1 /* $Id: extract.c,v 1.123 2002-08-29 10:00:15 adam Exp $
2 Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
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 Zebra; see the file LICENSE.zebra. If not, write to the
19 Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
37 #if _FILE_OFFSET_BITS == 64
38 #define PRINTF_OFF_T "%Ld"
40 #define PRINTF_OFF_T "%ld"
43 #define USE_SHELLSORT 0
46 static void shellsort(void *ar, int r, size_t s,
47 int (*cmp)(const void *a, const void *b))
52 static const int incs[16] = { 1391376, 463792, 198768, 86961, 33936,
53 13776, 4592, 1968, 861, 336,
54 112, 48, 21, 7, 3, 1 };
55 for ( k = 0; k < 16; k++)
56 for (h = incs[k], i = h; i < r; i++)
60 while (j > h && (*cmp)(a + s*(j-h), v) > 0)
62 memcpy (a + s*j, a + s*(j-h), s);
70 static void logRecord (ZebraHandle zh)
72 ++zh->records_processed;
73 if (!(zh->records_processed % 1000))
75 logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d",
76 zh->records_processed, zh->records_inserted, zh->records_updated,
81 static void extract_init (struct recExtractCtrl *p, RecWord *w)
83 w->zebra_maps = p->zebra_maps;
85 w->attrSet = VAL_BIB1;
91 static const char **searchRecordKey (ZebraHandle zh,
92 struct recKeys *reckeys,
93 int attrSetS, int attrUseS)
95 static const char *ws[32];
107 for (i = 0; i<32; i++)
111 chS = zebraExplain_lookupSU (zh->reg->zei, attrSetS, attrUseS);
115 while (off < reckeys->buf_used)
118 const char *src = reckeys->buf + off;
126 memcpy (&ch, src, sizeof(ch));
132 memcpy (&attrSet, src, sizeof(attrSet));
133 src += sizeof(attrSet);
137 memcpy (&attrUse, src, sizeof(attrUse));
138 src += sizeof(attrUse);
145 seqno += ((lead>>2) & 15)-1;
148 memcpy (&seqno, src, sizeof(seqno));
149 src += sizeof(seqno);
155 attrUseS == attrUse && attrSetS == attrSet
164 woff = seqno - startSeq;
165 if (woff >= 0 && woff < 31)
169 off = src - reckeys->buf;
171 assert (off == reckeys->buf_used);
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 */
185 static struct file_read_info *file_read_start (int fd)
187 struct file_read_info *fi = (struct file_read_info *)
188 xmalloc (sizeof(*fi));
192 fi->file_moffset = 0;
198 static void file_read_stop (struct file_read_info *fi)
203 static off_t file_seek (void *handle, off_t offset)
205 struct file_read_info *p = (struct file_read_info *) handle;
206 p->file_offset = offset;
209 return lseek (p->fd, offset, SEEK_SET);
212 static off_t file_tell (void *handle)
214 struct file_read_info *p = (struct file_read_info *) handle;
215 return p->file_offset;
218 static int file_read (void *handle, char *buf, size_t count)
220 struct file_read_info *p = (struct file_read_info *) handle;
226 if (r > p->sdrmax - p->file_offset)
227 r = p->sdrmax - p->file_offset;
229 memcpy (buf, p->sdrbuf + p->file_offset, r);
232 r = read (fd, buf, count);
236 if (p->file_offset > p->file_max)
237 p->file_max = p->file_offset;
242 static void file_begin (void *handle)
244 struct file_read_info *p = (struct file_read_info *) handle;
246 p->file_offset = p->file_moffset;
247 if (!p->sdrbuf && p->file_moffset)
248 lseek (p->fd, p->file_moffset, SEEK_SET);
252 static void file_end (void *handle, off_t offset)
254 struct file_read_info *p = (struct file_read_info *) handle;
256 assert (p->file_more == 0);
258 p->file_moffset = offset;
261 static char *fileMatchStr (ZebraHandle zh,
262 struct recKeys *reckeys, struct recordGroup *rGroup,
263 const char *fname, const char *spec)
265 static char dstBuf[2048]; /* static here ??? */
267 const char *s = spec;
268 static const char **w;
272 while (*s == ' ' || *s == '\t')
278 char attset_str[64], attname_str[64];
279 data1_attset *attset;
282 int attSet = 1, attUse = 1;
286 for (i = 0; *s && *s != ',' && *s != ')'; s++)
288 attset_str[i++] = *s;
289 attset_str[i] = '\0';
294 for (i = 0; *s && *s != ')'; s++)
296 attname_str[i++] = *s;
297 attname_str[i] = '\0';
300 if ((attset = data1_get_attset (zh->reg->dh, attset_str)))
303 attSet = attset->reference;
304 att = data1_getattbyname(zh->reg->dh, attset, attname_str);
308 attUse = atoi (attname_str);
310 w = searchRecordKey (zh, reckeys, attSet, attUse);
315 for (i = 0; i<32; i++)
320 logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
321 spec, rGroup->groupName ? rGroup->groupName : "none");
326 for (i = 0; i<32; i++)
327 if (matchFlag[i] && w[i])
339 logf (LOG_WARN, "Record didn't contain match"
