1 /* $Id: extract.c,v 1.232 2006-10-29 20:35:58 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 = 0;
47 static int log_level_initialized = 0;
49 static void zebra_init_log_level(void)
51 if (!log_level_initialized)
53 log_level = yaz_log_module_level("extract");
54 log_level_initialized = 1;
58 static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
59 int cmd, zebra_rec_keys_t reckeys,
61 static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
62 int cmd, zebra_rec_keys_t skp);
63 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid);
64 static void extract_token_add (RecWord *p);
66 static void encode_key_init (struct encode_info *i);
67 static void encode_key_write (char *k, struct encode_info *i, FILE *outf);
68 static void encode_key_flush (struct encode_info *i, FILE *outf);
70 #define USE_SHELLSORT 0
73 static void shellsort(void *ar, int r, size_t s,
74 int (*cmp)(const void *a, const void *b))
79 static const int incs[16] = { 1391376, 463792, 198768, 86961, 33936,
80 13776, 4592, 1968, 861, 336,
81 112, 48, 21, 7, 3, 1 };
82 for ( k = 0; k < 16; k++)
83 for (h = incs[k], i = h; i < r; i++)
87 while (j > h && (*cmp)(a + s*(j-h), v) > 0)
89 memcpy (a + s*j, a + s*(j-h), s);
97 static void logRecord (ZebraHandle zh)
99 ++zh->records_processed;
100 if (!(zh->records_processed % 1000))
102 yaz_log(YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
103 ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT,
104 zh->records_processed, zh->records_inserted,
105 zh->records_updated, zh->records_deleted);
109 static void extract_add_index_string (RecWord *p,
110 zinfo_index_category_t cat,
111 const char *str, int length);
113 static void extract_set_store_data_prepare(struct recExtractCtrl *p);
115 static void extract_init(struct recExtractCtrl *p, RecWord *w)
118 w->index_name = "any";
126 static void searchRecordKey(ZebraHandle zh,
127 zebra_rec_keys_t reckeys,
128 const char *index_name,
129 const char **ws, int ws_length)
133 zinfo_index_category_t cat = zinfo_index_category_index;
135 for (i = 0; i<ws_length; i++)
139 ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, '0', index_name);
141 ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'p', index_name);
143 ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'w', index_name);
148 if (zebra_rec_keys_rewind(reckeys))
155 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
157 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
159 seqno = key.mem[key.len-1];
161 if (key.mem[0] == ch)
167 woff = seqno - startSeq;
168 if (woff >= 0 && woff < ws_length)
175 #define FILE_MATCH_BLANK "\t "
177 static char *fileMatchStr (ZebraHandle zh,
178 zebra_rec_keys_t reckeys,
179 const char *fname, const char *spec)
181 static char dstBuf[2048]; /* static here ??? */
183 const char *s = spec;
187 for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
194 char attset_str[64], attname_str[64];
198 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
200 for (i = 0; *s && *s != ',' && *s != ')' &&
201 !strchr(FILE_MATCH_BLANK, *s); s++)
202 if (i+1 < sizeof(attset_str))
203 attset_str[i++] = *s;
204 attset_str[i] = '\0';
206 for (; strchr(FILE_MATCH_BLANK, *s); s++)
209 strcpy(attname_str, attset_str);
212 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
214 for (i = 0; *s && *s != ')' &&
215 !strchr(FILE_MATCH_BLANK, *s); s++)
216 if (i+1 < sizeof(attname_str))
217 attname_str[i++] = *s;
218 attname_str[i] = '\0';
221 searchRecordKey (zh, reckeys, attname_str, ws, 32);
225 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
226 spec, zh->m_group ? zh->m_group : "none");
231 for (i = 0; i<32; i++)
240 dst += strlen(ws[i]);
