1 /* $Id: extract.c,v 1.237 2006-11-16 11:11:36 adam Exp $
2 Copyright (C) 1995-2006
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 #define ENCODE_BUFLEN 768
43 char buf[ENCODE_BUFLEN];
46 static int log_level_extract = 0;
47 static int log_level_details = 0;
48 static int log_level_initialized = 0;
50 static void zebra_init_log_level(void)
52 if (!log_level_initialized)
54 log_level_initialized = 1;
56 log_level_extract = yaz_log_module_level("extract");
57 log_level_details = yaz_log_module_level("indexdetails");
61 static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
62 int cmd, zebra_rec_keys_t reckeys,
64 static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
65 int cmd, zebra_rec_keys_t skp);
66 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid);
67 static void extract_token_add (RecWord *p);
69 static void encode_key_init (struct encode_info *i);
70 static void encode_key_write (char *k, struct encode_info *i, FILE *outf);
71 static void encode_key_flush (struct encode_info *i, FILE *outf);
73 #define USE_SHELLSORT 0
76 static void shellsort(void *ar, int r, size_t s,
77 int (*cmp)(const void *a, const void *b))
82 static const int incs[16] = { 1391376, 463792, 198768, 86961, 33936,
83 13776, 4592, 1968, 861, 336,
84 112, 48, 21, 7, 3, 1 };
85 for ( k = 0; k < 16; k++)
86 for (h = incs[k], i = h; i < r; i++)
90 while (j > h && (*cmp)(a + s*(j-h), v) > 0)
92 memcpy (a + s*j, a + s*(j-h), s);
100 static void logRecord (ZebraHandle zh)
102 ++zh->records_processed;
103 if (!(zh->records_processed % 1000))
105 yaz_log(YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
106 ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT,
107 zh->records_processed, zh->records_inserted,
108 zh->records_updated, zh->records_deleted);
112 static void extract_add_index_string (RecWord *p,
113 zinfo_index_category_t cat,
114 const char *str, int length);
116 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
118 static void extract_init(struct recExtractCtrl *p, RecWord *w)
121 w->index_name = "any";
129 static void searchRecordKey(ZebraHandle zh,
130 zebra_rec_keys_t reckeys,
131 const char *index_name,
132 const char **ws, int ws_length)
136 zinfo_index_category_t cat = zinfo_index_category_index;
138 for (i = 0; i<ws_length; i++)
142 ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, '0', index_name);
144 ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'p', index_name);
146 ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'w', index_name);
151 if (zebra_rec_keys_rewind(reckeys))
158 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
160 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
162 seqno = key.mem[key.len-1];
164 if (key.mem[0] == ch)
170 woff = seqno - startSeq;
171 if (woff >= 0 && woff < ws_length)
178 #define FILE_MATCH_BLANK "\t "
180 static char *fileMatchStr (ZebraHandle zh,
181 zebra_rec_keys_t reckeys,
182 const char *fname, const char *spec)
184 static char dstBuf[2048]; /* static here ??? */
186 const char *s = spec;
190 for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
197 char attset_str[64], attname_str[64];
201 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
203 for (i = 0; *s && *s != ',' && *s != ')' &&
204 !strchr(FILE_MATCH_BLANK, *s); s++)
205 if (i+1 < sizeof(attset_str))
206 attset_str[i++] = *s;
207 attset_str[i] = '\0';
209 for (; strchr(FILE_MATCH_BLANK, *s); s++)
212 strcpy(attname_str, attset_str);
215 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
217 for (i = 0; *s && *s != ')' &&
218 !strchr(FILE_MATCH_BLANK, *s); s++)
219 if (i+1 < sizeof(attname_str))
220 attname_str[i++] = *s;
221 attname_str[i] = '\0';
224 searchRecordKey (zh, reckeys, attname_str, ws, 32);
