1 /* $Id: extract.c,v 1.260 2007-08-21 13:27:04 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 /* 1 if we use eliminitate identical delete/insert keys */
44 /* eventually this the 0-case code will be removed */
47 void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
48 zebra_rec_keys_t ins_keys,
50 zebra_rec_keys_t del_keys,
53 static void zebra_init_log_level(void)
55 if (!log_level_initialized)
57 log_level_initialized = 1;
59 log_level_extract = yaz_log_module_level("extract");
60 log_level_details = yaz_log_module_level("indexdetails");
64 static void extract_flush_record_keys(ZebraHandle zh, zint sysno,
65 int cmd, zebra_rec_keys_t reckeys,
67 static void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
68 int cmd, zebra_rec_keys_t skp);
69 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid);
70 static void extract_token_add (RecWord *p);
72 static void check_log_limit(ZebraHandle zh)
74 if (zh->records_processed + zh->records_skipped == zh->m_file_verbose_limit)
76 yaz_log(YLOG_LOG, "More than %d file log entries. Omitting rest",
77 zh->m_file_verbose_limit);
81 static void logRecord (ZebraHandle zh)
84 ++zh->records_processed;
85 if (!(zh->records_processed % 1000))
87 yaz_log(YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
88 ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT,
89 zh->records_processed, zh->records_inserted,
90 zh->records_updated, zh->records_deleted);
94 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
97 for (i = 0; i<256; i++)
99 if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
104 ctrl->flagShowRecords = !zh->m_flag_rw;
108 static void extract_add_index_string (RecWord *p,
109 zinfo_index_category_t cat,
110 const char *str, int length);
112 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
114 static void extract_init(struct recExtractCtrl *p, RecWord *w)
117 w->index_name = "any";
125 struct snip_rec_info {
127 zebra_snippets *snippets;
131 static void snippet_add_complete_field(RecWord *p, int ord)
133 struct snip_rec_info *h = p->extractCtrl->handle;
134 ZebraHandle zh = h->zh;
136 const char *b = p->term_buf;
137 char buf[IT_MAX_WORD+1];
138 const char **map = 0;
139 int i = 0, remain = p->term_len;
140 const char *start = b;
141 const char *last = 0;
144 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
146 while (remain > 0 && i < IT_MAX_WORD)
148 while (map && *map && **map == *CHR_SPACE)
150 remain = p->term_len - (b - p->term_buf);
153 start = b; /* set to first non-ws area */
156 int first = i ? 0 : 1; /* first position */
158 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type,
167 if (i && i < IT_MAX_WORD)
168 buf[i++] = *CHR_SPACE;
169 while (map && *map && **map != *CHR_SPACE)
171 const char *cp = *map;
173 if (**map == *CHR_CUT)
179 if (i >= IT_MAX_WORD)
181 while (i < IT_MAX_WORD && *cp)
185 remain = p->term_len - (b - p->term_buf);
188 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
197 if (last && start != last)
198 zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
199 start, last - start);
202 static void snippet_add_incomplete_field(RecWord *p, int ord)
204 struct snip_rec_info *h = p->extractCtrl->handle;
205 ZebraHandle zh = h->zh;
206 const char *b = p->term_buf;
207 int remain = p->term_len;
209 const char **map = 0;
210 const char *start = b;
213 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
217 char buf[IT_MAX_WORD+1];
218 const char *last = b;
222 while (map && *map && **map == *CHR_SPACE)
224 remain = p->term_len - (b - p->term_buf);
227 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
236 zebra_snippets_appendn(h->snippets, p->seqno, 1, ord,
237 start, last - start);
243 while (map && *map && **map != *CHR_SPACE)
245 const char *cp = *map;
247 while (i < IT_MAX_WORD && *cp)
249 remain = p->term_len - (b - p->term_buf);
252 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
262 if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
264 /* first in field marker */
269 zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
270 start, last - start);
277 static void snippet_token_add(RecWord *p)
279 struct snip_rec_info *h = p->extractCtrl->handle;
280 ZebraHandle zh = h->zh;
282 if (zebra_maps_is_index(zh->reg->zebra_maps, p->index_type))
284 ZebraExplainInfo zei = zh->reg->zei;
285 int ch = zebraExplain_lookup_attr_str(
286 zei, zinfo_index_category_index, p->index_type, p->index_name);
288 if (zebra_maps_is_complete (h->zh->reg->zebra_maps, p->index_type))
289 snippet_add_complete_field (p, ch);
291 snippet_add_incomplete_field(p, ch);
295 static void snippet_schema_add(
296 struct recExtractCtrl *p, Odr_oid *oid)
