1 /* $Id: extract.c,v 1.227 2006-08-16 13:16:36 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 = 0;
47 static int log_level_initialized = 1;
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 struct file_read_info {
176 off_t file_max; /* maximum offset so far */
177 off_t file_offset; /* current offset */
178 off_t file_moffset; /* offset of rec/rec boundary */
183 static struct file_read_info *file_read_start (int fd)
185 struct file_read_info *fi = (struct file_read_info *)
186 xmalloc (sizeof(*fi));
190 fi->file_moffset = 0;
196 static void file_read_stop (struct file_read_info *fi)
201 static off_t file_seek (void *handle, off_t offset)
203 struct file_read_info *p = (struct file_read_info *) handle;
204 p->file_offset = offset;
205 return lseek (p->fd, offset, SEEK_SET);
208 static off_t file_tell (void *handle)
210 struct file_read_info *p = (struct file_read_info *) handle;
211 return p->file_offset;
214 static int file_read (void *handle, char *buf, size_t count)
216 struct file_read_info *p = (struct file_read_info *) handle;
219 r = read (fd, buf, count);
223 if (p->file_offset > p->file_max)
224 p->file_max = p->file_offset;
229 static void file_end (void *handle, off_t offset)
231 struct file_read_info *p = (struct file_read_info *) handle;
233 if (offset != p->file_moffset)
235 p->file_moffset = offset;
240 #define FILE_MATCH_BLANK "\t "
242 static char *fileMatchStr (ZebraHandle zh,
243 zebra_rec_keys_t reckeys,
244 const char *fname, const char *spec)
246 static char dstBuf[2048]; /* static here ??? */
248 const char *s = spec;
252 for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
259 char attset_str[64], attname_str[64];
263 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
265 for (i = 0; *s && *s != ',' && *s != ')' &&
266 !strchr(FILE_MATCH_BLANK, *s); s++)
267 if (i+1 < sizeof(attset_str))
268 attset_str[i++] = *s;
269 attset_str[i] = '\0';
271 for (; strchr(FILE_MATCH_BLANK, *s); s++)
274 strcpy(attname_str, attset_str);
277 for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
279 for (i = 0; *s && *s != ')' &&
280 !strchr(FILE_MATCH_BLANK, *s); s++)
281 if (i+1 < sizeof(attname_str))
282 attname_str[i++] = *s;
283 attname_str[i] = '\0';
286 searchRecordKey (zh, reckeys, attname_str, ws, 32);
290 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
291 spec, zh->m_group ? zh->m_group : "none");
296 for (i = 0; i<32; i++)
305 dst += strlen(ws[i]);
309 yaz_log (YLOG_WARN, "Record didn't contain match"
310 " fields in (%s,%s)", attset_str, attname_str);
318 const char *spec_src = NULL;
319 const char *s1 = ++s;
320 while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
324 if (spec_len > sizeof(special)-1)
325 spec_len = sizeof(special)-1;
326 memcpy (special, s, spec_len);
327 special[spec_len] = '\0';
330 if (!strcmp (special, "group"))
331 spec_src = zh->m_group;
332 else if (!strcmp (special, "database"))
333 spec_src = zh->basenames[0];
334 else if (!strcmp (special, "filename")) {
337 else if (!strcmp (special, "type"))
338 spec_src = zh->m_record_type;
343 strcpy (dst, spec_src);
344 dst += strlen (spec_src);
347 else if (*s == '\"' || *s == '\'')
349 int stopMarker = *s++;
353 while (*s && *s != stopMarker)
355 if (i+1 < sizeof(tmpString))
356 tmpString[i++] = *s++;
361 strcpy (dst, tmpString);
362 dst += strlen (tmpString);
366 yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
367 spec, zh->m_group ? zh->m_group : "none");
374 yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
375 fname, zh->m_group ? zh->m_group : "none");
382 struct recordLogInfo {
385 struct recordGroup *rGroup;
388 static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
391 for (i = 0; i<256; i++)
393 if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
398 ctrl->flagShowRecords = !zh->m_flag_rw;
401 static void all_matches_add(struct recExtractCtrl *ctrl)
404 extract_init(ctrl, &word);
405 word.index_name = "_ALLRECORDS";
406 word.index_type = 'w';
408 extract_add_index_string (&word, zinfo_index_category_alwaysmatches,
412 static ZEBRA_RES file_extract_record(ZebraHandle zh,
413 SYSNO *sysno, const char *fname,
415 struct file_read_info *fi,
418 void *recTypeClientData)
420 const char *match_str_to_print = "";
421 RecordAttr *recordAttr;
423 const char *matchStr = 0;
426 off_t recordOffset = 0;
427 struct recExtractCtrl extractCtrl;
429 /* announce database */
430 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
432 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
433 zh->m_explain_database))
439 /* we are going to read from a file, so prepare the extraction */
440 zebra_rec_keys_reset(zh->reg->keys);
442 zebra_rec_keys_reset(zh->reg->sortKeys);
443 recordOffset = fi->file_moffset;
444 extractCtrl.handle = zh;
445 extractCtrl.offset = fi->file_moffset;