228 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
229 spec, zh->m_group ? zh->m_group : "none");
234 for (i = 0; i<32; i++)
243 dst += strlen(ws[i]);
247 yaz_log (YLOG_WARN, "Record didn't contain match"
248 " fields in (%s,%s)", attset_str, attname_str);
256 const char *spec_src = NULL;
257 const char *s1 = ++s;
258 while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
262 if (spec_len > sizeof(special)-1)
263 spec_len = sizeof(special)-1;
264 memcpy (special, s, spec_len);
265 special[spec_len] = '\0';
268 if (!strcmp (special, "group"))
269 spec_src = zh->m_group;
270 else if (!strcmp (special, "database"))
271 spec_src = zh->basenames[0];
272 else if (!strcmp (special, "filename")) {
275 else if (!strcmp (special, "type"))
276 spec_src = zh->m_record_type;
281 strcpy (dst, spec_src);
282 dst += strlen (spec_src);
285 else if (*s == '\"' || *s == '\'')
287 int stopMarker = *s++;
291 while (*s && *s != stopMarker)
293 if (i+1 < sizeof(tmpString))
294 tmpString[i++] = *s++;
299 strcpy (dst, tmpString);
300 dst += strlen (tmpString);
304 yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
305 spec, zh->m_group ? zh->m_group : "none");
312 yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
313 fname, zh->m_group ? zh->m_group : "none");
320 struct recordLogInfo {
323 struct recordGroup *rGroup;
326 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
329 for (i = 0; i<256; i++)
331 if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
336 ctrl->flagShowRecords = !zh->m_flag_rw;
339 static void all_matches_add(struct recExtractCtrl *ctrl)
342 extract_init(ctrl, &word);
343 word.index_name = "_ALLRECORDS";
344 word.index_type = 'w';
346 extract_add_index_string (&word, zinfo_index_category_alwaysmatches,
350 ZEBRA_RES zebra_extract_file(ZebraHandle zh, SYSNO *sysno, const char *fname,
353 ZEBRA_RES r = ZEBRA_OK;
358 struct file_read_info *fi = 0;
359 const char *original_record_type = 0;
361 void *recTypeClientData;
362 struct ZebraRecStream stream, *streamp;
364 zebra_init_log_level();
366 if (!zh->m_group || !*zh->m_group)
369 sprintf (gprefix, "%s.", zh->m_group);
371 yaz_log(log_level_extract, "zebra_extract_file %s", fname);
373 /* determine file extension */
375 for (i = strlen(fname); --i >= 0; )
378 else if (fname[i] == '.')
380 strcpy (ext, fname+i+1);
383 /* determine file type - depending on extension */
384 original_record_type = zh->m_record_type;
385 if (!zh->m_record_type)
387 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
388 zh->m_record_type = res_get (zh->res, ext_res);
390 if (!zh->m_record_type)
392 if (zh->records_processed < zh->m_file_verbose_limit)
393 yaz_log (YLOG_LOG, "? %s", fname);
396 /* determine match criteria */
397 if (!zh->m_record_id)
399 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
400 zh->m_record_id = res_get (zh->res, ext_res);
404 recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
405 &recTypeClientData)))
407 yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
411 switch(recType->version)
416 yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
418 if (sysno && deleteFlag)
427 if (zh->path_reg && !yaz_is_abspath (fname))
429 strcpy (full_rep, zh->path_reg);
430 strcat (full_rep, "/");
431 strcat (full_rep, fname);
434 strcpy (full_rep, fname);
436 if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
438 yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
439 zh->m_record_type = original_record_type;
443 zebra_create_stream_fd(streamp, fd, 0);
447 r = zebra_extract_record_stream(zh, streamp,
452 0, /*match_criteria */
454 1, /* force_update */
455 1, /* allow_update */
456 recType, recTypeClientData);
467 stream.destroy(streamp);
468 zh->m_record_type = original_record_type;