340 " fields in (%s,%s)", attset_str, attname_str);
348 const char *spec_src = NULL;
349 const char *s1 = ++s;
350 while (*s1 && *s1 != ' ' && *s1 != '\t')
356 memcpy (special, s, spec_len);
357 special[spec_len] = '\0';
360 if (!strcmp (special, "group"))
361 spec_src = rGroup->groupName;
362 else if (!strcmp (special, "database"))
363 spec_src = rGroup->databaseName;
364 else if (!strcmp (special, "filename"))
366 else if (!strcmp (special, "type"))
367 spec_src = rGroup->recordType;
372 strcpy (dst, spec_src);
373 dst += strlen (spec_src);
376 else if (*s == '\"' || *s == '\'')
378 int stopMarker = *s++;
382 while (*s && *s != stopMarker)
385 tmpString[i++] = *s++;
390 strcpy (dst, tmpString);
391 dst += strlen (tmpString);
395 logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
396 spec, rGroup->groupName ? rGroup->groupName : "none");
403 logf (LOG_WARN, "No match criteria for record %s in group %s",
404 fname, rGroup->groupName ? rGroup->groupName : "none");
411 struct recordLogInfo {
414 struct recordGroup *rGroup;
417 static int recordExtract (ZebraHandle zh,
418 SYSNO *sysno, const char *fname,
419 struct recordGroup *rGroup, int deleteFlag,
420 struct file_read_info *fi,
421 RecType recType, char *subType, void *clientData)
423 RecordAttr *recordAttr;
428 off_t recordOffset = 0;
432 struct recExtractCtrl extractCtrl;
434 /* we are going to read from a file, so prepare the extraction */
437 zh->reg->keys.buf_used = 0;
438 zh->reg->keys.prevAttrUse = -1;
439 zh->reg->keys.prevAttrSet = -1;
440 zh->reg->keys.prevSeqNo = 0;
441 zh->reg->sortKeys = 0;
443 recordOffset = fi->file_moffset;
444 extractCtrl.offset = fi->file_moffset;
445 extractCtrl.readf = file_read;
446 extractCtrl.seekf = file_seek;
447 extractCtrl.tellf = file_tell;
448 extractCtrl.endf = file_end;
450 extractCtrl.subType = subType;
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.handle = zh;
456 for (i = 0; i<256; i++)
458 if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
459 extractCtrl.seqno[i] = 1;
461 extractCtrl.seqno[i] = 0;
463 extractCtrl.zebra_maps = zh->reg->zebra_maps;
464 extractCtrl.flagShowRecords = !rGroup->flagRw;
467 printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset);
471 sprintf (msg, "%s:" PRINTF_OFF_T , fname, recordOffset);
472 yaz_log_init_prefix2 (msg);
475 r = (*recType->extract)(clientData, &extractCtrl);
477 yaz_log_init_prefix2 (0);
478 if (r == RECCTRL_EXTRACT_EOF)
480 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
482 /* error occured during extraction ... */
483 if (rGroup->flagRw &&
484 zh->records_processed < rGroup->fileVerboseLimit)
486 logf (LOG_WARN, "fail %s %s " PRINTF_OFF_T, rGroup->recordType,
487 fname, recordOffset);
491 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
493 /* error occured during extraction ... */
494 if (rGroup->flagRw &&
495 zh->records_processed < rGroup->fileVerboseLimit)
497 logf (LOG_WARN, "no filter for %s %s "
498 PRINTF_OFF_T, rGroup->recordType,
499 fname, recordOffset);
503 if (zh->reg->keys.buf_used == 0)
505 /* the extraction process returned no information - the record
506 is probably empty - unless flagShowRecords is in use */
510 logf (LOG_WARN, "empty %s %s " PRINTF_OFF_T, rGroup->recordType,
511 fname, recordOffset);
516 /* perform match if sysno not known and if match criteria is specified */
523 if (rGroup->recordId && *rGroup->recordId)
527 matchStr = fileMatchStr (zh, &zh->reg->keys, rGroup, fname,
531 rinfo = dict_lookup (zh->reg->matchDict, matchStr);
533 memcpy (sysno, rinfo+1, sizeof(*sysno));
537 logf (LOG_WARN, "Bad match criteria");
548 logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T, rGroup->recordType,
549 fname, recordOffset);
550 logf (LOG_WARN, "cannot delete record above (seems new)");
553 if (zh->records_processed < rGroup->fileVerboseLimit)
554 logf (LOG_LOG, "add %s %s " PRINTF_OFF_T, rGroup->recordType,
555 fname, recordOffset);
556 rec = rec_new (zh->reg->records);
560 recordAttr = rec_init_attr (zh->reg->zei, rec);
564 dict_insert (zh->reg->matchDict, matchStr, sizeof(*sysno), sysno);
566 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
567 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
569 zh->records_inserted++;
573 /* record already exists */
574 struct recKeys delkeys;
576 rec = rec_get (zh->reg->records, *sysno);
579 recordAttr = rec_init_attr (zh->reg->zei, rec);
