/*
- * 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 1.80 1998-03-05 08:45:11 adam
+ * 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.
+ *
+ * Revision 1.83 1998/06/08 14:43:10 adam
+ * Added suport for EXPLAIN Proxy servers - added settings databasePath
+ * and explainDatabase to facilitate this. Increased maximum number
+ * of databases and attributes in one register.
+ *
+ * Revision 1.82 1998/05/20 10:12:15 adam
+ * Implemented automatic EXPLAIN database maintenance.
+ * Modified Zebra to work with ASN.1 compiled version of YAZ.
+ *
+ * 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.
*
*/
#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;
}
}
-int key_open (BFiles bfs, int mem, int rw, data1_handle dh)
+static int explain_extract (void *handle, Record drec, data1_node *n);
+
+int key_open (struct recordGroup *rGroup, int mem)
{
+ BFiles bfs = rGroup->bfs;
+ int rw = rGroup->flagRw;
+ data1_handle dh = rGroup->dh;
if (!mem)
mem = atoi(res_get_def (common_resource, "memMax", "4"))*1024*1024;
if (mem < 50000)
dict_close (matchDict);
return -1;
}
- zti = zebraExplain_open (records, dh, rw);
+ zti = zebraExplain_open (records, dh, common_resource,
+ rw, rGroup, explain_extract);
if (!zti)
{
rec_close (&records);
key_buf_used = 0;
}
-int key_close (int rw)
+int key_close (struct recordGroup *rGroup)
{
- key_flush ();
- xfree (key_buf);
+ int rw = rGroup->flagRw;
if (rw)
zebraExplain_runNumberIncrement (zti, 1);
- zebraExplain_close (zti, rw);
+ zebraExplain_close (zti, rw, 0);
+ key_flush ();
+ xfree (key_buf);
rec_close (&records);
dict_close (matchDict);
sortIdx_close (sortIdx);
static void wordInit (struct recExtractCtrl *p, RecWord *w)
{
w->zebra_maps = p->zebra_maps;
- w->attrSet = 1;
+ w->seqnos = p->seqno;
+ w->attrSet = VAL_BIB1;
w->attrUse = 1016;
w->reg_type = 'w';
}
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];
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)
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;
}
sortKeys = NULL;
}
-static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys,
- const char *databaseName)
+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;
- if (zebraExplain_curDatabase (zti, databaseName))
- {
- if (zebraExplain_newDatabase (zti, databaseName))
- abort ();
- }
zebraExplain_recordCountIncrement (zti, cmd ? 1 : -1);
while (off < reckeys->buf_used)
{
if (ch < 0)
ch = zebraExplain_addSU (zti, attrSet, attrUse);
assert (ch > 0);
- ((char*) key_buf) [key_buf_used++] = ch;
+ key_buf_used += key_SU_code (ch, ((char*)key_buf) + key_buf_used);
+
while (*src)
((char*)key_buf) [key_buf_used++] = *src++;
src++;
}
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;
}
static void file_read_stop (struct file_read_info *fi)
{
- assert (fi);
xfree (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;
}
p->file_moffset = offset;
}
-static int atois (const char **s)
-{
- int val = 0, c;
- while ( (c=**s) >= '0' && c <= '9')
- {
- val = val*10 + c - '0';
- ++(*s);
- }
- return val;
-}
-
static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
- const char *fname,
- const char *spec)
+ const char *fname, const char *spec)
{
static char dstBuf[2048];
char *dst = dstBuf;
const char *s = spec;
static const char **w;
- int i;
while (1)
{
break;
if (*s == '(')
{
+ char attset_str[64], attname_str[64];
+ data1_attset *attset;
+ int i;
char matchFlag[32];
- int attrSet, attrUse;
+ int attSet = 1, attUse = 1;
int first = 1;
s++;
- attrSet = atois (&s);
- if (*s != ',')
- {
- logf (LOG_WARN, "Missing , in match criteria %s in group %s",
- spec, rGroup->groupName ? rGroup->groupName : "none");
- return NULL;
- }
- s++;
- attrUse = atois (&s);
- w = searchRecordKey (reckeys, attrSet, attrUse);
+ for (i = 0; *s && *s != ',' && *s != ')'; s++)