301 void extract_snippet(ZebraHandle zh, zebra_snippets *sn,
302 struct ZebraRecStream *stream,
303 RecType rt, void *recTypeClientData)
305 struct recExtractCtrl extractCtrl;
306 struct snip_rec_info info;
309 extractCtrl.stream = stream;
310 extractCtrl.first_record = 1;
311 extractCtrl.init = extract_init;
312 extractCtrl.tokenAdd = snippet_token_add;
313 extractCtrl.schemaAdd = snippet_schema_add;
317 extractCtrl.dh = zh->reg->dh;
321 extractCtrl.handle = &info;
322 extractCtrl.match_criteria[0] = '\0';
323 extractCtrl.staticrank = 0;
324 extractCtrl.action = action_insert;
326 init_extractCtrl(zh, &extractCtrl);
328 r = (*rt->extract)(recTypeClientData, &extractCtrl);
332 static void searchRecordKey(ZebraHandle zh,
333 zebra_rec_keys_t reckeys,
334 const char *index_name,
335 const char **ws, int ws_length)
339 zinfo_index_category_t cat = zinfo_index_category_index;
341 for (i = 0; i<ws_length; i++)
345 ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, '0', index_name);
347 ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'p', index_name);
349 ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'w', index_name);
354 if (zebra_rec_keys_rewind(reckeys))
361 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
363 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
365 seqno = key.mem[key.len-1];
367 if (key.mem[0] == ch)
373 woff = seqno - startSeq;
374 if (woff >= 0 && woff < ws_length)
381 #define FILE_MATCH_BLANK "\t "
383 static char *get_match_from_spec(ZebraHandle zh,
384 zebra_rec_keys_t reckeys,
385 const char *fname, const char *spec)
387 static char dstBuf[2048]; /* static here ??? */
389 const char *s = spec;
393 for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
400 char attset_str[64], attname_str[64];
404 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
406 for (i = 0; *s && *s != ',' && *s != ')' &&
407 !strchr(FILE_MATCH_BLANK, *s); s++)
408 if (i+1 < sizeof(attset_str))
409 attset_str[i++] = *s;
410 attset_str[i] = '\0';
412 for (; strchr(FILE_MATCH_BLANK, *s); s++)
415 strcpy(attname_str, attset_str);
418 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
420 for (i = 0; *s && *s != ')' &&
421 !strchr(FILE_MATCH_BLANK, *s); s++)
422 if (i+1 < sizeof(attname_str))
423 attname_str[i++] = *s;
424 attname_str[i] = '\0';
427 searchRecordKey (zh, reckeys, attname_str, ws, 32);
431 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
432 spec, zh->m_group ? zh->m_group : "none");
437 for (i = 0; i<32; i++)
446 dst += strlen(ws[i]);
450 yaz_log (YLOG_WARN, "Record didn't contain match"
451 " fields in (%s,%s)", attset_str, attname_str);
459 const char *spec_src = NULL;
460 const char *s1 = ++s;
461 while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
465 if (spec_len > sizeof(special)-1)
466 spec_len = sizeof(special)-1;
467 memcpy (special, s, spec_len);
468 special[spec_len] = '\0';
471 if (!strcmp (special, "group"))
472 spec_src = zh->m_group;
473 else if (!strcmp (special, "database"))
474 spec_src = zh->basenames[0];
475 else if (!strcmp (special, "filename")) {
478 else if (!strcmp (special, "type"))
479 spec_src = zh->m_record_type;
484 strcpy (dst, spec_src);
485 dst += strlen (spec_src);
488 else if (*s == '\"' || *s == '\'')
490 int stopMarker = *s++;
494 while (*s && *s != stopMarker)
496 if (i+1 < sizeof(tmpString))
497 tmpString[i++] = *s++;
502 strcpy (dst, tmpString);
503 dst += strlen (tmpString);
507 yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
508 spec, zh->m_group ? zh->m_group : "none");
515 yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
516 fname, zh->m_group ? zh->m_group : "none");
523 struct recordLogInfo {
526 struct recordGroup *rGroup;
529 static void all_matches_add(struct recExtractCtrl *ctrl)
532 extract_init(ctrl, &word);
533 word.index_name = "_ALLRECORDS";
534 word.index_type = 'w';
536 extract_add_index_string (&word, zinfo_index_category_alwaysmatches,
540 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
541 struct ZebraRecStream *stream,
542 enum zebra_recctrl_action_t action,
544 const char *recordType,
546 const char *match_criteria,
549 void *recTypeClientData);
552 ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
555 ZEBRA_RES r = ZEBRA_OK;
560 struct file_read_info *fi = 0;
561 const char *original_record_type = 0;
563 void *recTypeClientData;
564 struct ZebraRecStream stream, *streamp;
566 zebra_init_log_level();
568 if (!zh->m_group || !*zh->m_group)
571 sprintf (gprefix, "%s.", zh->m_group);
573 yaz_log(log_level_extract, "zebra_extract_file %s", fname);
575 /* determine file extension */
577 for (i = strlen(fname); --i >= 0; )
580 else if (fname[i] == '.')