581 if (recordAttr->runNumber ==
582 zebraExplain_runNumberIncrement (zh->reg->zei, 0))
584 yaz_log (LOG_LOG, "run number = %d", recordAttr->runNumber);
585 yaz_log (LOG_LOG, "skipped %s %s " PRINTF_OFF_T,
586 rGroup->recordType, fname, recordOffset);
587 extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
592 delkeys.buf_used = rec->size[recInfo_delKeys];
593 delkeys.buf = rec->info[recInfo_delKeys];
594 extract_flushSortKeys (zh, *sysno, 0, &zh->reg->sortKeys);
595 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
598 /* record going to be deleted */
599 if (!delkeys.buf_used)
601 logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T,
602 rGroup->recordType, fname, recordOffset);
603 logf (LOG_WARN, "cannot delete file above, storeKeys false");
607 if (zh->records_processed < rGroup->fileVerboseLimit)
608 logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T,
609 rGroup->recordType, fname, recordOffset);
610 zh->records_deleted++;
612 dict_delete (zh->reg->matchDict, matchStr);
613 rec_del (zh->reg->records, &rec);
621 /* record going to be updated */
622 if (!delkeys.buf_used)
624 logf (LOG_LOG, "update %s %s " PRINTF_OFF_T,
625 rGroup->recordType, fname, recordOffset);
626 logf (LOG_WARN, "cannot update file above, storeKeys false");
630 if (zh->records_processed < rGroup->fileVerboseLimit)
631 logf (LOG_LOG, "update %s %s " PRINTF_OFF_T,
632 rGroup->recordType, fname, recordOffset);
633 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
634 zh->records_updated++;
638 /* update file type */
639 xfree (rec->info[recInfo_fileType]);
640 rec->info[recInfo_fileType] =
641 rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
643 /* update filename */
644 xfree (rec->info[recInfo_filename]);
645 rec->info[recInfo_filename] =
646 rec_strdup (fname, &rec->size[recInfo_filename]);
648 /* update delete keys */
649 xfree (rec->info[recInfo_delKeys]);
650 if (zh->reg->keys.buf_used > 0 && rGroup->flagStoreKeys == 1)
653 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
654 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
655 zh->reg->keys.buf = NULL;
656 zh->reg->keys.buf_max = 0;
658 rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
659 rec->size[recInfo_delKeys] = reckeys.buf_used;
660 memcpy (rec->info[recInfo_delKeys], reckeys.buf,
661 rec->size[recInfo_delKeys]);
666 rec->info[recInfo_delKeys] = NULL;
667 rec->size[recInfo_delKeys] = 0;
670 /* save file size of original record */
671 zebraExplain_recordBytesIncrement (zh->reg->zei,
672 - recordAttr->recordSize);
673 recordAttr->recordSize = fi->file_moffset - recordOffset;
674 if (!recordAttr->recordSize)
675 recordAttr->recordSize = fi->file_max - recordOffset;
676 zebraExplain_recordBytesIncrement (zh->reg->zei,
677 recordAttr->recordSize);
679 /* set run-number for this record */
680 recordAttr->runNumber = zebraExplain_runNumberIncrement (zh->reg->zei,
683 /* update store data */
684 xfree (rec->info[recInfo_storeData]);
685 if (rGroup->flagStoreData == 1)
687 rec->size[recInfo_storeData] = recordAttr->recordSize;
688 rec->info[recInfo_storeData] = (char *)
689 xmalloc (recordAttr->recordSize);
690 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
692 logf (LOG_ERRNO|LOG_FATAL, "seek to " PRINTF_OFF_T " in %s",
693 recordOffset, fname);
696 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
697 < recordAttr->recordSize)
699 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
700 recordAttr->recordSize, fname);
706 rec->info[recInfo_storeData] = NULL;
707 rec->size[recInfo_storeData] = 0;
709 /* update database name */
710 xfree (rec->info[recInfo_databaseName]);
711 rec->info[recInfo_databaseName] =
712 rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
715 recordAttr->recordOffset = recordOffset;
717 /* commit this record */
718 rec_put (zh->reg->records, &rec);
723 int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
724 const struct recordGroup *rGroupP, int deleteFlag)
732 struct recordGroup rGroupM;
733 struct recordGroup *rGroup = &rGroupM;
734 struct file_read_info *fi;
737 memcpy (rGroup, rGroupP, sizeof(*rGroupP));
739 if (!rGroup->groupName || !*rGroup->groupName)
742 sprintf (gprefix, "%s.", rGroup->groupName);
744 logf (LOG_DEBUG, "fileExtract %s", fname);
746 /* determine file extension */
748 for (i = strlen(fname); --i >= 0; )
751 else if (fname[i] == '.')