244 yaz_log (YLOG_WARN, "Record didn't contain match"
245 " fields in (%s,%s)", attset_str, attname_str);
253 const char *spec_src = NULL;
254 const char *s1 = ++s;
255 while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
259 if (spec_len > sizeof(special)-1)
260 spec_len = sizeof(special)-1;
261 memcpy (special, s, spec_len);
262 special[spec_len] = '\0';
265 if (!strcmp (special, "group"))
266 spec_src = zh->m_group;
267 else if (!strcmp (special, "database"))
268 spec_src = zh->basenames[0];
269 else if (!strcmp (special, "filename")) {
272 else if (!strcmp (special, "type"))
273 spec_src = zh->m_record_type;
278 strcpy (dst, spec_src);
279 dst += strlen (spec_src);
282 else if (*s == '\"' || *s == '\'')
284 int stopMarker = *s++;
288 while (*s && *s != stopMarker)
290 if (i+1 < sizeof(tmpString))
291 tmpString[i++] = *s++;
296 strcpy (dst, tmpString);
297 dst += strlen (tmpString);
301 yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
302 spec, zh->m_group ? zh->m_group : "none");
309 yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
310 fname, zh->m_group ? zh->m_group : "none");
317 struct recordLogInfo {
320 struct recordGroup *rGroup;
323 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
326 for (i = 0; i<256; i++)
328 if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
333 ctrl->flagShowRecords = !zh->m_flag_rw;
336 static void all_matches_add(struct recExtractCtrl *ctrl)
339 extract_init(ctrl, &word);
340 word.index_name = "_ALLRECORDS";
341 word.index_type = 'w';
343 extract_add_index_string (&word, zinfo_index_category_alwaysmatches,
347 ZEBRA_RES zebra_extract_file(ZebraHandle zh, SYSNO *sysno, const char *fname,
350 ZEBRA_RES r = ZEBRA_OK;
355 struct file_read_info *fi = 0;
356 const char *original_record_type = 0;
358 void *recTypeClientData;
359 struct ZebraRecStream stream, *streamp;
361 zebra_init_log_level();
363 if (!zh->m_group || !*zh->m_group)
366 sprintf (gprefix, "%s.", zh->m_group);
368 yaz_log(log_level, "zebra_extract_file %s", fname);
370 /* determine file extension */
372 for (i = strlen(fname); --i >= 0; )
375 else if (fname[i] == '.')
377 strcpy (ext, fname+i+1);
380 /* determine file type - depending on extension */
381 original_record_type = zh->m_record_type;
382 if (!zh->m_record_type)
384 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
385 zh->m_record_type = res_get (zh->res, ext_res);
387 if (!zh->m_record_type)
389 if (zh->records_processed < zh->m_file_verbose_limit)
390 yaz_log (YLOG_LOG, "? %s", fname);
393 /* determine match criteria */
394 if (!zh->m_record_id)
396 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
397 zh->m_record_id = res_get (zh->res, ext_res);
401 recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
402 &recTypeClientData)))
404 yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
408 switch(recType->version)
413 yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
415 if (sysno && deleteFlag)
424 if (zh->path_reg && !yaz_is_abspath (fname))
426 strcpy (full_rep, zh->path_reg);
427 strcat (full_rep, "/");
428 strcat (full_rep, fname);
431 strcpy (full_rep, fname);
433 if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
435 yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
436 zh->m_record_type = original_record_type;
440 zebra_create_stream_fd(streamp, fd, 0);
444 off_t prev_off = streamp->tellf(streamp);
446 r = zebra_extract_record_stream(zh, streamp,
451 0, /*match_criteria */
453 1, /* force_update */
454 1, /* allow_update */
455 recType, recTypeClientData);
464 if (prev_off == streamp->tellf(streamp))
468 stream.destroy(streamp);
469 zh->m_record_type = original_record_type;