473 If sysno is provided, then it's used to identify the reocord.
474 If not, and match_criteria is provided, then sysno is guessed
475 If not, and a record is provided, then sysno is got from there
479 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
480 const char *buf, size_t buf_size,
483 const char *recordType,
485 const char *match_criteria,
490 struct ZebraRecStream stream;
495 if (recordType && *recordType)
497 yaz_log(log_level_extract,
498 "Record type explicitly specified: %s", recordType);
499 recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
504 if (!(zh->m_record_type))
506 yaz_log (YLOG_WARN, "No such record type defined");
509 yaz_log(log_level_extract, "Get record type from rgroup: %s",
511 recType = recType_byName (zh->reg->recTypes, zh->res,
512 zh->m_record_type, &clientData);
513 recordType = zh->m_record_type;
518 yaz_log (YLOG_WARN, "No such record type: %s", recordType);
524 zebra_create_stream_mem(&stream, buf, buf_size);
526 res = zebra_extract_record_stream(zh, &stream,
535 recType, clientData);
536 stream.destroy(&stream);
541 ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
542 struct ZebraRecStream *stream,
545 const char *recordType,
547 const char *match_criteria,
552 void *recTypeClientData)
556 RecordAttr *recordAttr;
557 struct recExtractCtrl extractCtrl;
559 const char *matchStr = 0;
561 off_t start_offset = 0;
562 const char *pr_fname = fname; /* filename to print .. */
563 int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
565 zebra_init_log_level();
568 pr_fname = "<no file>"; /* make it printable if file is omitted */
570 zebra_rec_keys_reset(zh->reg->keys);
571 zebra_rec_keys_reset(zh->reg->sortKeys);
573 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
575 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
576 zh->m_explain_database))
582 off_t null_offset = 0;
583 extractCtrl.stream = stream;
585 start_offset = stream->tellf(stream);
587 extractCtrl.first_record = start_offset ? 0 : 1;
589 stream->endf(stream, &null_offset);;
591 extractCtrl.init = extract_init;
592 extractCtrl.tokenAdd = extract_token_add;
593 extractCtrl.schemaAdd = extract_schema_add;
594 extractCtrl.dh = zh->reg->dh;
595 extractCtrl.handle = zh;
596 extractCtrl.match_criteria[0] = '\0';
597 extractCtrl.staticrank = 0;
600 init_extractCtrl(zh, &extractCtrl);
602 extract_set_store_data_prepare(&extractCtrl);
604 r = (*recType->extract)(recTypeClientData, &extractCtrl);
606 if (r == RECCTRL_EXTRACT_EOF)
608 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
610 /* error occured during extraction ... */
611 yaz_log (YLOG_WARN, "extract error: generic");
614 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
616 /* error occured during extraction ... */
617 yaz_log (YLOG_WARN, "extract error: no such filter");
621 all_matches_add(&extractCtrl);
623 if (extractCtrl.match_criteria[0])
624 match_criteria = extractCtrl.match_criteria;
630 if (match_criteria && *match_criteria) {
631 matchStr = match_criteria;
633 if (zh->m_record_id && *zh->m_record_id) {
634 matchStr = fileMatchStr (zh, zh->reg->keys, pr_fname,
638 yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
645 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
646 char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
650 assert(*rinfo == sizeof(*sysno));
651 memcpy (sysno, rinfo+1, sizeof(*sysno));
655 if (zebra_rec_keys_empty(zh->reg->keys))
657 /* the extraction process returned no information - the record
658 is probably empty - unless flagShowRecords is in use */
668 yaz_log (YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
669 pr_fname, (zint) start_offset);
670 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
674 yaz_log (YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
675 (zint) start_offset);
676 rec = rec_new (zh->reg->records);
680 recordAttr = rec_init_attr (zh->reg->zei, rec);
681 recordAttr->staticrank = extractCtrl.staticrank;
685 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
686 dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
687 sizeof(*sysno), sysno);
691 extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
692 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
693 recordAttr->staticrank);
694 zh->records_inserted++;