753 strcpy (ext, fname+i+1);
756 /* determine file type - depending on extension */
757 if (!rGroup->recordType)
759 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
760 if (!(rGroup->recordType = res_get (zh->res, ext_res)))
762 sprintf (ext_res, "%srecordType", gprefix);
763 rGroup->recordType = res_get (zh->res, ext_res);
766 if (!rGroup->recordType)
768 if (zh->records_processed < rGroup->fileVerboseLimit)
769 logf (LOG_LOG, "? %s", fname);
772 if (!*rGroup->recordType)
775 recType_byName (zh->reg->recTypes, rGroup->recordType, subType,
778 logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
782 /* determine match criteria */
783 if (!rGroup->recordId)
785 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
786 rGroup->recordId = res_get (zh->res, ext_res);
789 /* determine database name */
790 if (!rGroup->databaseName)
792 sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
793 if (!(rGroup->databaseName = res_get (zh->res, ext_res)))
795 sprintf (ext_res, "%sdatabase", gprefix);
796 rGroup->databaseName = res_get (zh->res, ext_res);
799 if (!rGroup->databaseName)
800 rGroup->databaseName = "Default";
802 /* determine if explain database */
804 sprintf (ext_res, "%sexplainDatabase", gprefix);
805 rGroup->explainDatabase =
806 atoi (res_get_def (zh->res, ext_res, "0"));
808 /* announce database */
809 if (zebraExplain_curDatabase (zh->reg->zei, rGroup->databaseName))
811 if (zebraExplain_newDatabase (zh->reg->zei, rGroup->databaseName,
812 rGroup->explainDatabase))
816 if (rGroup->flagStoreData == -1)
819 sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
820 if (!(sval = res_get (zh->res, ext_res)))
822 sprintf (ext_res, "%sstoreData", gprefix);
823 sval = res_get (zh->res, ext_res);
826 rGroup->flagStoreData = atoi (sval);
828 if (rGroup->flagStoreData == -1)
829 rGroup->flagStoreData = 0;
831 if (rGroup->flagStoreKeys == -1)
835 sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
836 sval = res_get (zh->res, ext_res);
839 sprintf (ext_res, "%sstoreKeys", gprefix);
840 sval = res_get (zh->res, ext_res);
843 sval = res_get (zh->res, "storeKeys");
845 rGroup->flagStoreKeys = atoi (sval);
847 if (rGroup->flagStoreKeys == -1)
848 rGroup->flagStoreKeys = 0;
850 if (sysno && deleteFlag)
856 if (zh->path_reg && !yaz_is_abspath (fname))
858 strcpy (full_rep, zh->path_reg);
859 strcat (full_rep, "/");
860 strcat (full_rep, fname);
863 strcpy (full_rep, fname);
866 if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
868 logf (LOG_WARN|LOG_ERRNO, "open %s", full_rep);
872 fi = file_read_start (fd);
876 r = recordExtract (zh, sysno, fname, rGroup, deleteFlag, fi,
877 recType, subType, clientData);
878 } while (r && !sysno && fi->file_more);
886 int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
887 const char *buf, size_t buf_size,
888 const char *databaseName, int delete_flag,
889 int test_mode, int *sysno,
890 int store_keys, int store_data,
891 const char *match_criteria)
893 RecordAttr *recordAttr;
894 struct recExtractCtrl extractCtrl;
900 const char *fname = "<no file>";
902 long recordOffset = 0;
903 struct zebra_fetch_control fc;
906 fc.record_int_buf = buf;
907 fc.record_int_len = buf_size;
908 fc.record_int_pos = 0;
910 fc.record_offset = 0;
912 extractCtrl.offset = 0;
913 extractCtrl.readf = zebra_record_int_read;
914 extractCtrl.seekf = zebra_record_int_seek;
915 extractCtrl.tellf = zebra_record_int_tell;
916 extractCtrl.endf = zebra_record_int_end;
917 extractCtrl.fh = &fc;
919 /* announce database */
920 if (zebraExplain_curDatabase (zh->reg->zei, databaseName))
922 if (zebraExplain_newDatabase (zh->reg->zei, databaseName, 0))
926 recType_byName (zh->reg->recTypes, recordType, subType,
929 logf (LOG_WARN, "No such record type: %s", recordType);
933 zh->reg->keys.buf_used = 0;
934 zh->reg->keys.prevAttrUse = -1;
935 zh->reg->keys.prevAttrSet = -1;
936 zh->reg->keys.prevSeqNo = 0;
937 zh->reg->sortKeys = 0;
939 extractCtrl.subType = subType;
