/*
- * Copyright (C) 1994-1998, Index Data I/S
+ * Copyright (C) 1994-1999, Index Data
* All rights reserved.
* Sebastian Hammer, Adam Dickmeiss
*
* $Log: extract.c,v $
- * Revision 1.84 1998-06-11 15:42:22 adam
+ * Revision 1.95 1999-05-21 12:00:17 adam
+ * Better diagnostics for extraction process.
+ *
+ * Revision 1.94 1999/05/20 12:57:18 adam
+ * Implemented TCL filter. Updated recctrl system.
+ *
+ * Revision 1.93 1999/05/15 14:36:38 adam
+ * Updated dictionary. Implemented "compression" of dictionary.
+ *
+ * Revision 1.92 1999/03/09 16:27:49 adam
+ * More work on SDRKit integration.
+ *
+ * Revision 1.91 1999/02/12 13:29:22 adam
+ * Implemented position-flag for registers.
+ *
+ * Revision 1.90 1999/02/02 14:50:52 adam
+ * Updated WIN32 code specific sections. Changed header.
+ *
+ * Revision 1.89 1998/10/28 10:54:38 adam
+ * SDRKit integration.
+ *
+ * Revision 1.88 1998/10/16 08:14:29 adam
+ * Updated record control system.
+ *
+ * Revision 1.87 1998/10/15 13:10:33 adam
+ * Fixed bug in Zebra that caused it to stop indexing when empty
+ * record was read.
+ *
+ * Revision 1.86 1998/10/13 20:33:53 adam
+ * Fixed one log message and change use ordinal to be an unsigned char.
+ *
+ * Revision 1.85 1998/09/22 10:03:41 adam
+ * Changed result sets to be persistent in the sense that they can
+ * be re-searched if needed.
+ * Fixed memory leak in rsm_or.
+ *
+ * Revision 1.84 1998/06/11 15:42:22 adam
* Changed the way use attributes are specified in the recordId
* specification.
*
*/
#include <stdio.h>
#include <assert.h>
-#ifdef WINDOWS
+#ifdef WIN32
#include <io.h>
#else
#include <unistd.h>
#include "zinfo.h"
+#ifndef ZEBRASDR
+#define ZEBRASDR 0
+#endif
+
+#if ZEBRASDR
+#include "zebrasdr.h"
+#endif
+
static Dict matchDict;
static Records records = NULL;
key_buf_used = 0;
key_file_no = 0;
- if (!(matchDict = dict_open (bfs, GMATCH_DICT, 50, rw)))
+ if (!(matchDict = dict_open (bfs, GMATCH_DICT, 50, rw, 0)))
{
logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT);
return -1;
struct encode_info {
int sysno;
int seqno;
+ int cmd;
char buf[768];
};
{
i->sysno = 0;
i->seqno = 0;
+ i->cmd = -1;
}
char *encode_key_int (int d, char *bp)
i->sysno = key.sysno;
i->seqno = 0;
}
+ else if (!i->seqno && !key.seqno && i->cmd == *k)
+ return;
bp = encode_key_int (key.seqno - i->seqno, bp);
i->seqno = key.seqno;
+ i->cmd = *k;
if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
{
logf (LOG_FATAL|LOG_ERRNO, "fwrite");
static void addIndexString (RecWord *p, const char *string, int length)
{
char *dst;
- char attrSet;
- short attrUse;
+ unsigned char attrSet;
+ unsigned short attrUse;
int lead = 0;
int diff = 0;
int *pseqno = &p->seqnos[p->reg_type];
dst += sizeof(*pseqno);
}
reckeys.buf_used = dst - reckeys.buf;
- (*pseqno)++;
+ if (*pseqno)
+ (*pseqno)++;
}
static void addSortString (RecWord *p, const char *string, int length)
sk->next = sortKeys;
sortKeys = sk;
- sk->string = xmalloc (p->length);
- sk->length = p->length;
- memcpy (sk->string, p->string, p->length);
+ sk->string = xmalloc (length);
+ sk->length = length;
+ memcpy (sk->string, string, length);
+
sk->attrSet = p->attrSet;
sk->attrUse = p->attrUse;
}
static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys)
{
- char attrSet = -1;
- short attrUse = -1;
+ unsigned char attrSet = (unsigned char) -1;
+ unsigned short attrUse = (unsigned short) -1;
int seqno = 0;
int off = 0;
}
struct file_read_info {
- off_t file_max;
- off_t file_offset;
- off_t file_moffset;
+ off_t file_max; /* maximum offset so far */
+ off_t file_offset; /* current offset */
+ off_t file_moffset; /* offset of rec/rec boundary */
int file_more;
int fd;
+ char *sdrbuf;
+ int sdrmax;
};
static struct file_read_info *file_read_start (int fd)
fi->fd = fd;
fi->file_max = 0;
fi->file_moffset = 0;
+ fi->sdrbuf = 0;
+ fi->sdrmax = 0;
return fi;
}
{
struct file_read_info *p = handle;