446 extractCtrl.readf = file_read;
447 extractCtrl.seekf = file_seek;
448 extractCtrl.tellf = file_tell;
449 extractCtrl.endf = file_end;
451 extractCtrl.init = extract_init;
452 extractCtrl.tokenAdd = extract_token_add;
453 extractCtrl.schemaAdd = extract_schema_add;
454 extractCtrl.dh = zh->reg->dh;
455 extractCtrl.match_criteria[0] = '\0';
456 extractCtrl.staticrank = 0;
458 extractCtrl.first_record = fi->file_offset ? 0 : 1;
460 extract_set_store_data_prepare(&extractCtrl);
462 init_extractCtrl(zh, &extractCtrl);
465 printf ("File: %s " ZINT_FORMAT "\n", fname, (zint)recordOffset);
469 sprintf (msg, "%s:" ZINT_FORMAT , fname, (zint)recordOffset);
470 yaz_log_init_prefix2 (msg);
473 r = (*recType->extract)(recTypeClientData, &extractCtrl);
475 yaz_log_init_prefix2 (0);
476 if (r == RECCTRL_EXTRACT_EOF)
478 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
480 /* error occured during extraction ... */
482 zh->records_processed < zh->m_file_verbose_limit)
484 yaz_log (YLOG_WARN, "fail %s %s " ZINT_FORMAT,
486 fname, (zint) recordOffset);
490 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
492 /* error occured during extraction ... */
494 zh->records_processed < zh->m_file_verbose_limit)
496 yaz_log (YLOG_WARN, "no filter for %s %s "
497 ZINT_FORMAT, zh->m_record_type,
498 fname, (zint) recordOffset);
502 all_matches_add(&extractCtrl);
503 if (extractCtrl.match_criteria[0])
504 matchStr = extractCtrl.match_criteria;
507 /* if matchStr is set now - we assume it's printable .
508 For internal matchStr (see below) we don't print */
510 match_str_to_print = matchStr;
512 /* perform internal match if sysno not known and if match criteria is
519 if (matchStr == 0 && zh->m_record_id && *zh->m_record_id)
521 matchStr = fileMatchStr (zh, zh->reg->keys, fname,
525 yaz_log(YLOG_WARN, "Bad match criteria");
527 if (zebra_rec_keys_empty(zh->reg->keys))
529 yaz_log(YLOG_WARN, "And no index keys");
536 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
537 char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
541 assert(*rinfo == sizeof(*sysno));
542 memcpy (sysno, rinfo+1, sizeof(*sysno));
546 if (! *sysno && zebra_rec_keys_empty(zh->reg->keys) )
548 /* the extraction process returned no information - the record
549 is probably empty - unless flagShowRecords is in use */
553 if (zh->records_processed < zh->m_file_verbose_limit)
554 yaz_log(YLOG_WARN, "empty %s %s " ZINT_FORMAT, zh->m_record_type,
555 fname, (zint)recordOffset);
564 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, zh->m_record_type,
565 fname, (zint)recordOffset);
566 yaz_log(YLOG_WARN, "cannot delete record above (seems new)");
570 rec = rec_new (zh->reg->records);
574 if (zh->records_processed < zh->m_file_verbose_limit)
576 yaz_log(YLOG_LOG, "add %s %s " ZINT_FORMAT
577 " " ZINT_FORMAT " %s" ,
579 fname, (zint) recordOffset, *sysno, match_str_to_print);
581 recordAttr = rec_init_attr (zh->reg->zei, rec);
582 recordAttr->staticrank = extractCtrl.staticrank;
586 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
587 dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
588 sizeof(*sysno), sysno);
592 extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
593 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
594 recordAttr->staticrank);
595 zh->records_inserted++;
599 /* record already exists */
600 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
602 zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
604 rec = rec_get (zh->reg->records, *sysno);
607 recordAttr = rec_init_attr (zh->reg->zei, rec);
609 zebra_rec_keys_set_buf(delkeys,
610 rec->info[recInfo_delKeys],
611 rec->size[recInfo_delKeys],
614 zebra_rec_keys_set_buf(sortKeys,
615 rec->info[recInfo_sortKeys],
616 rec->size[recInfo_sortKeys],
618 extract_flushSortKeys (zh, *sysno, 0, sortKeys);
619 extract_flushRecordKeys (zh, *sysno, 0, delkeys,
620 recordAttr->staticrank); /* old values */
623 /* record going to be deleted */
624 if (zebra_rec_keys_empty(delkeys))
626 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT
628 zh->m_record_type, fname, (zint)recordOffset, *sysno);
629 yaz_log(YLOG_WARN, "cannot delete file above, storeKeys false (1)");
633 if (zh->records_processed < zh->m_file_verbose_limit)
635 yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT
636 " " ZINT_FORMAT " %s" ,
637 zh->m_record_type, fname, (zint) recordOffset,
638 *sysno, match_str_to_print);
640 zh->records_deleted++;
643 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
644 dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
646 rec_del (zh->reg->records, &rec);
654 /* flush new keys for sort&search etc */
655 if (zh->records_processed < zh->m_file_verbose_limit)
657 yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT
658 " " ZINT_FORMAT " %s" ,
659 zh->m_record_type, fname, (zint) recordOffset,