474 If sysno is provided, then it's used to identify the reocord.
475 If not, and match_criteria is provided, then sysno is guessed
476 If not, and a record is provided, then sysno is got from there
480 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
481 const char *buf, size_t buf_size,
484 const char *recordType,
486 const char *match_criteria,
491 struct ZebraRecStream stream;
496 if (recordType && *recordType)
498 yaz_log(log_level, "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, "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;
828 else if (zh->m_store_data)
830 off_t cur_offset = stream->tellf(stream);
832 rec->size[recInfo_storeData] = recordAttr->recordSize;
833 rec->info[recInfo_storeData] = (char *)
834 xmalloc (recordAttr->recordSize);
835 stream->seekf(stream, start_offset);
836 stream->readf(stream, rec->info[recInfo_storeData],
837 recordAttr->recordSize);
838 stream->seekf(stream, cur_offset);
842 rec->info[recInfo_storeData] = NULL;
843 rec->size[recInfo_storeData] = 0;
845 /* update database name */
846 xfree (rec->info[recInfo_databaseName]);
847 rec->info[recInfo_databaseName] =
848 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
851 recordAttr->recordOffset = start_offset;
853 /* commit this record */
854 rec_put (zh->reg->records, &rec);
859 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
861 ZebraHandle zh = (ZebraHandle) handle;
862 struct recExtractCtrl extractCtrl;
864 if (zebraExplain_curDatabase (zh->reg->zei,
865 rec->info[recInfo_databaseName]))
868 if (zebraExplain_newDatabase (zh->reg->zei,
869 rec->info[recInfo_databaseName], 0))
873 zebra_rec_keys_reset(zh->reg->keys);
874 zebra_rec_keys_reset(zh->reg->sortKeys);
876 extractCtrl.init = extract_init;
877 extractCtrl.tokenAdd = extract_token_add;
878 extractCtrl.schemaAdd = extract_schema_add;
879 extractCtrl.dh = zh->reg->dh;
881 init_extractCtrl(zh, &extractCtrl);
883 extractCtrl.flagShowRecords = 0;
884 extractCtrl.match_criteria[0] = '\0';
885 extractCtrl.staticrank = 0;
886 extractCtrl.handle = handle;
887 extractCtrl.first_record = 1;
889 extract_set_store_data_prepare(&extractCtrl);
892 grs_extract_tree(&extractCtrl, n);
894 if (rec->size[recInfo_delKeys])
896 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
898 zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
900 zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
901 rec->size[recInfo_delKeys],
903 extract_flushRecordKeys (zh, rec->sysno, 0, delkeys, 0);
904 zebra_rec_keys_close(delkeys);
906 zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
907 rec->size[recInfo_sortKeys],
910 extract_flushSortKeys (zh, rec->sysno, 0, sortkeys);
911 zebra_rec_keys_close(sortkeys);
913 extract_flushRecordKeys (zh, rec->sysno, 1, zh->reg->keys, 0);
914 extract_flushSortKeys (zh, rec->sysno, 1, zh->reg->sortKeys);
916 xfree (rec->info[recInfo_delKeys]);
917 zebra_rec_keys_get_buf(zh->reg->keys,
918 &rec->info[recInfo_delKeys],
919 &rec->size[recInfo_delKeys]);
921 xfree (rec->info[recInfo_sortKeys]);
922 zebra_rec_keys_get_buf(zh->reg->sortKeys,
923 &rec->info[recInfo_sortKeys],
924 &rec->size[recInfo_sortKeys]);
928 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
929 zebra_rec_keys_t reckeys)
931 ZebraExplainInfo zei = zh->reg->zei;
935 struct ord_stat *next;
938 if (zebra_rec_keys_rewind(reckeys))
940 struct ord_stat *ord_list = 0;
944 struct it_key key_in;
945 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
947 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
949 for (p = ord_list; p ; p = p->next)
957 p = xmalloc(sizeof(*p));
968 struct ord_stat *p1 = p;
971 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
973 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
980 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
982 zebra_rec_keys_t reckeys,
985 ZebraExplainInfo zei = zh->reg->zei;
987 extract_rec_keys_adjust(zh, cmd, reckeys);
989 if (!zh->reg->key_buf)
991 int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
994 yaz_log(YLOG_WARN, "Invalid memory setting, using default 8 MB");
997 /* FIXME: That "8" should be in a default settings include */
998 /* not hard-coded here! -H */
999 zh->reg->key_buf = (char**) xmalloc (mem);
1000 zh->reg->ptr_top = mem/sizeof(char*);
