1 /* $Id: extract.c,v 1.259 2007-08-21 11:06:47 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)
136 static void snippet_add_incomplete_field(RecWord *p, int ord)
138 struct snip_rec_info *h = p->extractCtrl->handle;
139 ZebraHandle zh = h->zh;
140 const char *b = p->term_buf;
141 int remain = p->term_len;
143 const char **map = 0;
144 const char *start = b;
147 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
151 char buf[IT_MAX_WORD+1];
152 const char *last = b;
156 while (map && *map && **map == *CHR_SPACE)
158 remain = p->term_len - (b - p->term_buf);
161 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
170 zebra_snippets_appendn(h->snippets, p->seqno, 1, ord,
171 start, last - start);
177 while (map && *map && **map != *CHR_SPACE)
179 const char *cp = *map;
181 while (i < IT_MAX_WORD && *cp)
183 remain = p->term_len - (b - p->term_buf);
186 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
196 if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
198 /* first in field marker */
203 zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
204 start, last - start);
211 static void snippet_token_add(RecWord *p)
213 struct snip_rec_info *h = p->extractCtrl->handle;
214 ZebraHandle zh = h->zh;
216 if (zebra_maps_is_index(zh->reg->zebra_maps, p->index_type))
218 ZebraExplainInfo zei = zh->reg->zei;
219 int ch = zebraExplain_lookup_attr_str(
220 zei, zinfo_index_category_index, p->index_type, p->index_name);
222 if (zebra_maps_is_complete (h->zh->reg->zebra_maps, p->index_type))
223 snippet_add_complete_field (p);
225 snippet_add_incomplete_field(p, ch);
229 static void snippet_schema_add(
230 struct recExtractCtrl *p, Odr_oid *oid)
235 void extract_snippet(ZebraHandle zh, zebra_snippets *sn,
236 struct ZebraRecStream *stream,
237 RecType rt, void *recTypeClientData)
239 struct recExtractCtrl extractCtrl;
240 struct snip_rec_info info;
243 extractCtrl.stream = stream;
244 extractCtrl.first_record = 1;
245 extractCtrl.init = extract_init;
246 extractCtrl.tokenAdd = snippet_token_add;
247 extractCtrl.schemaAdd = snippet_schema_add;
251 extractCtrl.dh = zh->reg->dh;
255 extractCtrl.handle = &info;
256 extractCtrl.match_criteria[0] = '\0';
257 extractCtrl.staticrank = 0;
258 extractCtrl.action = action_insert;
260 init_extractCtrl(zh, &extractCtrl);
262 r = (*rt->extract)(recTypeClientData, &extractCtrl);
266 static void searchRecordKey(ZebraHandle zh,
267 zebra_rec_keys_t reckeys,
268 const char *index_name,
269 const char **ws, int ws_length)
273 zinfo_index_category_t cat = zinfo_index_category_index;
275 for (i = 0; i<ws_length; i++)
279 ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, '0', index_name);
281 ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'p', index_name);
283 ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'w', index_name);
288 if (zebra_rec_keys_rewind(reckeys))
295 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
297 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
299 seqno = key.mem[key.len-1];
301 if (key.mem[0] == ch)
307 woff = seqno - startSeq;
308 if (woff >= 0 && woff < ws_length)
315 #define FILE_MATCH_BLANK "\t "
317 static char *get_match_from_spec(ZebraHandle zh,
318 zebra_rec_keys_t reckeys,
319 const char *fname, const char *spec)
321 static char dstBuf[2048]; /* static here ??? */
323 const char *s = spec;
327 for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
334 char attset_str[64], attname_str[64];
338 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
340 for (i = 0; *s && *s != ',' && *s != ')' &&
341 !strchr(FILE_MATCH_BLANK, *s); s++)
342 if (i+1 < sizeof(attset_str))
343 attset_str[i++] = *s;
344 attset_str[i] = '\0';
346 for (; strchr(FILE_MATCH_BLANK, *s); s++)
349 strcpy(attname_str, attset_str);
352 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
354 for (i = 0; *s && *s != ')' &&
355 !strchr(FILE_MATCH_BLANK, *s); s++)