660 *sysno, match_str_to_print);
662 recordAttr->staticrank = extractCtrl.staticrank;
663 extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
664 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
665 recordAttr->staticrank);
666 zh->records_updated++;
668 zebra_rec_keys_close(delkeys);
669 zebra_rec_keys_close(sortKeys);
671 /* update file type */
672 xfree (rec->info[recInfo_fileType]);
673 rec->info[recInfo_fileType] =
674 rec_strdup (zh->m_record_type, &rec->size[recInfo_fileType]);
676 /* update filename */
677 xfree (rec->info[recInfo_filename]);
678 rec->info[recInfo_filename] =
679 rec_strdup (fname, &rec->size[recInfo_filename]);
681 /* update delete keys */
682 xfree (rec->info[recInfo_delKeys]);
683 if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
685 zebra_rec_keys_get_buf(zh->reg->keys,
686 &rec->info[recInfo_delKeys],
687 &rec->size[recInfo_delKeys]);
691 rec->info[recInfo_delKeys] = NULL;
692 rec->size[recInfo_delKeys] = 0;
695 /* update sort keys */
696 xfree (rec->info[recInfo_sortKeys]);
698 zebra_rec_keys_get_buf(zh->reg->sortKeys,
699 &rec->info[recInfo_sortKeys],
700 &rec->size[recInfo_sortKeys]);
702 /* save file size of original record */
703 zebraExplain_recordBytesIncrement (zh->reg->zei,
704 - recordAttr->recordSize);
705 recordAttr->recordSize = fi->file_moffset - recordOffset;
706 if (!recordAttr->recordSize)
707 recordAttr->recordSize = fi->file_max - recordOffset;
708 zebraExplain_recordBytesIncrement (zh->reg->zei,
709 recordAttr->recordSize);
711 /* set run-number for this record */
712 recordAttr->runNumber = zebraExplain_runNumberIncrement (zh->reg->zei,
715 /* update store data */
716 xfree (rec->info[recInfo_storeData]);
717 if (zh->store_data_buf)
719 rec->size[recInfo_storeData] = zh->store_data_size;
720 rec->info[recInfo_storeData] = zh->store_data_buf;
721 zh->store_data_buf = 0;
723 else if (zh->m_store_data)
725 rec->size[recInfo_storeData] = recordAttr->recordSize;
726 rec->info[recInfo_storeData] = (char *)
727 xmalloc (recordAttr->recordSize);
728 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
730 yaz_log(YLOG_ERRNO|YLOG_FATAL, "seek to " ZINT_FORMAT " in %s",
731 (zint)recordOffset, fname);
734 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
735 < recordAttr->recordSize)
737 yaz_log (YLOG_ERRNO|YLOG_FATAL, "read %d bytes of %s",
738 recordAttr->recordSize, fname);
744 rec->info[recInfo_storeData] = NULL;
745 rec->size[recInfo_storeData] = 0;
747 /* update database name */
748 xfree (rec->info[recInfo_databaseName]);
749 rec->info[recInfo_databaseName] =
750 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
753 recordAttr->recordOffset = recordOffset;
755 /* commit this record */
756 rec_put (zh->reg->records, &rec);
761 ZEBRA_RES zebra_extract_file(ZebraHandle zh, SYSNO *sysno, const char *fname,
764 ZEBRA_RES r = ZEBRA_OK;
769 struct file_read_info *fi;
770 const char *original_record_type = 0;
772 void *recTypeClientData;
774 zebra_init_log_level();
776 if (!zh->m_group || !*zh->m_group)
779 sprintf (gprefix, "%s.", zh->m_group);
781 yaz_log(log_level, "zebra_extract_file %s", fname);
783 /* determine file extension */
785 for (i = strlen(fname); --i >= 0; )
788 else if (fname[i] == '.')
790 strcpy (ext, fname+i+1);
793 /* determine file type - depending on extension */
794 original_record_type = zh->m_record_type;
795 if (!zh->m_record_type)
797 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
798 zh->m_record_type = res_get (zh->res, ext_res);
800 if (!zh->m_record_type)
802 if (zh->records_processed < zh->m_file_verbose_limit)
803 yaz_log (YLOG_LOG, "? %s", fname);
806 /* determine match criteria */
807 if (!zh->m_record_id)
809 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
810 zh->m_record_id = res_get (zh->res, ext_res);
814 recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
815 &recTypeClientData)))
817 yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
821 switch(recType->version)
826 yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
828 if (sysno && deleteFlag)
834 if (zh->path_reg && !yaz_is_abspath (fname))
836 strcpy (full_rep, zh->path_reg);
837 strcat (full_rep, "/");
838 strcat (full_rep, fname);
841 strcpy (full_rep, fname);
843 if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
845 yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
846 zh->m_record_type = original_record_type;
850 fi = file_read_start (fd);
853 fi->file_moffset = fi->file_offset;
854 fi->file_more = 0; /* file_end not called (yet) */
855 r = file_extract_record (zh, sysno, fname, deleteFlag, fi, 1,
856 recType, recTypeClientData);
858 { /* file_end has been called so reset offset .. */
859 fi->file_offset = fi->file_moffset;
860 lseek(fi->fd, fi->file_moffset, SEEK_SET);
874 zh->m_record_type = original_record_type;
