/*
- * 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 1.104 2000-09-05 14:04:05 adam
+ * Updates for prefix 'yaz_' for YAZ log functions.
+ *
+ * Revision 1.103 2000/05/18 12:01:36 adam
+ * System call times(2) used again. More 64-bit fixes.
+ *
+ * Revision 1.102 2000/05/15 15:32:33 adam
+ * Added 64 bit file input.
+ *
+ * Revision 1.101 2000/05/15 13:02:39 adam
+ * Minor change.
+ *
+ * Revision 1.100 2000/03/20 19:08:36 adam
+ * Added remote record import using Z39.50 extended services and Segment
+ * Requests.
+ *
+ * Revision 1.99 2000/02/24 10:57:02 adam
+ * Sequence number incremented after each incomplete-field.
+ *
+ * Revision 1.98 1999/09/07 07:19:21 adam
+ * Work on character mapping. Implemented replace rules.
+ *
+ * Revision 1.97 1999/07/06 12:28:04 adam
+ * Updated record index structure. Format includes version ID. Compression
+ * algorithm ID is stored for each record block.
+ *
+ * Revision 1.96 1999/05/26 07:49:13 adam
+ * C++ compilation.
+ *
+ * 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"
+#if _FILE_OFFSET_BITS == 64
+#define PRINTF_OFF_T "%Ld"
+#else
+#define PRINTF_OFF_T "%ld"
+#endif
+
+#ifndef ZEBRASDR
+#define ZEBRASDR 0
+#endif
+
+#if ZEBRASDR
+#include "zebrasdr.h"
+#endif
+
static Dict matchDict;
static Records records = NULL;
BFiles bfs = rGroup->bfs;
int rw = rGroup->flagRw;
data1_handle dh = rGroup->dh;
+ char *recordCompression;
+ int record_compression = REC_COMPRESS_NONE;
if (!mem)
- mem = atoi(res_get_def (common_resource, "memMax", "4"))*1024*1024;
+ mem = atoi(res_get_def (common_resource, "memMax", "16"))*1024*1024;
if (mem < 50000)
mem = 50000;
- key_buf = xmalloc (mem);
+ key_buf = (char **) xmalloc (mem);
ptr_top = mem/sizeof(char*);
ptr_i = 0;
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;
}
assert (!records);
- records = rec_open (bfs, rw);
+ recordCompression = res_get_def (common_resource,
+ "recordCompression", "none");
+ if (!strcmp (recordCompression, "none"))
+ record_compression = REC_COMPRESS_NONE;
+ if (!strcmp (recordCompression, "bzip2"))
+ record_compression = REC_COMPRESS_BZIP2;
+ records = rec_open (bfs, rw, record_compression);
if (!records)
{
dict_close (matchDict);
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");
logf (LOG_LOG, "sorting section %d", key_file_no);
#if !SORT_EXTRA
qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
- getFnameTmp (out_fname, key_file_no);
+ getFnameTmp (common_resource, out_fname, key_file_no);
if (!(outf = fopen (out_fname, "wb")))
{
int rw = rGroup->flagRw;
if (rw)
zebraExplain_runNumberIncrement (zti, 1);
- zebraExplain_close (zti, rw, 0);
+ zebraExplain_close (zti, rw);
key_flush ();
xfree (key_buf);
rec_close (&records);
w->attrSet = VAL_BIB1;
w->attrUse = 1016;
w->reg_type = 'w';
+ w->extractCtrl = p;
}
static struct sortKey {
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];
{
char *b;
- b = xmalloc (reckeys.buf_max += 128000);
+ b = (char *) xmalloc (reckeys.buf_max += 128000);
if (reckeys.buf_used > 0)
memcpy (b, reckeys.buf, reckeys.buf_used);
xfree (reckeys.buf);
dst += sizeof(*pseqno);
}
reckeys.buf_used = dst - reckeys.buf;
- (*pseqno)++;
+ if (*pseqno)
+ (*pseqno)++;
}
static void addSortString (RecWord *p, const char *string, int length)
if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
return;
- sk = xmalloc (sizeof(*sk));
+ sk = (struct sortKey *) xmalloc (sizeof(*sk));
sk->next = sortKeys;
sortKeys = sk;
- sk->string = xmalloc (p->length);
- sk->length = p->length;
- memcpy (sk->string, p->string, p->length);
+ sk->string = (char *) xmalloc (length);
+ sk->length = length;
+ memcpy (sk->string, string, length);
+
sk->attrSet = p->attrSet;
sk->attrUse = p->attrUse;
}
return;
addString (p, buf, i);
}
+ (p->seqnos[p->reg_type])++; /* to separate this from next one */
}
static void addCompleteField (RecWord *p)
static void addRecordKey (RecWord *p)
{
+ WRBUF wrbuf;
+ if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
+ p->string, p->length)))
+ {
+ p->string = wrbuf_buf(wrbuf);
+ p->length = wrbuf_len(wrbuf);
+ }
if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
addCompleteField (p);
else
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)
{
- struct file_read_info *fi = xmalloc (sizeof(*fi));
+ struct file_read_info *fi = (struct file_read_info *)
+ xmalloc (sizeof(*fi));
fi->fd = fd;
fi->file_max = 0;
fi->file_moffset = 0;
+ fi->sdrbuf = 0;
+ fi->sdrmax = 0;
return fi;
}
static off_t file_seek (void *handle, off_t offset)
{
- struct file_read_info *p = handle;
+ struct file_read_info *p = (struct file_read_info *) handle;
p->file_offset = offset;
+ if (p->sdrbuf)
+ return offset;
return lseek (p->fd, offset, SEEK_SET);
}
static off_t file_tell (void *handle)
{
- struct file_read_info *p = handle;