356 if (i+1 < sizeof(attname_str))
357 attname_str[i++] = *s;
358 attname_str[i] = '\0';
361 searchRecordKey (zh, reckeys, attname_str, ws, 32);
365 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
366 spec, zh->m_group ? zh->m_group : "none");
371 for (i = 0; i<32; i++)
380 dst += strlen(ws[i]);
384 yaz_log (YLOG_WARN, "Record didn't contain match"
385 " fields in (%s,%s)", attset_str, attname_str);
393 const char *spec_src = NULL;
394 const char *s1 = ++s;
395 while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
399 if (spec_len > sizeof(special)-1)
400 spec_len = sizeof(special)-1;
401 memcpy (special, s, spec_len);
402 special[spec_len] = '\0';
405 if (!strcmp (special, "group"))
406 spec_src = zh->m_group;
407 else if (!strcmp (special, "database"))
408 spec_src = zh->basenames[0];
409 else if (!strcmp (special, "filename")) {
412 else if (!strcmp (special, "type"))
413 spec_src = zh->m_record_type;
418 strcpy (dst, spec_src);
419 dst += strlen (spec_src);
422 else if (*s == '\"' || *s == '\'')
424 int stopMarker = *s++;
428 while (*s && *s != stopMarker)
430 if (i+1 < sizeof(tmpString))
431 tmpString[i++] = *s++;
436 strcpy (dst, tmpString);
437 dst += strlen (tmpString);
441 yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
442 spec, zh->m_group ? zh->m_group : "none");
449 yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
450 fname, zh->m_group ? zh->m_group : "none");
457 struct recordLogInfo {
460 struct recordGroup *rGroup;
463 static void all_matches_add(struct recExtractCtrl *ctrl)
466 extract_init(ctrl, &word);
467 word.index_name = "_ALLRECORDS";
468 word.index_type = 'w';
470 extract_add_index_string (&word, zinfo_index_category_alwaysmatches,
474 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
475 struct ZebraRecStream *stream,
476 enum zebra_recctrl_action_t action,
478 const char *recordType,
480 const char *match_criteria,
483 void *recTypeClientData);
486 ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
489 ZEBRA_RES r = ZEBRA_OK;
494 struct file_read_info *fi = 0;
495 const char *original_record_type = 0;
497 void *recTypeClientData;
498 struct ZebraRecStream stream, *streamp;
500 zebra_init_log_level();
502 if (!zh->m_group || !*zh->m_group)
505 sprintf (gprefix, "%s.", zh->m_group);
507 yaz_log(log_level_extract, "zebra_extract_file %s", fname);
509 /* determine file extension */
511 for (i = strlen(fname); --i >= 0; )
514 else if (fname[i] == '.')
516 strcpy (ext, fname+i+1);
519 /* determine file type - depending on extension */
520 original_record_type = zh->m_record_type;
521 if (!zh->m_record_type)
523 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
524 zh->m_record_type = res_get (zh->res, ext_res);
526 if (!zh->m_record_type)
529 if (zh->records_processed + zh->records_skipped
530 < zh->m_file_verbose_limit)
531 yaz_log (YLOG_LOG, "? %s", fname);
532 zh->records_skipped++;
535 /* determine match criteria */
536 if (!zh->m_record_id)
538 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
539 zh->m_record_id = res_get (zh->res, ext_res);
543 recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
544 &recTypeClientData)))
546 yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
550 switch(recType->version)
555 yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
557 if (sysno && deleteFlag)
566 if (zh->path_reg && !yaz_is_abspath (fname))
568 strcpy (full_rep, zh->path_reg);
569 strcat (full_rep, "/");
570 strcat (full_rep, fname);
573 strcpy (full_rep, fname);
575 if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
577 yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
578 zh->m_record_type = original_record_type;
582 zebra_create_stream_fd(streamp, fd, 0);
584 r = zebra_extract_records_stream(zh, streamp,
586 action_delete : action_update,
590 0, /*match_criteria */
592 recType, recTypeClientData);
594 stream.destroy(streamp);
595 zh->m_record_type = original_record_type;