879 If sysno is provided, then it's used to identify the reocord.
880 If not, and match_criteria is provided, then sysno is guessed
881 If not, and a record is provided, then sysno is got from there
884 ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
885 const char *buf, size_t buf_size,
888 const char *recordType,
890 const char *match_criteria,
896 RecordAttr *recordAttr;
897 struct recExtractCtrl extractCtrl;
899 const char *matchStr = 0;
900 RecType recType = NULL;
903 long recordOffset = 0;
904 struct zebra_fetch_control fc;
905 const char *pr_fname = fname; /* filename to print .. */
906 int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
908 zebra_init_log_level();
911 pr_fname = "<no file>"; /* make it printable if file is omitted */
914 fc.record_int_buf = buf;
915 fc.record_int_len = buf_size;
916 fc.record_int_pos = 0;
918 fc.record_offset = 0;
920 extractCtrl.offset = 0;
921 extractCtrl.readf = zebra_record_int_read;
922 extractCtrl.seekf = zebra_record_int_seek;
923 extractCtrl.tellf = zebra_record_int_tell;
924 extractCtrl.endf = zebra_record_int_end;
925 extractCtrl.first_record = 1;
926 extractCtrl.fh = &fc;
928 zebra_rec_keys_reset(zh->reg->keys);
929 zebra_rec_keys_reset(zh->reg->sortKeys);
931 if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
933 if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
934 zh->m_explain_database))
938 if (recordType && *recordType)
940 yaz_log(log_level, "Record type explicitly specified: %s", recordType);
941 recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
946 if (!(zh->m_record_type))
948 yaz_log (YLOG_WARN, "No such record type defined");
951 yaz_log(log_level, "Get record type from rgroup: %s",
953 recType = recType_byName (zh->reg->recTypes, zh->res,
954 zh->m_record_type, &clientData);
955 recordType = zh->m_record_type;
960 yaz_log (YLOG_WARN, "No such record type: %s", recordType);
964 extractCtrl.init = extract_init;
965 extractCtrl.tokenAdd = extract_token_add;
966 extractCtrl.schemaAdd = extract_schema_add;
967 extractCtrl.dh = zh->reg->dh;
968 extractCtrl.handle = zh;
969 extractCtrl.match_criteria[0] = '\0';
970 extractCtrl.staticrank = 0;
972 init_extractCtrl(zh, &extractCtrl);
974 extract_set_store_data_prepare(&extractCtrl);
976 r = (*recType->extract)(clientData, &extractCtrl);
978 if (r == RECCTRL_EXTRACT_EOF)
980 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
982 /* error occured during extraction ... */
983 yaz_log (YLOG_WARN, "extract error: generic");
986 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
988 /* error occured during extraction ... */
989 yaz_log (YLOG_WARN, "extract error: no such filter");
993 all_matches_add(&extractCtrl);
995 if (extractCtrl.match_criteria[0])
996 match_criteria = extractCtrl.match_criteria;
1002 if (match_criteria && *match_criteria) {
1003 matchStr = match_criteria;
1005 if (zh->m_record_id && *zh->m_record_id) {
1006 matchStr = fileMatchStr (zh, zh->reg->keys, pr_fname,
1010 yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
1017 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
1018 char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
1022 assert(*rinfo == sizeof(*sysno));
1023 memcpy (sysno, rinfo+1, sizeof(*sysno));
1027 if (zebra_rec_keys_empty(zh->reg->keys))
1029 /* the extraction process returned no information - the record
1030 is probably empty - unless flagShowRecords is in use */
1040 yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
1041 pr_fname, (long) recordOffset);
1042 yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
1046 yaz_log (YLOG_LOG, "add %s %s %ld", recordType, pr_fname,
1047 (long) recordOffset);
1048 rec = rec_new (zh->reg->records);
1050 *sysno = rec->sysno;
1052 recordAttr = rec_init_attr (zh->reg->zei, rec);
1053 recordAttr->staticrank = extractCtrl.staticrank;
1057 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
1058 dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
1059 sizeof(*sysno), sysno);
1063 extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
1064 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
1065 recordAttr->staticrank);
1066 zh->records_inserted++;
1070 /* record already exists */
1071 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1072 zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
1075 yaz_log (YLOG_LOG, "skipped %s %s %ld",
1076 recordType, pr_fname, (long) recordOffset);
1081 rec = rec_get (zh->reg->records, *sysno);
1084 recordAttr = rec_init_attr (zh->reg->zei, rec);
1086 zebra_rec_keys_set_buf(delkeys,
1087 rec->info[recInfo_delKeys],
1088 rec->size[recInfo_delKeys],
1090 zebra_rec_keys_set_buf(sortKeys,
1091 rec->info[recInfo_sortKeys],
1092 rec->size[recInfo_sortKeys],
1095 extract_flushSortKeys (zh, *sysno, 0, sortKeys);