940 extractCtrl.init = extract_init;
941 extractCtrl.tokenAdd = extract_token_add;
942 extractCtrl.schemaAdd = extract_schema_add;
943 extractCtrl.dh = zh->reg->dh;
944 extractCtrl.handle = zh;
945 extractCtrl.zebra_maps = zh->reg->zebra_maps;
946 extractCtrl.flagShowRecords = 0;
947 for (i = 0; i<256; i++)
949 if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
950 extractCtrl.seqno[i] = 1;
952 extractCtrl.seqno[i] = 0;
955 r = (*recType->extract)(clientData, &extractCtrl);
957 if (r == RECCTRL_EXTRACT_EOF)
959 else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
961 /* error occured during extraction ... */
962 yaz_log (LOG_WARN, "extract error: generic");
965 else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
967 /* error occured during extraction ... */
968 yaz_log (LOG_WARN, "extract error: no such filter");
971 if (zh->reg->keys.buf_used == 0)
973 /* the extraction process returned no information - the record
974 is probably empty - unless flagShowRecords is in use */
977 logf (LOG_WARN, "No keys generated for record");
978 logf (LOG_WARN, " The file is probably empty");
988 logf (LOG_LOG, "delete %s %s %ld", recordType,
989 fname, (long) recordOffset);
990 logf (LOG_WARN, "cannot delete record above (seems new)");
993 logf (LOG_LOG, "add %s %s %ld", recordType, fname,
994 (long) recordOffset);
995 rec = rec_new (zh->reg->records);
999 recordAttr = rec_init_attr (zh->reg->zei, rec);
1003 dict_insert (zh->reg->matchDict, matchStr,
1004 sizeof(*sysno), sysno);
1006 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
1007 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
1011 /* record already exists */
1012 struct recKeys delkeys;
1014 rec = rec_get (zh->reg->records, *sysno);
1017 recordAttr = rec_init_attr (zh->reg->zei, rec);
1019 if (recordAttr->runNumber ==
1020 zebraExplain_runNumberIncrement (zh->reg->zei, 0))
1022 logf (LOG_LOG, "skipped %s %s %ld", recordType,
1023 fname, (long) recordOffset);
1027 delkeys.buf_used = rec->size[recInfo_delKeys];
1028 delkeys.buf = rec->info[recInfo_delKeys];
1029 extract_flushSortKeys (zh, *sysno, 0, &zh->reg->sortKeys);
1030 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1033 /* record going to be deleted */
1034 if (!delkeys.buf_used)
1036 logf (LOG_LOG, "delete %s %s %ld", recordType,
1037 fname, (long) recordOffset);
1038 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1042 logf (LOG_LOG, "delete %s %s %ld", recordType,
1043 fname, (long) recordOffset);
1046 dict_delete (matchDict, matchStr);
1048 rec_del (zh->reg->records, &rec);
1055 /* record going to be updated */
1056 if (!delkeys.buf_used)
1058 logf (LOG_LOG, "update %s %s %ld", recordType,
1059 fname, (long) recordOffset);
1060 logf (LOG_WARN, "cannot update file above, storeKeys false");
1064 logf (LOG_LOG, "update %s %s %ld", recordType,
1065 fname, (long) recordOffset);
1066 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
1070 /* update file type */
1071 xfree (rec->info[recInfo_fileType]);
1072 rec->info[recInfo_fileType] =
1073 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1075 /* update filename */
1076 xfree (rec->info[recInfo_filename]);
1077 rec->info[recInfo_filename] =
1078 rec_strdup (fname, &rec->size[recInfo_filename]);
1080 /* update delete keys */
1081 xfree (rec->info[recInfo_delKeys]);
1082 if (zh->reg->keys.buf_used > 0 && store_keys == 1)
1084 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
1085 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
1086 zh->reg->keys.buf = NULL;
1087 zh->reg->keys.buf_max = 0;
1091 rec->info[recInfo_delKeys] = NULL;
1092 rec->size[recInfo_delKeys] = 0;
1095 /* save file size of original record */
1096 zebraExplain_recordBytesIncrement (zh->reg->zei,
1097 - recordAttr->recordSize);
1099 recordAttr->recordSize = fi->file_moffset - recordOffset;
1100 if (!recordAttr->recordSize)
1101 recordAttr->recordSize = fi->file_max - recordOffset;
