2 * Copyright (C) 1994-2002, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
6 * $Id: extract.c,v 1.118 2002-05-03 13:49:04 adam Exp $
21 #if _FILE_OFFSET_BITS == 64
22 #define PRINTF_OFF_T "%Ld"
24 #define PRINTF_OFF_T "%ld"
27 static void shellsort(void *ar, int r, size_t s,
28 int (*cmp)(const void *a, const void *b))
33 static const int incs[16] = { 1391376, 463792, 198768, 86961, 33936,
34 13776, 4592, 1968, 861, 336,
35 112, 48, 21, 7, 3, 1 };
36 for ( k = 0; k < 16; k++)
37 for (h = incs[k], i = h; i < r; i++)
41 while (j > h && (*cmp)(a + s*(j-h), v) > 0)
43 memcpy (a + s*j, a + s*(j-h), s);
51 static void logRecord (ZebraHandle zh)
53 ++zh->records_processed;
54 if (!(zh->records_processed % 1000))
56 logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d",
57 zh->records_processed, zh->records_inserted, zh->records_updated,
62 static void extract_init (struct recExtractCtrl *p, RecWord *w)
64 w->zebra_maps = p->zebra_maps;
66 w->attrSet = VAL_BIB1;
72 static const char **searchRecordKey (ZebraHandle zh,
73 struct recKeys *reckeys,
74 int attrSetS, int attrUseS)
76 static const char *ws[32];
88 for (i = 0; i<32; i++)
92 chS = zebraExplain_lookupSU (zh->reg->zei, attrSetS, attrUseS);
96 while (off < reckeys->buf_used)
99 const char *src = reckeys->buf + off;
107 memcpy (&ch, src, sizeof(ch));
113 memcpy (&attrSet, src, sizeof(attrSet));
114 src += sizeof(attrSet);
118 memcpy (&attrUse, src, sizeof(attrUse));
119 src += sizeof(attrUse);
126 seqno += ((lead>>2) & 15)-1;
129 memcpy (&seqno, src, sizeof(seqno));
130 src += sizeof(seqno);
136 attrUseS == attrUse && attrSetS == attrSet
145 woff = seqno - startSeq;
146 if (woff >= 0 && woff < 31)
150 off = src - reckeys->buf;
152 assert (off == reckeys->buf_used);
156 struct file_read_info {
157 off_t file_max; /* maximum offset so far */
158 off_t file_offset; /* current offset */
159 off_t file_moffset; /* offset of rec/rec boundary */
166 static struct file_read_info *file_read_start (int fd)
168 struct file_read_info *fi = (struct file_read_info *)
169 xmalloc (sizeof(*fi));
173 fi->file_moffset = 0;
179 static void file_read_stop (struct file_read_info *fi)
184 static off_t file_seek (void *handle, off_t offset)
186 struct file_read_info *p = (struct file_read_info *) handle;
187 p->file_offset = offset;
190 return lseek (p->fd, offset, SEEK_SET);
193 static off_t file_tell (void *handle)
195 struct file_read_info *p = (struct file_read_info *) handle;
196 return p->file_offset;
199 static int file_read (void *handle, char *buf, size_t count)
201 struct file_read_info *p = (struct file_read_info *) handle;
207 if (r > p->sdrmax - p->file_offset)
208 r = p->sdrmax - p->file_offset;
210 memcpy (buf, p->sdrbuf + p->file_offset, r);
213 r = read (fd, buf, count);
217 if (p->file_offset > p->file_max)
218 p->file_max = p->file_offset;
223 static void file_begin (void *handle)
225 struct file_read_info *p = (struct file_read_info *) handle;
227 p->file_offset = p->file_moffset;
228 if (!p->sdrbuf && p->file_moffset)
229 lseek (p->fd, p->file_moffset, SEEK_SET);
233 static void file_end (void *handle, off_t offset)
235 struct file_read_info *p = (struct file_read_info *) handle;
237 assert (p->file_more == 0);
239 p->file_moffset = offset;
242 static char *fileMatchStr (ZebraHandle zh,
243 struct recKeys *reckeys, struct recordGroup *rGroup,
244 const char *fname, const char *spec)
246 static char dstBuf[2048]; /* static here ??? */
248 const char *s = spec;
249 static const char **w;
253 while (*s == ' ' || *s == '\t')
259 char attset_str[64], attname_str[64];
260 data1_attset *attset;
263 int attSet = 1, attUse = 1;
267 for (i = 0; *s && *s != ',' && *s != ')'; s++)
269 attset_str[i++] = *s;
270 attset_str[i] = '\0';
275 for (i = 0; *s && *s != ')'; s++)
277 attname_str[i++] = *s;
278 attname_str[i] = '\0';
281 if ((attset = data1_get_attset (zh->reg->dh, attset_str)))
284 attSet = attset->reference;
285 att = data1_getattbyname(zh->reg->dh, attset, attname_str);
289 attUse = atoi (attname_str);
291 w = searchRecordKey (zh, reckeys, attSet, attUse);
296 for (i = 0; i<32; i++)
301 logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
302 spec, rGroup->groupName ? rGroup->groupName : "none");
307 for (i = 0; i<32; i++)
308 if (matchFlag[i] && w[i])
320 logf (LOG_WARN, "Record didn't contain match"
321 " fields in (%s,%s)", attset_str, attname_str);
329 const char *spec_src = NULL;