1096 extract_flushRecordKeys (zh, *sysno, 0, delkeys,
1097 recordAttr->staticrank);
1100 /* record going to be deleted */
1101 if (zebra_rec_keys_empty(delkeys))
1103 yaz_log(YLOG_LOG, "delete %s %s %ld", recordType,
1104 pr_fname, (long) recordOffset);
1105 yaz_log(YLOG_WARN, "cannot delete file above, "
1106 "storeKeys false (3)");
1111 yaz_log(YLOG_LOG, "delete %s %s %ld", recordType,
1112 pr_fname, (long) recordOffset);
1113 zh->records_deleted++;
1116 int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
1117 dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
1119 rec_del (zh->reg->records, &rec);
1128 yaz_log(YLOG_LOG, "update %s %s %ld", recordType,
1129 pr_fname, (long) recordOffset);
1130 recordAttr->staticrank = extractCtrl.staticrank;
1131 extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
1132 extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
1133 recordAttr->staticrank);
1134 zh->records_updated++;
1136 zebra_rec_keys_close(delkeys);
1137 zebra_rec_keys_close(sortKeys);
1139 /* update file type */
1140 xfree (rec->info[recInfo_fileType]);
1141 rec->info[recInfo_fileType] =
1142 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1144 /* update filename */
1145 xfree (rec->info[recInfo_filename]);
1146 rec->info[recInfo_filename] =
1147 rec_strdup (fname, &rec->size[recInfo_filename]);
1149 /* update delete keys */
1150 xfree (rec->info[recInfo_delKeys]);
1151 if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
1153 zebra_rec_keys_get_buf(zh->reg->keys,
1154 &rec->info[recInfo_delKeys],
1155 &rec->size[recInfo_delKeys]);
1159 rec->info[recInfo_delKeys] = NULL;
1160 rec->size[recInfo_delKeys] = 0;
1162 /* update sort keys */
1163 xfree (rec->info[recInfo_sortKeys]);
1165 zebra_rec_keys_get_buf(zh->reg->sortKeys,
1166 &rec->info[recInfo_sortKeys],
1167 &rec->size[recInfo_sortKeys]);
1169 /* save file size of original record */
1170 zebraExplain_recordBytesIncrement (zh->reg->zei,
1171 - recordAttr->recordSize);
1173 recordAttr->recordSize = fi->file_moffset - recordOffset;
1174 if (!recordAttr->recordSize)
1175 recordAttr->recordSize = fi->file_max - recordOffset;
1177 recordAttr->recordSize = buf_size;
1179 zebraExplain_recordBytesIncrement (zh->reg->zei,
1180 recordAttr->recordSize);
1182 /* set run-number for this record */
1183 recordAttr->runNumber =
1184 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
1186 /* update store data */
1187 xfree (rec->info[recInfo_storeData]);
1189 /* update store data */
1190 if (zh->store_data_buf)
1192 rec->size[recInfo_storeData] = zh->store_data_size;
1193 rec->info[recInfo_storeData] = zh->store_data_buf;
1194 zh->store_data_buf = 0;
1196 else if (zh->m_store_data)
1198 rec->size[recInfo_storeData] = recordAttr->recordSize;
1199 rec->info[recInfo_storeData] = (char *)
1200 xmalloc (recordAttr->recordSize);
1201 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1205 rec->info[recInfo_storeData] = NULL;
1206 rec->size[recInfo_storeData] = 0;
1208 /* update database name */
1209 xfree (rec->info[recInfo_databaseName]);
1210 rec->info[recInfo_databaseName] =
1211 rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]);
1214 recordAttr->recordOffset = recordOffset;
1216 /* commit this record */
1217 rec_put (zh->reg->records, &rec);
1222 ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
1224 ZebraHandle zh = (ZebraHandle) handle;
1225 struct recExtractCtrl extractCtrl;
1227 if (zebraExplain_curDatabase (zh->reg->zei,
1228 rec->info[recInfo_databaseName]))
1231 if (zebraExplain_newDatabase (zh->reg->zei,
1232 rec->info[recInfo_databaseName], 0))
1236 zebra_rec_keys_reset(zh->reg->keys);
1237 zebra_rec_keys_reset(zh->reg->sortKeys);
1239 extractCtrl.init = extract_init;
1240 extractCtrl.tokenAdd = extract_token_add;
1241 extractCtrl.schemaAdd = extract_schema_add;
1242 extractCtrl.dh = zh->reg->dh;
1244 init_extractCtrl(zh, &extractCtrl);
1246 extractCtrl.flagShowRecords = 0;
1247 extractCtrl.match_criteria[0] = '\0';
1248 extractCtrl.staticrank = 0;
1249 extractCtrl.handle = handle;
1250 extractCtrl.first_record = 1;
1252 extract_set_store_data_prepare(&extractCtrl);
1255 grs_extract_tree(&extractCtrl, n);
1257 if (rec->size[recInfo_delKeys])
1259 zebra_rec_keys_t delkeys = zebra_rec_keys_open();
1261 zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
1263 zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
1264 rec->size[recInfo_delKeys],
1266 extract_flushRecordKeys (zh, rec->sysno, 0, delkeys, 0);
1267 zebra_rec_keys_close(delkeys);
1269 zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
1270 rec->size[recInfo_sortKeys],
1273 extract_flushSortKeys (zh, rec->sysno, 0, sortkeys);
1274 zebra_rec_keys_close(sortkeys);
