1 /* $Id: extract.c,v 1.230 2006-09-15 10:45:12 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()
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 r = zebra_extract_record_stream(zh, streamp,
449 0, /*match_criteria */
451 1, /* force_update */
452 1, /* allow_update */
453 recType, recTypeClientData);
464 stream.destroy(streamp);
465 zh->m_record_type = original_record_type;
470 If sysno is provided, then it's used to identify the reocord.
471 If not, and match_criteria is provided, then sysno is guessed
472 If not, and a record is provided, then sysno is got from there
476 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
477 const char *buf, size_t buf_size,
480 const char *recordType,
482 const char *match_criteria,
487 struct ZebraRecStream stream;
492 if (recordType && *recordType)
494 yaz_log(log_level, "Record type explicitly specified: %s", recordType);
495 recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
500 if (!(zh->m_record_type))
502 yaz_log (YLOG_WARN, "No such record type defined");
505 yaz_log(log_level, "Get record type from rgroup: %s",
507 recType = recType_byName (zh->reg->recTypes, zh->res,
508 zh->m_record_type, &clientData);
509 recordType = zh->m_record_type;
514 yaz_log (YLOG_WARN, "No such record type: %s", recordType);
520 zebra_create_stream_mem(&stream, buf, buf_size);
522 res = zebra_extract_record_stream(zh, &stream,
531 recType, clientData);
532 stream.destroy(&stream);
537 ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
538 struct ZebraRecStream *stream,
541 const char *recordType,
543 const char *match_criteria,
548 void *recTypeClientData)
552 RecordAttr *recordAttr;
553 struct recExtractCtrl extractCtrl;
555 const char *matchStr = 0;
557 off_t start_offset = 0;
558 const char *pr_fname = fname; /* filename to print .. */
559 int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
561 zebra_init_log_level();
564 pr_fname = "<no file>"; /* make it printable if file is omitted */
566 zebra_rec_keys_reset(zh->reg->keys);
567 zebra_rec_keys_reset(zh->reg->sortKeys);
569 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
571 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
572 zh->m_explain_database))
578 off_t null_offset = 0;
579 extractCtrl.stream = stream;
581 start_offset = stream->tellf(stream);
583 extractCtrl.first_record = start_offset ? 0 : 1;
585 stream->endf(stream, &null_offset);;
587 extractCtrl.init = extract_init;
588 extractCtrl.tokenAdd = extract_token_add;
589 extractCtrl.schemaAdd = extract_schema_add;
590 extractCtrl.dh = zh->reg->dh;
591 extractCtrl.handle = zh;
592 extractCtrl.match_criteria[0] = '\0';
593 extractCtrl.staticrank = 0;
596 init_extractCtrl(zh, &extractCtrl);
598 extract_set_store_data_prepare(&extractCtrl);
600 r = (*recType->extract)(recTypeClientData, &extractCtrl);
602 if (r == RECCTRL_EXTRACT_EOF)
604 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
606 /* error occured during extraction ... */
607 yaz_log (YLOG_WARN, "extract error: generic");
610 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
612 /* error occured during extraction ... */
613 yaz_log (YLOG_WARN, "extract error: no such filter");
617 all_matches_add(&extractCtrl);
619 if (extractCtrl.match_criteria[0])
620 match_criteria = extractCtrl.match_criteria;
626 if (match_criteria && *match_criteria) {
627 matchStr = match_criteria;
629 if (zh->m_record_id && *zh->m_record_id) {
630 matchStr = fileMatchStr (zh, zh->reg->keys, pr_fname,
634 yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
641 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
642 char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
646 assert(*rinfo == sizeof(*sysno));
647 memcpy (sysno, rinfo+1, sizeof(*sysno));
651 if (zebra_rec_keys_empty(zh->reg->keys))
653 /* the extraction process returned no information - the record
654 is probably empty - unless flagShowRecords is in use */
664 yaz_log (YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
665 pr_fname, (zint) start_offset);