600 If sysno is provided, then it's used to identify the reocord.
601 If not, and match_criteria is provided, then sysno is guessed
602 If not, and a record is provided, then sysno is got from there
606 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
607 const char *buf, size_t buf_size,
608 enum zebra_recctrl_action_t action,
610 const char *recordType,
612 const char *match_criteria,
615 struct ZebraRecStream stream;
620 if (recordType && *recordType)
622 yaz_log(log_level_extract,
623 "Record type explicitly specified: %s", recordType);
624 recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
629 if (!(zh->m_record_type))
631 yaz_log (YLOG_WARN, "No such record type defined");
634 yaz_log(log_level_extract, "Get record type from rgroup: %s",
636 recType = recType_byName (zh->reg->recTypes, zh->res,
637 zh->m_record_type, &clientData);
638 recordType = zh->m_record_type;
643 yaz_log (YLOG_WARN, "No such record type: %s", recordType);
647 zebra_create_stream_mem(&stream, buf, buf_size);
649 res = zebra_extract_records_stream(zh, &stream,
656 recType, clientData);
657 stream.destroy(&stream);
661 ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
662 struct ZebraRecStream *stream,
663 enum zebra_recctrl_action_t action,
665 const char *recordType,
667 const char *match_criteria,
670 void *recTypeClientData)
672 ZEBRA_RES res = ZEBRA_OK;
676 res = zebra_extract_record_stream(zh, stream,
683 recType, recTypeClientData, &more);
698 static WRBUF wrbuf_hex_str(const char *cstr)
701 WRBUF w = wrbuf_alloc();
702 for (i = 0; cstr[i]; i++)
704 if (cstr[i] < ' ' || cstr[i] > 126)
705 wrbuf_printf(w, "\\%02X", cstr[i] & 0xff);
707 wrbuf_putc(w, cstr[i]);
712 ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
713 struct ZebraRecStream *stream,
714 enum zebra_recctrl_action_t action,
716 const char *recordType,
718 const char *match_criteria,
721 void *recTypeClientData,
726 RecordAttr *recordAttr;
727 struct recExtractCtrl extractCtrl;
729 const char *matchStr = 0;
731 off_t start_offset = 0, end_offset = 0;
732 const char *pr_fname = fname; /* filename to print .. */
733 int show_progress = zh->records_processed + zh->records_skipped
734 < zh->m_file_verbose_limit ? 1:0;
736 zebra_init_log_level();
739 pr_fname = "<no file>"; /* make it printable if file is omitted */
741 zebra_rec_keys_reset(zh->reg->keys);
742 zebra_rec_keys_reset(zh->reg->sortKeys);
744 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
746 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
747 zh->m_explain_database))
753 off_t null_offset = 0;
754 extractCtrl.stream = stream;
756 start_offset = stream->tellf(stream);
758 extractCtrl.first_record = start_offset ? 0 : 1;
760 stream->endf(stream, &null_offset);;
762 extractCtrl.init = extract_init;
763 extractCtrl.tokenAdd = extract_token_add;
764 extractCtrl.schemaAdd = extract_schema_add;
765 extractCtrl.dh = zh->reg->dh;
766 extractCtrl.handle = zh;
767 extractCtrl.match_criteria[0] = '\0';
768 extractCtrl.staticrank = 0;
769 extractCtrl.action = action;
771 init_extractCtrl(zh, &extractCtrl);
773 extract_set_store_data_prepare(&extractCtrl);
775 r = (*recType->extract)(recTypeClientData, &extractCtrl);
777 if (action == action_update)
779 action = extractCtrl.action;
784 case RECCTRL_EXTRACT_EOF:
786 case RECCTRL_EXTRACT_ERROR_GENERIC:
787 /* error occured during extraction ... */
788 yaz_log (YLOG_WARN, "extract error: generic");
790 case RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER:
791 /* error occured during extraction ... */
792 yaz_log (YLOG_WARN, "extract error: no such filter");
794 case RECCTRL_EXTRACT_SKIP:
796 yaz_log (YLOG_LOG, "skip %s %s " ZINT_FORMAT,
797 recordType, pr_fname, (zint) start_offset);
800 end_offset = stream->endf(stream, 0);
802 stream->seekf(stream, end_offset);
805 case RECCTRL_EXTRACT_OK:
808 yaz_log (YLOG_WARN, "extract error: unknown error: %d", r);
811 end_offset = stream->endf(stream, 0);
813 stream->seekf(stream, end_offset);