330 const char *s1 = ++s;
331 while (*s1 && *s1 != ' ' && *s1 != '\t')
337 memcpy (special, s, spec_len);
338 special[spec_len] = '\0';
341 if (!strcmp (special, "group"))
342 spec_src = rGroup->groupName;
343 else if (!strcmp (special, "database"))
344 spec_src = rGroup->databaseName;
345 else if (!strcmp (special, "filename"))
347 else if (!strcmp (special, "type"))
348 spec_src = rGroup->recordType;
353 strcpy (dst, spec_src);
354 dst += strlen (spec_src);
357 else if (*s == '\"' || *s == '\'')
359 int stopMarker = *s++;
363 while (*s && *s != stopMarker)
366 tmpString[i++] = *s++;
371 strcpy (dst, tmpString);
372 dst += strlen (tmpString);
376 logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
377 spec, rGroup->groupName ? rGroup->groupName : "none");
384 logf (LOG_WARN, "No match criteria for record %s in group %s",
385 fname, rGroup->groupName ? rGroup->groupName : "none");
392 struct recordLogInfo {
395 struct recordGroup *rGroup;
398 static void recordLogPreamble (int level, const char *msg, void *info)
400 struct recordLogInfo *p = (struct recordLogInfo *) info;
401 FILE *outf = yaz_log_file ();
405 fprintf (outf, "File %s, offset %d, type %s\n",
406 p->rGroup->recordType, p->recordOffset, p->fname);
407 log_event_start (NULL, NULL);
411 static int recordExtract (ZebraHandle zh,
412 SYSNO *sysno, const char *fname,
413 struct recordGroup *rGroup, int deleteFlag,
414 struct file_read_info *fi,
415 RecType recType, char *subType, void *clientData)
417 RecordAttr *recordAttr;
422 struct recordLogInfo logInfo;
423 off_t recordOffset = 0;
427 struct recExtractCtrl extractCtrl;
429 /* we are going to read from a file, so prepare the extraction */
432 zh->reg->keys.buf_used = 0;
433 zh->reg->keys.prevAttrUse = -1;
434 zh->reg->keys.prevAttrSet = -1;
435 zh->reg->keys.prevSeqNo = 0;
436 zh->reg->sortKeys = 0;
438 recordOffset = fi->file_moffset;
439 extractCtrl.offset = fi->file_moffset;
440 extractCtrl.readf = file_read;
441 extractCtrl.seekf = file_seek;
442 extractCtrl.tellf = file_tell;
443 extractCtrl.endf = file_end;
445 extractCtrl.subType = subType;
446 extractCtrl.init = extract_init;
447 extractCtrl.tokenAdd = extract_token_add;
448 extractCtrl.schemaAdd = extract_schema_add;
449 extractCtrl.dh = zh->reg->dh;
450 extractCtrl.handle = zh;
451 for (i = 0; i<256; i++)
453 if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
454 extractCtrl.seqno[i] = 1;
456 extractCtrl.seqno[i] = 0;
458 extractCtrl.zebra_maps = zh->reg->zebra_maps;
459 extractCtrl.flagShowRecords = !rGroup->flagRw;
462 printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset);
464 logInfo.fname = fname;
465 logInfo.recordOffset = recordOffset;
466 logInfo.rGroup = rGroup;
467 log_event_start (recordLogPreamble, &logInfo);
469 r = (*recType->extract)(clientData, &extractCtrl);
471 log_event_start (NULL, NULL);
473 if (r == RECCTRL_EXTRACT_EOF)
475 else if (r == RECCTRL_EXTRACT_ERROR)
477 /* error occured during extraction ... */
478 if (rGroup->flagRw &&
479 zh->records_processed < rGroup->fileVerboseLimit)
481 logf (LOG_WARN, "fail %s %s " PRINTF_OFF_T, rGroup->recordType,
482 fname, recordOffset);
486 if (zh->reg->keys.buf_used == 0)
488 /* the extraction process returned no information - the record
489 is probably empty - unless flagShowRecords is in use */
493 logf (LOG_WARN, "empty %s %s " PRINTF_OFF_T, rGroup->recordType,
494 fname, recordOffset);
499 /* perform match if sysno not known and if match criteria is specified */
506 if (rGroup->recordId && *rGroup->recordId)
510 matchStr = fileMatchStr (zh, &zh->reg->keys, rGroup, fname,
514 rinfo = dict_lookup (zh->reg->matchDict, matchStr);
516 memcpy (sysno, rinfo+1, sizeof(*sysno));
520 logf (LOG_WARN, "Bad match criteria");
531 logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T, rGroup->recordType,
532 fname, recordOffset);
533 logf (LOG_WARN, "cannot delete record above (seems new)");
536 if (zh->records_processed < rGroup->fileVerboseLimit)
537 logf (LOG_LOG, "add %s %s " PRINTF_OFF_T, rGroup->recordType,
538 fname, recordOffset);
539 rec = rec_new (zh->reg->records);
543 recordAttr = rec_init_attr (zh->reg->zei, rec);
547 dict_insert (zh->reg->matchDict, matchStr, sizeof(*sysno), sysno);
549 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
550 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
552 zh->records_inserted++;
556 /* record already exists */