+ if (i < 63)
+ attset_str[i++] = *s;
+ attset_str[i] = '\0';
+
+ if (*s == ',')
+ {
+ s++;
+ for (i = 0; *s && *s != ')'; s++)
+ if (i < 63)
+ attname_str[i++] = *s;
+ attname_str[i] = '\0';
+ }
+
+ if ((attset = data1_get_attset (rGroup->dh, attset_str)))
+ {
+ data1_att *att;
+ attSet = attset->reference;
+ att = data1_getattbyname(rGroup->dh, attset, attname_str);
+ if (att)
+ attUse = att->value;
+ else
+ attUse = atoi (attname_str);
+ }
+ w = searchRecordKey (reckeys, attSet, attUse);
assert (w);
if (*s == ')')
if (first)
{
logf (LOG_WARN, "Record didn't contain match"
- " fields in (%d,%d)", attrSet, attrUse);
+ " fields in (%s,%s)", attset_str, attname_str);
return NULL;
}
}
log_event_start (NULL, NULL);
}
+void addSchema (struct recExtractCtrl *p, Odr_oid *oid)
+{
+ zebraExplain_addSchema (zti, oid);
+}
+
static int recordExtract (SYSNO *sysno, const char *fname,
struct recordGroup *rGroup, int deleteFlag,
- struct file_read_info *fi, RecType recType,
- char *subType)
+ struct file_read_info *fi,
+ RecType recType, char *subType)
{
- struct recExtractCtrl extractCtrl;
RecordAttr *recordAttr;
int r;
char *matchStr;
SYSNO sysnotmp;
- off_t recordOffset = 0;
Record rec;
struct recordLogInfo logInfo;
+ off_t recordOffset = 0;
if (fi->fd != -1)
{
+ struct recExtractCtrl extractCtrl;
+
/* we are going to read from a file, so prepare the extraction */
- extractCtrl.fh = fi;
- extractCtrl.subType = subType;
- extractCtrl.init = wordInit;
- extractCtrl.add = addRecordKey;
- extractCtrl.dh = rGroup->dh;
+ int i;
- reckeys.buf_used = 0;
- reckeys.prevAttrUse = -1;
- reckeys.prevAttrSet = -1;
- reckeys.prevSeqNo = 0;
-
- recordOffset = fi->file_moffset;
- extractCtrl.offset = recordOffset;
- extractCtrl.readf = file_read;
- extractCtrl.seekf = file_seek;
- extractCtrl.tellf = file_tell;
- extractCtrl.endf = file_end;
+ reckeys.buf_used = 0;
+ reckeys.prevAttrUse = -1;
+ reckeys.prevAttrSet = -1;
+ reckeys.prevSeqNo = 0;
+
+ recordOffset = fi->file_moffset;
+ extractCtrl.offset = fi->file_moffset;
+ extractCtrl.readf = file_read;
+ extractCtrl.seekf = file_seek;
+ extractCtrl.tellf = file_tell;
+ extractCtrl.endf = file_end;
+ extractCtrl.fh = fi;
+ extractCtrl.subType = subType;
+ extractCtrl.init = wordInit;
+ extractCtrl.addWord = addRecordKey;
+ extractCtrl.addSchema = addSchema;
+ extractCtrl.dh = rGroup->dh;
+ for (i = 0; i<256; i++)
+ extractCtrl.seqno[i] = 0;
extractCtrl.zebra_maps = rGroup->zebra_maps;
- extractCtrl.flagShowRecords = !rGroup->flagRw;
+ extractCtrl.flagShowRecords = !rGroup->flagRw;
+
if (!rGroup->flagRw)
printf ("File: %s %ld\n", fname, (long) recordOffset);
return 1;
logf (LOG_WARN, "No keys generated for file %s", fname);
logf (LOG_WARN, " The file is probably empty");
- return 0;
+ return 1;
}
}
{
dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
}
- flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName);
+ flushRecordKeys (*sysno, 1, &reckeys);
flushSortKeys (*sysno, 1);
records_inserted++;
delkeys.buf_used = rec->size[recInfo_delKeys];
delkeys.buf = rec->info[recInfo_delKeys];
flushSortKeys (*sysno, 0);
- flushRecordKeys (*sysno, 0, &delkeys, rec->info[recInfo_databaseName]);
+ flushRecordKeys (*sysno, 0, &delkeys);
if (deleteFlag)
{
/* record going to be deleted */
if (records_processed < rGroup->fileVerboseLimit)
logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
fname, (long) recordOffset);
- flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName);
+ flushRecordKeys (*sysno, 1, &reckeys);
records_updated++;
}
}
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)
logf (LOG_DEBUG, "fileExtract %s", fname);
/* determine file extension */
+ *ext = '\0';
for (i = strlen(fname); --i >= 0; )
if (fname[i] == '/')
- {
- strcpy (ext, "");
break;
- }
else if (fname[i] == '.')