582 strcpy (ext, fname+i+1);
585 /* determine file type - depending on extension */
586 original_record_type = zh->m_record_type;
587 if (!zh->m_record_type)
589 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
590 zh->m_record_type = res_get (zh->res, ext_res);
592 if (!zh->m_record_type)
595 if (zh->records_processed + zh->records_skipped
596 < zh->m_file_verbose_limit)
597 yaz_log (YLOG_LOG, "? %s", fname);
598 zh->records_skipped++;
601 /* determine match criteria */
602 if (!zh->m_record_id)
604 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
605 zh->m_record_id = res_get (zh->res, ext_res);
609 recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
610 &recTypeClientData)))
612 yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
616 switch(recType->version)
621 yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
623 if (sysno && deleteFlag)
632 if (zh->path_reg && !yaz_is_abspath (fname))
634 strcpy (full_rep, zh->path_reg);
635 strcat (full_rep, "/");
636 strcat (full_rep, fname);
639 strcpy (full_rep, fname);
641 if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
643 yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
644 zh->m_record_type = original_record_type;
648 zebra_create_stream_fd(streamp, fd, 0);
650 r = zebra_extract_records_stream(zh, streamp,
652 action_delete : action_update,
656 0, /*match_criteria */
658 recType, recTypeClientData);
660 stream.destroy(streamp);
661 zh->m_record_type = original_record_type;
666 If sysno is provided, then it's used to identify the reocord.
667 If not, and match_criteria is provided, then sysno is guessed
668 If not, and a record is provided, then sysno is got from there
672 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
673 const char *buf, size_t buf_size,
674 enum zebra_recctrl_action_t action,
676 const char *recordType,
678 const char *match_criteria,
681 struct ZebraRecStream stream;
686 if (recordType && *recordType)
688 yaz_log(log_level_extract,
689 "Record type explicitly specified: %s", recordType);
690 recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
695 if (!(zh->m_record_type))
697 yaz_log (YLOG_WARN, "No such record type defined");
700 yaz_log(log_level_extract, "Get record type from rgroup: %s",
702 recType = recType_byName (zh->reg->recTypes, zh->res,
703 zh->m_record_type, &clientData);
704 recordType = zh->m_record_type;
709 yaz_log (YLOG_WARN, "No such record type: %s", recordType);
713 zebra_create_stream_mem(&stream, buf, buf_size);
715 res = zebra_extract_records_stream(zh, &stream,
722 recType, clientData);
723 stream.destroy(&stream);
727 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
728 struct ZebraRecStream *stream,
729 enum zebra_recctrl_action_t action,
731 const char *recordType,
733 const char *match_criteria,
736 void *recTypeClientData)
738 ZEBRA_RES res = ZEBRA_OK;
742 res = zebra_extract_record_stream(zh, stream,
749 recType, recTypeClientData, &more);
764 static WRBUF wrbuf_hex_str(const char *cstr)
767 WRBUF w = wrbuf_alloc();
768 for (i = 0; cstr[i]; i++)
770 if (cstr[i] < ' ' || cstr[i] > 126)
771 wrbuf_printf(w, "\\%02X", cstr[i] & 0xff);
773 wrbuf_putc(w, cstr[i]);
778 ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
779 struct ZebraRecStream *stream,
780 enum zebra_recctrl_action_t action,
782 const char *recordType,
784 const char *match_criteria,
787 void *recTypeClientData,
792 RecordAttr *recordAttr;
793 struct recExtractCtrl extractCtrl;
795 const char *matchStr = 0;
797 off_t start_offset = 0, end_offset = 0;
798 const char *pr_fname = fname; /* filename to print .. */
799 int show_progress = zh->records_processed + zh->records_skipped
800 < zh->m_file_verbose_limit ? 1:0;
802 zebra_init_log_level();
805 pr_fname = "<no file>"; /* make it printable if file is omitted */
807 zebra_rec_keys_reset(zh->reg->keys);
808 zebra_rec_keys_reset(zh->reg->sortKeys);
810 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
812 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
813 zh->m_explain_database))
819 off_t null_offset = 0;
820 extractCtrl.stream = stream;
822 start_offset = stream->tellf(stream);
824 extractCtrl.first_record = start_offset ? 0 : 1;
826 stream->endf(stream, &null_offset);;
828 extractCtrl.init = extract_init;