557 struct recKeys delkeys;
559 rec = rec_get (zh->reg->records, *sysno);
562 recordAttr = rec_init_attr (zh->reg->zei, rec);
564 if (recordAttr->runNumber ==
565 zebraExplain_runNumberIncrement (zh->reg->zei, 0))
567 logf (LOG_LOG, "skipped %s %s " PRINTF_OFF_T, rGroup->recordType,
568 fname, recordOffset);
569 extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
574 delkeys.buf_used = rec->size[recInfo_delKeys];
575 delkeys.buf = rec->info[recInfo_delKeys];
576 extract_flushSortKeys (zh, *sysno, 0, &zh->reg->sortKeys);
577 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
580 /* record going to be deleted */
581 if (!delkeys.buf_used)
583 logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T,
584 rGroup->recordType, fname, recordOffset);
585 logf (LOG_WARN, "cannot delete file above, storeKeys false");
589 if (zh->records_processed < rGroup->fileVerboseLimit)
590 logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T,
591 rGroup->recordType, fname, recordOffset);
592 zh->records_deleted++;
594 dict_delete (zh->reg->matchDict, matchStr);
595 rec_del (zh->reg->records, &rec);
603 /* record going to be updated */
604 if (!delkeys.buf_used)
606 logf (LOG_LOG, "update %s %s " PRINTF_OFF_T,
607 rGroup->recordType, fname, recordOffset);
608 logf (LOG_WARN, "cannot update file above, storeKeys false");
612 if (zh->records_processed < rGroup->fileVerboseLimit)
613 logf (LOG_LOG, "update %s %s " PRINTF_OFF_T,
614 rGroup->recordType, fname, recordOffset);
615 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
616 zh->records_updated++;
620 /* update file type */
621 xfree (rec->info[recInfo_fileType]);
622 rec->info[recInfo_fileType] =
623 rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
625 /* update filename */
626 xfree (rec->info[recInfo_filename]);
627 rec->info[recInfo_filename] =
628 rec_strdup (fname, &rec->size[recInfo_filename]);
630 /* update delete keys */
631 xfree (rec->info[recInfo_delKeys]);
632 if (zh->reg->keys.buf_used > 0 && rGroup->flagStoreKeys == 1)
635 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
636 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
637 zh->reg->keys.buf = NULL;
638 zh->reg->keys.buf_max = 0;
640 rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
641 rec->size[recInfo_delKeys] = reckeys.buf_used;
642 memcpy (rec->info[recInfo_delKeys], reckeys.buf,
643 rec->size[recInfo_delKeys]);
648 rec->info[recInfo_delKeys] = NULL;
649 rec->size[recInfo_delKeys] = 0;
652 /* save file size of original record */
653 zebraExplain_recordBytesIncrement (zh->reg->zei,
654 - recordAttr->recordSize);
655 recordAttr->recordSize = fi->file_moffset - recordOffset;
656 if (!recordAttr->recordSize)
657 recordAttr->recordSize = fi->file_max - recordOffset;
658 zebraExplain_recordBytesIncrement (zh->reg->zei,
659 recordAttr->recordSize);
661 /* set run-number for this record */
662 recordAttr->runNumber = zebraExplain_runNumberIncrement (zh->reg->zei,
665 /* update store data */
666 xfree (rec->info[recInfo_storeData]);
667 if (rGroup->flagStoreData == 1)
669 rec->size[recInfo_storeData] = recordAttr->recordSize;
670 rec->info[recInfo_storeData] = (char *)
671 xmalloc (recordAttr->recordSize);
672 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
674 logf (LOG_ERRNO|LOG_FATAL, "seek to " PRINTF_OFF_T " in %s",
675 recordOffset, fname);
678 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
679 < recordAttr->recordSize)
681 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
682 recordAttr->recordSize, fname);
688 rec->info[recInfo_storeData] = NULL;
689 rec->size[recInfo_storeData] = 0;
691 /* update database name */
692 xfree (rec->info[recInfo_databaseName]);
693 rec->info[recInfo_databaseName] =
694 rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
697 recordAttr->recordOffset = recordOffset;
699 /* commit this record */
700 rec_put (zh->reg->records, &rec);
705 int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
706 const struct recordGroup *rGroupP, int deleteFlag)
714 struct recordGroup rGroupM;
715 struct recordGroup *rGroup = &rGroupM;
716 struct file_read_info *fi;
719 memcpy (rGroup, rGroupP, sizeof(*rGroupP));
721 if (!rGroup->groupName || !*rGroup->groupName)
724 sprintf (gprefix, "%s.", rGroup->groupName);
726 logf (LOG_DEBUG, "fileExtract %s", fname);
728 /* determine file extension */
730 for (i = strlen(fname); --i >= 0; )
733 else if (fname[i] == '.')
