1 /* $Id: extract.c,v 1.254 2007-03-20 22:07:21 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;
603 extractCtrl.action = action;
605 init_extractCtrl(zh, &extractCtrl);
607 extract_set_store_data_prepare(&extractCtrl);
609 r = (*recType->extract)(recTypeClientData, &extractCtrl);
611 if (action == action_update)
613 action = extractCtrl.action;
618 case RECCTRL_EXTRACT_EOF:
620 case RECCTRL_EXTRACT_ERROR_GENERIC:
621 /* error occured during extraction ... */
622 yaz_log (YLOG_WARN, "extract error: generic");
624 case RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER:
625 /* error occured during extraction ... */
626 yaz_log (YLOG_WARN, "extract error: no such filter");
628 case RECCTRL_EXTRACT_SKIP:
630 yaz_log (YLOG_LOG, "skip %s %s " ZINT_FORMAT,
631 recordType, pr_fname, (zint) start_offset);
634 end_offset = stream->endf(stream, 0);
636 stream->seekf(stream, end_offset);
639 case RECCTRL_EXTRACT_OK:
642 yaz_log (YLOG_WARN, "extract error: unknown error: %d", r);
645 end_offset = stream->endf(stream, 0);
647 stream->seekf(stream, end_offset);
649 end_offset = stream->tellf(stream);
651 all_matches_add(&extractCtrl);
653 if (extractCtrl.match_criteria[0])
654 match_criteria = extractCtrl.match_criteria;
662 if (match_criteria && *match_criteria) {
663 matchStr = match_criteria;
665 if (zh->m_record_id && *zh->m_record_id) {
666 matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname,
670 yaz_log (YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
671 pr_fname, (zint) start_offset);
678 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
679 char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
683 assert(*rinfo == sizeof(*sysno));
684 memcpy (sysno, rinfo+1, sizeof(*sysno));
688 if (zebra_rec_keys_empty(zh->reg->keys))
690 /* the extraction process returned no information - the record
691 is probably empty - unless flagShowRecords is in use */
699 if (action == action_delete)
701 yaz_log (YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
702 pr_fname, (zint) start_offset);
703 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
706 else if (action == action_replace)
708 yaz_log (YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
709 pr_fname, (zint) start_offset);
710 yaz_log (YLOG_WARN, "cannot update record above (seems new)");
714 yaz_log (YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
715 (zint) start_offset);
716 rec = rec_new (zh->reg->records);
720 recordAttr = rec_init_attr (zh->reg->zei, rec);
721 if (extractCtrl.staticrank < 0)
723 yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
724 extractCtrl.staticrank = 0;
726 recordAttr->staticrank = extractCtrl.staticrank;
730 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
731 dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
732 sizeof(*sysno), sysno);
735 extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
736 extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
737 recordAttr->staticrank);
738 zh->records_inserted++;
742 /* record already exists */
743 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
744 zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
745 if (action == action_insert)
747 yaz_log (YLOG_LOG, "skipped %s %s " ZINT_FORMAT,
748 recordType, pr_fname, (zint) start_offset);
753 rec = rec_get (zh->reg->records, *sysno);
756 recordAttr = rec_init_attr (zh->reg->zei, rec);
758 /* decrease total size */
759 zebraExplain_recordBytesIncrement (zh->reg->zei,
760 - recordAttr->recordSize);
762 zebra_rec_keys_set_buf(delkeys,
763 rec->info[recInfo_delKeys],
764 rec->size[recInfo_delKeys],
766 zebra_rec_keys_set_buf(sortKeys,
767 rec->info[recInfo_sortKeys],
768 rec->size[recInfo_sortKeys],
771 extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
772 extract_flush_record_keys(zh, *sysno, 0, delkeys,
773 recordAttr->staticrank);
774 if (action == action_delete)
776 /* record going to be deleted */
777 if (zebra_rec_keys_empty(delkeys))
779 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
780 pr_fname, (zint) start_offset);
781 yaz_log(YLOG_WARN, "cannot delete file above, "
782 "storeKeys false (3)");
787 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
788 pr_fname, (zint) start_offset);
789 zh->records_deleted++;
792 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
793 dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
795 rec_del (zh->reg->records, &rec);
802 { /* update or special_update */
804 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
805 pr_fname, (zint) start_offset);
806 recordAttr->staticrank = extractCtrl.staticrank;
807 extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