698 /* record already exists */
699 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
700 zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
703 yaz_log (YLOG_LOG, "skipped %s %s " ZINT_FORMAT,
704 recordType, pr_fname, (zint) start_offset);
709 rec = rec_get (zh->reg->records, *sysno);
712 recordAttr = rec_init_attr (zh->reg->zei, rec);
714 zebra_rec_keys_set_buf(delkeys,
715 rec->info[recInfo_delKeys],
716 rec->size[recInfo_delKeys],
718 zebra_rec_keys_set_buf(sortKeys,
719 rec->info[recInfo_sortKeys],
720 rec->size[recInfo_sortKeys],
723 extract_flushSortKeys (zh, *sysno, 0, sortKeys);
724 extract_flushRecordKeys (zh, *sysno, 0, delkeys,
725 recordAttr->staticrank);
728 /* record going to be deleted */
729 if (zebra_rec_keys_empty(delkeys))
731 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
732 pr_fname, (zint) start_offset);
733 yaz_log(YLOG_WARN, "cannot delete file above, "
734 "storeKeys false (3)");
739 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
740 pr_fname, (zint) start_offset);
741 zh->records_deleted++;
744 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
745 dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
747 rec_del (zh->reg->records, &rec);
756 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
757 pr_fname, (zint) ZINT_FORMAT);
758 recordAttr->staticrank = extractCtrl.staticrank;
759 extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
760 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
761 recordAttr->staticrank);
762 zh->records_updated++;
764 zebra_rec_keys_close(delkeys);
765 zebra_rec_keys_close(sortKeys);
767 /* update file type */
768 xfree (rec->info[recInfo_fileType]);
769 rec->info[recInfo_fileType] =
770 rec_strdup (recordType, &rec->size[recInfo_fileType]);
772 /* update filename */
773 xfree (rec->info[recInfo_filename]);
774 rec->info[recInfo_filename] =
775 rec_strdup (fname, &rec->size[recInfo_filename]);
777 /* update delete keys */
778 xfree (rec->info[recInfo_delKeys]);
779 if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
781 zebra_rec_keys_get_buf(zh->reg->keys,
782 &rec->info[recInfo_delKeys],
783 &rec->size[recInfo_delKeys]);
787 rec->info[recInfo_delKeys] = NULL;
788 rec->size[recInfo_delKeys] = 0;
790 /* update sort keys */
791 xfree (rec->info[recInfo_sortKeys]);
793 zebra_rec_keys_get_buf(zh->reg->sortKeys,
794 &rec->info[recInfo_sortKeys],
795 &rec->size[recInfo_sortKeys]);
797 /* save file size of original record */
798 zebraExplain_recordBytesIncrement (zh->reg->zei,
799 - recordAttr->recordSize);
802 off_t end_offset = stream->endf(stream, 0);
805 end_offset = stream->tellf(stream);
807 stream->seekf(stream, end_offset);
809 recordAttr->recordSize = end_offset - start_offset;
810 zebraExplain_recordBytesIncrement(zh->reg->zei,
811 recordAttr->recordSize);
814 /* set run-number for this record */
815 recordAttr->runNumber =
816 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
818 /* update store data */
819 xfree (rec->info[recInfo_storeData]);
821 /* update store data */
822 if (zh->store_data_buf)
824 rec->size[recInfo_storeData] = zh->store_data_size;
825 rec->info[recInfo_storeData] = zh->store_data_buf;
826 zh->store_data_buf = 0;
827 recordAttr->recordSize = zh->store_data_size;
829 else if (zh->m_store_data)
831 off_t cur_offset = stream->tellf(stream);
833 rec->size[recInfo_storeData] = recordAttr->recordSize;
834 rec->info[recInfo_storeData] = (char *)
835 xmalloc (recordAttr->recordSize);
836 stream->seekf(stream, start_offset);
837 stream->readf(stream, rec->info[recInfo_storeData],
838 recordAttr->recordSize);
839 stream->seekf(stream, cur_offset);
843 rec->info[recInfo_storeData] = NULL;
844 rec->size[recInfo_storeData] = 0;
846 /* update database name */
847 xfree (rec->info[recInfo_databaseName]);
848 rec->info[recInfo_databaseName] =
849 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
852 recordAttr->recordOffset = start_offset;
854 /* commit this record */
855 rec_put (zh->reg->records, &rec);
860 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
862 ZebraHandle zh = (ZebraHandle) handle;
863 struct recExtractCtrl extractCtrl;
865 if (zebraExplain_curDatabase (zh->reg->zei,
866 rec->info[recInfo_databaseName]))
869 if (zebraExplain_newDatabase (zh->reg->zei,