735 strcpy (ext, fname+i+1);
738 /* determine file type - depending on extension */
739 if (!rGroup->recordType)
741 sprintf (ext_res, "%srecordType.%s", gprefix, ext);
742 if (!(rGroup->recordType = res_get (zh->res, ext_res)))
744 sprintf (ext_res, "%srecordType", gprefix);
745 rGroup->recordType = res_get (zh->res, ext_res);
748 if (!rGroup->recordType)
750 if (zh->records_processed < rGroup->fileVerboseLimit)
751 logf (LOG_LOG, "? %s", fname);
754 if (!*rGroup->recordType)
757 recType_byName (zh->reg->recTypes, rGroup->recordType, subType,
760 logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
764 /* determine match criteria */
765 if (!rGroup->recordId)
767 sprintf (ext_res, "%srecordId.%s", gprefix, ext);
768 rGroup->recordId = res_get (zh->res, ext_res);
771 /* determine database name */
772 if (!rGroup->databaseName)
774 sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
775 if (!(rGroup->databaseName = res_get (zh->res, ext_res)))
777 sprintf (ext_res, "%sdatabase", gprefix);
778 rGroup->databaseName = res_get (zh->res, ext_res);
781 if (!rGroup->databaseName)
782 rGroup->databaseName = "Default";
784 /* determine if explain database */
786 sprintf (ext_res, "%sexplainDatabase", gprefix);
787 rGroup->explainDatabase =
788 atoi (res_get_def (zh->res, ext_res, "0"));
790 /* announce database */
791 if (zebraExplain_curDatabase (zh->reg->zei, rGroup->databaseName))
793 if (zebraExplain_newDatabase (zh->reg->zei, rGroup->databaseName,
794 rGroup->explainDatabase))
798 if (rGroup->flagStoreData == -1)
801 sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
802 if (!(sval = res_get (zh->res, ext_res)))
804 sprintf (ext_res, "%sstoreData", gprefix);
805 sval = res_get (zh->res, ext_res);
808 rGroup->flagStoreData = atoi (sval);
810 if (rGroup->flagStoreData == -1)
811 rGroup->flagStoreData = 0;
813 if (rGroup->flagStoreKeys == -1)
817 sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
818 sval = res_get (zh->res, ext_res);
821 sprintf (ext_res, "%sstoreKeys", gprefix);
822 sval = res_get (zh->res, ext_res);
825 sval = res_get (zh->res, "storeKeys");
827 rGroup->flagStoreKeys = atoi (sval);
829 if (rGroup->flagStoreKeys == -1)
830 rGroup->flagStoreKeys = 0;
832 if (sysno && deleteFlag)
838 if (zh->path_reg && !yaz_is_abspath (fname))
840 strcpy (full_rep, zh->path_reg);
841 strcat (full_rep, "/");
842 strcat (full_rep, fname);
845 strcpy (full_rep, fname);
848 if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
850 logf (LOG_WARN|LOG_ERRNO, "open %s", full_rep);
854 fi = file_read_start (fd);
858 r = recordExtract (zh, sysno, fname, rGroup, deleteFlag, fi,
859 recType, subType, clientData);
860 } while (r && !sysno && fi->file_more);
868 int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
869 const char *buf, size_t buf_size,
870 const char *databaseName, int delete_flag,
871 int test_mode, int *sysno,
872 int store_keys, int store_data,
873 const char *match_criteria)
875 RecordAttr *recordAttr;
876 struct recExtractCtrl extractCtrl;
882 const char *fname = "<no file>";
884 long recordOffset = 0;
885 struct zebra_fetch_control fc;
888 fc.record_int_buf = buf;
889 fc.record_int_len = buf_size;
890 fc.record_int_pos = 0;
892 fc.record_offset = 0;
894 extractCtrl.offset = 0;
895 extractCtrl.readf = zebra_record_int_read;
896 extractCtrl.seekf = zebra_record_int_seek;
897 extractCtrl.tellf = zebra_record_int_tell;
898 extractCtrl.endf = zebra_record_int_end;
899 extractCtrl.fh = &fc;
901 /* announce database */
902 if (zebraExplain_curDatabase (zh->reg->zei, databaseName))
904 if (zebraExplain_newDatabase (zh->reg->zei, databaseName, 0))
908 recType_byName (zh->reg->recTypes, recordType, subType,
911 logf (LOG_WARN, "No such record type: %s", recordType);
915 zh->reg->keys.buf_used = 0;
916 zh->reg->keys.prevAttrUse = -1;
917 zh->reg->keys.prevAttrSet = -1;
918 zh->reg->keys.prevSeqNo = 0;
919 zh->reg->sortKeys = 0;
921 extractCtrl.subType = subType;