1103 recordAttr->recordSize = buf_size;
1105 zebraExplain_recordBytesIncrement (zh->reg->zei,
1106 recordAttr->recordSize);
1108 /* set run-number for this record */
1109 recordAttr->runNumber =
1110 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
1112 /* update store data */
1113 xfree (rec->info[recInfo_storeData]);
1114 if (store_data == 1)
1116 rec->size[recInfo_storeData] = recordAttr->recordSize;
1117 rec->info[recInfo_storeData] = (char *)
1118 xmalloc (recordAttr->recordSize);
1120 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1122 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1124 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1125 (long) recordOffset, fname);
1128 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1129 < recordAttr->recordSize)
1131 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1132 recordAttr->recordSize, fname);
1139 rec->info[recInfo_storeData] = NULL;
1140 rec->size[recInfo_storeData] = 0;
1142 /* update database name */
1143 xfree (rec->info[recInfo_databaseName]);
1144 rec->info[recInfo_databaseName] =
1145 rec_strdup (databaseName, &rec->size[recInfo_databaseName]);
1148 recordAttr->recordOffset = recordOffset;
1150 /* commit this record */
1151 rec_put (zh->reg->records, &rec);
1156 int explain_extract (void *handle, Record rec, data1_node *n)
1158 ZebraHandle zh = (ZebraHandle) handle;
1159 struct recExtractCtrl extractCtrl;
1162 if (zebraExplain_curDatabase (zh->reg->zei,
1163 rec->info[recInfo_databaseName]))
1166 if (zebraExplain_newDatabase (zh->reg->zei,
1167 rec->info[recInfo_databaseName], 0))
1171 zh->reg->keys.buf_used = 0;
1172 zh->reg->keys.prevAttrUse = -1;
1173 zh->reg->keys.prevAttrSet = -1;
1174 zh->reg->keys.prevSeqNo = 0;
1175 zh->reg->sortKeys = 0;
1177 extractCtrl.init = extract_init;
1178 extractCtrl.tokenAdd = extract_token_add;
1179 extractCtrl.schemaAdd = extract_schema_add;
1180 extractCtrl.dh = zh->reg->dh;
1181 for (i = 0; i<256; i++)
1182 extractCtrl.seqno[i] = 0;
1183 extractCtrl.zebra_maps = zh->reg->zebra_maps;
1184 extractCtrl.flagShowRecords = 0;
1185 extractCtrl.handle = handle;
1187 grs_extract_tree(&extractCtrl, n);
1189 if (rec->size[recInfo_delKeys])
1191 struct recKeys delkeys;
1192 struct sortKey *sortKeys = 0;
1194 delkeys.buf_used = rec->size[recInfo_delKeys];
1195 delkeys.buf = rec->info[recInfo_delKeys];
1196 extract_flushSortKeys (zh, rec->sysno, 0, &sortKeys);
1197 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1199 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->reg->keys);
1200 extract_flushSortKeys (zh, rec->sysno, 1, &zh->reg->sortKeys);
1202 xfree (rec->info[recInfo_delKeys]);
1203 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
1204 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
1205 zh->reg->keys.buf = NULL;
1206 zh->reg->keys.buf_max = 0;
1210 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1211 int cmd, struct recKeys *reckeys)
1215 unsigned char attrSet = (unsigned char) -1;
1216 unsigned short attrUse = (unsigned short) -1;
1221 ZebraExplainInfo zei = zh->reg->zei;
1223 if (!zh->reg->key_buf)
1225 int mem = 8*1024*1024;
1226 zh->reg->key_buf = (char**) xmalloc (mem);
1227 zh->reg->ptr_top = mem/sizeof(char*);
1229 zh->reg->key_buf_used = 0;
1230 zh->reg->key_file_no = 0;
1232 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1233 while (off < reckeys->buf_used)
1235 const char *src = reckeys->buf + off;
1244 memcpy (&ch, src, sizeof(ch));
1250 memcpy (&attrSet, src, sizeof(attrSet));
1251 src += sizeof(attrSet);
1255 memcpy (&attrUse, src, sizeof(attrUse));
1256 src += sizeof(attrUse);
1259 if (zh->reg->key_buf_used + 1024 >
1260 (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
1261 extract_flushWriteKeys (zh);
1263 (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
1264 (char*)zh->reg->key_buf + zh->reg->key_buf_used;
1267 ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
1269 ch = zebraExplain_addSU (zei, attrSet, attrUse);