1276 extract_flushRecordKeys (zh, rec->sysno, 1, zh->reg->keys, 0);
1277 extract_flushSortKeys (zh, rec->sysno, 1, zh->reg->sortKeys);
1279 xfree (rec->info[recInfo_delKeys]);
1280 zebra_rec_keys_get_buf(zh->reg->keys,
1281 &rec->info[recInfo_delKeys],
1282 &rec->size[recInfo_delKeys]);
1284 xfree (rec->info[recInfo_sortKeys]);
1285 zebra_rec_keys_get_buf(zh->reg->sortKeys,
1286 &rec->info[recInfo_sortKeys],
1287 &rec->size[recInfo_sortKeys]);
1291 void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
1292 zebra_rec_keys_t reckeys)
1294 ZebraExplainInfo zei = zh->reg->zei;
1298 struct ord_stat *next;
1301 if (zebra_rec_keys_rewind(reckeys))
1303 struct ord_stat *ord_list = 0;
1307 struct it_key key_in;
1308 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1310 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1312 for (p = ord_list; p ; p = p->next)
1320 p = xmalloc(sizeof(*p));
1331 struct ord_stat *p1 = p;
1334 zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
1336 zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
1343 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1345 zebra_rec_keys_t reckeys,
1348 ZebraExplainInfo zei = zh->reg->zei;
1350 extract_rec_keys_adjust(zh, cmd, reckeys);
1352 if (!zh->reg->key_buf)
1354 int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
1357 yaz_log(YLOG_WARN, "Invalid memory setting, using default 8 MB");
1360 /* FIXME: That "8" should be in a default settings include */
1361 /* not hard-coded here! -H */
1362 zh->reg->key_buf = (char**) xmalloc (mem);
1363 zh->reg->ptr_top = mem/sizeof(char*);
1365 zh->reg->key_buf_used = 0;
1366 zh->reg->key_file_no = 0;
1368 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1370 if (zebra_rec_keys_rewind(reckeys))
1374 struct it_key key_in;
1375 while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1379 struct it_key key_out;
1381 assert(key_in.len >= 2);
1382 assert(key_in.len <= IT_KEY_LEVEL_MAX);
1384 /* check for buffer overflow */
1385 if (zh->reg->key_buf_used + 1024 >
1386 (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
1387 extract_flushWriteKeys (zh, 0);
1390 assert(zh->reg->ptr_i > 0);
1391 (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
1392 (char*)zh->reg->key_buf + zh->reg->key_buf_used;
1394 /* key_in.mem[0] ord/ch */
1395 /* key_in.mem[1] filter specified record ID */
1397 /* encode the ordinal value (field/use/attribute) .. */
1398 ch = CAST_ZINT_TO_INT(key_in.mem[0]);
1399 zh->reg->key_buf_used +=
1400 key_SU_encode(ch, (char*)zh->reg->key_buf +
1401 zh->reg->key_buf_used);
1403 /* copy the 0-terminated stuff from str to output */
1404 memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used, str, slen);
1405 zh->reg->key_buf_used += slen;
1406 ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = '\0';
1408 /* the delete/insert indicator */
1409 ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = cmd;
1411 if (zh->m_staticrank) /* rank config enabled ? */
1415 yaz_log(YLOG_WARN, "staticrank = %ld. Setting to 0",
1419 key_out.mem[j++] = staticrank;
1422 if (key_in.mem[1]) /* filter specified record ID */
1423 key_out.mem[j++] = key_in.mem[1];
1425 key_out.mem[j++] = sysno;
1426 for (i = 2; i < key_in.len; i++)
1427 key_out.mem[j++] = key_in.mem[i];
1430 memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used,
1431 &key_out, sizeof(key_out));
1432 (zh->reg->key_buf_used) += sizeof(key_out);
1437 void extract_flushWriteKeys (ZebraHandle zh, int final)
1438 /* optimizing: if final=1, and no files written yet */
1439 /* push the keys directly to merge, sidestepping the */
1440 /* temp file altogether. Speeds small updates */
1443 char out_fname[200];
1445 struct encode_info encode_info;
1446 int ptr_i = zh->reg->ptr_i;
1451 if (!zh->reg->key_buf || ptr_i <= 0)
1453 yaz_log(log_level, " nothing to flush section=%d buf=%p i=%d",
1454 zh->reg->key_file_no, zh->reg->key_buf, ptr_i);
1458 (zh->reg->key_file_no)++;
1459 yaz_log (YLOG_LOG, "sorting section %d", (zh->reg->key_file_no));
1460 yaz_log(log_level, " sort_buff at %p n=%d",
1461 zh->reg->key_buf + zh->reg->ptr_top - ptr_i,ptr_i);
1463 qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
1464 sizeof(char*), key_qsort_compare);
1466 /* zebra.cfg: tempfiles:
1467 Y: always use temp files (old way)
1468 A: use temp files, if more than one (auto)
1469 = if this is both the last and the first
1470 N: never bother with temp files (new) */
1472 temp_policy=toupper(res_get_def(zh->res,"tempfiles","auto")[0]);
1473 if (temp_policy != 'Y' && temp_policy != 'N' && temp_policy != 'A') {
1474 yaz_log (YLOG_WARN, "Illegal tempfiles setting '%c'. using 'Auto' ",
1479 if ( ( temp_policy =='N' ) || /* always from memory */
1480 ( ( temp_policy =='A' ) && /* automatic */
1481 (zh->reg->key_file_no == 1) && /* this is first time */