666 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
670 yaz_log (YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
671 (zint) start_offset);
672 rec = rec_new (zh->reg->records);
676 recordAttr = rec_init_attr (zh->reg->zei, rec);
677 recordAttr->staticrank = extractCtrl.staticrank;
681 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
682 dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
683 sizeof(*sysno), sysno);
687 extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
688 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
689 recordAttr->staticrank);
690 zh->records_inserted++;
694 /* record already exists */
695 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
696 zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
699 yaz_log (YLOG_LOG, "skipped %s %s " ZINT_FORMAT,
700 recordType, pr_fname, (zint) start_offset);
705 rec = rec_get (zh->reg->records, *sysno);
708 recordAttr = rec_init_attr (zh->reg->zei, rec);
710 zebra_rec_keys_set_buf(delkeys,
711 rec->info[recInfo_delKeys],
712 rec->size[recInfo_delKeys],
714 zebra_rec_keys_set_buf(sortKeys,
715 rec->info[recInfo_sortKeys],
716 rec->size[recInfo_sortKeys],
719 extract_flushSortKeys (zh, *sysno, 0, sortKeys);
720 extract_flushRecordKeys (zh, *sysno, 0, delkeys,
721 recordAttr->staticrank);
724 /* record going to be deleted */
725 if (zebra_rec_keys_empty(delkeys))
727 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
728 pr_fname, (zint) start_offset);
729 yaz_log(YLOG_WARN, "cannot delete file above, "
730 "storeKeys false (3)");
735 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
736 pr_fname, (zint) start_offset);
737 zh->records_deleted++;
740 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
741 dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
743 rec_del (zh->reg->records, &rec);
752 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
753 pr_fname, (zint) ZINT_FORMAT);
754 recordAttr->staticrank = extractCtrl.staticrank;
755 extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
756 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
757 recordAttr->staticrank);
758 zh->records_updated++;
760 zebra_rec_keys_close(delkeys);
761 zebra_rec_keys_close(sortKeys);
763 /* update file type */
764 xfree (rec->info[recInfo_fileType]);
765 rec->info[recInfo_fileType] =
766 rec_strdup (recordType, &rec->size[recInfo_fileType]);
768 /* update filename */
769 xfree (rec->info[recInfo_filename]);
770 rec->info[recInfo_filename] =
771 rec_strdup (fname, &rec->size[recInfo_filename]);
773 /* update delete keys */
774 xfree (rec->info[recInfo_delKeys]);
775 if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
777 zebra_rec_keys_get_buf(zh->reg->keys,
778 &rec->info[recInfo_delKeys],
779 &rec->size[recInfo_delKeys]);
783 rec->info[recInfo_delKeys] = NULL;
784 rec->size[recInfo_delKeys] = 0;
786 /* update sort keys */
787 xfree (rec->info[recInfo_sortKeys]);
789 zebra_rec_keys_get_buf(zh->reg->sortKeys,
790 &rec->info[recInfo_sortKeys],
791 &rec->size[recInfo_sortKeys]);
793 /* save file size of original record */
794 zebraExplain_recordBytesIncrement (zh->reg->zei,
795 - recordAttr->recordSize);
798 off_t end_offset = stream->endf(stream, 0);
801 end_offset = stream->tellf(stream);
803 stream->seekf(stream, end_offset);
805 recordAttr->recordSize = end_offset - start_offset;
806 zebraExplain_recordBytesIncrement(zh->reg->zei,
807 recordAttr->recordSize);
810 /* set run-number for this record */
811 recordAttr->runNumber =
812 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
814 /* update store data */
815 xfree (rec->info[recInfo_storeData]);
817 /* update store data */
818 if (zh->store_data_buf)
820 rec->size[recInfo_storeData] = zh->store_data_size;
821 rec->info[recInfo_storeData] = zh->store_data_buf;
822 zh->store_data_buf = 0;
824 else if (zh->m_store_data)
826 off_t cur_offset = stream->tellf(stream);
828 rec->size[recInfo_storeData] = recordAttr->recordSize;
829 rec->info[recInfo_storeData] = (char *)
830 xmalloc (recordAttr->recordSize);
831 stream->seekf(stream, start_offset);