829 extractCtrl.tokenAdd = extract_token_add;
830 extractCtrl.schemaAdd = extract_schema_add;
831 extractCtrl.dh = zh->reg->dh;
832 extractCtrl.handle = zh;
833 extractCtrl.match_criteria[0] = '\0';
834 extractCtrl.staticrank = 0;
835 extractCtrl.action = action;
837 init_extractCtrl(zh, &extractCtrl);
839 extract_set_store_data_prepare(&extractCtrl);
841 r = (*recType->extract)(recTypeClientData, &extractCtrl);
843 if (action == action_update)
845 action = extractCtrl.action;
850 case RECCTRL_EXTRACT_EOF:
852 case RECCTRL_EXTRACT_ERROR_GENERIC:
853 /* error occured during extraction ... */
854 yaz_log (YLOG_WARN, "extract error: generic");
856 case RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER:
857 /* error occured during extraction ... */
858 yaz_log (YLOG_WARN, "extract error: no such filter");
860 case RECCTRL_EXTRACT_SKIP:
862 yaz_log (YLOG_LOG, "skip %s %s " ZINT_FORMAT,
863 recordType, pr_fname, (zint) start_offset);
866 end_offset = stream->endf(stream, 0);
868 stream->seekf(stream, end_offset);
871 case RECCTRL_EXTRACT_OK:
874 yaz_log (YLOG_WARN, "extract error: unknown error: %d", r);
877 end_offset = stream->endf(stream, 0);
879 stream->seekf(stream, end_offset);
881 end_offset = stream->tellf(stream);
883 all_matches_add(&extractCtrl);
885 if (extractCtrl.match_criteria[0])
886 match_criteria = extractCtrl.match_criteria;
894 if (match_criteria && *match_criteria) {
895 matchStr = match_criteria;
897 if (zh->m_record_id && *zh->m_record_id) {
898 matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname,
902 yaz_log (YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
903 pr_fname, (zint) start_offset);
910 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
911 char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
915 if (log_level_extract)
917 WRBUF w = wrbuf_hex_str(matchStr);
918 yaz_log(log_level_extract, "matchStr: %s", wrbuf_cstr(w));
923 assert(*rinfo == sizeof(*sysno));
924 memcpy (sysno, rinfo+1, sizeof(*sysno));
928 if (zebra_rec_keys_empty(zh->reg->keys))
930 /* the extraction process returned no information - the record
931 is probably empty - unless flagShowRecords is in use */
939 if (action == action_delete)
941 yaz_log (YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
942 pr_fname, (zint) start_offset);
943 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
946 else if (action == action_replace)
948 yaz_log (YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
949 pr_fname, (zint) start_offset);
950 yaz_log (YLOG_WARN, "cannot update record above (seems new)");
954 yaz_log (YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
955 (zint) start_offset);
956 rec = rec_new (zh->reg->records);
960 recordAttr = rec_init_attr (zh->reg->zei, rec);
961 if (extractCtrl.staticrank < 0)
963 yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
964 extractCtrl.staticrank = 0;
969 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
970 dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
971 sizeof(*sysno), sysno);
974 extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
976 extract_flush_record_keys2(zh, *sysno,
977 zh->reg->keys, extractCtrl.staticrank,
978 0, recordAttr->staticrank);
980 extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
981 extractCtrl.staticrank);
983 recordAttr->staticrank = extractCtrl.staticrank;
984 zh->records_inserted++;
988 /* record already exists */
989 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
990 zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
991 if (action == action_insert)
993 yaz_log (YLOG_LOG, "skipped %s %s " ZINT_FORMAT,
994 recordType, pr_fname, (zint) start_offset);
999 rec = rec_get (zh->reg->records, *sysno);
1002 recordAttr = rec_init_attr (zh->reg->zei, rec);
1004 /* decrease total size */
1005 zebraExplain_recordBytesIncrement (zh->reg->zei,
1006 - recordAttr->recordSize);
1008 zebra_rec_keys_set_buf(delkeys,
1009 rec->info[recInfo_delKeys],
1010 rec->size[recInfo_delKeys],
1012 zebra_rec_keys_set_buf(sortKeys,
1013 rec->info[recInfo_sortKeys],
1014 rec->size[recInfo_sortKeys],
1017 extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
1019 extract_flush_record_keys(zh, *sysno, 0, delkeys,
1020 recordAttr->staticrank);