1002 zh->reg->key_buf_used = 0;
1003 zh->reg->key_file_no = 0;
1005 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1007 if (zebra_rec_keys_rewind(reckeys))
1011 struct it_key key_in;
1012 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1016 struct it_key key_out;
1018 assert(key_in.len >= 2);
1019 assert(key_in.len <= IT_KEY_LEVEL_MAX);
1021 /* check for buffer overflow */
1022 if (zh->reg->key_buf_used + 1024 >
1023 (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
1024 extract_flushWriteKeys (zh, 0);
1027 assert(zh->reg->ptr_i > 0);
1028 (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
1029 (char*)zh->reg->key_buf + zh->reg->key_buf_used;
1031 /* key_in.mem[0] ord/ch */
1032 /* key_in.mem[1] filter specified record ID */
1034 /* encode the ordinal value (field/use/attribute) .. */
1035 ch = CAST_ZINT_TO_INT(key_in.mem[0]);
1036 zh->reg->key_buf_used +=
1037 key_SU_encode(ch, (char*)zh->reg->key_buf +
1038 zh->reg->key_buf_used);
1040 /* copy the 0-terminated stuff from str to output */
1041 memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used, str, slen);
1042 zh->reg->key_buf_used += slen;
1043 ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = '\0';
1045 /* the delete/insert indicator */
1046 ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = cmd;
1048 if (zh->m_staticrank) /* rank config enabled ? */
1052 yaz_log(YLOG_WARN, "staticrank = %ld. Setting to 0",
1056 key_out.mem[j++] = staticrank;
1059 if (key_in.mem[1]) /* filter specified record ID */
1060 key_out.mem[j++] = key_in.mem[1];
1062 key_out.mem[j++] = sysno;
1063 for (i = 2; i < key_in.len; i++)
1064 key_out.mem[j++] = key_in.mem[i];
1067 memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used,
1068 &key_out, sizeof(key_out));
1069 (zh->reg->key_buf_used) += sizeof(key_out);
1074 void extract_flushWriteKeys (ZebraHandle zh, int final)
1075 /* optimizing: if final=1, and no files written yet */
1076 /* push the keys directly to merge, sidestepping the */
1077 /* temp file altogether. Speeds small updates */
1080 char out_fname[200];
1082 struct encode_info encode_info;
1083 int ptr_i = zh->reg->ptr_i;
1088 if (!zh->reg->key_buf || ptr_i <= 0)
1090 yaz_log(log_level, " nothing to flush section=%d buf=%p i=%d",
1091 zh->reg->key_file_no, zh->reg->key_buf, ptr_i);
1095 (zh->reg->key_file_no)++;
1096 yaz_log (YLOG_LOG, "sorting section %d", (zh->reg->key_file_no));
1097 yaz_log(log_level, " sort_buff at %p n=%d",
1098 zh->reg->key_buf + zh->reg->ptr_top - ptr_i,ptr_i);
1100 qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
1101 sizeof(char*), key_qsort_compare);
1103 /* zebra.cfg: tempfiles:
1104 Y: always use temp files (old way)
1105 A: use temp files, if more than one (auto)
1106 = if this is both the last and the first
1107 N: never bother with temp files (new) */
1109 temp_policy=toupper(res_get_def(zh->res,"tempfiles","auto")[0]);
1110 if (temp_policy != 'Y' && temp_policy != 'N' && temp_policy != 'A') {
1111 yaz_log (YLOG_WARN, "Illegal tempfiles setting '%c'. using 'Auto' ",
1116 if ( ( temp_policy =='N' ) || /* always from memory */
1117 ( ( temp_policy =='A' ) && /* automatic */
1118 (zh->reg->key_file_no == 1) && /* this is first time */
1119 (final) ) ) /* and last (=only) time */
1120 { /* go directly from memory */
1121 zh->reg->key_file_no =0; /* signal not to read files */
1122 zebra_index_merge(zh);
1124 zh->reg->key_buf_used = 0;
1128 /* Not doing directly from memory, write into a temp file */
1129 extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
1131 if (!(outf = fopen (out_fname, "wb")))
1133 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1136 yaz_log (YLOG_LOG, "writing section %d", zh->reg->key_file_no);
1137 prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1139 encode_key_init (&encode_info);
1140 encode_key_write (cp, &encode_info, outf);
1144 cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1145 if (strcmp (cp, prevcp))
1147 encode_key_flush ( &encode_info, outf);
1148 encode_key_init (&encode_info);
1149 encode_key_write (cp, &encode_info, outf);
1153 encode_key_write (cp + strlen(cp), &encode_info, outf);
1155 encode_key_flush ( &encode_info, outf);