832 stream->readf(stream, rec->info[recInfo_storeData],
833 recordAttr->recordSize);
834 stream->seekf(stream, cur_offset);
838 rec->info[recInfo_storeData] = NULL;
839 rec->size[recInfo_storeData] = 0;
841 /* update database name */
842 xfree (rec->info[recInfo_databaseName]);
843 rec->info[recInfo_databaseName] =
844 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
847 recordAttr->recordOffset = start_offset;
849 /* commit this record */
850 rec_put (zh->reg->records, &rec);
855 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
857 ZebraHandle zh = (ZebraHandle) handle;
858 struct recExtractCtrl extractCtrl;
860 if (zebraExplain_curDatabase (zh->reg->zei,
861 rec->info[recInfo_databaseName]))
864 if (zebraExplain_newDatabase (zh->reg->zei,
865 rec->info[recInfo_databaseName], 0))
869 zebra_rec_keys_reset(zh->reg->keys);
870 zebra_rec_keys_reset(zh->reg->sortKeys);
872 extractCtrl.init = extract_init;
873 extractCtrl.tokenAdd = extract_token_add;
874 extractCtrl.schemaAdd = extract_schema_add;
875 extractCtrl.dh = zh->reg->dh;
877 init_extractCtrl(zh, &extractCtrl);
879 extractCtrl.flagShowRecords = 0;
880 extractCtrl.match_criteria[0] = '\0';
881 extractCtrl.staticrank = 0;
882 extractCtrl.handle = handle;
883 extractCtrl.first_record = 1;
885 extract_set_store_data_prepare(&extractCtrl);
888 grs_extract_tree(&extractCtrl, n);
890 if (rec->size[recInfo_delKeys])
892 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
894 zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
896 zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
897 rec->size[recInfo_delKeys],
899 extract_flushRecordKeys (zh, rec->sysno, 0, delkeys, 0);
900 zebra_rec_keys_close(delkeys);
902 zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
903 rec->size[recInfo_sortKeys],
906 extract_flushSortKeys (zh, rec->sysno, 0, sortkeys);
907 zebra_rec_keys_close(sortkeys);
909 extract_flushRecordKeys (zh, rec->sysno, 1, zh->reg->keys, 0);
910 extract_flushSortKeys (zh, rec->sysno, 1, zh->reg->sortKeys);
912 xfree (rec->info[recInfo_delKeys]);
913 zebra_rec_keys_get_buf(zh->reg->keys,
914 &rec->info[recInfo_delKeys],
915 &rec->size[recInfo_delKeys]);
917 xfree (rec->info[recInfo_sortKeys]);
918 zebra_rec_keys_get_buf(zh->reg->sortKeys,
919 &rec->info[recInfo_sortKeys],
920 &rec->size[recInfo_sortKeys]);
924 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
925 zebra_rec_keys_t reckeys)
927 ZebraExplainInfo zei = zh->reg->zei;
931 struct ord_stat *next;
934 if (zebra_rec_keys_rewind(reckeys))
936 struct ord_stat *ord_list = 0;
940 struct it_key key_in;
941 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
943 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
945 for (p = ord_list; p ; p = p->next)
953 p = xmalloc(sizeof(*p));
964 struct ord_stat *p1 = p;
967 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
969 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
976 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
978 zebra_rec_keys_t reckeys,
981 ZebraExplainInfo zei = zh->reg->zei;
983 extract_rec_keys_adjust(zh, cmd, reckeys);
985 if (!zh->reg->key_buf)
987 int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
990 yaz_log(YLOG_WARN, "Invalid memory setting, using default 8 MB");
993 /* FIXME: That "8" should be in a default settings include */
994 /* not hard-coded here! -H */
995 zh->reg->key_buf = (char**) xmalloc (mem);
996 zh->reg->ptr_top = mem/sizeof(char*);
998 zh->reg->key_buf_used = 0;
999 zh->reg->key_file_no = 0;
1001 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1003 if (zebra_rec_keys_rewind(reckeys))
1007 struct it_key key_in;
1008 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1012 struct it_key key_out;
1014 assert(key_in.len >= 2);
1015 assert(key_in.len <= IT_KEY_LEVEL_MAX);
1017 /* check for buffer overflow */
1018 if (zh->reg->key_buf_used + 1024 >
1019 (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
1020 extract_flushWriteKeys (zh, 0);
1023 assert(zh->reg->ptr_i > 0);
1024 (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