1482 (final) ) ) /* and last (=only) time */
1483 { /* go directly from memory */
1484 zh->reg->key_file_no =0; /* signal not to read files */
1485 zebra_index_merge(zh);
1487 zh->reg->key_buf_used = 0;
1491 /* Not doing directly from memory, write into a temp file */
1492 extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
1494 if (!(outf = fopen (out_fname, "wb")))
1496 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1499 yaz_log (YLOG_LOG, "writing section %d", zh->reg->key_file_no);
1500 prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1502 encode_key_init (&encode_info);
1503 encode_key_write (cp, &encode_info, outf);
1507 cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1508 if (strcmp (cp, prevcp))
1510 encode_key_flush ( &encode_info, outf);
1511 encode_key_init (&encode_info);
1512 encode_key_write (cp, &encode_info, outf);
1516 encode_key_write (cp + strlen(cp), &encode_info, outf);
1518 encode_key_flush ( &encode_info, outf);
1520 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1521 extract_get_fname_tmp (out_fname, key_file_no);
1523 if (!(outf = fopen (out_fname, "wb")))
1525 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
1528 yaz_log (YLOG_LOG, "writing section %d", key_file_no);
1530 prevcp = key_buf[ptr_top-i];
1532 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1534 key_y_len = strlen(prevcp)+1;
1536 yaz_log (YLOG_LOG, "key_y_len: %2d %02x %02x %s",
1537 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1539 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1540 sizeof(char*), key_y_compare);
1541 cp = key_buf[ptr_top-ptr_i];
1543 encode_key_init (&encode_info);
1544 encode_key_write (cp, &encode_info, outf);
1547 cp = key_buf[ptr_top-ptr_i];
1548 encode_key_write (cp+key_y_len, &encode_info, outf);
1550 encode_key_flush ( &encode_info, outf);
1553 prevcp = key_buf[ptr_top-ptr_i];
1558 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fclose %s", out_fname);
1561 yaz_log (YLOG_LOG, "finished section %d", zh->reg->key_file_no);
1563 zh->reg->key_buf_used = 0;
1566 ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh,
1567 zebra_rec_keys_t reckeys,
1568 zebra_snippets *snippets)
1570 NMEM nmem = nmem_create();
1571 if (zebra_rec_keys_rewind(reckeys))
1576 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1578 char dst_buf[IT_MAX_WORD];
1579 char *dst_term = dst_buf;
1584 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1585 seqno = key.mem[key.len-1];
1586 ord = CAST_ZINT_TO_INT(key.mem[0]);
1588 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
1589 0/* db */, 0 /* string_index */);
1591 zebra_term_untrans_iconv(zh, nmem, index_type,
1593 zebra_snippets_append(snippets, seqno, ord, dst_term);
1601 void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
1603 yaz_log(YLOG_LOG, "print_rec_keys");
1604 if (zebra_rec_keys_rewind(reckeys))
1609 while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
1611 char dst_buf[IT_MAX_WORD];
1614 int ord = CAST_ZINT_TO_INT(key.mem[0]);
1616 assert(key.len <= IT_KEY_LEVEL_MAX && key.len > 2);
1618 zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
1620 seqno = key.mem[key.len-1];
1622 zebra_term_untrans(zh, index_type, dst_buf, str);
1624 yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
1625 " term=%s", ord, seqno, dst_buf);
1630 static void extract_add_index_string(RecWord *p, zinfo_index_category_t cat,
1631 const char *str, int length)
1634 ZebraHandle zh = p->extractCtrl->handle;
1635 ZebraExplainInfo zei = zh->reg->zei;
1638 ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1640 ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1644 key.mem[i++] = p->record_id;
1645 key.mem[i++] = p->section_id;
1647 if (zh->m_segment_indexing)
1648 key.mem[i++] = p->segment;
1649 key.mem[i++] = p->seqno;
1652 zebra_rec_keys_write(zh->reg->keys, str, length, &key);
1655 static void extract_add_sort_string(RecWord *p, const char *str, int length)
1658 ZebraHandle zh = p->extractCtrl->handle;
1659 ZebraExplainInfo zei = zh->reg->zei;
1661 zinfo_index_category_t cat = zinfo_index_category_sort;
1663 ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name);
1665 ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name);
1668 key.mem[1] = p->record_id;
1670 zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
1673 static void extract_add_string(RecWord *p, const char *string, int length)
1675 ZebraHandle zh = p->extractCtrl->handle;
1676 assert (length > 0);
1681 if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
1682 extract_add_sort_string(p, string, length);
1685 extract_add_index_string(p, zinfo_index_category_index,
1687 if (zebra_maps_is_alwaysmatches(zh->reg->zebra_maps, p->index_type))
1690 memcpy(&word, p, sizeof(word));