815 end_offset = stream->tellf(stream);
817 all_matches_add(&extractCtrl);
819 if (extractCtrl.match_criteria[0])
820 match_criteria = extractCtrl.match_criteria;
828 if (match_criteria && *match_criteria) {
829 matchStr = match_criteria;
831 if (zh->m_record_id && *zh->m_record_id) {
832 matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname,
836 yaz_log (YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType,
837 pr_fname, (zint) start_offset);
844 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
845 char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
849 if (log_level_extract)
851 WRBUF w = wrbuf_hex_str(matchStr);
852 yaz_log(log_level_extract, "matchStr: %s", wrbuf_cstr(w));
857 assert(*rinfo == sizeof(*sysno));
858 memcpy (sysno, rinfo+1, sizeof(*sysno));
862 if (zebra_rec_keys_empty(zh->reg->keys))
864 /* the extraction process returned no information - the record
865 is probably empty - unless flagShowRecords is in use */
873 if (action == action_delete)
875 yaz_log (YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
876 pr_fname, (zint) start_offset);
877 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
880 else if (action == action_replace)
882 yaz_log (YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
883 pr_fname, (zint) start_offset);
884 yaz_log (YLOG_WARN, "cannot update record above (seems new)");
888 yaz_log (YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
889 (zint) start_offset);
890 rec = rec_new (zh->reg->records);
894 recordAttr = rec_init_attr (zh->reg->zei, rec);
895 if (extractCtrl.staticrank < 0)
897 yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
898 extractCtrl.staticrank = 0;
903 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
904 dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
905 sizeof(*sysno), sysno);
908 extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
910 extract_flush_record_keys2(zh, *sysno,
911 zh->reg->keys, extractCtrl.staticrank,
912 0, recordAttr->staticrank);
914 extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
915 extractCtrl.staticrank);
917 recordAttr->staticrank = extractCtrl.staticrank;
918 zh->records_inserted++;
922 /* record already exists */
923 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
924 zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
925 if (action == action_insert)
927 yaz_log (YLOG_LOG, "skipped %s %s " ZINT_FORMAT,
928 recordType, pr_fname, (zint) start_offset);
933 rec = rec_get (zh->reg->records, *sysno);
936 recordAttr = rec_init_attr (zh->reg->zei, rec);
938 /* decrease total size */
939 zebraExplain_recordBytesIncrement (zh->reg->zei,
940 - recordAttr->recordSize);
942 zebra_rec_keys_set_buf(delkeys,
943 rec->info[recInfo_delKeys],
944 rec->size[recInfo_delKeys],
946 zebra_rec_keys_set_buf(sortKeys,
947 rec->info[recInfo_sortKeys],
948 rec->size[recInfo_sortKeys],
951 extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
953 extract_flush_record_keys(zh, *sysno, 0, delkeys,
954 recordAttr->staticrank);
956 if (action == action_delete)
958 /* record going to be deleted */
960 extract_flush_record_keys2(zh, *sysno, 0, recordAttr->staticrank,
961 delkeys, recordAttr->staticrank);
963 if (zebra_rec_keys_empty(delkeys))
965 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
966 pr_fname, (zint) start_offset);
967 yaz_log(YLOG_WARN, "cannot delete file above, "
968 "storeKeys false (3)");
973 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
974 pr_fname, (zint) start_offset);
975 zh->records_deleted++;
978 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
979 dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
981 rec_del (zh->reg->records, &rec);
983 zebra_rec_keys_close(delkeys);
984 zebra_rec_keys_close(sortKeys);
990 { /* update or special_update */
992 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
993 pr_fname, (zint) start_offset);
994 extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
997 extract_flush_record_keys2(zh, *sysno,