1025 (char*)zh->reg->key_buf + zh->reg->key_buf_used;
1027 /* key_in.mem[0] ord/ch */
1028 /* key_in.mem[1] filter specified record ID */
1030 /* encode the ordinal value (field/use/attribute) .. */
1031 ch = CAST_ZINT_TO_INT(key_in.mem[0]);
1032 zh->reg->key_buf_used +=
1033 key_SU_encode(ch, (char*)zh->reg->key_buf +
1034 zh->reg->key_buf_used);
1036 /* copy the 0-terminated stuff from str to output */
1037 memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used, str, slen);
1038 zh->reg->key_buf_used += slen;
1039 ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = '\0';
1041 /* the delete/insert indicator */
1042 ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = cmd;
1044 if (zh->m_staticrank) /* rank config enabled ? */
1048 yaz_log(YLOG_WARN, "staticrank = %ld. Setting to 0",
1052 key_out.mem[j++] = staticrank;
1055 if (key_in.mem[1]) /* filter specified record ID */
1056 key_out.mem[j++] = key_in.mem[1];
1058 key_out.mem[j++] = sysno;
1059 for (i = 2; i < key_in.len; i++)
1060 key_out.mem[j++] = key_in.mem[i];
1063 memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used,
1064 &key_out, sizeof(key_out));
1065 (zh->reg->key_buf_used) += sizeof(key_out);
1070 void extract_flushWriteKeys (ZebraHandle zh, int final)
1071 /* optimizing: if final=1, and no files written yet */
1072 /* push the keys directly to merge, sidestepping the */
1073 /* temp file altogether. Speeds small updates */
1076 char out_fname[200];
1078 struct encode_info encode_info;
1079 int ptr_i = zh->reg->ptr_i;
1084 if (!zh->reg->key_buf || ptr_i <= 0)
1086 yaz_log(log_level, " nothing to flush section=%d buf=%p i=%d",
1087 zh->reg->key_file_no, zh->reg->key_buf, ptr_i);
1091 (zh->reg->key_file_no)++;
1092 yaz_log (YLOG_LOG, "sorting section %d", (zh->reg->key_file_no));
1093 yaz_log(log_level, " sort_buff at %p n=%d",
1094 zh->reg->key_buf + zh->reg->ptr_top - ptr_i,ptr_i);
1096 qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
1097 sizeof(char*), key_qsort_compare);
1099 /* zebra.cfg: tempfiles:
1100 Y: always use temp files (old way)
1101 A: use temp files, if more than one (auto)
1102 = if this is both the last and the first
1103 N: never bother with temp files (new) */
1105 temp_policy=toupper(res_get_def(zh->res,"tempfiles","auto")[0]);
1106 if (temp_policy != 'Y' && temp_policy != 'N' && temp_policy != 'A') {
1107 yaz_log (YLOG_WARN, "Illegal tempfiles setting '%c'. using 'Auto' ",
1112 if ( ( temp_policy =='N' ) || /* always from memory */
1113 ( ( temp_policy =='A' ) && /* automatic */
1114 (zh->reg->key_file_no == 1) && /* this is first time */
1115 (final) ) ) /* and last (=only) time */
1116 { /* go directly from memory */
1117 zh->reg->key_file_no =0; /* signal not to read files */
1118 zebra_index_merge(zh);
1120 zh->reg->key_buf_used = 0;
1124 /* Not doing directly from memory, write into a temp file */
1125 extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
1127 if (!(outf = fopen (out_fname, "wb")))
1129 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1132 yaz_log (YLOG_LOG, "writing section %d", zh->reg->key_file_no);
1133 prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1135 encode_key_init (&encode_info);
1136 encode_key_write (cp, &encode_info, outf);
1140 cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1141 if (strcmp (cp, prevcp))
1143 encode_key_flush ( &encode_info, outf);
1144 encode_key_init (&encode_info);
1145 encode_key_write (cp, &encode_info, outf);
1149 encode_key_write (cp + strlen(cp), &encode_info, outf);
1151 encode_key_flush ( &encode_info, outf);
1153 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1154 extract_get_fname_tmp (out_fname, key_file_no);
1156 if (!(outf = fopen (out_fname, "wb")))
1158 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1161 yaz_log (YLOG_LOG, "writing section %d", key_file_no);
1163 prevcp = key_buf[ptr_top-i];
1165 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1167 key_y_len = strlen(prevcp)+1;
1169 yaz_log (YLOG_LOG, "key_y_len: %2d %02x %02x %s",