1693 extract_add_index_string(
1694 &word, zinfo_index_category_alwaysmatches, "", 0);
1699 static void extract_add_incomplete_field (RecWord *p)
1701 ZebraHandle zh = p->extractCtrl->handle;
1702 const char *b = p->term_buf;
1703 int remain = p->term_len;
1704 const char **map = 0;
1707 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1711 char buf[IT_MAX_WORD+1];
1715 while (map && *map && **map == *CHR_SPACE)
1717 remain = p->term_len - (b - p->term_buf);
1719 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b,
1727 while (map && *map && **map != *CHR_SPACE)
1729 const char *cp = *map;
1731 while (i < IT_MAX_WORD && *cp)
1733 remain = p->term_len - (b - p->term_buf);
1735 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0);
1741 extract_add_string (p, buf, i);
1746 static void extract_add_complete_field (RecWord *p)
1748 ZebraHandle zh = p->extractCtrl->handle;
1749 const char *b = p->term_buf;
1750 char buf[IT_MAX_WORD+1];
1751 const char **map = 0;
1752 int i = 0, remain = p->term_len;
1755 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b, remain, 1);
1757 while (remain > 0 && i < IT_MAX_WORD)
1759 while (map && *map && **map == *CHR_SPACE)
1761 remain = p->term_len - (b - p->term_buf);
1765 int first = i ? 0 : 1; /* first position */
1766 map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, first);
1774 if (i && i < IT_MAX_WORD)
1775 buf[i++] = *CHR_SPACE;
1776 while (map && *map && **map != *CHR_SPACE)
1778 const char *cp = *map;
1780 if (**map == *CHR_CUT)
1786 if (i >= IT_MAX_WORD)
1788 while (i < IT_MAX_WORD && *cp)
1791 remain = p->term_len - (b - p->term_buf);
1794 map = zebra_maps_input (zh->reg->zebra_maps, p->index_type, &b,
1803 extract_add_string (p, buf, i);
1806 static void extract_token_add(RecWord *p)
1808 ZebraHandle zh = p->extractCtrl->handle;
1811 yaz_log(log_level, "extract_token_add "
1812 "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
1813 p->index_type, p->index_name,
1814 p->seqno, p->term_len, p->term_buf);
1815 if ((wrbuf = zebra_replace(zh->reg->zebra_maps, p->index_type, 0,
1816 p->term_buf, p->term_len)))
1818 p->term_buf = wrbuf_buf(wrbuf);
1819 p->term_len = wrbuf_len(wrbuf);
1821 if (zebra_maps_is_complete (zh->reg->zebra_maps, p->index_type))
1822 extract_add_complete_field (p);
1824 extract_add_incomplete_field(p);
1827 static void extract_set_store_data_cb(struct recExtractCtrl *p,
1828 void *buf, size_t sz)
1830 ZebraHandle zh = (ZebraHandle) p->handle;
1832 xfree(zh->store_data_buf);
1833 zh->store_data_buf = 0;
1834 zh->store_data_size = 0;
1837 zh->store_data_buf = xmalloc(sz);
1838 zh->store_data_size = sz;
1839 memcpy(zh->store_data_buf, buf, sz);
1843 static void extract_set_store_data_prepare(struct recExtractCtrl *p)
1845 ZebraHandle zh = (ZebraHandle) p->handle;
1846 xfree(zh->store_data_buf);
1847 zh->store_data_buf = 0;
1848 zh->store_data_size = 0;
1849 p->setStoreData = extract_set_store_data_cb;
1852 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1854 ZebraHandle zh = (ZebraHandle) p->handle;
1855 zebraExplain_addSchema (zh->reg->zei, oid);
1858 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1859 int cmd, zebra_rec_keys_t reckeys)
1861 if (zebra_rec_keys_rewind(reckeys))
1863 SortIdx sortIdx = zh->reg->sortIdx;
1866 struct it_key key_in;
1868 sortIdx_sysno (sortIdx, sysno);
1870 while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
1872 int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
1874 sortIdx_type(sortIdx, ord);
1876 sortIdx_add(sortIdx, str, slen);
1878 sortIdx_add(sortIdx, "", 1);
1883 static void encode_key_init(struct encode_info *i)
1885 i->encode_handle = iscz1_start();
1886 i->decode_handle = iscz1_start();
1889 static void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1892 char *bp = i->buf, *bp0;
1893 const char *src = (char *) &key;
1895 /* copy term to output buf */
1896 while ((*bp++ = *k++))
1898 /* and copy & align key so we can mangle */
1899 memcpy (&key, k+1, sizeof(struct it_key)); /* *k is insert/delete */
1903 key_logdump_txt(YLOG_LOG, &key, *k ? "i" : "d");
1905 assert(key.mem[0] >= 0);
1908 iscz1_encode(i->encode_handle, &bp, &src);
1910 *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */
1911 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1913 yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
1922 const char *src = bp0+1;
1923 char *dst = (char*) &key2;
1924 iscz1_decode(i->decode_handle, &dst, &src);
1926 key_logdump_txt(YLOG_LOG, &key2, *k ? "i" : "d");
1928 assert(key2.mem[1]);
1933 static void encode_key_flush (struct encode_info *i, FILE *outf)
1935 iscz1_stop(i->encode_handle);
1936 iscz1_stop(i->decode_handle);
1942 * indent-tabs-mode: nil
1944 * vim: shiftwidth=4 tabstop=8 expandtab