870 rec->info[recInfo_databaseName], 0))
874 zebra_rec_keys_reset(zh->reg->keys);
875 zebra_rec_keys_reset(zh->reg->sortKeys);
877 extractCtrl.init = extract_init;
878 extractCtrl.tokenAdd = extract_token_add;
879 extractCtrl.schemaAdd = extract_schema_add;
880 extractCtrl.dh = zh->reg->dh;
882 init_extractCtrl(zh, &extractCtrl);
884 extractCtrl.flagShowRecords = 0;
885 extractCtrl.match_criteria[0] = '\0';
886 extractCtrl.staticrank = 0;
887 extractCtrl.handle = handle;
888 extractCtrl.first_record = 1;
890 extract_set_store_data_prepare(&extractCtrl);
893 grs_extract_tree(&extractCtrl, n);
895 if (rec->size[recInfo_delKeys])
897 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
899 zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
901 zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
902 rec->size[recInfo_delKeys],
904 extract_flushRecordKeys (zh, rec->sysno, 0, delkeys, 0);
905 zebra_rec_keys_close(delkeys);
907 zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
908 rec->size[recInfo_sortKeys],
911 extract_flushSortKeys (zh, rec->sysno, 0, sortkeys);
912 zebra_rec_keys_close(sortkeys);
914 extract_flushRecordKeys (zh, rec->sysno, 1, zh->reg->keys, 0);
915 extract_flushSortKeys (zh, rec->sysno, 1, zh->reg->sortKeys);
917 xfree (rec->info[recInfo_delKeys]);
918 zebra_rec_keys_get_buf(zh->reg->keys,
919 &rec->info[recInfo_delKeys],
920 &rec->size[recInfo_delKeys]);
922 xfree (rec->info[recInfo_sortKeys]);
923 zebra_rec_keys_get_buf(zh->reg->sortKeys,
924 &rec->info[recInfo_sortKeys],
925 &rec->size[recInfo_sortKeys]);
929 void extract_rec_keys_log(ZebraHandle zh, int is_insert,
930 zebra_rec_keys_t reckeys,
933 if (zebra_rec_keys_rewind(reckeys))
938 NMEM nmem = nmem_create();
940 while(zebra_rec_keys_read(reckeys, &str, &slen, &key))
942 char keystr[200]; /* room for zints to print */
944 int ord = CAST_ZINT_TO_INT(key.mem[0]);
946 const char *string_index;
948 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
949 0/* db */, &string_index);
951 zebra_term_untrans_iconv(zh, nmem, index_type,
954 for (i = 0; i<key.len; i++)
956 sprintf(keystr + strlen(keystr), ZINT_FORMAT " ", key.mem[i]);
959 if (*str < CHR_BASE_CHAR)
962 char dst_buf[200]; /* room for special chars */
964 strcpy(dst_buf , "?");
966 if (!strcmp(str, ""))
967 strcpy(dst_buf, "alwaysmatches");
968 if (!strcmp(str, FIRST_IN_FIELD_STR))
969 strcpy(dst_buf, "firstinfield");
970 else if (!strcmp(str, CHR_UNKNOWN))
971 strcpy(dst_buf, "unknown");
972 else if (!strcmp(str, CHR_SPACE))
973 strcpy(dst_buf, "space");
975 for (i = 0; i<slen; i++)
977 sprintf(dst_buf + strlen(dst_buf), " %d", str[i] & 0xff);
979 yaz_log(level, "%s%c %s %s", keystr, index_type,
980 string_index, dst_buf);
984 yaz_log(level, "%s%c %s \"%s\"", keystr, index_type,
985 string_index, dst_term);
993 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
994 zebra_rec_keys_t reckeys)
996 ZebraExplainInfo zei = zh->reg->zei;
1000 struct ord_stat *next;
1003 if (zebra_rec_keys_rewind(reckeys))
1005 struct ord_stat *ord_list = 0;
1009 struct it_key key_in;
1010 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1012 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1014 for (p = ord_list; p ; p = p->next)
1022 p = xmalloc(sizeof(*p));
1033 struct ord_stat *p1 = p;
1036 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
1038 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
1045 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1047 zebra_rec_keys_t reckeys,
1050 ZebraExplainInfo zei = zh->reg->zei;
1052 extract_rec_keys_adjust(zh, cmd, reckeys);
1054 if (log_level_details)
1056 yaz_log(log_level_details, "Keys for record " ZINT_FORMAT " %s",
1057 sysno, cmd ? "insert" : "delete");
1058 extract_rec_keys_log(zh, cmd, reckeys, log_level_details);
1061 if (!zh->reg->key_buf)
1063 int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
1066 yaz_log(YLOG_WARN, "Invalid memory setting, using default 8 MB");
1069 /* FIXME: That "8" should be in a default settings include */
1070 /* not hard-coded here! -H */
1071 zh->reg->key_buf = (char**) xmalloc (mem);
1072 zh->reg->ptr_top = mem/sizeof(char*);
1074 zh->reg->key_buf_used = 0;
1075 zh->reg->key_file_no = 0;
1077 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1079 if (zebra_rec_keys_rewind(reckeys))