1170 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1172 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1173 sizeof(char*), key_y_compare);
1174 cp = key_buf[ptr_top-ptr_i];
1176 encode_key_init (&encode_info);
1177 encode_key_write (cp, &encode_info, outf);
1180 cp = key_buf[ptr_top-ptr_i];
1181 encode_key_write (cp+key_y_len, &encode_info, outf);
1183 encode_key_flush ( &encode_info, outf);
1186 prevcp = key_buf[ptr_top-ptr_i];
1191 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fclose %s", out_fname);
1194 yaz_log (YLOG_LOG, "finished section %d", zh->reg->key_file_no);
1196 zh->reg->key_buf_used = 0;
1199 ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh,
1200 zebra_rec_keys_t reckeys,
1201 zebra_snippets *snippets)
1203 NMEM nmem = nmem_create();
1204 if (zebra_rec_keys_rewind(reckeys))
1209 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1211 char dst_buf[IT_MAX_WORD];
1212 char *dst_term = dst_buf;
1217 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1218 seqno = key.mem[key.len-1];
1219 ord = CAST_ZINT_TO_INT(key.mem[0]);
1221 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1222 0/* db */, 0 /* string_index */);
1224 zebra_term_untrans_iconv(zh, nmem, index_type,
1226 zebra_snippets_append(snippets, seqno, ord, dst_term);
1234 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1236 yaz_log(YLOG_LOG, "print_rec_keys");
1237 if (zebra_rec_keys_rewind(reckeys))
1242 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1244 char dst_buf[IT_MAX_WORD];
1247 int ord = CAST_ZINT_TO_INT(key.mem[0]);
1249 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1251 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1253 seqno = key.mem[key.len-1];
1255 zebra_term_untrans(zh, index_type, dst_buf, str);
1257 yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
1258 " term=%s", ord, seqno, dst_buf);
1263 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1264 const char *str, int length)
1267 ZebraHandle zh = p->extractCtrl->handle;
1268 ZebraExplainInfo zei = zh->reg->zei;
1271 ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1273 ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1277 key.mem[i++] = p->record_id;
1278 key.mem[i++] = p->section_id;
1280 if (zh->m_segment_indexing)
1281 key.mem[i++] = p->segment;
1282 key.mem[i++] = p->seqno;
1285 zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1288 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1291 ZebraHandle zh = p->extractCtrl->handle;
1292 ZebraExplainInfo zei = zh->reg->zei;
1294 zinfo_index_category_t cat = zinfo_index_category_sort;
1296 ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1298 ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1301 key.mem[1] = p->record_id;
1303 zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1306 static void extract_add_string(RecWord *p, const char *string, int length)
1308 ZebraHandle zh = p->extractCtrl->handle;
1309 assert (length > 0);
1314 if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
1315 extract_add_sort_string(p, string, length);
1318 extract_add_index_string(p, zinfo_index_category_index,
1320 if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type))
1323 memcpy(&word, p, sizeof(word));
1326 extract_add_index_string(
1327 &word, zinfo_index_category_alwaysmatches, "", 0);
1332 static void extract_add_incomplete_field(RecWord *p)
1334 ZebraHandle zh = p->extractCtrl->handle;
1335 const char *b = p->term_buf;
1336 int remain = p->term_len;
1337 const char **map = 0;
1340 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1344 if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type))
1346 /* first in field marker */
1347 extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN);
1353 char buf[IT_MAX_WORD+1];
1357 while (map && *map && **map == *CHR_SPACE)
1359 remain = p->term_len - (b - p->term_buf);
1361 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
1369 while (map && *map && **map != *CHR_SPACE)
1371 const char *cp = *map;
1373 while (i < IT_MAX_WORD && *cp)