1022 if (action == action_delete)
1024 /* record going to be deleted */
1026 extract_flush_record_keys2(zh, *sysno, 0, recordAttr->staticrank,
1027 delkeys, recordAttr->staticrank);
1029 if (zebra_rec_keys_empty(delkeys))
1031 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
1032 pr_fname, (zint) start_offset);
1033 yaz_log(YLOG_WARN, "cannot delete file above, "
1034 "storeKeys false (3)");
1039 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
1040 pr_fname, (zint) start_offset);
1041 zh->records_deleted++;
1044 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
1045 dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
1047 rec_del (zh->reg->records, &rec);
1049 zebra_rec_keys_close(delkeys);
1050 zebra_rec_keys_close(sortKeys);
1056 { /* update or special_update */
1058 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
1059 pr_fname, (zint) start_offset);
1060 extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
1063 extract_flush_record_keys2(zh, *sysno,
1064 zh->reg->keys, extractCtrl.staticrank,
1065 delkeys, recordAttr->staticrank);
1067 extract_flush_record_keys(zh, *sysno, 1,
1068 zh->reg->keys, extractCtrl.staticrank);
1070 recordAttr->staticrank = extractCtrl.staticrank;
1071 zh->records_updated++;
1073 zebra_rec_keys_close(delkeys);
1074 zebra_rec_keys_close(sortKeys);
1076 /* update file type */
1077 xfree (rec->info[recInfo_fileType]);
1078 rec->info[recInfo_fileType] =
1079 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1081 /* update filename */
1082 xfree (rec->info[recInfo_filename]);
1083 rec->info[recInfo_filename] =
1084 rec_strdup (fname, &rec->size[recInfo_filename]);
1086 /* update delete keys */
1087 xfree (rec->info[recInfo_delKeys]);
1088 if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
1090 zebra_rec_keys_get_buf(zh->reg->keys,
1091 &rec->info[recInfo_delKeys],
1092 &rec->size[recInfo_delKeys]);
1096 rec->info[recInfo_delKeys] = NULL;
1097 rec->size[recInfo_delKeys] = 0;
1099 /* update sort keys */
1100 xfree (rec->info[recInfo_sortKeys]);
1102 zebra_rec_keys_get_buf(zh->reg->sortKeys,
1103 &rec->info[recInfo_sortKeys],
1104 &rec->size[recInfo_sortKeys]);
1108 recordAttr->recordSize = end_offset - start_offset;
1109 zebraExplain_recordBytesIncrement(zh->reg->zei,
1110 recordAttr->recordSize);
1113 /* set run-number for this record */
1114 recordAttr->runNumber =
1115 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
1117 /* update store data */
1118 xfree (rec->info[recInfo_storeData]);
1120 /* update store data */
1121 if (zh->store_data_buf)
1123 rec->size[recInfo_storeData] = zh->store_data_size;
1124 rec->info[recInfo_storeData] = zh->store_data_buf;
1125 zh->store_data_buf = 0;
1126 recordAttr->recordSize = zh->store_data_size;
1128 else if (zh->m_store_data)
1130 off_t cur_offset = stream->tellf(stream);
1132 rec->size[recInfo_storeData] = recordAttr->recordSize;
1133 rec->info[recInfo_storeData] = (char *)
1134 xmalloc (recordAttr->recordSize);
1135 stream->seekf(stream, start_offset);
1136 stream->readf(stream, rec->info[recInfo_storeData],
1137 recordAttr->recordSize);
1138 stream->seekf(stream, cur_offset);
1142 rec->info[recInfo_storeData] = NULL;
1143 rec->size[recInfo_storeData] = 0;
1145 /* update database name */
1146 xfree (rec->info[recInfo_databaseName]);
1147 rec->info[recInfo_databaseName] =
1148 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
1151 recordAttr->recordOffset = start_offset;
1153 /* commit this record */
1154 rec_put (zh->reg->records, &rec);
1159 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
1161 ZebraHandle zh = (ZebraHandle) handle;
1162 struct recExtractCtrl extractCtrl;
1164 if (zebraExplain_curDatabase (zh->reg->zei,
1165 rec->info[recInfo_databaseName]))
1168 if (zebraExplain_newDatabase (zh->reg->zei,
1169 rec->info[recInfo_databaseName], 0))
1173 zebra_rec_keys_reset(zh->reg->keys);
1174 zebra_rec_keys_reset(zh->reg->sortKeys);
1176 extractCtrl.init = extract_init;
1177 extractCtrl.tokenAdd = extract_token_add;
1178 extractCtrl.schemaAdd = extract_schema_add;
1179 extractCtrl.dh = zh->reg->dh;
1181 init_extractCtrl(zh, &extractCtrl);
1183 extractCtrl.flagShowRecords = 0;
1184 extractCtrl.match_criteria[0] = '\0';
1185 extractCtrl.staticrank = 0;