1083 struct it_key key_in;
1084 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1088 struct it_key key_out;
1090 assert(key_in.len >= 2);
1091 assert(key_in.len <= IT_KEY_LEVEL_MAX);
1093 /* check for buffer overflow */
1094 if (zh->reg->key_buf_used + 1024 >
1095 (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
1096 extract_flushWriteKeys (zh, 0);
1099 assert(zh->reg->ptr_i > 0);
1100 (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
1101 (char*)zh->reg->key_buf + zh->reg->key_buf_used;
1103 /* key_in.mem[0] ord/ch */
1104 /* key_in.mem[1] filter specified record ID */
1106 /* encode the ordinal value (field/use/attribute) .. */
1107 ch = CAST_ZINT_TO_INT(key_in.mem[0]);
1108 zh->reg->key_buf_used +=
1109 key_SU_encode(ch, (char*)zh->reg->key_buf +
1110 zh->reg->key_buf_used);
1112 /* copy the 0-terminated stuff from str to output */
1113 memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used, str, slen);
1114 zh->reg->key_buf_used += slen;
1115 ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = '\0';
1117 /* the delete/insert indicator */
1118 ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = cmd;
1120 if (zh->m_staticrank) /* rank config enabled ? */
1124 yaz_log(YLOG_WARN, "staticrank = %ld. Setting to 0",
1128 key_out.mem[j++] = staticrank;
1131 if (key_in.mem[1]) /* filter specified record ID */
1132 key_out.mem[j++] = key_in.mem[1];
1134 key_out.mem[j++] = sysno;
1135 for (i = 2; i < key_in.len; i++)
1136 key_out.mem[j++] = key_in.mem[i];
1139 memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used,
1140 &key_out, sizeof(key_out));
1141 (zh->reg->key_buf_used) += sizeof(key_out);
1146 void extract_flushWriteKeys (ZebraHandle zh, int final)
1147 /* optimizing: if final=1, and no files written yet */
1148 /* push the keys directly to merge, sidestepping the */
1149 /* temp file altogether. Speeds small updates */
1152 char out_fname[200];
1154 struct encode_info encode_info;
1155 int ptr_i = zh->reg->ptr_i;
1160 if (!zh->reg->key_buf || ptr_i <= 0)
1162 yaz_log(log_level_extract, " nothing to flush section=%d buf=%p i=%d",
1163 zh->reg->key_file_no, zh->reg->key_buf, ptr_i);
1167 (zh->reg->key_file_no)++;
1168 yaz_log (YLOG_LOG, "sorting section %d", (zh->reg->key_file_no));
1169 yaz_log(log_level_extract, " sort_buff at %p n=%d",
1170 zh->reg->key_buf + zh->reg->ptr_top - ptr_i,ptr_i);
1172 qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
1173 sizeof(char*), key_qsort_compare);
1175 /* zebra.cfg: tempfiles:
1176 Y: always use temp files (old way)
1177 A: use temp files, if more than one (auto)
1178 = if this is both the last and the first
1179 N: never bother with temp files (new) */
1181 temp_policy=toupper(res_get_def(zh->res,"tempfiles","auto")[0]);
1182 if (temp_policy != 'Y' && temp_policy != 'N' && temp_policy != 'A') {
1183 yaz_log (YLOG_WARN, "Illegal tempfiles setting '%c'. using 'Auto' ",
1188 if ( ( temp_policy =='N' ) || /* always from memory */
1189 ( ( temp_policy =='A' ) && /* automatic */
1190 (zh->reg->key_file_no == 1) && /* this is first time */
1191 (final) ) ) /* and last (=only) time */
1192 { /* go directly from memory */
1193 zh->reg->key_file_no =0; /* signal not to read files */
1194 zebra_index_merge(zh);
1196 zh->reg->key_buf_used = 0;
1200 /* Not doing directly from memory, write into a temp file */
1201 extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
1203 if (!(outf = fopen (out_fname, "wb")))
1205 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1206 zebra_exit("extract_flushWriteKeys");
1208 yaz_log (YLOG_LOG, "writing section %d", zh->reg->key_file_no);
1209 prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1211 encode_key_init (&encode_info);
1212 encode_key_write (cp, &encode_info, outf);
1216 cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1217 if (strcmp (cp, prevcp))
1219 encode_key_flush ( &encode_info, outf);
1220 encode_key_init (&encode_info);
1221 encode_key_write (cp, &encode_info, outf);
1225 encode_key_write (cp + strlen(cp), &encode_info, outf);
1227 encode_key_flush ( &encode_info, outf);
1229 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1230 extract_get_fname_tmp (out_fname, key_file_no);
1232 if (!(outf = fopen (out_fname, "wb")))
