* Sebastian Hammer, Adam Dickmeiss
*
* $Log: extract.c,v $
- * Revision 1.79 1998-02-17 10:32:52 adam
+ * Revision 1.81 1998-03-11 11:19:04 adam
+ * Changed the way sequence numbers are generated.
+ *
+ * Revision 1.80 1998/03/05 08:45:11 adam
+ * New result set model and modular ranking system. Moved towards
+ * descent server API. System information stored as "SGML" records.
+ *
+ * Revision 1.79 1998/02/17 10:32:52 adam
* Fixed bug: binary files weren't opened with flag b on NT.
*
* Revision 1.78 1998/02/10 12:03:05 adam
static int records_deleted = 0;
static int records_processed = 0;
-static ZebTargetInfo *zti = NULL;
+static ZebraExplainInfo zti = NULL;
static void logRecord (int showFlag)
{
}
}
-int key_open (BFiles bfs, int mem, int rw)
+int key_open (BFiles bfs, int mem, int rw, data1_handle dh)
{
if (!mem)
mem = atoi(res_get_def (common_resource, "memMax", "4"))*1024*1024;
dict_close (matchDict);
return -1;
}
- zti = zebTargetInfo_open (records, rw);
+ zti = zebraExplain_open (records, dh, rw);
if (!zti)
{
rec_close (&records);
key_buf_used = 0;
}
-int key_close ()
+int key_close (int rw)
{
key_flush ();
xfree (key_buf);
-#if 1
- zebTargetInfo_close (zti, 1);
-#endif
+ if (rw)
+ zebraExplain_runNumberIncrement (zti, 1);
+ zebraExplain_close (zti, rw);
rec_close (&records);
dict_close (matchDict);
sortIdx_close (sortIdx);
static void wordInit (struct recExtractCtrl *p, RecWord *w)
{
w->zebra_maps = p->zebra_maps;
+ w->seqnos = p->seqno;
w->attrSet = 1;
w->attrUse = 1016;
w->reg_type = 'w';
short attrUse;
int lead = 0;
int diff = 0;
+ int *pseqno = &p->seqnos[p->reg_type];
if (reckeys.buf_used+1024 > reckeys.buf_max)
{
else
reckeys.prevAttrUse = attrUse;
#if 1
- diff = 1 + p->seqno - reckeys.prevSeqNo;
+ diff = 1 + *pseqno - reckeys.prevSeqNo;
if (diff >= 1 && diff <= 15)
lead |= (diff << 2);
else
diff = 0;
#endif
- reckeys.prevSeqNo = p->seqno;
-
+ reckeys.prevSeqNo = *pseqno;
+
*dst++ = lead;
if (!(lead & 1))
if (!diff)
{
- memcpy (dst, &p->seqno, sizeof(p->seqno));
- dst += sizeof(p->seqno);
+ memcpy (dst, pseqno, sizeof(*pseqno));
+ dst += sizeof(*pseqno);
}
reckeys.buf_used = dst - reckeys.buf;
- (p->seqno)++;
+ (*pseqno)++;
}
static void addSortString (RecWord *p, const char *string, int length)
int seqno = 0;
int off = 0;
- if (zebTargetInfo_curDatabase (zti, databaseName))
+ if (zebraExplain_curDatabase (zti, databaseName))
{
- if (zebTargetInfo_newDatabase (zti, databaseName))
+ if (zebraExplain_newDatabase (zti, databaseName))
abort ();
}
+ zebraExplain_recordCountIncrement (zti, cmd ? 1 : -1);
while (off < reckeys->buf_used)
{
const char *src = reckeys->buf + off;
++ptr_i;
key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
- ch = zebTargetInfo_lookupSU (zti, attrSet, attrUse);
+ ch = zebraExplain_lookupSU (zti, attrSet, attrUse);
if (ch < 0)
- ch = zebTargetInfo_addSU (zti, attrSet, attrUse);
+ ch = zebraExplain_addSU (zti, attrSet, attrUse);
assert (ch > 0);
((char*) key_buf) [key_buf_used++] = ch;
while (*src)
}
static const char **searchRecordKey (struct recKeys *reckeys,