922 extractCtrl.init = extract_init;
923 extractCtrl.tokenAdd = extract_token_add;
924 extractCtrl.schemaAdd = extract_schema_add;
925 extractCtrl.dh = zh->reg->dh;
926 extractCtrl.handle = zh;
927 extractCtrl.zebra_maps = zh->reg->zebra_maps;
928 extractCtrl.flagShowRecords = 0;
929 for (i = 0; i<256; i++)
931 if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
932 extractCtrl.seqno[i] = 1;
934 extractCtrl.seqno[i] = 0;
937 r = (*recType->extract)(clientData, &extractCtrl);
939 if (r == RECCTRL_EXTRACT_EOF)
941 else if (r == RECCTRL_EXTRACT_ERROR)
943 /* error occured during extraction ... */
945 yaz_log (LOG_WARN, "extract error");
947 if (rGroup->flagRw &&
948 zh->records_processed < rGroup->fileVerboseLimit)
950 logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
951 fname, (long) recordOffset);
956 if (zh->reg->keys.buf_used == 0)
958 /* the extraction process returned no information - the record
959 is probably empty - unless flagShowRecords is in use */
962 logf (LOG_WARN, "No keys generated for record");
963 logf (LOG_WARN, " The file is probably empty");
973 logf (LOG_LOG, "delete %s %s %ld", recordType,
974 fname, (long) recordOffset);
975 logf (LOG_WARN, "cannot delete record above (seems new)");
978 logf (LOG_LOG, "add %s %s %ld", recordType, fname,
979 (long) recordOffset);
980 rec = rec_new (zh->reg->records);
984 recordAttr = rec_init_attr (zh->reg->zei, rec);
988 dict_insert (zh->reg->matchDict, matchStr,
989 sizeof(*sysno), sysno);
991 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
992 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
996 /* record already exists */
997 struct recKeys delkeys;
999 rec = rec_get (zh->reg->records, *sysno);
1002 recordAttr = rec_init_attr (zh->reg->zei, rec);
1004 if (recordAttr->runNumber ==
1005 zebraExplain_runNumberIncrement (zh->reg->zei, 0))
1007 logf (LOG_LOG, "skipped %s %s %ld", recordType,
1008 fname, (long) recordOffset);
1012 delkeys.buf_used = rec->size[recInfo_delKeys];
1013 delkeys.buf = rec->info[recInfo_delKeys];
1014 extract_flushSortKeys (zh, *sysno, 0, &zh->reg->sortKeys);
1015 extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
1018 /* record going to be deleted */
1019 if (!delkeys.buf_used)
1021 logf (LOG_LOG, "delete %s %s %ld", recordType,
1022 fname, (long) recordOffset);
1023 logf (LOG_WARN, "cannot delete file above, storeKeys false");
1027 logf (LOG_LOG, "delete %s %s %ld", recordType,
1028 fname, (long) recordOffset);
1031 dict_delete (matchDict, matchStr);
1033 rec_del (zh->reg->records, &rec);
1040 /* record going to be updated */
1041 if (!delkeys.buf_used)
1043 logf (LOG_LOG, "update %s %s %ld", recordType,
1044 fname, (long) recordOffset);
1045 logf (LOG_WARN, "cannot update file above, storeKeys false");
1049 logf (LOG_LOG, "update %s %s %ld", recordType,
1050 fname, (long) recordOffset);
1051 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
1055 /* update file type */
1056 xfree (rec->info[recInfo_fileType]);
1057 rec->info[recInfo_fileType] =
1058 rec_strdup (recordType, &rec->size[recInfo_fileType]);
1060 /* update filename */
1061 xfree (rec->info[recInfo_filename]);
1062 rec->info[recInfo_filename] =
1063 rec_strdup (fname, &rec->size[recInfo_filename]);
1065 /* update delete keys */
1066 xfree (rec->info[recInfo_delKeys]);
1067 if (zh->reg->keys.buf_used > 0 && store_keys == 1)
1069 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
1070 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
1071 zh->reg->keys.buf = NULL;
1072 zh->reg->keys.buf_max = 0;
1076 rec->info[recInfo_delKeys] = NULL;
1077 rec->size[recInfo_delKeys] = 0;
1080 /* save file size of original record */
1081 zebraExplain_recordBytesIncrement (zh->reg->zei,
1082 - recordAttr->recordSize);
1084 recordAttr->recordSize = fi->file_moffset - recordOffset;
1085 if (!recordAttr->recordSize)
1086 recordAttr->recordSize = fi->file_max - recordOffset;
1088 recordAttr->recordSize = buf_size;