808 extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
809 recordAttr->staticrank);
810 zh->records_updated++;
812 zebra_rec_keys_close(delkeys);
813 zebra_rec_keys_close(sortKeys);
815 /* update file type */
816 xfree (rec->info[recInfo_fileType]);
817 rec->info[recInfo_fileType] =
818 rec_strdup (recordType, &rec->size[recInfo_fileType]);
820 /* update filename */
821 xfree (rec->info[recInfo_filename]);
822 rec->info[recInfo_filename] =
823 rec_strdup (fname, &rec->size[recInfo_filename]);
825 /* update delete keys */
826 xfree (rec->info[recInfo_delKeys]);
827 if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
829 zebra_rec_keys_get_buf(zh->reg->keys,
830 &rec->info[recInfo_delKeys],
831 &rec->size[recInfo_delKeys]);
835 rec->info[recInfo_delKeys] = NULL;
836 rec->size[recInfo_delKeys] = 0;
838 /* update sort keys */
839 xfree (rec->info[recInfo_sortKeys]);
841 zebra_rec_keys_get_buf(zh->reg->sortKeys,
842 &rec->info[recInfo_sortKeys],
843 &rec->size[recInfo_sortKeys]);
847 recordAttr->recordSize = end_offset - start_offset;
848 zebraExplain_recordBytesIncrement(zh->reg->zei,
849 recordAttr->recordSize);
852 /* set run-number for this record */
853 recordAttr->runNumber =
854 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
856 /* update store data */
857 xfree (rec->info[recInfo_storeData]);
859 /* update store data */
860 if (zh->store_data_buf)
862 rec->size[recInfo_storeData] = zh->store_data_size;
863 rec->info[recInfo_storeData] = zh->store_data_buf;
864 zh->store_data_buf = 0;
865 recordAttr->recordSize = zh->store_data_size;
867 else if (zh->m_store_data)
869 off_t cur_offset = stream->tellf(stream);
871 rec->size[recInfo_storeData] = recordAttr->recordSize;
872 rec->info[recInfo_storeData] = (char *)
873 xmalloc (recordAttr->recordSize);
874 stream->seekf(stream, start_offset);
875 stream->readf(stream, rec->info[recInfo_storeData],
876 recordAttr->recordSize);
877 stream->seekf(stream, cur_offset);
881 rec->info[recInfo_storeData] = NULL;
882 rec->size[recInfo_storeData] = 0;
884 /* update database name */
885 xfree (rec->info[recInfo_databaseName]);
886 rec->info[recInfo_databaseName] =
887 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
890 recordAttr->recordOffset = start_offset;
892 /* commit this record */
893 rec_put (zh->reg->records, &rec);
898 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
900 ZebraHandle zh = (ZebraHandle) handle;
901 struct recExtractCtrl extractCtrl;
903 if (zebraExplain_curDatabase (zh->reg->zei,
904 rec->info[recInfo_databaseName]))
907 if (zebraExplain_newDatabase (zh->reg->zei,
908 rec->info[recInfo_databaseName], 0))
912 zebra_rec_keys_reset(zh->reg->keys);
913 zebra_rec_keys_reset(zh->reg->sortKeys);
915 extractCtrl.init = extract_init;
916 extractCtrl.tokenAdd = extract_token_add;
917 extractCtrl.schemaAdd = extract_schema_add;
918 extractCtrl.dh = zh->reg->dh;
920 init_extractCtrl(zh, &extractCtrl);
922 extractCtrl.flagShowRecords = 0;
923 extractCtrl.match_criteria[0] = '\0';
924 extractCtrl.staticrank = 0;
925 extractCtrl.action = action_update;
927 extractCtrl.handle = handle;
928 extractCtrl.first_record = 1;
930 extract_set_store_data_prepare(&extractCtrl);
933 grs_extract_tree(&extractCtrl, n);
935 if (rec->size[recInfo_delKeys])
937 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
939 zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
941 zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
942 rec->size[recInfo_delKeys],
944 extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
945 zebra_rec_keys_close(delkeys);
947 zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
948 rec->size[recInfo_sortKeys],
951 extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
952 zebra_rec_keys_close(sortkeys);
954 extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
955 extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
957 xfree (rec->info[recInfo_delKeys]);
958 zebra_rec_keys_get_buf(zh->reg->keys,
959 &rec->info[recInfo_delKeys],
960 &rec->size[recInfo_delKeys]);
962 xfree (rec->info[recInfo_sortKeys]);
963 zebra_rec_keys_get_buf(zh->reg->sortKeys,
964 &rec->info[recInfo_sortKeys],
965 &rec->size[recInfo_sortKeys]);
969 void extract_rec_keys_log(ZebraHandle zh, int is_insert,
970 zebra_rec_keys_t reckeys,
973 if (zebra_rec_keys_rewind(reckeys))
978 NMEM nmem = nmem_create();
980 while(zebra_rec_keys_read(reckeys, &str, &slen, &key))
982 char keystr[200]; /* room for zints to print */
984 int ord = CAST_ZINT_TO_INT(key.mem[0]);
986 const char *string_index;
988 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
989 0/* db */, &string_index);