998 zh->reg->keys, extractCtrl.staticrank,
999 delkeys, recordAttr->staticrank);
1001 extract_flush_record_keys(zh, *sysno, 1,
1002 zh->reg->keys, extractCtrl.staticrank);
1004 recordAttr->staticrank = extractCtrl.staticrank;
1005 zh->records_updated++;
1007 zebra_rec_keys_close(delkeys);
1008 zebra_rec_keys_close(sortKeys);
1010 /* update file type */
1011 xfree (rec->info[recInfo_fileType]);
1012 rec->info[recInfo_fileType] =
1013 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1015 /* update filename */
1016 xfree (rec->info[recInfo_filename]);
1017 rec->info[recInfo_filename] =
1018 rec_strdup (fname, &rec->size[recInfo_filename]);
1020 /* update delete keys */
1021 xfree (rec->info[recInfo_delKeys]);
1022 if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
1024 zebra_rec_keys_get_buf(zh->reg->keys,
1025 &rec->info[recInfo_delKeys],
1026 &rec->size[recInfo_delKeys]);
1030 rec->info[recInfo_delKeys] = NULL;
1031 rec->size[recInfo_delKeys] = 0;
1033 /* update sort keys */
1034 xfree (rec->info[recInfo_sortKeys]);
1036 zebra_rec_keys_get_buf(zh->reg->sortKeys,
1037 &rec->info[recInfo_sortKeys],
1038 &rec->size[recInfo_sortKeys]);
1042 recordAttr->recordSize = end_offset - start_offset;
1043 zebraExplain_recordBytesIncrement(zh->reg->zei,
1044 recordAttr->recordSize);
1047 /* set run-number for this record */
1048 recordAttr->runNumber =
1049 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
1051 /* update store data */
1052 xfree (rec->info[recInfo_storeData]);
1054 /* update store data */
1055 if (zh->store_data_buf)
1057 rec->size[recInfo_storeData] = zh->store_data_size;
1058 rec->info[recInfo_storeData] = zh->store_data_buf;
1059 zh->store_data_buf = 0;
1060 recordAttr->recordSize = zh->store_data_size;
1062 else if (zh->m_store_data)
1064 off_t cur_offset = stream->tellf(stream);
1066 rec->size[recInfo_storeData] = recordAttr->recordSize;
1067 rec->info[recInfo_storeData] = (char *)
1068 xmalloc (recordAttr->recordSize);
1069 stream->seekf(stream, start_offset);
1070 stream->readf(stream, rec->info[recInfo_storeData],
1071 recordAttr->recordSize);
1072 stream->seekf(stream, cur_offset);
1076 rec->info[recInfo_storeData] = NULL;
1077 rec->size[recInfo_storeData] = 0;
1079 /* update database name */
1080 xfree (rec->info[recInfo_databaseName]);
1081 rec->info[recInfo_databaseName] =
1082 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
1085 recordAttr->recordOffset = start_offset;
1087 /* commit this record */
1088 rec_put (zh->reg->records, &rec);
1093 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
1095 ZebraHandle zh = (ZebraHandle) handle;
1096 struct recExtractCtrl extractCtrl;
1098 if (zebraExplain_curDatabase (zh->reg->zei,
1099 rec->info[recInfo_databaseName]))
1102 if (zebraExplain_newDatabase (zh->reg->zei,
1103 rec->info[recInfo_databaseName], 0))
1107 zebra_rec_keys_reset(zh->reg->keys);
1108 zebra_rec_keys_reset(zh->reg->sortKeys);
1110 extractCtrl.init = extract_init;
1111 extractCtrl.tokenAdd = extract_token_add;
1112 extractCtrl.schemaAdd = extract_schema_add;
1113 extractCtrl.dh = zh->reg->dh;
1115 init_extractCtrl(zh, &extractCtrl);
1117 extractCtrl.flagShowRecords = 0;
1118 extractCtrl.match_criteria[0] = '\0';
1119 extractCtrl.staticrank = 0;
1120 extractCtrl.action = action_update;
1122 extractCtrl.handle = handle;
1123 extractCtrl.first_record = 1;
1125 extract_set_store_data_prepare(&extractCtrl);
1128 grs_extract_tree(&extractCtrl, n);
1130 if (rec->size[recInfo_delKeys])
1132 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1134 zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
1136 zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
1137 rec->size[recInfo_delKeys],
1140 extract_flush_record_keys2(zh, rec->sysno,
1141 zh->reg->keys, 0, delkeys, 0);
1143 extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
1144 extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
1146 zebra_rec_keys_close(delkeys);
1148 zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