p->file_offset = offset;
+ if (p->sdrbuf)
+ return offset;
return lseek (p->fd, offset, SEEK_SET);
}
struct file_read_info *p = handle;
int fd = p->fd;
int r;
- r = read (fd, buf, count);
+ if (p->sdrbuf)
+ {
+ r = count;
+ if (r > p->sdrmax - p->file_offset)
+ r = p->sdrmax - p->file_offset;
+ if (r)
+ memcpy (buf, p->sdrbuf + p->file_offset, r);
+ }
+ else
+ r = read (fd, buf, count);
if (r > 0)
{
p->file_offset += r;
struct file_read_info *p = handle;
p->file_offset = p->file_moffset;
- if (p->file_moffset)
+ if (!p->sdrbuf && p->file_moffset)
lseek (p->fd, p->file_moffset, SEEK_SET);
p->file_more = 0;
}
static int recordExtract (SYSNO *sysno, const char *fname,
struct recordGroup *rGroup, int deleteFlag,
struct file_read_info *fi,
- RecType recType, char *subType)
+ RecType recType, char *subType, void *clientData)
{
RecordAttr *recordAttr;
int r;
extractCtrl.addSchema = addSchema;
extractCtrl.dh = rGroup->dh;
for (i = 0; i<256; i++)
- extractCtrl.seqno[i] = 0;
+ {
+ if (zebra_maps_is_positioned(rGroup->zebra_maps, i))
+ extractCtrl.seqno[i] = 1;
+ else
+ extractCtrl.seqno[i] = 0;
+ }
extractCtrl.zebra_maps = rGroup->zebra_maps;
extractCtrl.flagShowRecords = !rGroup->flagRw;
logInfo.rGroup = rGroup;
log_event_start (recordLogPreamble, &logInfo);
- r = (*recType->extract)(&extractCtrl);
+ r = (*recType->extract)(clientData, &extractCtrl);
log_event_start (NULL, NULL);
- if (r)
- {
+ if (r == RECCTRL_EXTRACT_EOF)
+ return 0;
+ else if (r == RECCTRL_EXTRACT_ERROR)
+ {
/* error occured during extraction ... */
if (rGroup->flagRw &&
records_processed < rGroup->fileVerboseLimit)
{
- logf (LOG_WARN, "fail %s %s %ld code = %d", rGroup->recordType,
- fname, (long) recordOffset, r);
+ logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
+ fname, (long) recordOffset);
}
return 0;
}
return 1;
logf (LOG_WARN, "No keys generated for file %s", fname);
logf (LOG_WARN, " The file is probably empty");
- return 0;
+ return 1;
}
}
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);
+ logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s",
+ (long) recordOffset, fname);
exit (1);
}
if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
struct recordGroup rGroupM;
struct recordGroup *rGroup = &rGroupM;
struct file_read_info *fi;
+ void *clientData;
memcpy (rGroup, rGroupP, sizeof(*rGroupP));
}
if (!*rGroup->recordType)
return 0;
- if (!(recType = recType_byName (rGroup->recordType, subType)))
+ if (!(recType =
+ recType_byName (rGroup->recTypes, rGroup->recordType, subType,
+ &clientData)))
{
logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
return 0;
if (rGroup->flagStoreKeys == -1)
rGroup->flagStoreKeys = 0;
+#if ZEBRASDR
+ if (rGroup->useSDR)
+ {
+ ZebraSdrHandle h;
+ char xname[128], *xp;
+
+ strncpy (xname, fname, 127);
+ if (!(xp = strchr (xname, '.')))
+ return 0;
+ *xp = '\0';
+ if (strcmp (xp+1, "sdr.bits"))
+ return 0;
+
+ h = zebraSdr_open (xname);
+ if (!h)
+ {
+ logf (LOG_WARN, "sdr open %s", xname);
+ return 0;
+ }
+ for (;;)
+ {
+ unsigned char *buf;
+ char sdr_name[128];
+ int r, segmentno;
+
+ segmentno = zebraSdr_segment (h, 0);
+ sprintf (sdr_name, "%%%s.%d", xname, segmentno);
+
+#if 0
+ if (segmentno > 20)
+ break;
+#endif
+ r = zebraSdr_read (h, &buf);
+
+ if (!r)
+ break;
+
+ fi = file_read_start (0);
+ fi->sdrbuf = buf;
+ fi->sdrmax = r;
+ do
+ {
+ file_begin (fi);
+ r = recordExtract (sysno, sdr_name, rGroup, deleteFlag, fi,
+ recType, subType);
+ } while (r && !sysno && fi->file_more);
+ file_read_stop (fi);
+ free (buf);
+ }
+ zebraSdr_close (h);
+ return 1;
+ }
+#endif
if (sysno && deleteFlag)
fd = -1;
else
{
file_begin (fi);
r = recordExtract (sysno, fname, rGroup, deleteFlag, fi,
- recType, subType);
+ recType, subType, clientData);
} while (r && !sysno && fi->file_more);
file_read_stop (fi);
if (fd != -1)
projects / idzebra-moved-to-github.git / blobdiff