1090 zebraExplain_recordBytesIncrement (zh->reg->zei,
1091 recordAttr->recordSize);
1093 /* set run-number for this record */
1094 recordAttr->runNumber =
1095 zebraExplain_runNumberIncrement (zh->reg->zei, 0);
1097 /* update store data */
1098 xfree (rec->info[recInfo_storeData]);
1099 if (store_data == 1)
1101 rec->size[recInfo_storeData] = recordAttr->recordSize;
1102 rec->info[recInfo_storeData] = (char *)
1103 xmalloc (recordAttr->recordSize);
1105 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize);
1107 if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
1109 logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
1110 (long) recordOffset, fname);
1113 if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
1114 < recordAttr->recordSize)
1116 logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
1117 recordAttr->recordSize, fname);
1124 rec->info[recInfo_storeData] = NULL;
1125 rec->size[recInfo_storeData] = 0;
1127 /* update database name */
1128 xfree (rec->info[recInfo_databaseName]);
1129 rec->info[recInfo_databaseName] =
1130 rec_strdup (databaseName, &rec->size[recInfo_databaseName]);
1133 recordAttr->recordOffset = recordOffset;
1135 /* commit this record */
1136 rec_put (zh->reg->records, &rec);
1141 int explain_extract (void *handle, Record rec, data1_node *n)
1143 ZebraHandle zh = (ZebraHandle) handle;
1144 struct recExtractCtrl extractCtrl;
1147 if (zebraExplain_curDatabase (zh->reg->zei,
1148 rec->info[recInfo_databaseName]))
1151 if (zebraExplain_newDatabase (zh->reg->zei,
1152 rec->info[recInfo_databaseName], 0))
1156 zh->reg->keys.buf_used = 0;
1157 zh->reg->keys.prevAttrUse = -1;
1158 zh->reg->keys.prevAttrSet = -1;
1159 zh->reg->keys.prevSeqNo = 0;
1160 zh->reg->sortKeys = 0;
1162 extractCtrl.init = extract_init;
1163 extractCtrl.tokenAdd = extract_token_add;
1164 extractCtrl.schemaAdd = extract_schema_add;
1165 extractCtrl.dh = zh->reg->dh;
1166 for (i = 0; i<256; i++)
1167 extractCtrl.seqno[i] = 0;
1168 extractCtrl.zebra_maps = zh->reg->zebra_maps;
1169 extractCtrl.flagShowRecords = 0;
1170 extractCtrl.handle = handle;
1172 grs_extract_tree(&extractCtrl, n);
1174 if (rec->size[recInfo_delKeys])
1176 struct recKeys delkeys;
1177 struct sortKey *sortKeys = 0;
1179 delkeys.buf_used = rec->size[recInfo_delKeys];
1180 delkeys.buf = rec->info[recInfo_delKeys];
1181 extract_flushSortKeys (zh, rec->sysno, 0, &sortKeys);
1182 extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
1184 extract_flushRecordKeys (zh, rec->sysno, 1, &zh->reg->keys);
1185 extract_flushSortKeys (zh, rec->sysno, 1, &zh->reg->sortKeys);
1187 xfree (rec->info[recInfo_delKeys]);
1188 rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
1189 rec->info[recInfo_delKeys] = zh->reg->keys.buf;
1190 zh->reg->keys.buf = NULL;
1191 zh->reg->keys.buf_max = 0;
1195 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
1196 int cmd, struct recKeys *reckeys)
1200 unsigned char attrSet = (unsigned char) -1;
1201 unsigned short attrUse = (unsigned short) -1;
1206 ZebraExplainInfo zei = zh->reg->zei;
1208 if (!zh->reg->key_buf)
1210 int mem = 8*1024*1024;
1211 zh->reg->key_buf = (char**) xmalloc (mem);
1212 zh->reg->ptr_top = mem/sizeof(char*);
1214 zh->reg->key_buf_used = 0;
1215 zh->reg->key_file_no = 0;
1217 zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
1218 while (off < reckeys->buf_used)
1220 const char *src = reckeys->buf + off;
1229 memcpy (&ch, src, sizeof(ch));
1235 memcpy (&attrSet, src, sizeof(attrSet));
1236 src += sizeof(attrSet);
1240 memcpy (&attrUse, src, sizeof(attrUse));
1241 src += sizeof(attrUse);
1244 if (zh->reg->key_buf_used + 1024 >
1245 (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
1246 extract_flushWriteKeys (zh);
1248 (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
1249 (char*)zh->reg->key_buf + zh->reg->key_buf_used;
1252 ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
1254 ch = zebraExplain_addSU (zei, attrSet, attrUse);