1186 extractCtrl.action = action_update;
1188 extractCtrl.handle = handle;
1189 extractCtrl.first_record = 1;
1191 extract_set_store_data_prepare(&extractCtrl);
1194 grs_extract_tree(&extractCtrl, n);
1196 if (rec->size[recInfo_delKeys])
1198 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1200 zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
1202 zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
1203 rec->size[recInfo_delKeys],
1206 extract_flush_record_keys2(zh, rec->sysno,
1207 zh->reg->keys, 0, delkeys, 0);
1209 extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
1210 extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
1212 zebra_rec_keys_close(delkeys);
1214 zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
1215 rec->size[recInfo_sortKeys],
1218 extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
1219 zebra_rec_keys_close(sortkeys);
1224 extract_flush_record_keys2(zh, rec->sysno, zh->reg->keys, 0, 0, 0);
1226 extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
1229 extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
1231 xfree (rec->info[recInfo_delKeys]);
1232 zebra_rec_keys_get_buf(zh->reg->keys,
1233 &rec->info[recInfo_delKeys],
1234 &rec->size[recInfo_delKeys]);
1236 xfree (rec->info[recInfo_sortKeys]);
1237 zebra_rec_keys_get_buf(zh->reg->sortKeys,
1238 &rec->info[recInfo_sortKeys],
1239 &rec->size[recInfo_sortKeys]);
1243 void extract_rec_keys_log(ZebraHandle zh, int is_insert,
1244 zebra_rec_keys_t reckeys,
1247 if (zebra_rec_keys_rewind(reckeys))
1252 NMEM nmem = nmem_create();
1254 while(zebra_rec_keys_read(reckeys, &str, &slen, &key))
1256 char keystr[200]; /* room for zints to print */
1258 int ord = CAST_ZINT_TO_INT(key.mem[0]);
1260 const char *string_index;
1262 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1263 0/* db */, &string_index);
1265 zebra_term_untrans_iconv(zh, nmem, index_type,
1268 for (i = 0; i<key.len; i++)
1270 sprintf(keystr + strlen(keystr), ZINT_FORMAT " ", key.mem[i]);
1273 if (*str < CHR_BASE_CHAR)
1276 char dst_buf[200]; /* room for special chars */
1278 strcpy(dst_buf , "?");
1280 if (!strcmp(str, ""))
1281 strcpy(dst_buf, "alwaysmatches");
1282 if (!strcmp(str, FIRST_IN_FIELD_STR))
1283 strcpy(dst_buf, "firstinfield");
1284 else if (!strcmp(str, CHR_UNKNOWN))
1285 strcpy(dst_buf, "unknown");
1286 else if (!strcmp(str, CHR_SPACE))
1287 strcpy(dst_buf, "space");
1289 for (i = 0; i<slen; i++)
1291 sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
1293 yaz_log(level, "%s%c %s %s", keystr, index_type,
1294 string_index, dst_buf);
1298 yaz_log(level, "%s%c %s \"%s\"", keystr, index_type,
1299 string_index, dst_term);
1307 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
1308 zebra_rec_keys_t reckeys)
1310 ZebraExplainInfo zei = zh->reg->zei;
1314 struct ord_stat *next;
1317 if (zebra_rec_keys_rewind(reckeys))
1319 struct ord_stat *ord_list = 0;
1323 struct it_key key_in;
1324 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1326 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1328 for (p = ord_list; p ; p = p->next)
1336 p = xmalloc(sizeof(*p));
1347 struct ord_stat *p1 = p;
1350 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
1352 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
1359 void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
1360 zebra_rec_keys_t ins_keys, zint ins_rank,
1361 zebra_rec_keys_t del_keys, zint del_rank)
1363 ZebraExplainInfo zei = zh->reg->zei;
1367 if (!zh->reg->key_block)
1369 int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1370 const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", ".");
1371 int use_threads = atoi(res_get_def (zh->res, "threads", "1"));
1372 zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1377 extract_rec_keys_adjust(zh, 1, ins_keys);
1379 zebraExplain_recordCountIncrement (zei, 1);
1380 zebra_rec_keys_rewind(ins_keys);
1384 extract_rec_keys_adjust(zh, 0, del_keys);
1386 zebraExplain_recordCountIncrement (zei, -1);
1387 zebra_rec_keys_rewind(del_keys);
1393 const char *del_str;
1394 struct it_key del_key_in;
1398 const char *ins_str;
1399 struct it_key ins_key_in;
1403 del = zebra_rec_keys_read(del_keys, &del_str, &del_slen,
1406 ins = zebra_rec_keys_read(ins_keys, &ins_str, &ins_slen,