1375 remain = p->term_len - (b - p->term_buf);
1377 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1383 extract_add_string (p, buf, i);
1388 static void extract_add_complete_field (RecWord *p)
1390 ZebraHandle zh = p->extractCtrl->handle;
1391 const char *b = p->term_buf;
1392 char buf[IT_MAX_WORD+1];
1393 const char **map = 0;
1394 int i = 0, remain = p->term_len;
1397 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
1399 while (remain > 0 && i < IT_MAX_WORD)
1401 while (map && *map && **map == *CHR_SPACE)
1403 remain = p->term_len - (b - p->term_buf);
1407 int first = i ? 0 : 1; /* first position */
1408 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, first);
1416 if (i && i < IT_MAX_WORD)
1417 buf[i++] = *CHR_SPACE;
1418 while (map && *map && **map != *CHR_SPACE)
1420 const char *cp = *map;
1422 if (**map == *CHR_CUT)
1428 if (i >= IT_MAX_WORD)
1430 while (i < IT_MAX_WORD && *cp)
1433 remain = p->term_len - (b - p->term_buf);
1436 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
1445 extract_add_string (p, buf, i);
1448 static void extract_token_add(RecWord *p)
1450 ZebraHandle zh = p->extractCtrl->handle;
1455 yaz_log(log_level, "extract_token_add "
1456 "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
1457 p->index_type, p->index_name,
1458 p->seqno, p->term_len, p->term_buf);
1460 if ((wrbuf = zebra_replace(zh->reg->zebra_maps, p->index_type, 0,
1461 p->term_buf, p->term_len)))
1463 p->term_buf = wrbuf_buf(wrbuf);
1464 p->term_len = wrbuf_len(wrbuf);
1466 if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type))
1467 extract_add_complete_field (p);
1469 extract_add_incomplete_field(p);
1472 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1473 void *buf, size_t sz)
1475 ZebraHandle zh = (ZebraHandle) p->handle;
1477 xfree(zh->store_data_buf);
1478 zh->store_data_buf = 0;
1479 zh->store_data_size = 0;
1482 zh->store_data_buf = xmalloc(sz);
1483 zh->store_data_size = sz;
1484 memcpy(zh->store_data_buf, buf, sz);
1488 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1490 ZebraHandle zh = (ZebraHandle) p->handle;
1491 xfree(zh->store_data_buf);
1492 zh->store_data_buf = 0;
1493 zh->store_data_size = 0;
1494 p->setStoreData = extract_set_store_data_cb;
1497 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1499 ZebraHandle zh = (ZebraHandle) p->handle;
1500 zebraExplain_addSchema (zh->reg->zei, oid);
1503 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1504 int cmd, zebra_rec_keys_t reckeys)
1506 if (zebra_rec_keys_rewind(reckeys))
1508 SortIdx sortIdx = zh->reg->sortIdx;
1511 struct it_key key_in;
1513 sortIdx_sysno (sortIdx, sysno);
1515 while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1517 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1519 sortIdx_type(sortIdx, ord);
1521 sortIdx_add(sortIdx, str, slen);
1523 sortIdx_add(sortIdx, "", 1);
1528 static void encode_key_init(struct encode_info *i)
1530 i->encode_handle = iscz1_start();
1531 i->decode_handle = iscz1_start();
1534 static void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1537 char *bp = i->buf, *bp0;
1538 const char *src = (char *) &key;
1540 /* copy term to output buf */
1541 while ((*bp++ = *k++))
1543 /* and copy & align key so we can mangle */
1544 memcpy (&key, k+1, sizeof(struct it_key)); /* *k is insert/delete */
1548 key_logdump_txt(YLOG_LOG, &key, *k ? "i" : "d");
1550 assert(key.mem[0] >= 0);
1553 iscz1_encode(i->encode_handle, &bp, &src);
1555 *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */
1556 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1558 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
1567 const char *src = bp0+1;
1568 char *dst = (char*) &key2;
1569 iscz1_decode(i->decode_handle, &dst, &src);
1571 key_logdump_txt(YLOG_LOG, &key2, *k ? "i" : "d");
1573 assert(key2.mem[1]);
1578 static void encode_key_flush (struct encode_info *i, FILE *outf)
1580 iscz1_stop(i->encode_handle);
1581 iscz1_stop(i->decode_handle);
1587 * indent-tabs-mode: nil
1589 * vim: shiftwidth=4 tabstop=8 expandtab