991 zebra_term_untrans_iconv(zh, nmem, index_type,
994 for (i = 0; i<key.len; i++)
996 sprintf(keystr + strlen(keystr), ZINT_FORMAT " ", key.mem[i]);
999 if (*str < CHR_BASE_CHAR)
1002 char dst_buf[200]; /* room for special chars */
1004 strcpy(dst_buf , "?");
1006 if (!strcmp(str, ""))
1007 strcpy(dst_buf, "alwaysmatches");
1008 if (!strcmp(str, FIRST_IN_FIELD_STR))
1009 strcpy(dst_buf, "firstinfield");
1010 else if (!strcmp(str, CHR_UNKNOWN))
1011 strcpy(dst_buf, "unknown");
1012 else if (!strcmp(str, CHR_SPACE))
1013 strcpy(dst_buf, "space");
1015 for (i = 0; i<slen; i++)
1017 sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
1019 yaz_log(level, "%s%c %s %s", keystr, index_type,
1020 string_index, dst_buf);
1024 yaz_log(level, "%s%c %s \"%s\"", keystr, index_type,
1025 string_index, dst_term);
1033 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
1034 zebra_rec_keys_t reckeys)
1036 ZebraExplainInfo zei = zh->reg->zei;
1040 struct ord_stat *next;
1043 if (zebra_rec_keys_rewind(reckeys))
1045 struct ord_stat *ord_list = 0;
1049 struct it_key key_in;
1050 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1052 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1054 for (p = ord_list; p ; p = p->next)
1062 p = xmalloc(sizeof(*p));
1073 struct ord_stat *p1 = p;
1076 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
1078 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
1085 void extract_flush_record_keys(ZebraHandle zh, zint sysno, int cmd,
1086 zebra_rec_keys_t reckeys,
1089 ZebraExplainInfo zei = zh->reg->zei;
1091 extract_rec_keys_adjust(zh, cmd, reckeys);
1093 if (log_level_details)
1095 yaz_log(log_level_details, "Keys for record " ZINT_FORMAT " %s",
1096 sysno, cmd ? "insert" : "delete");
1097 extract_rec_keys_log(zh, cmd, reckeys, log_level_details);
1100 if (!zh->reg->key_block)
1102 int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1103 const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", ".");
1104 int use_threads = atoi(res_get_def (zh->res, "threads", "1"));
1105 zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1107 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1109 if (zebra_rec_keys_rewind(reckeys))
1113 struct it_key key_in;
1114 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1116 key_block_write(zh->reg->key_block, sysno,
1117 &key_in, cmd, str, slen,
1118 staticrank, zh->m_staticrank);
1124 ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
1125 zebra_rec_keys_t reckeys,
1126 zebra_snippets *snippets)
1128 NMEM nmem = nmem_create();
1129 if (zebra_rec_keys_rewind(reckeys))
1134 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1141 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1142 seqno = key.mem[key.len-1];
1143 ord = CAST_ZINT_TO_INT(key.mem[0]);
1145 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1146 0/* db */, 0 /* string_index */);
1148 zebra_term_untrans_iconv(zh, nmem, index_type,
1150 zebra_snippets_append(snippets, seqno, ord, dst_term);
1158 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1160 yaz_log(YLOG_LOG, "print_rec_keys");
1161 if (zebra_rec_keys_rewind(reckeys))
1166 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1168 char dst_buf[IT_MAX_WORD];
1171 int ord = CAST_ZINT_TO_INT(key.mem[0]);
1173 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1175 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1177 seqno = key.mem[key.len-1];
1179 zebra_term_untrans(zh, index_type, dst_buf, str);
1181 yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
1182 " term=%s", ord, seqno, dst_buf);
1187 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1188 const char *str, int length)
1191 ZebraHandle zh = p->extractCtrl->handle;
1192 ZebraExplainInfo zei = zh->reg->zei;
1195 ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1197 ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1201 key.mem[i++] = p->record_id;
1202 key.mem[i++] = p->section_id;
1204 if (zh->m_segment_indexing)
1205 key.mem[i++] = p->segment;
1206 key.mem[i++] = p->seqno;
1209 zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1212 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1215 ZebraHandle zh = p->extractCtrl->handle;
1216 ZebraExplainInfo zei = zh->reg->zei;
1218 zinfo_index_category_t cat = zinfo_index_category_sort;
1220 ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1222 ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1225 key.mem[1] = p->record_id;
1227 zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1230 static void extract_add_staticrank_string(RecWord *p,
1231 const char *str, int length)
1234 struct recExtractCtrl *ctrl = p->extractCtrl;
1236 if (length > sizeof(valz)-1)