1409 if (del && ins && ins_rank == del_rank
1410 && !key_compare(&del_key_in, &ins_key_in)
1411 && ins_slen == del_slen && !memcmp(del_str, ins_str, del_slen))
1421 key_block_write(zh->reg->key_block, sysno,
1422 &del_key_in, 0, del_str, del_slen,
1423 del_rank, zh->m_staticrank);
1425 key_block_write(zh->reg->key_block, sysno,
1426 &ins_key_in, 1, ins_str, ins_slen,
1427 ins_rank, zh->m_staticrank);
1429 yaz_log(log_level_extract, "normal=%d optimized=%d", normal, optimized);
1432 void extract_flush_record_keys(ZebraHandle zh, zint sysno, int cmd,
1433 zebra_rec_keys_t reckeys,
1436 ZebraExplainInfo zei = zh->reg->zei;
1438 extract_rec_keys_adjust(zh, cmd, reckeys);
1440 if (log_level_details)
1442 yaz_log(log_level_details, "Keys for record " ZINT_FORMAT " %s",
1443 sysno, cmd ? "insert" : "delete");
1444 extract_rec_keys_log(zh, cmd, reckeys, log_level_details);
1447 if (!zh->reg->key_block)
1449 int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1450 const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", ".");
1451 int use_threads = atoi(res_get_def (zh->res, "threads", "1"));
1452 zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1454 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1457 yaz_log(YLOG_LOG, "sysno=" ZINT_FORMAT " cmd=%d", sysno, cmd);
1458 print_rec_keys(zh, reckeys);
1460 if (zebra_rec_keys_rewind(reckeys))
1464 struct it_key key_in;
1465 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1467 key_block_write(zh->reg->key_block, sysno,
1468 &key_in, cmd, str, slen,
1469 staticrank, zh->m_staticrank);
1474 ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
1475 zebra_rec_keys_t reckeys,
1476 zebra_snippets *snippets)
1478 NMEM nmem = nmem_create();
1479 if (zebra_rec_keys_rewind(reckeys))
1484 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1491 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1492 seqno = key.mem[key.len-1];
1493 ord = CAST_ZINT_TO_INT(key.mem[0]);
1495 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1496 0/* db */, 0 /* string_index */);
1498 zebra_term_untrans_iconv(zh, nmem, index_type,
1500 zebra_snippets_append(snippets, seqno, 0, ord, dst_term);
1508 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1510 yaz_log(YLOG_LOG, "print_rec_keys");
1511 if (zebra_rec_keys_rewind(reckeys))
1516 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1518 char dst_buf[IT_MAX_WORD];
1521 int ord = CAST_ZINT_TO_INT(key.mem[0]);
1523 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1525 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1527 seqno = key.mem[key.len-1];
1529 zebra_term_untrans(zh, index_type, dst_buf, str);
1531 yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
1532 " term=%s", ord, seqno, dst_buf);
1537 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1538 const char *str, int length)
1541 ZebraHandle zh = p->extractCtrl->handle;
1542 ZebraExplainInfo zei = zh->reg->zei;
1545 ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1547 ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1551 key.mem[i++] = p->record_id;
1552 key.mem[i++] = p->section_id;
1554 if (zh->m_segment_indexing)
1555 key.mem[i++] = p->segment;
1556 key.mem[i++] = p->seqno;
1559 zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1562 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1565 ZebraHandle zh = p->extractCtrl->handle;
1566 ZebraExplainInfo zei = zh->reg->zei;
1568 zinfo_index_category_t cat = zinfo_index_category_sort;
1570 ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1572 ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1575 key.mem[1] = p->record_id;
1577 zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1580 static void extract_add_staticrank_string(RecWord *p,
1581 const char *str, int length)
1584 struct recExtractCtrl *ctrl = p->extractCtrl;
1586 if (length > sizeof(valz)-1)
1587 length = sizeof(valz)-1;
1589 memcpy(valz, str, length);
1590 valz[length] = '\0';
1591 ctrl->staticrank = atozint(valz);
1594 static void extract_add_string(RecWord *p, const char *string, int length)
1596 ZebraHandle zh = p->extractCtrl->handle;
1597 assert (length > 0);
1602 if (zebra_maps_is_index(zh->reg->zebra_maps, p->index_type))
1604 extract_add_index_string(p, zinfo_index_category_index,