1234 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1235 zebra_exit("extract_flushWriteKeys");
1237 yaz_log (YLOG_LOG, "writing section %d", key_file_no);
1239 prevcp = key_buf[ptr_top-i];
1241 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1243 key_y_len = strlen(prevcp)+1;
1245 yaz_log (YLOG_LOG, "key_y_len: %2d %02x %02x %s",
1246 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1248 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1249 sizeof(char*), key_y_compare);
1250 cp = key_buf[ptr_top-ptr_i];
1252 encode_key_init (&encode_info);
1253 encode_key_write (cp, &encode_info, outf);
1256 cp = key_buf[ptr_top-ptr_i];
1257 encode_key_write (cp+key_y_len, &encode_info, outf);
1259 encode_key_flush ( &encode_info, outf);
1262 prevcp = key_buf[ptr_top-ptr_i];
1267 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fclose %s", out_fname);
1268 zebra_exit("extract_flushWriteKeys");
1270 yaz_log (YLOG_LOG, "finished section %d", zh->reg->key_file_no);
1272 zh->reg->key_buf_used = 0;
1275 ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
1276 zebra_rec_keys_t reckeys,
1277 zebra_snippets *snippets)
1279 NMEM nmem = nmem_create();
1280 if (zebra_rec_keys_rewind(reckeys))
1285 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1287 char dst_buf[IT_MAX_WORD];
1288 char *dst_term = dst_buf;
1293 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1294 seqno = key.mem[key.len-1];
1295 ord = CAST_ZINT_TO_INT(key.mem[0]);
1297 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1298 0/* db */, 0 /* string_index */);
1300 zebra_term_untrans_iconv(zh, nmem, index_type,
1302 zebra_snippets_append(snippets, seqno, ord, dst_term);
1310 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1312 yaz_log(YLOG_LOG, "print_rec_keys");
1313 if (zebra_rec_keys_rewind(reckeys))
1318 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1320 char dst_buf[IT_MAX_WORD];
1323 int ord = CAST_ZINT_TO_INT(key.mem[0]);
1325 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1327 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1329 seqno = key.mem[key.len-1];
1331 zebra_term_untrans(zh, index_type, dst_buf, str);
1333 yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
1334 " term=%s", ord, seqno, dst_buf);
1339 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1340 const char *str, int length)
1343 ZebraHandle zh = p->extractCtrl->handle;
1344 ZebraExplainInfo zei = zh->reg->zei;
1347 ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1349 ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1353 key.mem[i++] = p->record_id;
1354 key.mem[i++] = p->section_id;
1356 if (zh->m_segment_indexing)
1357 key.mem[i++] = p->segment;
1358 key.mem[i++] = p->seqno;
1361 zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1364 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1367 ZebraHandle zh = p->extractCtrl->handle;
1368 ZebraExplainInfo zei = zh->reg->zei;
1370 zinfo_index_category_t cat = zinfo_index_category_sort;
1372 ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1374 ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1377 key.mem[1] = p->record_id;
1379 zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1382 static void extract_add_string(RecWord *p, const char *string, int length)
1384 ZebraHandle zh = p->extractCtrl->handle;
1385 assert (length > 0);
1390 if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
1391 extract_add_sort_string(p, string, length);
1394 extract_add_index_string(p, zinfo_index_category_index,
1396 if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type))
1399 memcpy(&word, p, sizeof(word));
1402 extract_add_index_string(
1403 &word, zinfo_index_category_alwaysmatches, "", 0);
1408 static void extract_add_incomplete_field(RecWord *p)
1410 ZebraHandle zh = p->extractCtrl->handle;
1411 const char *b = p->term_buf;
1412 int remain = p->term_len;
1413 const char **map = 0;
1416 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1420 if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
1422 /* first in field marker */
1423 extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1429 char buf[IT_MAX_WORD+1];
1433 while (map && *map && **map == *CHR_SPACE)
1435 remain = p->term_len - (b - p->term_buf);