+ struct file_read_info *p = (struct file_read_info *) handle;
return p->file_offset;
}
static int file_read (void *handle, char *buf, size_t count)
{
- struct file_read_info *p = handle;
+ struct file_read_info *p = (struct file_read_info *) 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;
static void file_begin (void *handle)
{
- struct file_read_info *p = handle;
+ struct file_read_info *p = (struct file_read_info *) 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 void file_end (void *handle, off_t offset)
{
- struct file_read_info *p = handle;
+ struct file_read_info *p = (struct file_read_info *) handle;
assert (p->file_more == 0);
p->file_more = 1;
static void recordLogPreamble (int level, const char *msg, void *info)
{
- struct recordLogInfo *p = info;
- FILE *outf = log_file ();
+ struct recordLogInfo *p = (struct recordLogInfo *) info;
+ FILE *outf = yaz_log_file ();
if (level & LOG_LOG)
return ;
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.fh = fi;
extractCtrl.subType = subType;
extractCtrl.init = wordInit;
- extractCtrl.addWord = addRecordKey;
- extractCtrl.addSchema = addSchema;
+ extractCtrl.tokenAdd = addRecordKey;
+ extractCtrl.schemaAdd = 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;
if (!rGroup->flagRw)
- printf ("File: %s %ld\n", fname, (long) recordOffset);
+ printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset);
logInfo.fname = fname;
logInfo.recordOffset = recordOffset;
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 " PRINTF_OFF_T, rGroup->recordType,
+ fname, recordOffset);
}
return 0;
}
is probably empty - unless flagShowRecords is in use */
if (!rGroup->flagRw)
return 1;
- logf (LOG_WARN, "No keys generated for file %s", fname);
- logf (LOG_WARN, " The file is probably empty");
- return 0;
+
+ logf (LOG_WARN, "empty %s %s " PRINTF_OFF_T, rGroup->recordType,
+ fname, recordOffset);
+ return 1;
}
}
/* new record */
if (deleteFlag)
{
- logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
- fname, (long) recordOffset);
+ logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T, rGroup->recordType,
+ fname, recordOffset);
logf (LOG_WARN, "cannot delete record above (seems new)");
return 1;
}
if (records_processed < rGroup->fileVerboseLimit)
- logf (LOG_LOG, "add %s %s %ld", rGroup->recordType,
- fname, (long) recordOffset);
+ logf (LOG_LOG, "add %s %s " PRINTF_OFF_T, rGroup->recordType,
+ fname, recordOffset);
rec = rec_new (records);
*sysno = rec->sysno;
if (recordAttr->runNumber == zebraExplain_runNumberIncrement (zti, 0))
{
- logf (LOG_LOG, "skipped %s %s %ld", rGroup->recordType,
- fname, (long) recordOffset);
+ logf (LOG_LOG, "skipped %s %s " PRINTF_OFF_T, rGroup->recordType,
+ fname, recordOffset);
rec_rm (&rec);
logRecord (0);
return 1;
/* record going to be deleted */
if (!delkeys.buf_used)
{
- logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
- fname, (long) recordOffset);
+ logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T,
+ rGroup->recordType, fname, recordOffset);
logf (LOG_WARN, "cannot delete file above, storeKeys false");
}
else
{
if (records_processed < rGroup->fileVerboseLimit)
- logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
- fname, (long) recordOffset);
+ logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T,
+ rGroup->recordType, fname, recordOffset);
records_deleted++;
if (matchStr)
dict_delete (matchDict, matchStr);
/* record going to be updated */
if (!delkeys.buf_used)
{
- logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
- fname, (long) recordOffset);
+ logf (LOG_LOG, "update %s %s " PRINTF_OFF_T,
+ rGroup->recordType, fname, recordOffset);
logf (LOG_WARN, "cannot update file above, storeKeys false");
}
else
{
if (records_processed < rGroup->fileVerboseLimit)
- logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
- fname, (long) recordOffset);
+ logf (LOG_LOG, "update %s %s " PRINTF_OFF_T,
+ rGroup->recordType, fname, recordOffset);
flushRecordKeys (*sysno, 1, &reckeys);
records_updated++;
}
if (rGroup->flagStoreData == 1)
{
rec->size[recInfo_storeData] = recordAttr->recordSize;
- rec->info[recInfo_storeData] = xmalloc (recordAttr->recordSize);
+ rec->info[recInfo_storeData] = (char *)
+ 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 " PRINTF_OFF_T " in %s",
+ 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 (zebraExplain_newDatabase (zti, rGroup->databaseName,
rGroup->explainDatabase))
- abort ();
+ return 0;
}
if (rGroup->flagStoreData == -1)
{
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)
reckeys.prevSeqNo = 0;
extractCtrl.init = wordInit;
- extractCtrl.addWord = addRecordKey;
- extractCtrl.addSchema = addSchema;
+ extractCtrl.tokenAdd = addRecordKey;
+ extractCtrl.schemaAdd = addSchema;
extractCtrl.dh = rGroup->dh;
for (i = 0; i<256; i++)
extractCtrl.seqno[i] = 0;