1257 zh->reg->key_buf_used +=
1258 key_SU_encode (ch,((char*)zh->reg->key_buf) +
1259 zh->reg->key_buf_used);
1262 ((char*)zh->reg->key_buf) [(zh->reg->key_buf_used)++] = *src++;
1264 ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = '\0';
1265 ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = cmd;
1268 seqno += ((lead>>2) & 15)-1;
1271 memcpy (&seqno, src, sizeof(seqno));
1272 src += sizeof(seqno);
1276 memcpy ((char*)zh->reg->key_buf + zh->reg->key_buf_used, &key, sizeof(key));
1277 (zh->reg->key_buf_used) += sizeof(key);
1278 off = src - reckeys->buf;
1280 assert (off == reckeys->buf_used);
1283 void extract_flushWriteKeys (ZebraHandle zh)
1286 char out_fname[200];
1288 struct encode_info encode_info;
1289 int ptr_i = zh->reg->ptr_i;
1293 if (!zh->reg->key_buf || ptr_i <= 0)
1296 (zh->reg->key_file_no)++;
1297 logf (LOG_LOG, "sorting section %d", (zh->reg->key_file_no));
1299 qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
1300 sizeof(char*), key_qsort_compare);
1301 extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
1303 if (!(outf = fopen (out_fname, "wb")))
1305 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1308 logf (LOG_LOG, "writing section %d", zh->reg->key_file_no);
1309 prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1311 encode_key_init (&encode_info);
1312 encode_key_write (cp, &encode_info, outf);
1316 cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
1317 if (strcmp (cp, prevcp))
1319 encode_key_init (&encode_info);
1320 encode_key_write (cp, &encode_info, outf);
1324 encode_key_write (cp + strlen(cp), &encode_info, outf);
1327 qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
1328 extract_get_fname_tmp (out_fname, key_file_no);
1330 if (!(outf = fopen (out_fname, "wb")))
1332 logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
1335 logf (LOG_LOG, "writing section %d", key_file_no);
1337 prevcp = key_buf[ptr_top-i];
1339 if (!--i || strcmp (prevcp, key_buf[ptr_top-i]))
1341 key_y_len = strlen(prevcp)+1;
1343 logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
1344 key_y_len, prevcp[0], prevcp[1], 2+prevcp);
1346 qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
1347 sizeof(char*), key_y_compare);
1348 cp = key_buf[ptr_top-ptr_i];
1350 encode_key_init (&encode_info);
1351 encode_key_write (cp, &encode_info, outf);
1354 cp = key_buf[ptr_top-ptr_i];
1355 encode_key_write (cp+key_y_len, &encode_info, outf);
1359 prevcp = key_buf[ptr_top-ptr_i];
1364 logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
1367 logf (LOG_LOG, "finished section %d", zh->reg->key_file_no);
1369 zh->reg->key_buf_used = 0;
1372 void extract_add_index_string (RecWord *p, const char *string,
1376 unsigned char attrSet;
1377 unsigned short attrUse;
1380 int *pseqno = &p->seqno;
1381 ZebraHandle zh = p->extractCtrl->handle;
1382 ZebraExplainInfo zei = zh->reg->zei;
1383 struct recKeys *keys = &zh->reg->keys;
1385 if (keys->buf_used+1024 > keys->buf_max)
1389 b = (char *) xmalloc (keys->buf_max += 128000);
1390 if (keys->buf_used > 0)
1391 memcpy (b, keys->buf, keys->buf_used);
1395 dst = keys->buf + keys->buf_used;
1397 attrSet = p->attrSet;
1398 if (keys->buf_used > 0 && keys->prevAttrSet == attrSet)
1401 keys->prevAttrSet = attrSet;
1402 attrUse = p->attrUse;
1403 if (keys->buf_used > 0 && keys->prevAttrUse == attrUse)
1406 keys->prevAttrUse = attrUse;
1408 diff = 1 + *pseqno - keys->prevSeqNo;
1409 if (diff >= 1 && diff <= 15)
1410 lead |= (diff << 2);
1414 keys->prevSeqNo = *pseqno;
1421 int ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
1424 ch = zebraExplain_addSU (zei, attrSet, attrUse);
1425 yaz_log (LOG_DEBUG, "addSU set=%d use=%d SU=%d",
1426 attrSet, attrUse, ch);
1429 memcpy (dst, &ch, sizeof(ch));
1435 memcpy (dst, &attrSet, sizeof(attrSet));
1436 dst += sizeof(attrSet);
1440 memcpy (dst, &attrUse, sizeof(attrUse));
1441 dst += sizeof(attrUse);
1444 *dst++ = p->reg_type;
1445 memcpy (dst, string, length);