1157 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1158 extract_get_fname_tmp (out_fname, key_file_no);
1160 if (!(outf = fopen (out_fname, "wb")))
1162 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1165 yaz_log (YLOG_LOG, "writing section %d", key_file_no);
1167 prevcp = key_buf[ptr_top-i];
1169 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1171 key_y_len = strlen(prevcp)+1;
1173 yaz_log (YLOG_LOG, "key_y_len: %2d %02x %02x %s",
1174 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1176 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1177 sizeof(char*), key_y_compare);
1178 cp = key_buf[ptr_top-ptr_i];
1180 encode_key_init (&encode_info);
1181 encode_key_write (cp, &encode_info, outf);
1184 cp = key_buf[ptr_top-ptr_i];
1185 encode_key_write (cp+key_y_len, &encode_info, outf);
1187 encode_key_flush ( &encode_info, outf);
1190 prevcp = key_buf[ptr_top-ptr_i];
1195 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fclose %s", out_fname);
1198 yaz_log (YLOG_LOG, "finished section %d", zh->reg->key_file_no);
1200 zh->reg->key_buf_used = 0;
1203 ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh,
1204 zebra_rec_keys_t reckeys,
1205 zebra_snippets *snippets)
1207 NMEM nmem = nmem_create();
1208 if (zebra_rec_keys_rewind(reckeys))
1213 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1215 char dst_buf[IT_MAX_WORD];
1216 char *dst_term = dst_buf;
1221 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1222 seqno = key.mem[key.len-1];
1223 ord = CAST_ZINT_TO_INT(key.mem[0]);
1225 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1226 0/* db */, 0 /* string_index */);
1228 zebra_term_untrans_iconv(zh, nmem, index_type,
1230 zebra_snippets_append(snippets, seqno, ord, dst_term);
1238 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1240 yaz_log(YLOG_LOG, "print_rec_keys");
1241 if (zebra_rec_keys_rewind(reckeys))
1246 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1248 char dst_buf[IT_MAX_WORD];
1251 int ord = CAST_ZINT_TO_INT(key.mem[0]);
1253 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1255 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1257 seqno = key.mem[key.len-1];
1259 zebra_term_untrans(zh, index_type, dst_buf, str);
1261 yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
1262 " term=%s", ord, seqno, dst_buf);
1267 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1268 const char *str, int length)
1271 ZebraHandle zh = p->extractCtrl->handle;
1272 ZebraExplainInfo zei = zh->reg->zei;
1275 ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1277 ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1281 key.mem[i++] = p->record_id;
1282 key.mem[i++] = p->section_id;
1284 if (zh->m_segment_indexing)
1285 key.mem[i++] = p->segment;
1286 key.mem[i++] = p->seqno;
1289 zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1292 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1295 ZebraHandle zh = p->extractCtrl->handle;
1296 ZebraExplainInfo zei = zh->reg->zei;
1298 zinfo_index_category_t cat = zinfo_index_category_sort;
1300 ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1302 ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1305 key.mem[1] = p->record_id;
1307 zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1310 static void extract_add_string(RecWord *p, const char *string, int length)
1312 ZebraHandle zh = p->extractCtrl->handle;
1313 assert (length > 0);
1318 if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
1319 extract_add_sort_string(p, string, length);
1322 extract_add_index_string(p, zinfo_index_category_index,
1324 if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type))
1327 memcpy(&word, p, sizeof(word));
1330 extract_add_index_string(
1331 &word, zinfo_index_category_alwaysmatches, "", 0);
1336 static void extract_add_incomplete_field(RecWord *p)
1338 ZebraHandle zh = p->extractCtrl->handle;
1339 const char *b = p->term_buf;
1340 int remain = p->term_len;
1341 const char **map = 0;
1344 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1348 if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
1350 /* first in field marker */
1351 extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1357 char buf[IT_MAX_WORD+1];
1361 while (map && *map && **map == *CHR_SPACE)