1272 zh->reg->key_buf_used +=
1273 key_SU_encode (ch,((char*)zh->reg->key_buf) +
1274 zh->reg->key_buf_used);
1277 ((char*)zh->reg->key_buf) [(zh->reg->key_buf_used)++] = *src++;
1279 ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = '\0';
1280 ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = cmd;
1283 seqno += ((lead>>2) & 15)-1;
1286 memcpy (&seqno, src, sizeof(seqno));
1287 src += sizeof(seqno);
1291 memcpy ((char*)zh->reg->key_buf + zh->reg->key_buf_used, &key, sizeof(key));
1292 (zh->reg->key_buf_used) += sizeof(key);
1293 off = src - reckeys->buf;
1295 assert (off == reckeys->buf_used);
1298 void extract_flushWriteKeys (ZebraHandle zh)
1301 char out_fname[200];
1303 struct encode_info encode_info;
1304 int ptr_i = zh->reg->ptr_i;
1308 if (!zh->reg->key_buf || ptr_i <= 0)
1311 (zh->reg->key_file_no)++;
1312 logf (LOG_LOG, "sorting section %d", (zh->reg->key_file_no));
1314 qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
1315 sizeof(char*), key_qsort_compare);
1316 extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
1318 if (!(outf = fopen (out_fname, "wb")))
1320 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1323 logf (LOG_LOG, "writing section %d", zh->reg->key_file_no);
1324 prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1326 encode_key_init (&encode_info);
1327 encode_key_write (cp, &encode_info, outf);
1331 cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1332 if (strcmp (cp, prevcp))
1334 encode_key_init (&encode_info);
1335 encode_key_write (cp, &encode_info, outf);
1339 encode_key_write (cp + strlen(cp), &encode_info, outf);
1342 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1343 extract_get_fname_tmp (out_fname, key_file_no);
1345 if (!(outf = fopen (out_fname, "wb")))
1347 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1350 logf (LOG_LOG, "writing section %d", key_file_no);
1352 prevcp = key_buf[ptr_top-i];
1354 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1356 key_y_len = strlen(prevcp)+1;
1358 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
1359 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1361 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1362 sizeof(char*), key_y_compare);
1363 cp = key_buf[ptr_top-ptr_i];
1365 encode_key_init (&encode_info);
1366 encode_key_write (cp, &encode_info, outf);
1369 cp = key_buf[ptr_top-ptr_i];
1370 encode_key_write (cp+key_y_len, &encode_info, outf);
1374 prevcp = key_buf[ptr_top-ptr_i];
1379 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
1382 logf (LOG_LOG, "finished section %d", zh->reg->key_file_no);
1384 zh->reg->key_buf_used = 0;
1387 void extract_add_index_string (RecWord *p, const char *string,
1391 unsigned char attrSet;
1392 unsigned short attrUse;
1395 int *pseqno = &p->seqno;
1396 ZebraHandle zh = p->extractCtrl->handle;
1397 ZebraExplainInfo zei = zh->reg->zei;
1398 struct recKeys *keys = &zh->reg->keys;
1400 if (keys->buf_used+1024 > keys->buf_max)
1404 b = (char *) xmalloc (keys->buf_max += 128000);
1405 if (keys->buf_used > 0)
1406 memcpy (b, keys->buf, keys->buf_used);
1410 dst = keys->buf + keys->buf_used;
1412 attrSet = p->attrSet;
1413 if (keys->buf_used > 0 && keys->prevAttrSet == attrSet)
1416 keys->prevAttrSet = attrSet;
1417 attrUse = p->attrUse;
1418 if (keys->buf_used > 0 && keys->prevAttrUse == attrUse)
1421 keys->prevAttrUse = attrUse;
1423 diff = 1 + *pseqno - keys->prevSeqNo;
1424 if (diff >= 1 && diff <= 15)
1425 lead |= (diff << 2);
1429 keys->prevSeqNo = *pseqno;
1436 int ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
1439 ch = zebraExplain_addSU (zei, attrSet, attrUse);
1440 yaz_log (LOG_DEBUG, "addSU set=%d use=%d SU=%d",
1441 attrSet, attrUse, ch);
1444 memcpy (dst, &ch, sizeof(ch));
1450 memcpy (dst, &attrSet, sizeof(attrSet));
1451 dst += sizeof(attrSet);
1455 memcpy (dst, &attrUse, sizeof(attrUse));
1456 dst += sizeof(attrUse);
1459 *dst++ = p->reg_type;
1460 memcpy (dst, string, length);