- int attrSetS, int attrUseS)
+ int attrSetS, int attrUseS)
{
static const char *ws[32];
int off = 0;
char *subType)
{
struct recExtractCtrl extractCtrl;
+ RecordAttr *recordAttr;
int r;
char *matchStr;
SYSNO sysnotmp;
if (fi->fd != -1)
{
+ int i;
/* we are going to read from a file, so prepare the extraction */
extractCtrl.fh = fi;
extractCtrl.subType = subType;
reckeys.prevAttrSet = -1;
reckeys.prevSeqNo = 0;
+ for (i = 0; i<256; i++)
+ extractCtrl.seqno[i] = 0;
recordOffset = fi->file_moffset;
extractCtrl.offset = recordOffset;
extractCtrl.readf = file_read;
logf (LOG_LOG, "add %s %s %ld", rGroup->recordType,
fname, (long) recordOffset);
rec = rec_new (records);
+
*sysno = rec->sysno;
+ recordAttr = rec_init_attr (zti, rec);
+
if (matchStr)
{
dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
rec = rec_get (records, *sysno);
assert (rec);
+
+ recordAttr = rec_init_attr (zti, rec);
+
+ if (recordAttr->runNumber == zebraExplain_runNumberIncrement (zti, 0))
+ {
+ logf (LOG_LOG, "skipped %s %s %ld", rGroup->recordType,
+ fname, (long) recordOffset);
+ rec_rm (&rec);
+ logRecord (0);
+ return 1;
+ }
delkeys.buf_used = rec->size[recInfo_delKeys];
delkeys.buf = rec->info[recInfo_delKeys];
flushSortKeys (*sysno, 0);
dict_delete (matchDict, matchStr);
rec_del (records, &rec);
}
+ rec_rm (&rec);
logRecord (0);
return 1;
}
rec->size[recInfo_delKeys] = 0;
}
+ /* save file size of original record */
+ zebraExplain_recordBytesIncrement (zti, - recordAttr->recordSize);
+ recordAttr->recordSize = fi->file_moffset - recordOffset;
+ if (!recordAttr->recordSize)
+ recordAttr->recordSize = fi->file_max - recordOffset;
+ zebraExplain_recordBytesIncrement (zti, recordAttr->recordSize);
+
+ /* set run-number for this record */
+ recordAttr->runNumber = zebraExplain_runNumberIncrement (zti, 0);
+
/* update store data */
xfree (rec->info[recInfo_storeData]);
if (rGroup->flagStoreData == 1)
{
- int size = fi->file_moffset - recordOffset;
- if (!size)
- size = fi->file_max - recordOffset;
- rec->size[recInfo_storeData] = size;
- rec->info[recInfo_storeData] = xmalloc (size);
+ rec->size[recInfo_storeData] = recordAttr->recordSize;
+ rec->info[recInfo_storeData] = xmalloc (recordAttr->recordSize);
if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
{
logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s", fname,
(long) recordOffset);
exit (1);
}
- if (read (fi->fd, rec->info[recInfo_storeData], size) < size)
+ if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
+ < recordAttr->recordSize)
{
logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
- fi->file_max, fname);
+ recordAttr->recordSize, fname);
exit (1);
}
}
rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
/* update offset */
- xfree (rec->info[recInfo_offset]);
-
- rec->size[recInfo_offset] = sizeof(recordOffset);
- rec->info[recInfo_offset] = xmalloc (sizeof(recordOffset));
- memcpy (rec->info[recInfo_offset], &recordOffset, sizeof(recordOffset));
+ recordAttr->recordOffset = recordOffset;
/* commit this record */
rec_put (records, &rec);