1149 rec->size[recInfo_sortKeys],
1152 extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
1153 zebra_rec_keys_close(sortkeys);
1158 extract_flush_record_keys2(zh, rec->sysno, zh->reg->keys, 0, 0, 0);
1160 extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
1163 extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
1165 xfree (rec->info[recInfo_delKeys]);
1166 zebra_rec_keys_get_buf(zh->reg->keys,
1167 &rec->info[recInfo_delKeys],
1168 &rec->size[recInfo_delKeys]);
1170 xfree (rec->info[recInfo_sortKeys]);
1171 zebra_rec_keys_get_buf(zh->reg->sortKeys,
1172 &rec->info[recInfo_sortKeys],
1173 &rec->size[recInfo_sortKeys]);
1177 void extract_rec_keys_log(ZebraHandle zh, int is_insert,
1178 zebra_rec_keys_t reckeys,
1181 if (zebra_rec_keys_rewind(reckeys))
1186 NMEM nmem = nmem_create();
1188 while(zebra_rec_keys_read(reckeys, &str, &slen, &key))
1190 char keystr[200]; /* room for zints to print */
1192 int ord = CAST_ZINT_TO_INT(key.mem[0]);
1194 const char *string_index;
1196 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1197 0/* db */, &string_index);
1199 zebra_term_untrans_iconv(zh, nmem, index_type,
1202 for (i = 0; i<key.len; i++)
1204 sprintf(keystr + strlen(keystr), ZINT_FORMAT " ", key.mem[i]);
1207 if (*str < CHR_BASE_CHAR)
1210 char dst_buf[200]; /* room for special chars */
1212 strcpy(dst_buf , "?");
1214 if (!strcmp(str, ""))
1215 strcpy(dst_buf, "alwaysmatches");
1216 if (!strcmp(str, FIRST_IN_FIELD_STR))
1217 strcpy(dst_buf, "firstinfield");
1218 else if (!strcmp(str, CHR_UNKNOWN))
1219 strcpy(dst_buf, "unknown");
1220 else if (!strcmp(str, CHR_SPACE))
1221 strcpy(dst_buf, "space");
1223 for (i = 0; i<slen; i++)
1225 sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
1227 yaz_log(level, "%s%c %s %s", keystr, index_type,
1228 string_index, dst_buf);
1232 yaz_log(level, "%s%c %s \"%s\"", keystr, index_type,
1233 string_index, dst_term);
1241 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
1242 zebra_rec_keys_t reckeys)
1244 ZebraExplainInfo zei = zh->reg->zei;
1248 struct ord_stat *next;
1251 if (zebra_rec_keys_rewind(reckeys))
1253 struct ord_stat *ord_list = 0;
1257 struct it_key key_in;
1258 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1260 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1262 for (p = ord_list; p ; p = p->next)
1270 p = xmalloc(sizeof(*p));
1281 struct ord_stat *p1 = p;
1284 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
1286 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
1293 void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
1294 zebra_rec_keys_t ins_keys, zint ins_rank,
1295 zebra_rec_keys_t del_keys, zint del_rank)
1297 ZebraExplainInfo zei = zh->reg->zei;
1301 if (!zh->reg->key_block)
1303 int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1304 const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", ".");
1305 int use_threads = atoi(res_get_def (zh->res, "threads", "1"));
1306 zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1311 extract_rec_keys_adjust(zh, 1, ins_keys);
1313 zebraExplain_recordCountIncrement (zei, 1);
1314 zebra_rec_keys_rewind(ins_keys);
1318 extract_rec_keys_adjust(zh, 0, del_keys);
1320 zebraExplain_recordCountIncrement (zei, -1);
1321 zebra_rec_keys_rewind(del_keys);
1327 const char *del_str;
1328 struct it_key del_key_in;
1332 const char *ins_str;
1333 struct it_key ins_key_in;
1337 del = zebra_rec_keys_read(del_keys, &del_str, &del_slen,
1340 ins = zebra_rec_keys_read(ins_keys, &ins_str, &ins_slen,
1343 if (del && ins && ins_rank == del_rank
1344 && !key_compare(&del_key_in, &ins_key_in)
1345 && ins_slen == del_slen && !memcmp(del_str, ins_str, del_slen))
1355 key_block_write(zh->reg->key_block, sysno,
1356 &del_key_in, 0, del_str, del_slen,
1357 del_rank, zh->m_staticrank);
1359 key_block_write(zh->reg->key_block, sysno,
1360 &ins_key_in, 1, ins_str, ins_slen,
1361 ins_rank, zh->m_staticrank);
1363 yaz_log(log_level_extract, "normal=%d optimized=%d", normal, optimized);