1466 memcpy (dst, pseqno, sizeof(*pseqno));
1467 dst += sizeof(*pseqno);
1469 keys->buf_used = dst - keys->buf;
1472 static void extract_add_sort_string (RecWord *p, const char *string,
1476 ZebraHandle zh = p->extractCtrl->handle;
1478 for (sk = zh->reg->sortKeys; sk; sk = sk->next)
1479 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
1482 sk = (struct sortKey *) xmalloc (sizeof(*sk));
1483 sk->next = zh->reg->sortKeys;
1484 zh->reg->sortKeys = sk;
1486 sk->string = (char *) xmalloc (length);
1487 sk->length = length;
1488 memcpy (sk->string, string, length);
1490 sk->attrSet = p->attrSet;
1491 sk->attrUse = p->attrUse;
1494 void extract_add_string (RecWord *p, const char *string, int length)
1496 assert (length > 0);
1497 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
1498 extract_add_sort_string (p, string, length);
1500 extract_add_index_string (p, string, length);
1503 static void extract_add_incomplete_field (RecWord *p)
1505 const char *b = p->string;
1506 int remain = p->length;
1507 const char **map = 0;
1510 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1514 char buf[IT_MAX_WORD+1];
1518 while (map && *map && **map == *CHR_SPACE)
1520 remain = p->length - (b - p->string);
1522 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1529 while (map && *map && **map != *CHR_SPACE)
1531 const char *cp = *map;
1533 while (i < IT_MAX_WORD && *cp)
1535 remain = p->length - (b - p->string);
1537 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1543 extract_add_string (p, buf, i);
1548 static void extract_add_complete_field (RecWord *p)
1550 const char *b = p->string;
1551 char buf[IT_MAX_WORD+1];
1552 const char **map = 0;
1553 int i = 0, remain = p->length;
1556 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
1558 while (remain > 0 && i < IT_MAX_WORD)
1560 while (map && *map && **map == *CHR_SPACE)
1562 remain = p->length - (b - p->string);
1564 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1571 if (i && i < IT_MAX_WORD)
1572 buf[i++] = *CHR_SPACE;
1573 while (map && *map && **map != *CHR_SPACE)
1575 const char *cp = *map;
1577 if (i >= IT_MAX_WORD)
1579 while (i < IT_MAX_WORD && *cp)
1581 remain = p->length - (b - p->string);
1583 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
1591 extract_add_string (p, buf, i);
1594 void extract_token_add (RecWord *p)
1597 if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
1598 p->string, p->length)))
1600 p->string = wrbuf_buf(wrbuf);
1601 p->length = wrbuf_len(wrbuf);
1603 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
1604 extract_add_complete_field (p);
1606 extract_add_incomplete_field(p);
1609 void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1611 ZebraHandle zh = (ZebraHandle) (p->handle);
1612 zebraExplain_addSchema (zh->reg->zei, oid);
1615 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1616 int cmd, struct sortKey **skp)
1618 struct sortKey *sk = *skp;
1619 SortIdx sortIdx = zh->reg->sortIdx;
1621 sortIdx_sysno (sortIdx, sysno);
1624 struct sortKey *sk_next = sk->next;
1625 sortIdx_type (sortIdx, sk->attrUse);
1626 sortIdx_add (sortIdx, sk->string, sk->length);
1634 void encode_key_init (struct encode_info *i)
1641 char *encode_key_int (int d, char *bp)
1645 else if (d <= 16383)
1647 *bp++ = 64 + (d>>8);
1650 else if (d <= 4194303)
1652 *bp++ = 128 + (d>>16);
1653 *bp++ = (d>>8) & 255;
1658 *bp++ = 192 + (d>>24);
1659 *bp++ = (d>>16) & 255;
1660 *bp++ = (d>>8) & 255;
1666 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1671 while ((*bp++ = *k++))
1673 memcpy (&key, k+1, sizeof(struct it_key));
1674 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
1675 if (i->sysno != key.sysno)
1677 i->sysno = key.sysno;
1680 else if (!i->seqno && !key.seqno && i->cmd == *k)
1682 bp = encode_key_int (key.seqno - i->seqno, bp);
1683 i->seqno = key.seqno;
1685 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1687 logf (LOG_FATAL|LOG_ERRNO, "fwrite");