1237 length = sizeof(valz)-1;
1239 memcpy(valz, str, length);
1240 valz[length] = '\0';
1241 ctrl->staticrank = atozint(valz);
1244 static void extract_add_string(RecWord *p, const char *string, int length)
1246 ZebraHandle zh = p->extractCtrl->handle;
1247 assert (length > 0);
1252 if (zebra_maps_is_index(zh->reg->zebra_maps, p->index_type))
1254 extract_add_index_string(p, zinfo_index_category_index,
1256 if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type))
1259 memcpy(&word, p, sizeof(word));
1262 extract_add_index_string(
1263 &word, zinfo_index_category_alwaysmatches, "", 0);
1266 else if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
1268 extract_add_sort_string(p, string, length);
1270 else if (zebra_maps_is_staticrank(zh->reg->zebra_maps, p->index_type))
1272 extract_add_staticrank_string(p, string, length);
1276 static void extract_add_incomplete_field(RecWord *p)
1278 ZebraHandle zh = p->extractCtrl->handle;
1279 const char *b = p->term_buf;
1280 int remain = p->term_len;
1282 const char **map = 0;
1285 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1289 char buf[IT_MAX_WORD+1];
1293 while (map && *map && **map == *CHR_SPACE)
1295 remain = p->term_len - (b - p->term_buf);
1297 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
1305 while (map && *map && **map != *CHR_SPACE)
1307 const char *cp = *map;
1309 while (i < IT_MAX_WORD && *cp)
1311 remain = p->term_len - (b - p->term_buf);
1313 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1323 if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
1325 /* first in field marker */
1326 extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1330 extract_add_string (p, buf, i);
1335 static void extract_add_complete_field (RecWord *p)
1337 ZebraHandle zh = p->extractCtrl->handle;
1338 const char *b = p->term_buf;
1339 char buf[IT_MAX_WORD+1];
1340 const char **map = 0;
1341 int i = 0, remain = p->term_len;
1344 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
1346 while (remain > 0 && i < IT_MAX_WORD)
1348 while (map && *map && **map == *CHR_SPACE)
1350 remain = p->term_len - (b - p->term_buf);
1354 int first = i ? 0 : 1; /* first position */
1355 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, first);
1363 if (i && i < IT_MAX_WORD)
1364 buf[i++] = *CHR_SPACE;
1365 while (map && *map && **map != *CHR_SPACE)
1367 const char *cp = *map;
1369 if (**map == *CHR_CUT)
1375 if (i >= IT_MAX_WORD)
1377 while (i < IT_MAX_WORD && *cp)
1380 remain = p->term_len - (b - p->term_buf);
1383 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
1392 extract_add_string (p, buf, i);
1395 static void extract_token_add(RecWord *p)
1397 ZebraHandle zh = p->extractCtrl->handle;
1400 if (log_level_extract)
1402 yaz_log(log_level_extract, "extract_token_add "
1403 "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
1404 p->index_type, p->index_name,
1405 p->seqno, p->term_len, p->term_buf);
1407 if ((wrbuf = zebra_replace(zh->reg->zebra_maps, p->index_type, 0,
1408 p->term_buf, p->term_len)))
1410 p->term_buf = wrbuf_buf(wrbuf);
1411 p->term_len = wrbuf_len(wrbuf);
1413 if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type))
1414 extract_add_complete_field (p);
1416 extract_add_incomplete_field(p);
1419 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1420 void *buf, size_t sz)
1422 ZebraHandle zh = (ZebraHandle) p->handle;
1424 xfree(zh->store_data_buf);
1425 zh->store_data_buf = 0;
1426 zh->store_data_size = 0;
1429 zh->store_data_buf = xmalloc(sz);
1430 zh->store_data_size = sz;
1431 memcpy(zh->store_data_buf, buf, sz);
1435 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1437 ZebraHandle zh = (ZebraHandle) p->handle;
1438 xfree(zh->store_data_buf);
1439 zh->store_data_buf = 0;
1440 zh->store_data_size = 0;
1441 p->setStoreData = extract_set_store_data_cb;
1444 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
1446 ZebraHandle zh = (ZebraHandle) p->handle;
1447 zebraExplain_addSchema (zh->reg->zei, oid);
1450 void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
1451 int cmd, zebra_rec_keys_t reckeys)
1454 yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT,
1456 extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG);
1459 if (zebra_rec_keys_rewind(reckeys))
1461 zebra_sort_index_t si = zh->reg->sort_index;
1464 struct it_key key_in;
1466 zebra_sort_sysno(si, sysno);
1468 while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1470 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1472 zebra_sort_type(si, ord);
1474 zebra_sort_add(si, str, slen);
1476 zebra_sort_delete(si);
1484 * indent-tabs-mode: nil
1486 * vim: shiftwidth=4 tabstop=8 expandtab