1366 void extract_flush_record_keys(ZebraHandle zh, zint sysno, int cmd,
1367 zebra_rec_keys_t reckeys,
1370 ZebraExplainInfo zei = zh->reg->zei;
1372 extract_rec_keys_adjust(zh, cmd, reckeys);
1374 if (log_level_details)
1376 yaz_log(log_level_details, "Keys for record " ZINT_FORMAT " %s",
1377 sysno, cmd ? "insert" : "delete");
1378 extract_rec_keys_log(zh, cmd, reckeys, log_level_details);
1381 if (!zh->reg->key_block)
1383 int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
1384 const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", ".");
1385 int use_threads = atoi(res_get_def (zh->res, "threads", "1"));
1386 zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
1388 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1391 yaz_log(YLOG_LOG, "sysno=" ZINT_FORMAT " cmd=%d", sysno, cmd);
1392 print_rec_keys(zh, reckeys);
1394 if (zebra_rec_keys_rewind(reckeys))
1398 struct it_key key_in;
1399 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1401 key_block_write(zh->reg->key_block, sysno,
1402 &key_in, cmd, str, slen,
1403 staticrank, zh->m_staticrank);
1408 ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
1409 zebra_rec_keys_t reckeys,
1410 zebra_snippets *snippets)
1412 NMEM nmem = nmem_create();
1413 if (zebra_rec_keys_rewind(reckeys))
1418 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1425 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1426 seqno = key.mem[key.len-1];
1427 ord = CAST_ZINT_TO_INT(key.mem[0]);
1429 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1430 0/* db */, 0 /* string_index */);
1432 zebra_term_untrans_iconv(zh, nmem, index_type,
1434 zebra_snippets_append(snippets, seqno, 0, ord, dst_term);
1442 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1444 yaz_log(YLOG_LOG, "print_rec_keys");
1445 if (zebra_rec_keys_rewind(reckeys))
1450 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1452 char dst_buf[IT_MAX_WORD];
1455 int ord = CAST_ZINT_TO_INT(key.mem[0]);
1457 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1459 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1461 seqno = key.mem[key.len-1];
1463 zebra_term_untrans(zh, index_type, dst_buf, str);
1465 yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
1466 " term=%s", ord, seqno, dst_buf);
1471 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1472 const char *str, int length)
1475 ZebraHandle zh = p->extractCtrl->handle;
1476 ZebraExplainInfo zei = zh->reg->zei;
1479 ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1481 ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1485 key.mem[i++] = p->record_id;
1486 key.mem[i++] = p->section_id;
1488 if (zh->m_segment_indexing)
1489 key.mem[i++] = p->segment;
1490 key.mem[i++] = p->seqno;
1493 zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1496 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1499 ZebraHandle zh = p->extractCtrl->handle;
1500 ZebraExplainInfo zei = zh->reg->zei;
1502 zinfo_index_category_t cat = zinfo_index_category_sort;
1504 ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1506 ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1509 key.mem[1] = p->record_id;
1511 zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1514 static void extract_add_staticrank_string(RecWord *p,
1515 const char *str, int length)
1518 struct recExtractCtrl *ctrl = p->extractCtrl;
1520 if (length > sizeof(valz)-1)
1521 length = sizeof(valz)-1;
1523 memcpy(valz, str, length);
1524 valz[length] = '\0';
1525 ctrl->staticrank = atozint(valz);
1528 static void extract_add_string(RecWord *p, const char *string, int length)
1530 ZebraHandle zh = p->extractCtrl->handle;
1531 assert (length > 0);
1536 if (zebra_maps_is_index(zh->reg->zebra_maps, p->index_type))
1538 extract_add_index_string(p, zinfo_index_category_index,
1540 if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type))
1543 memcpy(&word, p, sizeof(word));
1546 extract_add_index_string(
1547 &word, zinfo_index_category_alwaysmatches, "", 0);
1550 else if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