1606 if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type))
1609 memcpy(&word, p, sizeof(word));
1612 extract_add_index_string(
1613 &word, zinfo_index_category_alwaysmatches, "", 0);
1616 else if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
1618 extract_add_sort_string(p, string, length);
1620 else if (zebra_maps_is_staticrank(zh->reg->zebra_maps, p->index_type))
1622 extract_add_staticrank_string(p, string, length);
1626 static void extract_add_incomplete_field(RecWord *p)
1628 ZebraHandle zh = p->extractCtrl->handle;
1629 const char *b = p->term_buf;
1630 int remain = p->term_len;
1632 const char **map = 0;
1635 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1639 char buf[IT_MAX_WORD+1];
1643 while (map && *map && **map == *CHR_SPACE)
1645 remain = p->term_len - (b - p->term_buf);
1647 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
1655 while (map && *map && **map != *CHR_SPACE)
1657 const char *cp = *map;
1659 while (i < IT_MAX_WORD && *cp)
1661 remain = p->term_len - (b - p->term_buf);
1663 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1673 if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
1675 /* first in field marker */
1676 extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1680 extract_add_string (p, buf, i);
1685 static void extract_add_complete_field (RecWord *p)
1687 ZebraHandle zh = p->extractCtrl->handle;
1688 const char *b = p->term_buf;
1689 char buf[IT_MAX_WORD+1];
1690 const char **map = 0;
1691 int i = 0, remain = p->term_len;
1694 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
1696 while (remain > 0 && i < IT_MAX_WORD)
1698 while (map && *map && **map == *CHR_SPACE)
1700 remain = p->term_len - (b - p->term_buf);
1704 int first = i ? 0 : 1; /* first position */
1705 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, first);
1713 if (i && i < IT_MAX_WORD)
1714 buf[i++] = *CHR_SPACE;
1715 while (map && *map && **map != *CHR_SPACE)
1717 const char *cp = *map;
1719 if (**map == *CHR_CUT)
1725 if (i >= IT_MAX_WORD)
1727 while (i < IT_MAX_WORD && *cp)
1730 remain = p->term_len - (b - p->term_buf);
1733 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
1742 extract_add_string (p, buf, i);
1745 static void extract_token_add(RecWord *p)
1747 ZebraHandle zh = p->extractCtrl->handle;
1750 if (log_level_details)
1752 yaz_log(log_level_details, "extract_token_add "
1753 "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
1754 p->index_type, p->index_name,
1755 p->seqno, p->term_len, p->term_buf);
1757 if ((wrbuf = zebra_replace(zh->reg->zebra_maps, p->index_type, 0,
1758 p->term_buf, p->term_len)))
1760 p->term_buf = wrbuf_buf(wrbuf);
1761 p->term_len = wrbuf_len(wrbuf);
1763 if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type))
1764 extract_add_complete_field (p);
1766 extract_add_incomplete_field(p);
1769 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1770 void *buf, size_t sz)
1772 ZebraHandle zh = (ZebraHandle) p->handle;
1774 xfree(zh->store_data_buf);
1775 zh->store_data_buf = 0;
1776 zh->store_data_size = 0;
1779 zh->store_data_buf = xmalloc(sz);
1780 zh->store_data_size = sz;
1781 memcpy(zh->store_data_buf, buf, sz);
1785 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1787 ZebraHandle zh = (ZebraHandle) p->handle;
1788 xfree(zh->store_data_buf);
1789 zh->store_data_buf = 0;
1790 zh->store_data_size = 0;
1791 p->setStoreData = extract_set_store_data_cb;
1794 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
1796 ZebraHandle zh = (ZebraHandle) p->handle;
1797 zebraExplain_addSchema (zh->reg->zei, oid);
1800 void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
1801 int cmd, zebra_rec_keys_t reckeys)
1804 yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT,
1806 extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG);
1809 if (zebra_rec_keys_rewind(reckeys))
1811 zebra_sort_index_t si = zh->reg->sort_index;
1814 struct it_key key_in;
1816 zebra_sort_sysno(si, sysno);
1818 while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1820 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1822 zebra_sort_type(si, ord);
1824 zebra_sort_add(si, str, slen);
1826 zebra_sort_delete(si);
1834 * indent-tabs-mode: nil
1836 * vim: shiftwidth=4 tabstop=8 expandtab