1363 remain = p->term_len - (b - p->term_buf);
1365 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
1373 while (map && *map && **map != *CHR_SPACE)
1375 const char *cp = *map;
1377 while (i < IT_MAX_WORD && *cp)
1379 remain = p->term_len - (b - p->term_buf);
1381 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1387 extract_add_string (p, buf, i);
1392 static void extract_add_complete_field (RecWord *p)
1394 ZebraHandle zh = p->extractCtrl->handle;
1395 const char *b = p->term_buf;
1396 char buf[IT_MAX_WORD+1];
1397 const char **map = 0;
1398 int i = 0, remain = p->term_len;
1401 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
1403 while (remain > 0 && i < IT_MAX_WORD)
1405 while (map && *map && **map == *CHR_SPACE)
1407 remain = p->term_len - (b - p->term_buf);
1411 int first = i ? 0 : 1; /* first position */
1412 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, first);
1420 if (i && i < IT_MAX_WORD)
1421 buf[i++] = *CHR_SPACE;
1422 while (map && *map && **map != *CHR_SPACE)
1424 const char *cp = *map;
1426 if (**map == *CHR_CUT)
1432 if (i >= IT_MAX_WORD)
1434 while (i < IT_MAX_WORD && *cp)
1437 remain = p->term_len - (b - p->term_buf);
1440 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
1449 extract_add_string (p, buf, i);
1452 static void extract_token_add(RecWord *p)
1454 ZebraHandle zh = p->extractCtrl->handle;
1459 yaz_log(log_level, "extract_token_add "
1460 "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
1461 p->index_type, p->index_name,
1462 p->seqno, p->term_len, p->term_buf);
1464 if ((wrbuf = zebra_replace(zh->reg->zebra_maps, p->index_type, 0,
1465 p->term_buf, p->term_len)))
1467 p->term_buf = wrbuf_buf(wrbuf);
1468 p->term_len = wrbuf_len(wrbuf);
1470 if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type))
1471 extract_add_complete_field (p);
1473 extract_add_incomplete_field(p);
1476 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1477 void *buf, size_t sz)
1479 ZebraHandle zh = (ZebraHandle) p->handle;
1481 xfree(zh->store_data_buf);
1482 zh->store_data_buf = 0;
1483 zh->store_data_size = 0;
1486 zh->store_data_buf = xmalloc(sz);
1487 zh->store_data_size = sz;
1488 memcpy(zh->store_data_buf, buf, sz);
1492 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1494 ZebraHandle zh = (ZebraHandle) p->handle;
1495 xfree(zh->store_data_buf);
1496 zh->store_data_buf = 0;
1497 zh->store_data_size = 0;
1498 p->setStoreData = extract_set_store_data_cb;
1501 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1503 ZebraHandle zh = (ZebraHandle) p->handle;
1504 zebraExplain_addSchema (zh->reg->zei, oid);
1507 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1508 int cmd, zebra_rec_keys_t reckeys)
1510 if (zebra_rec_keys_rewind(reckeys))
1512 SortIdx sortIdx = zh->reg->sortIdx;
1515 struct it_key key_in;
1517 sortIdx_sysno (sortIdx, sysno);
1519 while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1521 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1523 sortIdx_type(sortIdx, ord);
1525 sortIdx_add(sortIdx, str, slen);
1527 sortIdx_add(sortIdx, "", 1);
1532 static void encode_key_init(struct encode_info *i)
1534 i->encode_handle = iscz1_start();
1535 i->decode_handle = iscz1_start();
1538 static void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1541 char *bp = i->buf, *bp0;
1542 const char *src = (char *) &key;
1544 /* copy term to output buf */
1545 while ((*bp++ = *k++))
1547 /* and copy & align key so we can mangle */
1548 memcpy (&key, k+1, sizeof(struct it_key)); /* *k is insert/delete */
1552 key_logdump_txt(YLOG_LOG, &key, *k ? "i" : "d");
1554 assert(key.mem[0] >= 0);
1557 iscz1_encode(i->encode_handle, &bp, &src);
1559 *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */
1560 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1562 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
1571 const char *src = bp0+1;
1572 char *dst = (char*) &key2;
1573 iscz1_decode(i->decode_handle, &dst, &src);
1575 key_logdump_txt(YLOG_LOG, &key2, *k ? "i" : "d");
1577 assert(key2.mem[1]);
1582 static void encode_key_flush (struct encode_info *i, FILE *outf)
1584 iscz1_stop(i->encode_handle);
1585 iscz1_stop(i->decode_handle);
1591 * indent-tabs-mode: nil
1593 * vim: shiftwidth=4 tabstop=8 expandtab