1552 extract_add_sort_string(p, string, length);
1554 else if (zebra_maps_is_staticrank(zh->reg->zebra_maps, p->index_type))
1556 extract_add_staticrank_string(p, string, length);
1560 static void extract_add_incomplete_field(RecWord *p)
1562 ZebraHandle zh = p->extractCtrl->handle;
1563 const char *b = p->term_buf;
1564 int remain = p->term_len;
1566 const char **map = 0;
1569 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1573 char buf[IT_MAX_WORD+1];
1577 while (map && *map && **map == *CHR_SPACE)
1579 remain = p->term_len - (b - p->term_buf);
1581 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
1589 while (map && *map && **map != *CHR_SPACE)
1591 const char *cp = *map;
1593 while (i < IT_MAX_WORD && *cp)
1595 remain = p->term_len - (b - p->term_buf);
1597 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1607 if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
1609 /* first in field marker */
1610 extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1614 extract_add_string (p, buf, i);
1619 static void extract_add_complete_field (RecWord *p)
1621 ZebraHandle zh = p->extractCtrl->handle;
1622 const char *b = p->term_buf;
1623 char buf[IT_MAX_WORD+1];
1624 const char **map = 0;
1625 int i = 0, remain = p->term_len;
1628 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
1630 while (remain > 0 && i < IT_MAX_WORD)
1632 while (map && *map && **map == *CHR_SPACE)
1634 remain = p->term_len - (b - p->term_buf);
1638 int first = i ? 0 : 1; /* first position */
1639 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, first);
1647 if (i && i < IT_MAX_WORD)
1648 buf[i++] = *CHR_SPACE;
1649 while (map && *map && **map != *CHR_SPACE)
1651 const char *cp = *map;
1653 if (**map == *CHR_CUT)
1659 if (i >= IT_MAX_WORD)
1661 while (i < IT_MAX_WORD && *cp)
1664 remain = p->term_len - (b - p->term_buf);
1667 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
1676 extract_add_string (p, buf, i);
1679 static void extract_token_add(RecWord *p)
1681 ZebraHandle zh = p->extractCtrl->handle;
1684 if (log_level_details)
1686 yaz_log(log_level_details, "extract_token_add "
1687 "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
1688 p->index_type, p->index_name,
1689 p->seqno, p->term_len, p->term_buf);
1691 if ((wrbuf = zebra_replace(zh->reg->zebra_maps, p->index_type, 0,
1692 p->term_buf, p->term_len)))
1694 p->term_buf = wrbuf_buf(wrbuf);
1695 p->term_len = wrbuf_len(wrbuf);
1697 if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type))
1698 extract_add_complete_field (p);
1700 extract_add_incomplete_field(p);
1703 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1704 void *buf, size_t sz)
1706 ZebraHandle zh = (ZebraHandle) p->handle;
1708 xfree(zh->store_data_buf);
1709 zh->store_data_buf = 0;
1710 zh->store_data_size = 0;
1713 zh->store_data_buf = xmalloc(sz);
1714 zh->store_data_size = sz;
1715 memcpy(zh->store_data_buf, buf, sz);
1719 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1721 ZebraHandle zh = (ZebraHandle) p->handle;
1722 xfree(zh->store_data_buf);
1723 zh->store_data_buf = 0;
1724 zh->store_data_size = 0;
1725 p->setStoreData = extract_set_store_data_cb;
1728 static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
1730 ZebraHandle zh = (ZebraHandle) p->handle;
1731 zebraExplain_addSchema (zh->reg->zei, oid);
1734 void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
1735 int cmd, zebra_rec_keys_t reckeys)
1738 yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT,
1740 extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG);
1743 if (zebra_rec_keys_rewind(reckeys))
1745 zebra_sort_index_t si = zh->reg->sort_index;
1748 struct it_key key_in;
1750 zebra_sort_sysno(si, sysno);
1752 while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1754 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1756 zebra_sort_type(si, ord);
1758 zebra_sort_add(si, str, slen);
1760 zebra_sort_delete(si);
1768 * indent-tabs-mode: nil
1770 * vim: shiftwidth=4 tabstop=8 expandtab