1437 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
1445 while (map && *map && **map != *CHR_SPACE)
1447 const char *cp = *map;
1449 while (i < IT_MAX_WORD && *cp)
1451 remain = p->term_len - (b - p->term_buf);
1453 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1459 extract_add_string (p, buf, i);
1464 static void extract_add_complete_field (RecWord *p)
1466 ZebraHandle zh = p->extractCtrl->handle;
1467 const char *b = p->term_buf;
1468 char buf[IT_MAX_WORD+1];
1469 const char **map = 0;
1470 int i = 0, remain = p->term_len;
1473 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
1475 while (remain > 0 && i < IT_MAX_WORD)
1477 while (map && *map && **map == *CHR_SPACE)
1479 remain = p->term_len - (b - p->term_buf);
1483 int first = i ? 0 : 1; /* first position */
1484 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, first);
1492 if (i && i < IT_MAX_WORD)
1493 buf[i++] = *CHR_SPACE;
1494 while (map && *map && **map != *CHR_SPACE)
1496 const char *cp = *map;
1498 if (**map == *CHR_CUT)
1504 if (i >= IT_MAX_WORD)
1506 while (i < IT_MAX_WORD && *cp)
1509 remain = p->term_len - (b - p->term_buf);
1512 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
1521 extract_add_string (p, buf, i);
1524 static void extract_token_add(RecWord *p)
1526 ZebraHandle zh = p->extractCtrl->handle;
1529 if (log_level_extract)
1531 yaz_log(log_level_extract, "extract_token_add "
1532 "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
1533 p->index_type, p->index_name,
1534 p->seqno, p->term_len, p->term_buf);
1536 if ((wrbuf = zebra_replace(zh->reg->zebra_maps, p->index_type, 0,
1537 p->term_buf, p->term_len)))
1539 p->term_buf = wrbuf_buf(wrbuf);
1540 p->term_len = wrbuf_len(wrbuf);
1542 if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type))
1543 extract_add_complete_field (p);
1545 extract_add_incomplete_field(p);
1548 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1549 void *buf, size_t sz)
1551 ZebraHandle zh = (ZebraHandle) p->handle;
1553 xfree(zh->store_data_buf);
1554 zh->store_data_buf = 0;
1555 zh->store_data_size = 0;
1558 zh->store_data_buf = xmalloc(sz);
1559 zh->store_data_size = sz;
1560 memcpy(zh->store_data_buf, buf, sz);
1564 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1566 ZebraHandle zh = (ZebraHandle) p->handle;
1567 xfree(zh->store_data_buf);
1568 zh->store_data_buf = 0;
1569 zh->store_data_size = 0;
1570 p->setStoreData = extract_set_store_data_cb;
1573 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1575 ZebraHandle zh = (ZebraHandle) p->handle;
1576 zebraExplain_addSchema (zh->reg->zei, oid);
1579 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1580 int cmd, zebra_rec_keys_t reckeys)
1582 if (zebra_rec_keys_rewind(reckeys))
1584 SortIdx sortIdx = zh->reg->sortIdx;
1587 struct it_key key_in;
1589 sortIdx_sysno (sortIdx, sysno);
1591 while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1593 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1595 sortIdx_type(sortIdx, ord);
1597 sortIdx_add(sortIdx, str, slen);
1599 sortIdx_add(sortIdx, "", 1);
1604 static void encode_key_init(struct encode_info *i)
1606 i->encode_handle = iscz1_start();
1607 i->decode_handle = iscz1_start();
1610 static void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1613 char *bp = i->buf, *bp0;
1614 const char *src = (char *) &key;
1616 /* copy term to output buf */
1617 while ((*bp++ = *k++))
1619 /* and copy & align key so we can mangle */
1620 memcpy (&key, k+1, sizeof(struct it_key)); /* *k is insert/delete */
1624 key_logdump_txt(YLOG_LOG, &key, *k ? "i" : "d");
1626 assert(key.mem[0] >= 0);
1629 iscz1_encode(i->encode_handle, &bp, &src);
1631 *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */
1632 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1634 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
1635 zebra_exit("encode_key_write");
1643 const char *src = bp0+1;
1644 char *dst = (char*) &key2;
1645 iscz1_decode(i->decode_handle, &dst, &src);
1647 key_logdump_txt(YLOG_LOG, &key2, *k ? "i" : "d");
1649 assert(key2.mem[1]);
1654 static void encode_key_flush (struct encode_info *i, FILE *outf)
1656 iscz1_stop(i->encode_handle);
1657 iscz1_stop(i->decode_handle);
1663 * indent-tabs-mode: nil
1665 * vim: shiftwidth=4 tabstop=8 expandtab