 0);
1258 else if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
1260 extract_add_sort_string(p, string, length);
1262 else if (zebra_maps_is_staticrank(zh->reg->zebra_maps, p->index_type))
1264 extract_add_staticrank_string(p, string, length);
1268 static void extract_add_incomplete_field(RecWord *p)
1270 ZebraHandle zh = p->extractCtrl->handle;
1271 const char *b = p->term_buf;
1272 int remain = p->term_len;
1274 const char **map = 0;
1277 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1281 char buf[IT_MAX_WORD+1];
1285 while (map && *map && **map == *CHR_SPACE)
1287 remain = p->term_len - (b - p->term_buf);
1289 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
1297 while (map && *map && **map != *CHR_SPACE)
1299 const char *cp = *map;
1301 while (i < IT_MAX_WORD && *cp)
1303 remain = p->term_len - (b - p->term_buf);
1305 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1315 if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
1317 /* first in field marker */
1318 extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1322 extract_add_string (p, buf, i);
1327 static void extract_add_complete_field (RecWord *p)
1329 ZebraHandle zh = p->extractCtrl->handle;
1330 const char *b = p->term_buf;
1331 char buf[IT_MAX_WORD+1];
1332 const char **map = 0;
1333 int i = 0, remain = p->term_len;
1336 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
1338 while (remain > 0 && i < IT_MAX_WORD)
1340 while (map && *map && **map == *CHR_SPACE)
1342 remain = p->term_len - (b - p->term_buf);
1346 int first = i ? 0 : 1; /* first position */
1347 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, first);
1355 if (i && i < IT_MAX_WORD)
1356 buf[i++] = *CHR_SPACE;
1357 while (map && *map && **map != *CHR_SPACE)
1359 const char *cp = *map;
1361 if (**map == *CHR_CUT)
1367 if (i >= IT_MAX_WORD)
1369 while (i < IT_MAX_WORD && *cp)
1372 remain = p->term_len - (b - p->term_buf);
1375 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
1384 extract_add_string (p, buf, i);
1387 static void extract_token_add(RecWord *p)
1389 ZebraHandle zh = p->extractCtrl->handle;
1392 if (log_level_extract)
1394 yaz_log(log_level_extract, "extract_token_add "
1395 "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
1396 p->index_type, p->index_name,
1397 p->seqno, p->term_len, p->term_buf);
1399 if ((wrbuf = zebra_replace(zh->reg->zebra_maps, p->index_type, 0,
1400 p->term_buf, p->term_len)))
1402 p->term_buf = wrbuf_buf(wrbuf);
1403 p->term_len = wrbuf_len(wrbuf);
1405 if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type))
1406 extract_add_complete_field (p);
1408 extract_add_incomplete_field(p);
1411 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1412 void *buf, size_t sz)
1414 ZebraHandle zh = (ZebraHandle) p->handle;
1416 xfree(zh->store_data_buf);
1417 zh->store_data_buf = 0;
1418 zh->store_data_size = 0;
1421 zh->store_data_buf = xmalloc(sz);
1422 zh->store_data_size = sz;
1423 memcpy(zh->store_data_buf, buf, sz);
1427 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1429 ZebraHandle zh = (ZebraHandle) p->handle;
1430 xfree(zh->store_data_buf);
1431 zh->store_data_buf = 0;
1432 zh->store_data_size = 0;
1433 p->setStoreData = extract_set_store_data_cb;
1436 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
1438 ZebraHandle zh = (ZebraHandle) p->handle;
1439 zebraExplain_addSchema (zh->reg->zei, oid);
1442 void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
1443 int cmd, zebra_rec_keys_t reckeys)
1446 yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT,
1448 extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG);
1451 if (zebra_rec_keys_rewind(reckeys))
1453 zebra_sort_index_t si = zh->reg->sort_index;
1456 struct it_key key_in;
1458 zebra_sort_sysno(si, sysno);
1460 while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1462 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1464 zebra_sort_type(si, ord);
1466 zebra_sort_add(si, str, slen);
1468 zebra_sort_delete(si);
1476 * indent-tabs-mode: nil
1478 * vim: shiftwidth=4 tabstop=8 expandtab