{
strcpy (ext, fname+i+1);
if (!(rGroup->recordType = res_get (common_resource, ext_res)))
{
sprintf (ext_res, "%srecordType", gprefix);
- if (!(rGroup->recordType = res_get (common_resource, ext_res)))
- {
- if (records_processed < rGroup->fileVerboseLimit)
- logf (LOG_LOG, "? %s", fname);
- return 0;
- }
+ rGroup->recordType = res_get (common_resource, ext_res);
}
}
if (!rGroup->recordType)
{
if (records_processed < rGroup->fileVerboseLimit)
- logf (LOG_LOG, "? record %s", fname);
+ logf (LOG_LOG, "? %s", fname);
return 0;
}
- if (!(recType = recType_byName (rGroup->recordType, subType)))
+ if (!*rGroup->recordType)
+ return 0;
+ if (!(recType =
+ recType_byName (rGroup->recTypes, rGroup->recordType, subType)))
{
logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
return 0;
if (!rGroup->databaseName)
rGroup->databaseName = "Default";
+ /* determine if explain database */
+
+ sprintf (ext_res, "%sexplainDatabase", gprefix);
+ rGroup->explainDatabase =
+ atoi (res_get_def (common_resource, ext_res, "0"));
+
+ /* announce database */
+ if (zebraExplain_curDatabase (zti, rGroup->databaseName))
+ {
+ if (zebraExplain_newDatabase (zti, rGroup->databaseName,
+ rGroup->explainDatabase))
+ abort ();
+ }
+
if (rGroup->flagStoreData == -1)
{
const char *sval;
if (rGroup->flagStoreKeys == -1)
rGroup->flagStoreKeys = 0;
+#if ZEBRASDR
+ if (1)
+ {
+ ZebraSdrHandle h;
+ char xname[128], *xp;
+
+ strncpy (xname, fname, 127);
+ if ((xp = strchr (xname, '.')))
+ *xp = '\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);
+ logf (LOG_LOG, "SDR: %s", sdr_name);
+
+#if 1
+ 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
return r;
}
+static int explain_extract (void *handle, Record rec, data1_node *n)
+{
+ struct recordGroup *rGroup = (struct recordGroup*) handle;
+ struct recExtractCtrl extractCtrl;
+ int i;
+
+ if (zebraExplain_curDatabase (zti, rec->info[recInfo_databaseName]))
+ {
+ if (zebraExplain_newDatabase (zti, rec->info[recInfo_databaseName], 0))
+ abort ();
+ }
+
+ reckeys.buf_used = 0;
+ reckeys.prevAttrUse = -1;
+ reckeys.prevAttrSet = -1;
+ reckeys.prevSeqNo = 0;
+
+ extractCtrl.init = wordInit;
+ extractCtrl.addWord = addRecordKey;
+ extractCtrl.addSchema = addSchema;
+ extractCtrl.dh = rGroup->dh;
+ for (i = 0; i<256; i++)
+ extractCtrl.seqno[i] = 0;
+ extractCtrl.zebra_maps = rGroup->zebra_maps;
+ extractCtrl.flagShowRecords = !rGroup->flagRw;
+
+ grs_extract_tree(&extractCtrl, n);
+
+ logf (LOG_DEBUG, "flush explain record, sysno=%d", rec->sysno);
+
+ if (rec->size[recInfo_delKeys])
+ {
+ struct recKeys delkeys;
+
+ delkeys.buf_used = rec->size[recInfo_delKeys];
+ delkeys.buf = rec->info[recInfo_delKeys];
+ flushSortKeys (rec->sysno, 0);
+ flushRecordKeys (rec->sysno, 0, &delkeys);
+ }
+ flushRecordKeys (rec->sysno, 1, &reckeys);
+ flushSortKeys (rec->sysno, 1);
+
+ xfree (rec->info[recInfo_delKeys]);
+ rec->size[recInfo_delKeys] = reckeys.buf_used;
+ rec->info[recInfo_delKeys] = reckeys.buf;
+ reckeys.buf = NULL;
+ reckeys.buf_max = 0;
+
+ return 0;
+}
projects / idzebra-moved-to-github.git / blobdiff