1451 memcpy (dst, pseqno, sizeof(*pseqno));
1452 dst += sizeof(*pseqno);
1454 keys->buf_used = dst - keys->buf;
1457 static void extract_add_sort_string (RecWord *p, const char *string,
1461 ZebraHandle zh = p->extractCtrl->handle;
1463 for (sk = zh->reg->sortKeys; sk; sk = sk->next)
1464 if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
1467 sk = (struct sortKey *) xmalloc (sizeof(*sk));
1468 sk->next = zh->reg->sortKeys;
1469 zh->reg->sortKeys = sk;
1471 sk->string = (char *) xmalloc (length);
1472 sk->length = length;
1473 memcpy (sk->string, string, length);
1475 sk->attrSet = p->attrSet;
1476 sk->attrUse = p->attrUse;
1479 void extract_add_string (RecWord *p, const char *string, int length)
1481 assert (length > 0);
1482 if (zebra_maps_is_sort (p->zebra_maps, p->reg_type))
1483 extract_add_sort_string (p, string, length);
1485 extract_add_index_string (p, string, length);
1488 static void extract_add_incomplete_field (RecWord *p)
1490 const char *b = p->string;
1491 int remain = p->length;
1492 const char **map = 0;
1495 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1499 char buf[IT_MAX_WORD+1];
1503 while (map && *map && **map == *CHR_SPACE)
1505 remain = p->length - (b - p->string);
1507 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1514 while (map && *map && **map != *CHR_SPACE)
1516 const char *cp = *map;
1518 while (i < IT_MAX_WORD && *cp)
1520 remain = p->length - (b - p->string);
1522 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1528 extract_add_string (p, buf, i);
1533 static void extract_add_complete_field (RecWord *p)
1535 const char *b = p->string;
1536 char buf[IT_MAX_WORD+1];
1537 const char **map = 0;
1538 int i = 0, remain = p->length;
1541 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
1543 while (remain > 0 && i < IT_MAX_WORD)
1545 while (map && *map && **map == *CHR_SPACE)
1547 remain = p->length - (b - p->string);
1549 map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
1556 if (i && i < IT_MAX_WORD)
1557 buf[i++] = *CHR_SPACE;
1558 while (map && *map && **map != *CHR_SPACE)
1560 const char *cp = *map;
1562 if (i >= IT_MAX_WORD)
1564 while (i < IT_MAX_WORD && *cp)
1566 remain = p->length - (b - p->string);
1568 map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
1576 extract_add_string (p, buf, i);
1579 void extract_token_add (RecWord *p)
1582 if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
1583 p->string, p->length)))
1585 p->string = wrbuf_buf(wrbuf);
1586 p->length = wrbuf_len(wrbuf);
1588 if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
1589 extract_add_complete_field (p);
1591 extract_add_incomplete_field(p);
1594 void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
1596 ZebraHandle zh = (ZebraHandle) (p->handle);
1597 zebraExplain_addSchema (zh->reg->zei, oid);
1600 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
1601 int cmd, struct sortKey **skp)
1603 struct sortKey *sk = *skp;
1604 SortIdx sortIdx = zh->reg->sortIdx;
1606 sortIdx_sysno (sortIdx, sysno);
1609 struct sortKey *sk_next = sk->next;
1610 sortIdx_type (sortIdx, sk->attrUse);
1611 sortIdx_add (sortIdx, sk->string, sk->length);
1619 void encode_key_init (struct encode_info *i)
1626 char *encode_key_int (int d, char *bp)
1630 else if (d <= 16383)
1632 *bp++ = 64 + (d>>8);
1635 else if (d <= 4194303)
1637 *bp++ = 128 + (d>>16);
1638 *bp++ = (d>>8) & 255;
1643 *bp++ = 192 + (d>>24);
1644 *bp++ = (d>>16) & 255;
1645 *bp++ = (d>>8) & 255;
1651 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
1656 while ((*bp++ = *k++))
1658 memcpy (&key, k+1, sizeof(struct it_key));
1659 bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
1660 if (i->sysno != key.sysno)
1662 i->sysno = key.sysno;
1665 else if (!i->seqno && !key.seqno && i->cmd == *k)
1667 bp = encode_key_int (key.seqno - i->seqno, bp);
1668 i->seqno = key.seqno;
1670 if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
1672 logf (LOG_FATAL|LOG_ERRNO, "fwrite");